├── .gitignore
├── LICENSE
├── README.md
├── clean_data
    ├── clean_data.py
    └── dev_clean_data.ipynb
├── data
    ├── FakePII.py
    └── create_text_with_fake_pii.ipynb
├── model
    ├── Doc2vec.ipynb
    ├── Dov2vec_visualizer_logistic.ipynb
    ├── build_vocab_Test.ipynb
    ├── doc2vec_different_vector_size.ipynb
    ├── doc2vec_index_100_logistic.pkl
    ├── doc2vec_label_100_logistic.pkl
    ├── doc2vec_label_100_logistic_no_weights.pkl
    ├── doc2vec_label_100_logistic_weights.pkl
    ├── fasttext_test.ipynb
    ├── inference_time.ipynb
    ├── model_test.ipynb
    ├── multiple_test_data_performance.ipynb
    ├── nlp-pipe.ipynb
    ├── pipeline_test_performance.ipynb
    ├── production_model_train.ipynb
    ├── test_data_peformance.ipynb
    ├── transfer_learning_word2vec.ipynb
    ├── visualizer.py
    ├── word2vec.ipynb
    ├── word2vec.png
    ├── word2vec_300_logistic.pkl
    ├── word2vec_300_logistic_with_weights.pkl
    ├── word2vec_from_cleaned_data.ipynb
    ├── word2vec_pipe_cv_test_performance.png
    ├── word2vec_visualizer_logistic_regression.ipynb
    └── word2vec_with_google_update.ipynb
├── requirements.txt
└── website
    ├── app.py
    ├── static
        ├── bootstrap.min.css
        └── cover.css
    ├── templates
        ├── API_page.html
        ├── Forget_password.html
        ├── Login_page.html
        ├── Register_page.html
        ├── home_page.html
        └── template.html
    ├── user_entry_forms.py
    └── word2vec_model.py


/.gitignore:
--------------------------------------------------------------------------------
 1 | *.csv
 2 | *.bin
 3 | *.npy
 4 | *.txt
 5 | enwik9
 6 | *.pkl
 7 | *tar.gz
 8 | *.whl
 9 | .ipynb_checkpoints/
10 | Untitled.ipynb
11 | __pycache__/
12 | 


--------------------------------------------------------------------------------
/LICENSE:
--------------------------------------------------------------------------------
  1 |                     GNU GENERAL PUBLIC LICENSE
  2 |                        Version 3, 29 June 2007
  3 | 
  4 |  Copyright (C) 2007 Free Software Foundation, Inc. <http://fsf.org/>
  5 |  Everyone is permitted to copy and distribute verbatim copies
  6 |  of this license document, but changing it is not allowed.
  7 | 
  8 |                             Preamble
  9 | 
 10 |   The GNU General Public License is a free, copyleft license for
 11 | software and other kinds of works.
 12 | 
 13 |   The licenses for most software and other practical works are designed
 14 | to take away your freedom to share and change the works.  By contrast,
 15 | the GNU General Public License is intended to guarantee your freedom to
 16 | share and change all versions of a program--to make sure it remains free
 17 | software for all its users.  We, the Free Software Foundation, use the
 18 | GNU General Public License for most of our software; it applies also to
 19 | any other work released this way by its authors.  You can apply it to
 20 | your programs, too.
 21 | 
 22 |   When we speak of free software, we are referring to freedom, not
 23 | price.  Our General Public Licenses are designed to make sure that you
 24 | have the freedom to distribute copies of free software (and charge for
 25 | them if you wish), that you receive source code or can get it if you
 26 | want it, that you can change the software or use pieces of it in new
 27 | free programs, and that you know you can do these things.
 28 | 
 29 |   To protect your rights, we need to prevent others from denying you
 30 | these rights or asking you to surrender the rights.  Therefore, you have
 31 | certain responsibilities if you distribute copies of the software, or if
 32 | you modify it: responsibilities to respect the freedom of others.
 33 | 
 34 |   For example, if you distribute copies of such a program, whether
 35 | gratis or for a fee, you must pass on to the recipients the same
 36 | freedoms that you received.  You must make sure that they, too, receive
 37 | or can get the source code.  And you must show them these terms so they
 38 | know their rights.
 39 | 
 40 |   Developers that use the GNU GPL protect your rights with two steps:
 41 | (1) assert copyright on the software, and (2) offer you this License
 42 | giving you legal permission to copy, distribute and/or modify it.
 43 | 
 44 |   For the developers' and authors' protection, the GPL clearly explains
 45 | that there is no warranty for this free software.  For both users' and
 46 | authors' sake, the GPL requires that modified versions be marked as
 47 | changed, so that their problems will not be attributed erroneously to
 48 | authors of previous versions.
 49 | 
 50 |   Some devices are designed to deny users access to install or run
 51 | modified versions of the software inside them, although the manufacturer
 52 | can do so.  This is fundamentally incompatible with the aim of
 53 | protecting users' freedom to change the software.  The systematic
 54 | pattern of such abuse occurs in the area of products for individuals to
 55 | use, which is precisely where it is most unacceptable.  Therefore, we
 56 | have designed this version of the GPL to prohibit the practice for those
 57 | products.  If such problems arise substantially in other domains, we
 58 | stand ready to extend this provision to those domains in future versions
 59 | of the GPL, as needed to protect the freedom of users.
 60 | 
 61 |   Finally, every program is threatened constantly by software patents.
 62 | States should not allow patents to restrict development and use of
 63 | software on general-purpose computers, but in those that do, we wish to
 64 | avoid the special danger that patents applied to a free program could
 65 | make it effectively proprietary.  To prevent this, the GPL assures that
 66 | patents cannot be used to render the program non-free.
 67 | 
 68 |   The precise terms and conditions for copying, distribution and
 69 | modification follow.
 70 | 
 71 |                        TERMS AND CONDITIONS
 72 | 
 73 |   0. Definitions.
 74 | 
 75 |   "This License" refers to version 3 of the GNU General Public License.
 76 | 
 77 |   "Copyright" also means copyright-like laws that apply to other kinds of
 78 | works, such as semiconductor masks.
 79 | 
 80 |   "The Program" refers to any copyrightable work licensed under this
 81 | License.  Each licensee is addressed as "you".  "Licensees" and
 82 | "recipients" may be individuals or organizations.
 83 | 
 84 |   To "modify" a work means to copy from or adapt all or part of the work
 85 | in a fashion requiring copyright permission, other than the making of an
 86 | exact copy.  The resulting work is called a "modified version" of the
 87 | earlier work or a work "based on" the earlier work.
 88 | 
 89 |   A "covered work" means either the unmodified Program or a work based
 90 | on the Program.
 91 | 
 92 |   To "propagate" a work means to do anything with it that, without
 93 | permission, would make you directly or secondarily liable for
 94 | infringement under applicable copyright law, except executing it on a
 95 | computer or modifying a private copy.  Propagation includes copying,
 96 | distribution (with or without modification), making available to the
 97 | public, and in some countries other activities as well.
 98 | 
 99 |   To "convey" a work means any kind of propagation that enables other
100 | parties to make or receive copies.  Mere interaction with a user through
101 | a computer network, with no transfer of a copy, is not conveying.
102 | 
103 |   An interactive user interface displays "Appropriate Legal Notices"
104 | to the extent that it includes a convenient and prominently visible
105 | feature that (1) displays an appropriate copyright notice, and (2)
106 | tells the user that there is no warranty for the work (except to the
107 | extent that warranties are provided), that licensees may convey the
108 | work under this License, and how to view a copy of this License.  If
109 | the interface presents a list of user commands or options, such as a
110 | menu, a prominent item in the list meets this criterion.
111 | 
112 |   1. Source Code.
113 | 
114 |   The "source code" for a work means the preferred form of the work
115 | for making modifications to it.  "Object code" means any non-source
116 | form of a work.
117 | 
118 |   A "Standard Interface" means an interface that either is an official
119 | standard defined by a recognized standards body, or, in the case of
120 | interfaces specified for a particular programming language, one that
121 | is widely used among developers working in that language.
122 | 
123 |   The "System Libraries" of an executable work include anything, other
124 | than the work as a whole, that (a) is included in the normal form of
125 | packaging a Major Component, but which is not part of that Major
126 | Component, and (b) serves only to enable use of the work with that
127 | Major Component, or to implement a Standard Interface for which an
128 | implementation is available to the public in source code form.  A
129 | "Major Component", in this context, means a major essential component
130 | (kernel, window system, and so on) of the specific operating system
131 | (if any) on which the executable work runs, or a compiler used to
132 | produce the work, or an object code interpreter used to run it.
133 | 
134 |   The "Corresponding Source" for a work in object code form means all
135 | the source code needed to generate, install, and (for an executable
136 | work) run the object code and to modify the work, including scripts to
137 | control those activities.  However, it does not include the work's
138 | System Libraries, or general-purpose tools or generally available free
139 | programs which are used unmodified in performing those activities but
140 | which are not part of the work.  For example, Corresponding Source
141 | includes interface definition files associated with source files for
142 | the work, and the source code for shared libraries and dynamically
143 | linked subprograms that the work is specifically designed to require,
144 | such as by intimate data communication or control flow between those
145 | subprograms and other parts of the work.
146 | 
147 |   The Corresponding Source need not include anything that users
148 | can regenerate automatically from other parts of the Corresponding
149 | Source.
150 | 
151 |   The Corresponding Source for a work in source code form is that
152 | same work.
153 | 
154 |   2. Basic Permissions.
155 | 
156 |   All rights granted under this License are granted for the term of
157 | copyright on the Program, and are irrevocable provided the stated
158 | conditions are met.  This License explicitly affirms your unlimited
159 | permission to run the unmodified Program.  The output from running a
160 | covered work is covered by this License only if the output, given its
161 | content, constitutes a covered work.  This License acknowledges your
162 | rights of fair use or other equivalent, as provided by copyright law.
163 | 
164 |   You may make, run and propagate covered works that you do not
165 | convey, without conditions so long as your license otherwise remains
166 | in force.  You may convey covered works to others for the sole purpose
167 | of having them make modifications exclusively for you, or provide you
168 | with facilities for running those works, provided that you comply with
169 | the terms of this License in conveying all material for which you do
170 | not control copyright.  Those thus making or running the covered works
171 | for you must do so exclusively on your behalf, under your direction
172 | and control, on terms that prohibit them from making any copies of
173 | your copyrighted material outside their relationship with you.
174 | 
175 |   Conveying under any other circumstances is permitted solely under
176 | the conditions stated below.  Sublicensing is not allowed; section 10
177 | makes it unnecessary.
178 | 
179 |   3. Protecting Users' Legal Rights From Anti-Circumvention Law.
180 | 
181 |   No covered work shall be deemed part of an effective technological
182 | measure under any applicable law fulfilling obligations under article
183 | 11 of the WIPO copyright treaty adopted on 20 December 1996, or
184 | similar laws prohibiting or restricting circumvention of such
185 | measures.
186 | 
187 |   When you convey a covered work, you waive any legal power to forbid
188 | circumvention of technological measures to the extent such circumvention
189 | is effected by exercising rights under this License with respect to
190 | the covered work, and you disclaim any intention to limit operation or
191 | modification of the work as a means of enforcing, against the work's
192 | users, your or third parties' legal rights to forbid circumvention of
193 | technological measures.
194 | 
195 |   4. Conveying Verbatim Copies.
196 | 
197 |   You may convey verbatim copies of the Program's source code as you
198 | receive it, in any medium, provided that you conspicuously and
199 | appropriately publish on each copy an appropriate copyright notice;
200 | keep intact all notices stating that this License and any
201 | non-permissive terms added in accord with section 7 apply to the code;
202 | keep intact all notices of the absence of any warranty; and give all
203 | recipients a copy of this License along with the Program.
204 | 
205 |   You may charge any price or no price for each copy that you convey,
206 | and you may offer support or warranty protection for a fee.
207 | 
208 |   5. Conveying Modified Source Versions.
209 | 
210 |   You may convey a work based on the Program, or the modifications to
211 | produce it from the Program, in the form of source code under the
212 | terms of section 4, provided that you also meet all of these conditions:
213 | 
214 |     a) The work must carry prominent notices stating that you modified
215 |     it, and giving a relevant date.
216 | 
217 |     b) The work must carry prominent notices stating that it is
218 |     released under this License and any conditions added under section
219 |     7.  This requirement modifies the requirement in section 4 to
220 |     "keep intact all notices".
221 | 
222 |     c) You must license the entire work, as a whole, under this
223 |     License to anyone who comes into possession of a copy.  This
224 |     License will therefore apply, along with any applicable section 7
225 |     additional terms, to the whole of the work, and all its parts,
226 |     regardless of how they are packaged.  This License gives no
227 |     permission to license the work in any other way, but it does not
228 |     invalidate such permission if you have separately received it.
229 | 
230 |     d) If the work has interactive user interfaces, each must display
231 |     Appropriate Legal Notices; however, if the Program has interactive
232 |     interfaces that do not display Appropriate Legal Notices, your
233 |     work need not make them do so.
234 | 
235 |   A compilation of a covered work with other separate and independent
236 | works, which are not by their nature extensions of the covered work,
237 | and which are not combined with it such as to form a larger program,
238 | in or on a volume of a storage or distribution medium, is called an
239 | "aggregate" if the compilation and its resulting copyright are not
240 | used to limit the access or legal rights of the compilation's users
241 | beyond what the individual works permit.  Inclusion of a covered work
242 | in an aggregate does not cause this License to apply to the other
243 | parts of the aggregate.
244 | 
245 |   6. Conveying Non-Source Forms.
246 | 
247 |   You may convey a covered work in object code form under the terms
248 | of sections 4 and 5, provided that you also convey the
249 | machine-readable Corresponding Source under the terms of this License,
250 | in one of these ways:
251 | 
252 |     a) Convey the object code in, or embodied in, a physical product
253 |     (including a physical distribution medium), accompanied by the
254 |     Corresponding Source fixed on a durable physical medium
255 |     customarily used for software interchange.
256 | 
257 |     b) Convey the object code in, or embodied in, a physical product
258 |     (including a physical distribution medium), accompanied by a
259 |     written offer, valid for at least three years and valid for as
260 |     long as you offer spare parts or customer support for that product
261 |     model, to give anyone who possesses the object code either (1) a
262 |     copy of the Corresponding Source for all the software in the
263 |     product that is covered by this License, on a durable physical
264 |     medium customarily used for software interchange, for a price no
265 |     more than your reasonable cost of physically performing this
266 |     conveying of source, or (2) access to copy the
267 |     Corresponding Source from a network server at no charge.
268 | 
269 |     c) Convey individual copies of the object code with a copy of the
270 |     written offer to provide the Corresponding Source.  This
271 |     alternative is allowed only occasionally and noncommercially, and
272 |     only if you received the object code with such an offer, in accord
273 |     with subsection 6b.
274 | 
275 |     d) Convey the object code by offering access from a designated
276 |     place (gratis or for a charge), and offer equivalent access to the
277 |     Corresponding Source in the same way through the same place at no
278 |     further charge.  You need not require recipients to copy the
279 |     Corresponding Source along with the object code.  If the place to
280 |     copy the object code is a network server, the Corresponding Source
281 |     may be on a different server (operated by you or a third party)
282 |     that supports equivalent copying facilities, provided you maintain
283 |     clear directions next to the object code saying where to find the
284 |     Corresponding Source.  Regardless of what server hosts the
285 |     Corresponding Source, you remain obligated to ensure that it is
286 |     available for as long as needed to satisfy these requirements.
287 | 
288 |     e) Convey the object code using peer-to-peer transmission, provided
289 |     you inform other peers where the object code and Corresponding
290 |     Source of the work are being offered to the general public at no
291 |     charge under subsection 6d.
292 | 
293 |   A separable portion of the object code, whose source code is excluded
294 | from the Corresponding Source as a System Library, need not be
295 | included in conveying the object code work.
296 | 
297 |   A "User Product" is either (1) a "consumer product", which means any
298 | tangible personal property which is normally used for personal, family,
299 | or household purposes, or (2) anything designed or sold for incorporation
300 | into a dwelling.  In determining whether a product is a consumer product,
301 | doubtful cases shall be resolved in favor of coverage.  For a particular
302 | product received by a particular user, "normally used" refers to a
303 | typical or common use of that class of product, regardless of the status
304 | of the particular user or of the way in which the particular user
305 | actually uses, or expects or is expected to use, the product.  A product
306 | is a consumer product regardless of whether the product has substantial
307 | commercial, industrial or non-consumer uses, unless such uses represent
308 | the only significant mode of use of the product.
309 | 
310 |   "Installation Information" for a User Product means any methods,
311 | procedures, authorization keys, or other information required to install
312 | and execute modified versions of a covered work in that User Product from
313 | a modified version of its Corresponding Source.  The information must
314 | suffice to ensure that the continued functioning of the modified object
315 | code is in no case prevented or interfered with solely because
316 | modification has been made.
317 | 
318 |   If you convey an object code work under this section in, or with, or
319 | specifically for use in, a User Product, and the conveying occurs as
320 | part of a transaction in which the right of possession and use of the
321 | User Product is transferred to the recipient in perpetuity or for a
322 | fixed term (regardless of how the transaction is characterized), the
323 | Corresponding Source conveyed under this section must be accompanied
324 | by the Installation Information.  But this requirement does not apply
325 | if neither you nor any third party retains the ability to install
326 | modified object code on the User Product (for example, the work has
327 | been installed in ROM).
328 | 
329 |   The requirement to provide Installation Information does not include a
330 | requirement to continue to provide support service, warranty, or updates
331 | for a work that has been modified or installed by the recipient, or for
332 | the User Product in which it has been modified or installed.  Access to a
333 | network may be denied when the modification itself materially and
334 | adversely affects the operation of the network or violates the rules and
335 | protocols for communication across the network.
336 | 
337 |   Corresponding Source conveyed, and Installation Information provided,
338 | in accord with this section must be in a format that is publicly
339 | documented (and with an implementation available to the public in
340 | source code form), and must require no special password or key for
341 | unpacking, reading or copying.
342 | 
343 |   7. Additional Terms.
344 | 
345 |   "Additional permissions" are terms that supplement the terms of this
346 | License by making exceptions from one or more of its conditions.
347 | Additional permissions that are applicable to the entire Program shall
348 | be treated as though they were included in this License, to the extent
349 | that they are valid under applicable law.  If additional permissions
350 | apply only to part of the Program, that part may be used separately
351 | under those permissions, but the entire Program remains governed by
352 | this License without regard to the additional permissions.
353 | 
354 |   When you convey a copy of a covered work, you may at your option
355 | remove any additional permissions from that copy, or from any part of
356 | it.  (Additional permissions may be written to require their own
357 | removal in certain cases when you modify the work.)  You may place
358 | additional permissions on material, added by you to a covered work,
359 | for which you have or can give appropriate copyright permission.
360 | 
361 |   Notwithstanding any other provision of this License, for material you
362 | add to a covered work, you may (if authorized by the copyright holders of
363 | that material) supplement the terms of this License with terms:
364 | 
365 |     a) Disclaiming warranty or limiting liability differently from the
366 |     terms of sections 15 and 16 of this License; or
367 | 
368 |     b) Requiring preservation of specified reasonable legal notices or
369 |     author attributions in that material or in the Appropriate Legal
370 |     Notices displayed by works containing it; or
371 | 
372 |     c) Prohibiting misrepresentation of the origin of that material, or
373 |     requiring that modified versions of such material be marked in
374 |     reasonable ways as different from the original version; or
375 | 
376 |     d) Limiting the use for publicity purposes of names of licensors or
377 |     authors of the material; or
378 | 
379 |     e) Declining to grant rights under trademark law for use of some
380 |     trade names, trademarks, or service marks; or
381 | 
382 |     f) Requiring indemnification of licensors and authors of that
383 |     material by anyone who conveys the material (or modified versions of
384 |     it) with contractual assumptions of liability to the recipient, for
385 |     any liability that these contractual assumptions directly impose on
386 |     those licensors and authors.
387 | 
388 |   All other non-permissive additional terms are considered "further
389 | restrictions" within the meaning of section 10.  If the Program as you
390 | received it, or any part of it, contains a notice stating that it is
391 | governed by this License along with a term that is a further
392 | restriction, you may remove that term.  If a license document contains
393 | a further restriction but permits relicensing or conveying under this
394 | License, you may add to a covered work material governed by the terms
395 | of that license document, provided that the further restriction does
396 | not survive such relicensing or conveying.
397 | 
398 |   If you add terms to a covered work in accord with this section, you
399 | must place, in the relevant source files, a statement of the
400 | additional terms that apply to those files, or a notice indicating
401 | where to find the applicable terms.
402 | 
403 |   Additional terms, permissive or non-permissive, may be stated in the
404 | form of a separately written license, or stated as exceptions;
405 | the above requirements apply either way.
406 | 
407 |   8. Termination.
408 | 
409 |   You may not propagate or modify a covered work except as expressly
410 | provided under this License.  Any attempt otherwise to propagate or
411 | modify it is void, and will automatically terminate your rights under
412 | this License (including any patent licenses granted under the third
413 | paragraph of section 11).
414 | 
415 |   However, if you cease all violation of this License, then your
416 | license from a particular copyright holder is reinstated (a)
417 | provisionally, unless and until the copyright holder explicitly and
418 | finally terminates your license, and (b) permanently, if the copyright
419 | holder fails to notify you of the violation by some reasonable means
420 | prior to 60 days after the cessation.
421 | 
422 |   Moreover, your license from a particular copyright holder is
423 | reinstated permanently if the copyright holder notifies you of the
424 | violation by some reasonable means, this is the first time you have
425 | received notice of violation of this License (for any work) from that
426 | copyright holder, and you cure the violation prior to 30 days after
427 | your receipt of the notice.
428 | 
429 |   Termination of your rights under this section does not terminate the
430 | licenses of parties who have received copies or rights from you under
431 | this License.  If your rights have been terminated and not permanently
432 | reinstated, you do not qualify to receive new licenses for the same
433 | material under section 10.
434 | 
435 |   9. Acceptance Not Required for Having Copies.
436 | 
437 |   You are not required to accept this License in order to receive or
438 | run a copy of the Program.  Ancillary propagation of a covered work
439 | occurring solely as a consequence of using peer-to-peer transmission
440 | to receive a copy likewise does not require acceptance.  However,
441 | nothing other than this License grants you permission to propagate or
442 | modify any covered work.  These actions infringe copyright if you do
443 | not accept this License.  Therefore, by modifying or propagating a
444 | covered work, you indicate your acceptance of this License to do so.
445 | 
446 |   10. Automatic Licensing of Downstream Recipients.
447 | 
448 |   Each time you convey a covered work, the recipient automatically
449 | receives a license from the original licensors, to run, modify and
450 | propagate that work, subject to this License.  You are not responsible
451 | for enforcing compliance by third parties with this License.
452 | 
453 |   An "entity transaction" is a transaction transferring control of an
454 | organization, or substantially all assets of one, or subdividing an
455 | organization, or merging organizations.  If propagation of a covered
456 | work results from an entity transaction, each party to that
457 | transaction who receives a copy of the work also receives whatever
458 | licenses to the work the party's predecessor in interest had or could
459 | give under the previous paragraph, plus a right to possession of the
460 | Corresponding Source of the work from the predecessor in interest, if
461 | the predecessor has it or can get it with reasonable efforts.
462 | 
463 |   You may not impose any further restrictions on the exercise of the
464 | rights granted or affirmed under this License.  For example, you may
465 | not impose a license fee, royalty, or other charge for exercise of
466 | rights granted under this License, and you may not initiate litigation
467 | (including a cross-claim or counterclaim in a lawsuit) alleging that
468 | any patent claim is infringed by making, using, selling, offering for
469 | sale, or importing the Program or any portion of it.
470 | 
471 |   11. Patents.
472 | 
473 |   A "contributor" is a copyright holder who authorizes use under this
474 | License of the Program or a work on which the Program is based.  The
475 | work thus licensed is called the contributor's "contributor version".
476 | 
477 |   A contributor's "essential patent claims" are all patent claims
478 | owned or controlled by the contributor, whether already acquired or
479 | hereafter acquired, that would be infringed by some manner, permitted
480 | by this License, of making, using, or selling its contributor version,
481 | but do not include claims that would be infringed only as a
482 | consequence of further modification of the contributor version.  For
483 | purposes of this definition, "control" includes the right to grant
484 | patent sublicenses in a manner consistent with the requirements of
485 | this License.
486 | 
487 |   Each contributor grants you a non-exclusive, worldwide, royalty-free
488 | patent license under the contributor's essential patent claims, to
489 | make, use, sell, offer for sale, import and otherwise run, modify and
490 | propagate the contents of its contributor version.
491 | 
492 |   In the following three paragraphs, a "patent license" is any express
493 | agreement or commitment, however denominated, not to enforce a patent
494 | (such as an express permission to practice a patent or covenant not to
495 | sue for patent infringement).  To "grant" such a patent license to a
496 | party means to make such an agreement or commitment not to enforce a
497 | patent against the party.
498 | 
499 |   If you convey a covered work, knowingly relying on a patent license,
500 | and the Corresponding Source of the work is not available for anyone
501 | to copy, free of charge and under the terms of this License, through a
502 | publicly available network server or other readily accessible means,
503 | then you must either (1) cause the Corresponding Source to be so
504 | available, or (2) arrange to deprive yourself of the benefit of the
505 | patent license for this particular work, or (3) arrange, in a manner
506 | consistent with the requirements of this License, to extend the patent
507 | license to downstream recipients.  "Knowingly relying" means you have
508 | actual knowledge that, but for the patent license, your conveying the
509 | covered work in a country, or your recipient's use of the covered work
510 | in a country, would infringe one or more identifiable patents in that
511 | country that you have reason to believe are valid.
512 | 
513 |   If, pursuant to or in connection with a single transaction or
514 | arrangement, you convey, or propagate by procuring conveyance of, a
515 | covered work, and grant a patent license to some of the parties
516 | receiving the covered work authorizing them to use, propagate, modify
517 | or convey a specific copy of the covered work, then the patent license
518 | you grant is automatically extended to all recipients of the covered
519 | work and works based on it.
520 | 
521 |   A patent license is "discriminatory" if it does not include within
522 | the scope of its coverage, prohibits the exercise of, or is
523 | conditioned on the non-exercise of one or more of the rights that are
524 | specifically granted under this License.  You may not convey a covered
525 | work if you are a party to an arrangement with a third party that is
526 | in the business of distributing software, under which you make payment
527 | to the third party based on the extent of your activity of conveying
528 | the work, and under which the third party grants, to any of the
529 | parties who would receive the covered work from you, a discriminatory
530 | patent license (a) in connection with copies of the covered work
531 | conveyed by you (or copies made from those copies), or (b) primarily
532 | for and in connection with specific products or compilations that
533 | contain the covered work, unless you entered into that arrangement,
534 | or that patent license was granted, prior to 28 March 2007.
535 | 
536 |   Nothing in this License shall be construed as excluding or limiting
537 | any implied license or other defenses to infringement that may
538 | otherwise be available to you under applicable patent law.
539 | 
540 |   12. No Surrender of Others' Freedom.
541 | 
542 |   If conditions are imposed on you (whether by court order, agreement or
543 | otherwise) that contradict the conditions of this License, they do not
544 | excuse you from the conditions of this License.  If you cannot convey a
545 | covered work so as to satisfy simultaneously your obligations under this
546 | License and any other pertinent obligations, then as a consequence you may
547 | not convey it at all.  For example, if you agree to terms that obligate you
548 | to collect a royalty for further conveying from those to whom you convey
549 | the Program, the only way you could satisfy both those terms and this
550 | License would be to refrain entirely from conveying the Program.
551 | 
552 |   13. Use with the GNU Affero General Public License.
553 | 
554 |   Notwithstanding any other provision of this License, you have
555 | permission to link or combine any covered work with a work licensed
556 | under version 3 of the GNU Affero General Public License into a single
557 | combined work, and to convey the resulting work.  The terms of this
558 | License will continue to apply to the part which is the covered work,
559 | but the special requirements of the GNU Affero General Public License,
560 | section 13, concerning interaction through a network will apply to the
561 | combination as such.
562 | 
563 |   14. Revised Versions of this License.
564 | 
565 |   The Free Software Foundation may publish revised and/or new versions of
566 | the GNU General Public License from time to time.  Such new versions will
567 | be similar in spirit to the present version, but may differ in detail to
568 | address new problems or concerns.
569 | 
570 |   Each version is given a distinguishing version number.  If the
571 | Program specifies that a certain numbered version of the GNU General
572 | Public License "or any later version" applies to it, you have the
573 | option of following the terms and conditions either of that numbered
574 | version or of any later version published by the Free Software
575 | Foundation.  If the Program does not specify a version number of the
576 | GNU General Public License, you may choose any version ever published
577 | by the Free Software Foundation.
578 | 
579 |   If the Program specifies that a proxy can decide which future
580 | versions of the GNU General Public License can be used, that proxy's
581 | public statement of acceptance of a version permanently authorizes you
582 | to choose that version for the Program.
583 | 
584 |   Later license versions may give you additional or different
585 | permissions.  However, no additional obligations are imposed on any
586 | author or copyright holder as a result of your choosing to follow a
587 | later version.
588 | 
589 |   15. Disclaimer of Warranty.
590 | 
591 |   THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY
592 | APPLICABLE LAW.  EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT
593 | HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY
594 | OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO,
595 | THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
596 | PURPOSE.  THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM
597 | IS WITH YOU.  SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF
598 | ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
599 | 
600 |   16. Limitation of Liability.
601 | 
602 |   IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
603 | WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS
604 | THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY
605 | GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE
606 | USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF
607 | DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD
608 | PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS),
609 | EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF
610 | SUCH DAMAGES.
611 | 
612 |   17. Interpretation of Sections 15 and 16.
613 | 
614 |   If the disclaimer of warranty and limitation of liability provided
615 | above cannot be given local legal effect according to their terms,
616 | reviewing courts shall apply local law that most closely approximates
617 | an absolute waiver of all civil liability in connection with the
618 | Program, unless a warranty or assumption of liability accompanies a
619 | copy of the Program in return for a fee.
620 | 
621 |                      END OF TERMS AND CONDITIONS
622 | 
623 |             How to Apply These Terms to Your New Programs
624 | 
625 |   If you develop a new program, and you want it to be of the greatest
626 | possible use to the public, the best way to achieve this is to make it
627 | free software which everyone can redistribute and change under these terms.
628 | 
629 |   To do so, attach the following notices to the program.  It is safest
630 | to attach them to the start of each source file to most effectively
631 | state the exclusion of warranty; and each file should have at least
632 | the "copyright" line and a pointer to where the full notice is found.
633 | 
634 |     <one line to give the program's name and a brief idea of what it does.>
635 |     Copyright (C) <year>  <name of author>
636 | 
637 |     This program is free software: you can redistribute it and/or modify
638 |     it under the terms of the GNU General Public License as published by
639 |     the Free Software Foundation, either version 3 of the License, or
640 |     (at your option) any later version.
641 | 
642 |     This program is distributed in the hope that it will be useful,
643 |     but WITHOUT ANY WARRANTY; without even the implied warranty of
644 |     MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
645 |     GNU General Public License for more details.
646 | 
647 |     You should have received a copy of the GNU General Public License
648 |     along with this program.  If not, see <http://www.gnu.org/licenses/>.
649 | 
650 | Also add information on how to contact you by electronic and paper mail.
651 | 
652 |   If the program does terminal interaction, make it output a short
653 | notice like this when it starts in an interactive mode:
654 | 
655 |     <program>  Copyright (C) <year>  <name of author>
656 |     This program comes with ABSOLUTELY NO WARRANTY; for details type `show w'.
657 |     This is free software, and you are welcome to redistribute it
658 |     under certain conditions; type `show c' for details.
659 | 
660 | The hypothetical commands `show w' and `show c' should show the appropriate
661 | parts of the General Public License.  Of course, your program's commands
662 | might be different; for a GUI interface, you would use an "about box".
663 | 
664 |   You should also get your employer (if you work as a programmer) or school,
665 | if any, to sign a "copyright disclaimer" for the program, if necessary.
666 | For more information on this, and how to apply and follow the GNU GPL, see
667 | <http://www.gnu.org/licenses/>.
668 | 
669 |   The GNU General Public License does not permit incorporating your program
670 | into proprietary programs.  If your program is a subroutine library, you
671 | may consider it more useful to permit linking proprietary applications with
672 | the library.  If this is what you want to do, use the GNU Lesser General
673 | Public License instead of this License.  But first, please read
674 | <http://www.gnu.org/philosophy/why-not-lgpl.html>.
675 | 


--------------------------------------------------------------------------------
/README.md:
--------------------------------------------------------------------------------
1 | # data-sentry
2 | A project to build a machine learning pipeline to detect sensitive data from text data.
3 | 
4 | A short description of the project is presented in [google slides](https://docs.google.com/presentation/d/181960blaOy3A8jPAhRxSpRZdn9v1dx6rcKquD4PMn2o/edit?usp=sharing) here.
5 | 


--------------------------------------------------------------------------------
/clean_data/clean_data.py:
--------------------------------------------------------------------------------
 1 | import re
 2 | def clean_text(text):
 3 |     # replace  . and a space with only a space, then amke all words lower case.
 4 |     text = text.replace(". "," ").replace(",","").lower()
 5 |     # get rid of the . at the end of each line. 
 6 |     cleaned_text = re.sub("\.$","",text)
 7 |     
 8 |     return cleaned_text
 9 | 
10 | 
11 | # extracting the number of special characters. 
12 | def extract_special_len(text):
13 |     # the complete special char list
14 |     all_special_char = '!"#$%&\'()*+,-./:;<=>?@[\\]^_`{|}~'
15 |     # return the special char if it is in the text
16 |     special_list = [special_char for special_char in all_special_char\
17 |                    if special_char in text]
18 |     
19 |     return len(special_list)   
20 | 
21 | # extracting the exact special characters.
22 | def extract_special_char(text):
23 |     # the complete special char list
24 |     all_special_char = '!"#$%&\'()*+,-./:;<=>?@[\\]^_`{|}~'
25 |     # return the special char if it is in the text
26 |     special_list = [special_char for special_char in all_special_char\
27 |                    if special_char in text]
28 |     
29 |     return special_list  
30 | 
31 | 
32 | # 0 denote no pii
33 | # 1 denote pii exists
34 | def binary_pii(label):
35 |     pii_label = 0
36 |     if label != "None":
37 |         pii_label = 1
38 |     return pii_label 
39 | 
40 | 
41 | if __name__ == "__main__":
42 |     import sys
43 |     assert len(sys.argv)>1, "Please enter a csv file to be cleaned"
44 |     input_file_name = sys.argv[1]
45 |     
46 |     # split the input file path 
47 |     input_file_name_list = sys.argv[1].split("/")
48 |     # combine the file path with the new file names
49 |     output_file_name = "./Cleaned_"+input_file_name_list[-1]
50 |     
51 |     import pandas as pd
52 |     print("Begin importing data for preprocessing")
53 |     input_data = pd.read_csv(input_file_name)
54 |     
55 |     input_data["Cleaned_text"] = input_data["Text"].apply(clean_text)
56 |     print("Finished text cleaning") 
57 |     input_data["Target"] = input_data['Labels'].apply(binary_pii)
58 |     print("Convert multi-class labels to binary 1/0 (1 for PII, 0 for no PII).")
59 |     
60 |     # save to disk 
61 |     input_data.to_csv(output_file_name, index = False)
62 |     print("Saving to disk {}".format(output_file_name))
63 |     
64 |     # showcase the number of special characters in the cleaned text
65 |     input_data['Special_char_num'] = input_data["Cleaned_text"].apply(extract_special_len)
66 |     print("Number of special characters for different PIIs after cleaning.")
67 |     print(input_data.groupby("Labels").agg({'Special_char_num':["min","max","mean",'median']}))
68 |     


--------------------------------------------------------------------------------
/clean_data/dev_clean_data.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "markdown",
  5 |    "metadata": {},
  6 |    "source": [
  7 |     "The goal of this notebook is to develop data cleaning methods. "
  8 |    ]
  9 |   },
 10 |   {
 11 |    "cell_type": "code",
 12 |    "execution_count": 16,
 13 |    "metadata": {},
 14 |    "outputs": [],
 15 |    "source": [
 16 |     "import pandas as pd\n",
 17 |     "import re"
 18 |    ]
 19 |   },
 20 |   {
 21 |    "cell_type": "code",
 22 |    "execution_count": 4,
 23 |    "metadata": {},
 24 |    "outputs": [],
 25 |    "source": [
 26 |     "train_data = pd.read_csv(\"../data/train_text_with_pii_2018_12_29_07_26_56_266227.csv\")\n",
 27 |     "test_data = pd.read_csv(\"../data/test_text_with_pii_2018_12_29_07_26_58_843238.csv\")"
 28 |    ]
 29 |   },
 30 |   {
 31 |    "cell_type": "code",
 32 |    "execution_count": 3,
 33 |    "metadata": {},
 34 |    "outputs": [
 35 |     {
 36 |      "data": {
 37 |       "text/html": [
 38 |        "<div>\n",
 39 |        "<style scoped>\n",
 40 |        "    .dataframe tbody tr th:only-of-type {\n",
 41 |        "        vertical-align: middle;\n",
 42 |        "    }\n",
 43 |        "\n",
 44 |        "    .dataframe tbody tr th {\n",
 45 |        "        vertical-align: top;\n",
 46 |        "    }\n",
 47 |        "\n",
 48 |        "    .dataframe thead th {\n",
 49 |        "        text-align: right;\n",
 50 |        "    }\n",
 51 |        "</style>\n",
 52 |        "<table border=\"1\" class=\"dataframe\">\n",
 53 |        "  <thead>\n",
 54 |        "    <tr style=\"text-align: right;\">\n",
 55 |        "      <th></th>\n",
 56 |        "      <th>Text</th>\n",
 57 |        "      <th>Labels</th>\n",
 58 |        "      <th>PII</th>\n",
 59 |        "    </tr>\n",
 60 |        "  </thead>\n",
 61 |        "  <tbody>\n",
 62 |        "    <tr>\n",
 63 |        "      <th>0</th>\n",
 64 |        "      <td>Return never place become nearly movement. No ...</td>\n",
 65 |        "      <td>Address</td>\n",
 66 |        "      <td>9479 Ward Creek Suite 474 Warnershire, NH 08434</td>\n",
 67 |        "    </tr>\n",
 68 |        "    <tr>\n",
 69 |        "      <th>1</th>\n",
 70 |        "      <td>Quality recently seek Suite 937 goal relations...</td>\n",
 71 |        "      <td>Address</td>\n",
 72 |        "      <td>Suite 937</td>\n",
 73 |        "    </tr>\n",
 74 |        "    <tr>\n",
 75 |        "      <th>2</th>\n",
 76 |        "      <td>Word expert reveal effect response forward. In...</td>\n",
 77 |        "      <td>Address</td>\n",
 78 |        "      <td>Apt. 925</td>\n",
 79 |        "    </tr>\n",
 80 |        "    <tr>\n",
 81 |        "      <th>3</th>\n",
 82 |        "      <td>Son specific think. Conference its home expert...</td>\n",
 83 |        "      <td>Address</td>\n",
 84 |        "      <td>15286 Ramirez Lakes Apt. 147 West Anthony, FL ...</td>\n",
 85 |        "    </tr>\n",
 86 |        "    <tr>\n",
 87 |        "      <th>4</th>\n",
 88 |        "      <td>Control trouble Apt. 223 keep action rule. Sol...</td>\n",
 89 |        "      <td>Address</td>\n",
 90 |        "      <td>Apt. 223</td>\n",
 91 |        "    </tr>\n",
 92 |        "  </tbody>\n",
 93 |        "</table>\n",
 94 |        "</div>"
 95 |       ],
 96 |       "text/plain": [
 97 |        "                                                Text   Labels  \\\n",
 98 |        "0  Return never place become nearly movement. No ...  Address   \n",
 99 |        "1  Quality recently seek Suite 937 goal relations...  Address   \n",
100 |        "2  Word expert reveal effect response forward. In...  Address   \n",
101 |        "3  Son specific think. Conference its home expert...  Address   \n",
102 |        "4  Control trouble Apt. 223 keep action rule. Sol...  Address   \n",
103 |        "\n",
104 |        "                                                 PII  \n",
105 |        "0    9479 Ward Creek Suite 474 Warnershire, NH 08434  \n",
106 |        "1                                          Suite 937  \n",
107 |        "2                                           Apt. 925  \n",
108 |        "3  15286 Ramirez Lakes Apt. 147 West Anthony, FL ...  \n",
109 |        "4                                           Apt. 223  "
110 |       ]
111 |      },
112 |      "execution_count": 3,
113 |      "metadata": {},
114 |      "output_type": "execute_result"
115 |     }
116 |    ],
117 |    "source": [
118 |     "train_data.head()"
119 |    ]
120 |   },
121 |   {
122 |    "cell_type": "code",
123 |    "execution_count": 90,
124 |    "metadata": {},
125 |    "outputs": [
126 |     {
127 |      "data": {
128 |       "text/plain": [
129 |        "Name                100000\n",
130 |        "SSN                 100000\n",
131 |        "Address             100000\n",
132 |        "None                100000\n",
133 |        "Email               100000\n",
134 |        "Plates              100000\n",
135 |        "Phone_number        100000\n",
136 |        "CreditCardNumber    100000\n",
137 |        "Name: Labels, dtype: int64"
138 |       ]
139 |      },
140 |      "execution_count": 90,
141 |      "metadata": {},
142 |      "output_type": "execute_result"
143 |     }
144 |    ],
145 |    "source": [
146 |     "train_data['Labels'].value_counts()"
147 |    ]
148 |   },
149 |   {
150 |    "cell_type": "code",
151 |    "execution_count": 98,
152 |    "metadata": {},
153 |    "outputs": [],
154 |    "source": [
155 |     "example = train_data[\"Text\"].loc[train_data.Labels == \"SSN\"].iloc[0:100]"
156 |    ]
157 |   },
158 |   {
159 |    "cell_type": "code",
160 |    "execution_count": 99,
161 |    "metadata": {},
162 |    "outputs": [],
163 |    "source": [
164 |     "def clean_text(text):\n",
165 |     "    # replace  . and a space with only a space, then amke all words lower case.\n",
166 |     "    text = text.replace(\". \",\" \").replace(\",\",\"\").lower()\n",
167 |     "    # get rid of the . at the end of each line. \n",
168 |     "    cleaned_text = re.sub(\"\\.$\",\"\",text)\n",
169 |     "    \n",
170 |     "    return cleaned_text"
171 |    ]
172 |   },
173 |   {
174 |    "cell_type": "code",
175 |    "execution_count": 127,
176 |    "metadata": {},
177 |    "outputs": [],
178 |    "source": [
179 |     "def extract_special_len(text):\n",
180 |     "    # the complete special char list\n",
181 |     "    all_special_char = '!\"#$%&\\'()*+,-./:;<=>?@[\\\\]^_`{|}~'\n",
182 |     "    # return the special char if it is in the text\n",
183 |     "    special_list = [special_char for special_char in all_special_char\\\n",
184 |     "                   if special_char in text]\n",
185 |     "    \n",
186 |     "    return len(special_list)   \n",
187 |     "\n",
188 |     "\n",
189 |     "def extract_special_char(text):\n",
190 |     "    # the complete special char list\n",
191 |     "    all_special_char = '!\"#$%&\\'()*+,-./:;<=>?@[\\\\]^_`{|}~'\n",
192 |     "    # return the special char if it is in the text\n",
193 |     "    special_list = [special_char for special_char in all_special_char\\\n",
194 |     "                   if special_char in text]\n",
195 |     "    \n",
196 |     "    return special_list  "
197 |    ]
198 |   },
199 |   {
200 |    "cell_type": "code",
201 |    "execution_count": 128,
202 |    "metadata": {},
203 |    "outputs": [],
204 |    "source": [
205 |     "train_data[\"Cleaned_text\"] = train_data[\"Text\"].apply(clean_text)"
206 |    ]
207 |   },
208 |   {
209 |    "cell_type": "code",
210 |    "execution_count": 129,
211 |    "metadata": {},
212 |    "outputs": [],
213 |    "source": [
214 |     "train_data['Special_char_num'] = train_data[\"Cleaned_text\"].apply(extract_special_len)"
215 |    ]
216 |   },
217 |   {
218 |    "cell_type": "code",
219 |    "execution_count": 130,
220 |    "metadata": {},
221 |    "outputs": [],
222 |    "source": [
223 |     "train_data['Special_chars'] = train_data[\"Cleaned_text\"].apply(extract_special_char)"
224 |    ]
225 |   },
226 |   {
227 |    "cell_type": "code",
228 |    "execution_count": 131,
229 |    "metadata": {},
230 |    "outputs": [
231 |     {
232 |      "data": {
233 |       "text/html": [
234 |        "<div>\n",
235 |        "<style scoped>\n",
236 |        "    .dataframe tbody tr th:only-of-type {\n",
237 |        "        vertical-align: middle;\n",
238 |        "    }\n",
239 |        "\n",
240 |        "    .dataframe tbody tr th {\n",
241 |        "        vertical-align: top;\n",
242 |        "    }\n",
243 |        "\n",
244 |        "    .dataframe thead tr th {\n",
245 |        "        text-align: left;\n",
246 |        "    }\n",
247 |        "\n",
248 |        "    .dataframe thead tr:last-of-type th {\n",
249 |        "        text-align: right;\n",
250 |        "    }\n",
251 |        "</style>\n",
252 |        "<table border=\"1\" class=\"dataframe\">\n",
253 |        "  <thead>\n",
254 |        "    <tr>\n",
255 |        "      <th></th>\n",
256 |        "      <th colspan=\"4\" halign=\"left\">Special_char_num</th>\n",
257 |        "    </tr>\n",
258 |        "    <tr>\n",
259 |        "      <th></th>\n",
260 |        "      <th>min</th>\n",
261 |        "      <th>max</th>\n",
262 |        "      <th>mean</th>\n",
263 |        "      <th>median</th>\n",
264 |        "    </tr>\n",
265 |        "    <tr>\n",
266 |        "      <th>Labels</th>\n",
267 |        "      <th></th>\n",
268 |        "      <th></th>\n",
269 |        "      <th></th>\n",
270 |        "      <th></th>\n",
271 |        "    </tr>\n",
272 |        "  </thead>\n",
273 |        "  <tbody>\n",
274 |        "    <tr>\n",
275 |        "      <th>Address</th>\n",
276 |        "      <td>0</td>\n",
277 |        "      <td>0</td>\n",
278 |        "      <td>0.00000</td>\n",
279 |        "      <td>0</td>\n",
280 |        "    </tr>\n",
281 |        "    <tr>\n",
282 |        "      <th>CreditCardNumber</th>\n",
283 |        "      <td>0</td>\n",
284 |        "      <td>0</td>\n",
285 |        "      <td>0.00000</td>\n",
286 |        "      <td>0</td>\n",
287 |        "    </tr>\n",
288 |        "    <tr>\n",
289 |        "      <th>Email</th>\n",
290 |        "      <td>2</td>\n",
291 |        "      <td>3</td>\n",
292 |        "      <td>2.16875</td>\n",
293 |        "      <td>2</td>\n",
294 |        "    </tr>\n",
295 |        "    <tr>\n",
296 |        "      <th>Name</th>\n",
297 |        "      <td>0</td>\n",
298 |        "      <td>0</td>\n",
299 |        "      <td>0.00000</td>\n",
300 |        "      <td>0</td>\n",
301 |        "    </tr>\n",
302 |        "    <tr>\n",
303 |        "      <th>None</th>\n",
304 |        "      <td>0</td>\n",
305 |        "      <td>0</td>\n",
306 |        "      <td>0.00000</td>\n",
307 |        "      <td>0</td>\n",
308 |        "    </tr>\n",
309 |        "    <tr>\n",
310 |        "      <th>Phone_number</th>\n",
311 |        "      <td>0</td>\n",
312 |        "      <td>3</td>\n",
313 |        "      <td>1.48627</td>\n",
314 |        "      <td>1</td>\n",
315 |        "    </tr>\n",
316 |        "    <tr>\n",
317 |        "      <th>Plates</th>\n",
318 |        "      <td>0</td>\n",
319 |        "      <td>1</td>\n",
320 |        "      <td>0.28694</td>\n",
321 |        "      <td>0</td>\n",
322 |        "    </tr>\n",
323 |        "    <tr>\n",
324 |        "      <th>SSN</th>\n",
325 |        "      <td>0</td>\n",
326 |        "      <td>1</td>\n",
327 |        "      <td>0.60662</td>\n",
328 |        "      <td>1</td>\n",
329 |        "    </tr>\n",
330 |        "  </tbody>\n",
331 |        "</table>\n",
332 |        "</div>"
333 |       ],
334 |       "text/plain": [
335 |        "                 Special_char_num                    \n",
336 |        "                              min max     mean median\n",
337 |        "Labels                                               \n",
338 |        "Address                         0   0  0.00000      0\n",
339 |        "CreditCardNumber                0   0  0.00000      0\n",
340 |        "Email                           2   3  2.16875      2\n",
341 |        "Name                            0   0  0.00000      0\n",
342 |        "None                            0   0  0.00000      0\n",
343 |        "Phone_number                    0   3  1.48627      1\n",
344 |        "Plates                          0   1  0.28694      0\n",
345 |        "SSN                             0   1  0.60662      1"
346 |       ]
347 |      },
348 |      "execution_count": 131,
349 |      "metadata": {},
350 |      "output_type": "execute_result"
351 |     }
352 |    ],
353 |    "source": [
354 |     "train_data.groupby(\"Labels\").agg({'Special_char_num':[\"min\",\"max\",\"mean\",'median']})"
355 |    ]
356 |   }
357 |  ],
358 |  "metadata": {
359 |   "kernelspec": {
360 |    "display_name": "Python 3",
361 |    "language": "python",
362 |    "name": "python3"
363 |   },
364 |   "language_info": {
365 |    "codemirror_mode": {
366 |     "name": "ipython",
367 |     "version": 3
368 |    },
369 |    "file_extension": ".py",
370 |    "mimetype": "text/x-python",
371 |    "name": "python",
372 |    "nbconvert_exporter": "python",
373 |    "pygments_lexer": "ipython3",
374 |    "version": "3.6.6"
375 |   }
376 |  },
377 |  "nbformat": 4,
378 |  "nbformat_minor": 2
379 | }
380 | 


--------------------------------------------------------------------------------
/data/FakePII.py:
--------------------------------------------------------------------------------
  1 | # for command line usage
  2 | import sys, getopt
  3 | 
  4 | 
  5 | # for creating fake values
  6 | from faker import Faker
  7 | # for random sampling
  8 | from numpy.random import choice as choose
  9 | from numpy.random import seed as setseed
 10 | # for progress bar
 11 | from tqdm import tqdm
 12 | # for generating file name with current time
 13 | from datetime import datetime
 14 | # for saving the final dataframe.
 15 | import pandas as pd
 16 | 
 17 | from faker import Faker
 18 | from numpy.random import choice as choose
 19 | from numpy.random import seed as setseed
 20 | 
 21 | from tqdm import tqdm
 22 | 
 23 | from datetime import datetime
 24 | 
 25 | import pandas as pd
 26 | 
 27 | 
 28 | 
 29 | def sep_change(text, init_sep, after_sep):
 30 |     """
 31 |     A function to separate the input text into strings by the init_sep variable
 32 |     
 33 |     and combine the strings again by the after_sep variable.
 34 |     """
 35 |     splitted_strings = text.split(init_sep)
 36 |     
 37 |     combined_strings = after_sep.join(splitted_strings)
 38 |     
 39 |     return combined_strings
 40 | 
 41 | def convert_datetime_underscore(data):
 42 |     """
 43 |     A Helper function to convert all separators in datetime.now() into underscore. 
 44 |     """
 45 |     now_string = str(data)
 46 |     
 47 |     now_string = sep_change(now_string, init_sep = "-", after_sep = "_")
 48 |     
 49 |     now_string = sep_change(now_string, init_sep = " ", after_sep = "_")
 50 |     
 51 |     now_string = sep_change(now_string, init_sep = ":", after_sep = "_")
 52 |     
 53 |     now_string = sep_change(now_string, init_sep = ".", after_sep = "_")
 54 |     
 55 |     return now_string
 56 |     
 57 | def _random_sep_change(data, init_sep = "-", after_sep = " ", percentage = 0.5 , seed = 7):
 58 |     """
 59 |     A function to randomly change the SSN data's separator. 
 60 |     
 61 |     The input data is a list.
 62 |     """
 63 |     setseed(seed)
 64 |     # generate the index for replacing separator.
 65 |     replacing_indexes = choose(range(len(data)), int(len(data)*percentage))
 66 |     
 67 |     for each_replacing_index in replacing_indexes:
 68 |         # change the ssn data's separator from init_sep to after_sep
 69 |         data[each_replacing_index] = sep_change(data[each_replacing_index], init_sep, after_sep)
 70 |         
 71 |     return data
 72 |         
 73 | 
 74 |     
 75 |     
 76 | 
 77 | class Fake_PII():
 78 |     '''
 79 |     A class to generate a number of fake profiles, training/testing text mixed with
 80 |     different types of fake PIIs.
 81 |     '''
 82 |     def __init__(self, n_profile = None,fake_profiles = None, seed = 7,\
 83 |                 pii_with_text = None, pii_labels = None, PII = None):
 84 | 
 85 |             
 86 |         
 87 |         # initialize the Faker from faker package for fake data generation.
 88 |         try:
 89 |             self.faker = Faker()
 90 |         except ImportError as error:
 91 |             print(error.__class__.__name__ + ": " + error.message)
 92 |             
 93 | 
 94 |         self.n_profile = n_profile
 95 |         self.pii_with_text = pii_with_text
 96 |         self.pii_labels = pii_labels
 97 |         self.fake_profiles = fake_profiles
 98 |         self.seed = seed 
 99 |         self.PII = PII 
100 |  
101 |         
102 |     def create_fake_profile(self, n_profile, verbose = False, ssn_sep_change = True):
103 |         
104 |         assert isinstance(n_profile, int), "Please enter an integer\
105 |         for the number of generated profiles."
106 |         self.n_profile = n_profile
107 |         
108 |         fake_profiles = dict()
109 |         # use faker package to generate either a full/last name/first name.
110 |         fake_profiles["Name"] = [choose([self.faker.name(),\
111 |                                          self.faker.last_name(),
112 |                                          self.faker.first_name()])\
113 |                                  for _ in range(self.n_profile)]
114 |         # use faker to generate either a full/secondary/street address
115 |         fake_profiles["Address"] = [choose([self.faker.address(),\
116 |                                             self.faker.street_address(),\
117 |                                             self.faker.secondary_address()])\
118 |                                    for _ in range(self.n_profile)]
119 |         
120 |         fake_profiles["SSN"] = [self.faker.ssn() for _ in range(self.n_profile)]
121 |         
122 |         fake_profiles["Email"] = [self.faker.email() for _ in range(self.n_profile)]
123 |                                  
124 |         fake_profiles["Plates"] = [self.faker.license_plate()\
125 |                                    for _ in range(self.n_profile)]
126 |                                         
127 |         
128 |         fake_profiles["CreditCardNumber"] = [self.faker.credit_card_number()\
129 |                                              for _ in range(self.n_profile)]
130 |                                      
131 |         
132 |         fake_profiles["Phone_number"] = [self.faker.phone_number()\
133 |                                          for _ in range(self.n_profile)]
134 |         
135 |         # change the separator in SSN data.
136 |         if ssn_sep_change:
137 |             fake_profiles["SSN"] = _random_sep_change(fake_profiles["SSN"])
138 |                                         
139 |         # change the separator in Address data from "/n" to " "
140 |         fake_profiles['Address'] = [sep_change(each_address, init_sep = "\n" , after_sep = " ")\
141 |                                    for each_address in fake_profiles['Address'] ]
142 |                                               
143 |         
144 |         
145 |         self.fake_profiles = fake_profiles
146 |         print("Finished creating fake profiles.")
147 |         
148 |         if verbose:
149 |             return self.fake_profiles 
150 |         
151 |     def _init_pii_gen_train(self):
152 |         
153 |         # generate the all possible PII implemented in the create_fake_profile methods
154 |         self._fake_labels = list(self.fake_profiles.keys())
155 |         
156 |         # generate the None labels 
157 |         self._none_pii_labels = ["None" for _ in range(self._n_text)]
158 |         
159 |         # generate the pii labels
160 |         self.pii_labels = sorted(self._fake_labels*self._n_text)
161 |         
162 |         # generate the test with no pii
163 |         self._fake_text_no_pii = [self.faker.paragraph() for _ in range(self._n_text)]
164 |         
165 |         # mutiply the no pii text with the number of PII types
166 |         self._init_fake_text_no_pii = self._fake_text_no_pii*len(self._fake_labels)     
167 |         
168 |         # initialize the text mixed with PII with all "None" strings. 
169 |         self.pii_with_text = ["None" for _ in range(len(self._fake_labels)*(self._n_text))]
170 |         
171 |         # initialize the PII with all "None" strings
172 |         self.PII = ["None" for _ in range(self._n_text*(len(self._fake_labels)+1))]
173 |         
174 |     def _random_pii_insert(self):
175 |         # randomly insert PII into the text
176 |         for index, PII in enumerate(tqdm(self.pii_labels)):
177 |             # choose a PII value from the dictionary according to the PII type.
178 |             PII_value = choose(self.fake_profiles[PII])
179 |             
180 |             original_fake_text = self._init_fake_text_no_pii[index]
181 |             
182 |             tokenized_fake_text = original_fake_text.split(" ")
183 |             
184 |             # generate the position to fill in the PII value
185 |             PII_position = choose(range(len(tokenized_fake_text)+1))
186 |             
187 |             tokenized_fake_text.insert(PII_position, PII_value)
188 |             
189 |             one_text_mixed_with_PII = " ".join(tokenized_fake_text)
190 |             
191 |             self.pii_with_text[index] = one_text_mixed_with_PII
192 |             self.PII[index] = PII_value
193 |         
194 |     
195 |     def create_pii_text_train(self, n_text = 10):
196 |         """
197 |         A method to create the training text randomly mixed with fake PII. This
198 |         method creates a text and mixed it with different kinds of PII, which leads
199 |         to a total number of (num_of_PII)*(n_text) rows.
200 |         
201 |         """
202 |         warning_text = "Please create fake profiles first with .create_fake_profile method."
203 |         assert self.fake_profiles is not None, warning_text
204 |         
205 |         self._n_text = n_text
206 |         # initialized a few variables for inserting PII values 
207 |         self._init_pii_gen_train()
208 |         
209 |         # randomly insert Pii text into the paragraph.
210 |         self._random_pii_insert()
211 |         
212 |         
213 |         self.pii_with_text.extend(self._fake_text_no_pii)
214 |         self.pii_labels.extend(self._none_pii_labels)
215 |         
216 |    
217 |         
218 |         return self.pii_labels, self.pii_with_text, self.PII
219 |         
220 |     def _init_pii_gen_test(self):
221 |         
222 |         # generate the all possible PII implemented in the create_fake_profile methods
223 |         self._fake_labels = list(self.fake_profiles.keys())
224 |         
225 |         # generate the None labels 
226 |         self._none_pii_labels = ["None" for _ in range(self._n_text)]
227 |         
228 |         # generate the pii labels
229 |         self.pii_labels = sorted(self._fake_labels*self._n_text)  
230 |         
231 |         total_num_pii_text = (1+len(self._fake_labels))*(self._n_text)
232 |         # generate the fake text with no pii
233 |         self._init_fake_text_no_pii = [self.faker.paragraph() for _ in range(total_num_pii_text)]
234 |         
235 |         # initialize the text mixed with PII with all "None" strings. 
236 |         self.pii_with_text = self._init_fake_text_no_pii
237 |         # initialize the PII with all "None" strings
238 |         self.PII = ["None" for _ in range(total_num_pii_text)]
239 |         
240 |     
241 |     def create_pii_text_test(self, n_text = 10):
242 |         """
243 |         A method to create the testing text randomly mixed with fake PII. 
244 |         This method creates a text and mixed it with a type of PII. 
245 |         
246 |         In the training text, a normal text is repeated used to insert different PIIs into
247 |         it. In the testing text, a normal text is not intentionally repeated to insert 
248 |         different PIIs. 
249 |         
250 |         """
251 |         warning_text = "Please create fake profiles first with .create_fake_profile method."
252 |         assert self.fake_profiles is not None, warning_text
253 |         
254 |         self._n_text = n_text
255 |         
256 |         # initialized a few variables for inserting PII values 
257 |         self._init_pii_gen_test()
258 |         
259 |         # randomly insert Pii text into the paragraph.
260 |         self._random_pii_insert()
261 |         # add the none labels 
262 |         self.pii_labels.extend(self._none_pii_labels)
263 |        
264 |         return self.pii_labels, self.pii_with_text, self.PII
265 |         
266 | 
267 |         
268 | def write_to_disk_train(requested_train_data_size):
269 |     
270 |     fake_ = Fake_PII()
271 |     fake_.create_fake_profile(requested_train_data_size)
272 |     train_labels, train_text, train_PII = fake_.create_pii_text_train(n_text = requested_train_data_size)
273 |     
274 |     # save training data to disk
275 |     train_text_with_pii = pd.DataFrame({"Text":train_text, "Labels":train_labels, "PII":train_PII})
276 |     train_file_name = "train_text_with_pii_" + convert_datetime_underscore(datetime.now()) + ".csv"
277 |     train_text_with_pii.to_csv(train_file_name,index=False)
278 | 
279 | def write_to_disk_test(requested_test_data_size):
280 |     fake_ = Fake_PII()
281 |     fake_.create_fake_profile(requested_test_data_size)
282 |     test_labels, test_text, test_PII = fake_.create_pii_text_test(n_text = requested_test_data_size)
283 |     
284 |     # save testing data to disk
285 |     test_text_with_pii = pd.DataFrame({"Text":test_text, "Labels":test_labels, "PII":test_PII})
286 |     test_file_name = "test_text_with_pii_" + convert_datetime_underscore(datetime.now()) + ".csv"
287 |     test_text_with_pii.to_csv(test_file_name,index=False)
288 |     
289 |     
290 |     
291 |     
292 | if __name__ == "__main__":
293 |     argument_list = sys.argv[1:]
294 |     
295 |     if "-train" in argument_list:
296 |         train_value_index = argument_list.index("-train") + 1
297 |         requested_train_data_size = int(argument_list[train_value_index])
298 |         
299 |         write_to_disk_train(requested_train_data_size)
300 |         
301 |     if "-test" in argument_list:
302 |         test_value_index = argument_list.index("-test") + 1
303 |         requested_test_data_size = int(argument_list[test_value_index])
304 |         
305 |         write_to_disk_test(requested_test_data_size)
306 |     
307 |     
308 | # example command line use
309 | 
310 | # python FakePII.py -train 1000 -test 100
311 | # python FakePII.py  -test 100
312 | # python FakePII.py -train 1000 
313 |        
314 |     
315 |     
316 |     
317 |     
318 |     
319 |     
320 |     
321 |     
322 | 


--------------------------------------------------------------------------------
/data/create_text_with_fake_pii.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "markdown",
  5 |    "metadata": {},
  6 |    "source": [
  7 |     "The goal of this notebook is to create some fake PII data and merge it with some blog or news for training. "
  8 |    ]
  9 |   },
 10 |   {
 11 |    "cell_type": "markdown",
 12 |    "metadata": {},
 13 |    "source": [
 14 |     "The generated PII type is SSN, Name (first or last name), address (street address or full address), email (personal or company), phone_number, credit card number, birthday.   "
 15 |    ]
 16 |   },
 17 |   {
 18 |    "cell_type": "code",
 19 |    "execution_count": 1,
 20 |    "metadata": {},
 21 |    "outputs": [],
 22 |    "source": [
 23 |     "from faker import Faker\n",
 24 |     "from numpy.random import choice as choose\n",
 25 |     "from numpy.random import seed as setseed\n",
 26 |     "\n",
 27 |     "from tqdm import tqdm\n",
 28 |     "\n",
 29 |     "from datetime import datetime\n",
 30 |     "\n",
 31 |     "import pandas as pd\n",
 32 |     "\n",
 33 |     "\n",
 34 |     "\n",
 35 |     "def sep_change(text, init_sep, after_sep):\n",
 36 |     "    \"\"\"\n",
 37 |     "    A function to separate the input text into strings by the init_sep variable\n",
 38 |     "    \n",
 39 |     "    and combine the strings again by the after_sep variable.\n",
 40 |     "    \"\"\"\n",
 41 |     "    splitted_strings = text.split(init_sep)\n",
 42 |     "    \n",
 43 |     "    combined_strings = after_sep.join(splitted_strings)\n",
 44 |     "    \n",
 45 |     "    return combined_strings\n",
 46 |     "\n",
 47 |     "def convert_datetime_underscore(data):\n",
 48 |     "    \"\"\"\n",
 49 |     "    A Helper function to convert all separators in datetime.now() into underscore. \n",
 50 |     "    \"\"\"\n",
 51 |     "    now_string = str(data)\n",
 52 |     "    \n",
 53 |     "    now_string = sep_change(now_string, init_sep = \"-\", after_sep = \"_\")\n",
 54 |     "    \n",
 55 |     "    now_string = sep_change(now_string, init_sep = \" \", after_sep = \"_\")\n",
 56 |     "    \n",
 57 |     "    now_string = sep_change(now_string, init_sep = \":\", after_sep = \"_\")\n",
 58 |     "    \n",
 59 |     "    now_string = sep_change(now_string, init_sep = \".\", after_sep = \"_\")\n",
 60 |     "    \n",
 61 |     "    return now_string\n",
 62 |     "    \n",
 63 |     "def _random_sep_change(data, init_sep = \"-\", after_sep = \" \", percentage = 0.5 , seed = 7):\n",
 64 |     "    \"\"\"\n",
 65 |     "    A function to randomly change the SSN data's separator. \n",
 66 |     "    \n",
 67 |     "    The input data is a list.\n",
 68 |     "    \"\"\"\n",
 69 |     "    setseed(seed)\n",
 70 |     "    # generate the index for replacing separator.\n",
 71 |     "    replacing_indexes = choose(range(len(data)), int(len(data)*percentage))\n",
 72 |     "    \n",
 73 |     "    for each_replacing_index in replacing_indexes:\n",
 74 |     "        # change the ssn data's separator from init_sep to after_sep\n",
 75 |     "        data[each_replacing_index] = sep_change(data[each_replacing_index], init_sep, after_sep)\n",
 76 |     "        \n",
 77 |     "    return data\n",
 78 |     "        \n",
 79 |     "\n",
 80 |     "    \n",
 81 |     "    \n",
 82 |     "\n",
 83 |     "class Fake_PII():\n",
 84 |     "    '''\n",
 85 |     "    A class to generate a number of fake profiles, training/testing text mixed with\n",
 86 |     "    different types of fake PIIs.\n",
 87 |     "    '''\n",
 88 |     "    def __init__(self, n_profile = None,fake_profiles = Nongne, seed = 7,\\\n",
 89 |     "                pii_with_text = None, pii_labels = None, PII = None):\n",
 90 |     "\n",
 91 |     "            \n",
 92 |     "        \n",
 93 |     "        # initialize the Faker from faker package for fake data generation.\n",
 94 |     "        try:\n",
 95 |     "            self.faker = Faker()\n",
 96 |     "        except ImportError as error:\n",
 97 |     "            print(error.__class__.__name__ + \": \" + error.message)\n",
 98 |     "            \n",
 99 |     "\n",
100 |     "        self.n_profile = n_profile\n",
101 |     "        self.pii_with_text = pii_with_text\n",
102 |     "        self.pii_labels = pii_labels\n",
103 |     "        self.fake_profiles = fake_profiles\n",
104 |     "        self.seed = seed \n",
105 |     "        self.PII = PII \n",
106 |     " \n",
107 |     "        \n",
108 |     "    def create_fake_profile(self, n_profile, verbose = False, ssn_sep_change = True):\n",
109 |     "        \n",
110 |     "        assert isinstance(n_profile, int), \"Please enter an integer\\\n",
111 |     "        for the number of generated profiles.\"\n",
112 |     "        self.n_profile = n_profile\n",
113 |     "        \n",
114 |     "        fake_profiles = dict()\n",
115 |     "        # use faker package to generate either a full/last name/first name.\n",
116 |     "        fake_profiles[\"Name\"] = [choose([self.faker.name(),\\\n",
117 |     "                                         self.faker.last_name(),\n",
118 |     "                                         self.faker.first_name()])\\\n",
119 |     "                                 for _ in range(self.n_profile)]\n",
120 |     "        # use faker to generate either a full/secondary/street address\n",
121 |     "        fake_profiles[\"Address\"] = [choose([self.faker.address(),\\\n",
122 |     "                                            self.faker.street_address(),\\\n",
123 |     "                                            self.faker.secondary_address()])\\\n",
124 |     "                                   for _ in range(self.n_profile)]\n",
125 |     "        \n",
126 |     "        fake_profiles[\"SSN\"] = [self.faker.ssn() for _ in range(self.n_profile)]\n",
127 |     "        \n",
128 |     "        fake_profiles[\"Email\"] = [self.faker.email() for _ in range(self.n_profile)]\n",
129 |     "                                 \n",
130 |     "        fake_profiles[\"Plates\"] = [self.faker.license_plate()\\\n",
131 |     "                                   for _ in range(self.n_profile)]\n",
132 |     "                                        \n",
133 |     "        \n",
134 |     "        fake_profiles[\"CreditCardNumber\"] = [self.faker.credit_card_number()\\\n",
135 |     "                                             for _ in range(self.n_profile)]\n",
136 |     "                                     \n",
137 |     "        \n",
138 |     "        fake_profiles[\"Phone_number\"] = [self.faker.phone_number()\\\n",
139 |     "                                         for _ in range(self.n_profile)]\n",
140 |     "        \n",
141 |     "        # change the separator in SSN data.\n",
142 |     "        if ssn_sep_change:\n",
143 |     "            fake_profiles[\"SSN\"] = _random_sep_change(fake_profiles[\"SSN\"])\n",
144 |     "                                        \n",
145 |     "        # change the separator in Address data from \"/n\" to \" \"\n",
146 |     "        fake_profiles['Address'] = [sep_change(each_address, init_sep = \"\\n\" , after_sep = \" \")\\\n",
147 |     "                                   for each_address in fake_profiles['Address'] ]\n",
148 |     "                                              \n",
149 |     "        \n",
150 |     "        \n",
151 |     "        self.fake_profiles = fake_profiles\n",
152 |     "        \n",
153 |     "        if verbose:\n",
154 |     "            return self.fake_profiles \n",
155 |     "        \n",
156 |     "    def _init_pii_gen_train(self):\n",
157 |     "        \n",
158 |     "        # generate the all possible PII implemented in the create_fake_profile methods\n",
159 |     "        self._fake_labels = list(self.fake_profiles.keys())\n",
160 |     "        \n",
161 |     "        # generate the None labels \n",
162 |     "        self._none_pii_labels = [\"None\" for _ in range(self._n_text)]\n",
163 |     "        \n",
164 |     "        # generate the pii labels\n",
165 |     "        self.pii_labels = sorted(self._fake_labels*self._n_text)\n",
166 |     "        \n",
167 |     "        # generate the test with no pii\n",
168 |     "        self._fake_text_no_pii = [self.faker.paragraph() for _ in range(self._n_text)]\n",
169 |     "        \n",
170 |     "        # mutiply the no pii text with the number of PII types\n",
171 |     "        self._init_fake_text_no_pii = self._fake_text_no_pii*len(self._fake_labels)     \n",
172 |     "        \n",
173 |     "        # initialize the text mixed with PII with all \"None\" strings. \n",
174 |     "        self.pii_with_text = [\"None\" for _ in range(len(self._fake_labels)*(self._n_text))]\n",
175 |     "        \n",
176 |     "        # initialize the PII with all \"None\" strings\n",
177 |     "        self.PII = [\"None\" for _ in range(self._n_text*(len(self._fake_labels)+1))]\n",
178 |     "        \n",
179 |     "    def _random_pii_insert(self):\n",
180 |     "        # randomly insert PII into the text\n",
181 |     "        for index, PII in enumerate(tqdm(self.pii_labels)):\n",
182 |     "            # choose a PII value from the dictionary according to the PII type.\n",
183 |     "            PII_value = choose(self.fake_profiles[PII])\n",
184 |     "            \n",
185 |     "            original_fake_text = self._init_fake_text_no_pii[index]\n",
186 |     "            \n",
187 |     "            tokenized_fake_text = original_fake_text.split(\" \")\n",
188 |     "            \n",
189 |     "            # generate the position to fill in the PII value\n",
190 |     "            PII_position = choose(range(len(tokenized_fake_text)+1))\n",
191 |     "            \n",
192 |     "            tokenized_fake_text.insert(PII_position, PII_value)\n",
193 |     "            \n",
194 |     "            one_text_mixed_with_PII = \" \".join(tokenized_fake_text)\n",
195 |     "            \n",
196 |     "            self.pii_with_text[index] = one_text_mixed_with_PII\n",
197 |     "            self.PII[index] = PII_value\n",
198 |     "        \n",
199 |     "    \n",
200 |     "    def create_pii_text_train(self, n_text = 10):\n",
201 |     "        \"\"\"\n",
202 |     "        A method to create the training text randomly mixed with fake PII. This\n",
203 |     "        method creates a text and mixed it with different kinds of PII, which leads\n",
204 |     "        to a total number of (num_of_PII)*(n_text) rows.\n",
205 |     "        \n",
206 |     "        \"\"\"\n",
207 |     "        warning_text = \"Please create fake profiles first with .create_fake_profile method.\"\n",
208 |     "        assert self.fake_profiles is not None, warning_text\n",
209 |     "        \n",
210 |     "        self._n_text = n_text\n",
211 |     "        # initialized a few variables for inserting PII values \n",
212 |     "        self._init_pii_gen_train()\n",
213 |     "        \n",
214 |     "        # randomly insert Pii text into the paragraph.\n",
215 |     "        self._random_pii_insert()\n",
216 |     "        \n",
217 |     "        \n",
218 |     "        self.pii_with_text.extend(self._fake_text_no_pii)\n",
219 |     "        self.pii_labels.extend(self._none_pii_labels)\n",
220 |     "        \n",
221 |     "   \n",
222 |     "        \n",
223 |     "        return self.pii_labels, self.pii_with_text, self.PII\n",
224 |     "        \n",
225 |     "    def _init_pii_gen_test(self):\n",
226 |     "        \n",
227 |     "        # generate the all possible PII implemented in the create_fake_profile methods\n",
228 |     "        self._fake_labels = list(self.fake_profiles.keys())\n",
229 |     "        \n",
230 |     "        # generate the None labels \n",
231 |     "        self._none_pii_labels = [\"None\" for _ in range(self._n_text)]\n",
232 |     "        \n",
233 |     "        # generate the pii labels\n",
234 |     "        self.pii_labels = sorted(self._fake_labels*self._n_text)  \n",
235 |     "        \n",
236 |     "        total_num_pii_text = (1+len(self._fake_labels))*(self._n_text)\n",
237 |     "        # generate the fake text with no pii\n",
238 |     "        self._init_fake_text_no_pii = [self.faker.paragraph() for _ in range(total_num_pii_text)]\n",
239 |     "        \n",
240 |     "        # initialize the text mixed with PII with all \"None\" strings. \n",
241 |     "        self.pii_with_text = self._init_fake_text_no_pii\n",
242 |     "        # initialize the PII with all \"None\" strings\n",
243 |     "        self.PII = [\"None\" for _ in range(total_num_pii_text)]\n",
244 |     "        \n",
245 |     "    \n",
246 |     "    def create_pii_text_test(self, n_text = 10):\n",
247 |     "        \"\"\"\n",
248 |     "        A method to create the testing text randomly mixed with fake PII. \n",
249 |     "        This method creates a text and mixed it with a type of PII. \n",
250 |     "        \n",
251 |     "        In the training text, a normal text is repeated used to insert different PIIs into\n",
252 |     "        it. In the testing text, a normal text is not intentionally repeated to insert \n",
253 |     "        different PIIs. \n",
254 |     "        \n",
255 |     "        \"\"\"\n",
256 |     "        warning_text = \"Please create fake profiles first with .create_fake_profile method.\"\n",
257 |     "        assert self.fake_profiles is not None, warning_text\n",
258 |     "        \n",
259 |     "        self._n_text = n_text\n",
260 |     "        \n",
261 |     "        # initialized a few variables for inserting PII values \n",
262 |     "        self._init_pii_gen_test()\n",
263 |     "        \n",
264 |     "        # randomly insert Pii text into the paragraph.\n",
265 |     "        self._random_pii_insert()\n",
266 |     "        # add the none labels \n",
267 |     "        self.pii_labels.extend(self._none_pii_labels)\n",
268 |     "       \n",
269 |     "        return self.pii_labels, self.pii_with_text, self.PII\n",
270 |     "        \n",
271 |     "    "
272 |    ]
273 |   },
274 |   {
275 |    "cell_type": "code",
276 |    "execution_count": 2,
277 |    "metadata": {},
278 |    "outputs": [
279 |     {
280 |      "name": "stderr",
281 |      "output_type": "stream",
282 |      "text": [
283 |       "100%|██████████| 70/70 [00:00<00:00, 35023.41it/s]\n",
284 |       "100%|██████████| 70/70 [00:00<00:00, 37010.12it/s]\n"
285 |      ]
286 |     }
287 |    ],
288 |    "source": [
289 |     "fake_ = Fake_PII()\n",
290 |     "fake_.create_fake_profile(10)\n",
291 |     "train_labels, train_text, train_PII = fake_.create_pii_text_train(n_text = 10)\n",
292 |     "test_labels, test_text, test_PII = fake_.create_pii_text_test(n_text = 10)"
293 |    ]
294 |   },
295 |   {
296 |    "cell_type": "code",
297 |    "execution_count": 3,
298 |    "metadata": {},
299 |    "outputs": [],
300 |    "source": [
301 |     "train_text_with_pii = pd.DataFrame({\"Text\":train_text, \"Labels\":train_labels, \"PII\":train_PII})\n",
302 |     "train_file_name = \"train_text_with_pii_\" + convert_datetime_underscore(datetime.now()) + \".csv\"\n",
303 |     "train_text_with_pii.to_csv(train_file_name,index=False)\n",
304 |     "\n",
305 |     "\n",
306 |     "\n",
307 |     "\n",
308 |     "\n",
309 |     "test_text_with_pii = pd.DataFrame({\"Text\":test_labels, \"Labels\":test_text, \"PII\":test_PII})\n",
310 |     "test_file_name = \"test_text_with_pii_\" + convert_datetime_underscore(datetime.now()) + \".csv\"\n",
311 |     "test_text_with_pii.to_csv(test_file_name,index=False)\n",
312 |     "\n"
313 |    ]
314 |   },
315 |   {
316 |    "cell_type": "code",
317 |    "execution_count": 4,
318 |    "metadata": {
319 |     "scrolled": false
320 |    },
321 |    "outputs": [
322 |     {
323 |      "data": {
324 |       "text/plain": [
325 |        "['Maybe site begin pass relationship natural style. Suite 927 Full actually arrive house someone per understand.',\n",
326 |        " 'Have specific top already under part avoid. 25468 Matthew Village Theresaland, AR 06964 Sit state south film east.',\n",
327 |        " 'Ahead bank several nothing product chair. Per probably eat 6958 Julie Harbors Apt. 486 take kid measure.',\n",
328 |        " 'Instead many appear beautiful now. Within candidate 6958 Julie Harbors Apt. 486 great alone task general ever. Indeed like bar.',\n",
329 |        " 'Analysis our policy produce 56342 William Isle direction morning. Trade energy newspaper price. Too he five daughter.',\n",
330 |        " 'Card point particularly table imagine exactly. Office 7047 Raymond Stream Suite 912 Port Stephen, RI 57413 identify civil sign rich form wish. Stay type to under somebody include individual ability.',\n",
331 |        " 'When feeling hair fund bill. Let card team meeting already start. 7047 Raymond Stream Suite 912 Port Stephen, RI 57413 Impact wear single start respond.',\n",
332 |        " 'Subject but study language toward. Industry Suite 927 theory company rule individual. Decade picture better evening too support.',\n",
333 |        " 'Mission voice hope foreign see artist way. Every idea sort article so Apt. 893 sound. Large should why director bar training board service.',\n",
334 |        " 'Catch tend report little continue. That vote hold 3576 Davis Extension Suite 359 maintain itself movement affect true.',\n",
335 |        " 'Maybe site begin pass relationship natural style. 5266437982583777 Full actually arrive house someone per understand.',\n",
336 |        " '676161251563 Have specific top already under part avoid. Sit state south film east.',\n",
337 |        " 'Ahead bank 582102647028 several nothing product chair. Per probably eat take kid measure.',\n",
338 |        " 'Instead many appear beautiful now. Within candidate great alone 5266437982583777 task general ever. Indeed like bar.',\n",
339 |        " 'Analysis our policy produce 3526823311192151 direction morning. Trade energy newspaper price. Too he five daughter.',\n",
340 |        " 'Card point particularly 5172225577125994 table imagine exactly. Office identify civil sign rich form wish. Stay type to under somebody include individual ability.',\n",
341 |        " 'When feeling hair fund bill. Let card team 3594510054984228 meeting already start. Impact wear single start respond.',\n",
342 |        " 'Subject but study language toward. Industry theory company rule individual. Decade picture better evening too support. 3594510054984228',\n",
343 |        " '4694718030901409 Mission voice hope foreign see artist way. Every idea sort article so sound. Large should why director bar training board service.',\n",
344 |        " 'Catch tend report little continue. That 582102647028 vote hold maintain itself movement affect true.',\n",
345 |        " 'Maybe site begin pass relationship natural style. djenkins@best.com Full actually arrive house someone per understand.',\n",
346 |        " 'Have specific top robertmathis@beasley.net already under part avoid. Sit state south film east.',\n",
347 |        " 'Ahead bank several nothing product chair. Per georgemiller@hotmail.com probably eat take kid measure.',\n",
348 |        " 'Instead many appear beautiful now. Within candidate great alone task general ever. djenkins@best.com Indeed like bar.',\n",
349 |        " 'djenkins@best.com Analysis our policy produce direction morning. Trade energy newspaper price. Too he five daughter.',\n",
350 |        " 'Card point particularly table antonio67@yahoo.com imagine exactly. Office identify civil sign rich form wish. Stay type to under somebody include individual ability.',\n",
351 |        " 'When megan82@gmail.com feeling hair fund bill. Let card team meeting already start. Impact wear single start respond.',\n",
352 |        " 'Subject megan82@gmail.com but study language toward. Industry theory company rule individual. Decade picture better evening too support.',\n",
353 |        " 'Mission voice hope foreign megan82@gmail.com see artist way. Every idea sort article so sound. Large should why director bar training board service.',\n",
354 |        " 'Catch megan82@gmail.com tend report little continue. That vote hold maintain itself movement affect true.',\n",
355 |        " 'Maybe site begin pass relationship King natural style. Full actually arrive house someone per understand.',\n",
356 |        " 'Have King specific top already under part avoid. Sit state south film east.',\n",
357 |        " 'Ahead bank several Fields nothing product chair. Per probably eat take kid measure.',\n",
358 |        " 'Instead many Fields appear beautiful now. Within candidate great alone task general ever. Indeed like bar.',\n",
359 |        " 'Analysis our policy produce direction morning. Trade Jason Harris energy newspaper price. Too he five daughter.',\n",
360 |        " 'Card point particularly table imagine exactly. Office identify civil sign rich Mack form wish. Stay type to under somebody include individual ability.',\n",
361 |        " 'When feeling hair fund bill. Let card team meeting already start. Impact Karen wear single start respond.',\n",
362 |        " 'Subject but study language toward. Industry theory company rule King individual. Decade picture better evening too support.',\n",
363 |        " 'Mission voice hope foreign see artist way. Every idea sort article so sound. Large Mack should why director bar training board service.',\n",
364 |        " 'Catch tend report little continue. Jessica Schmidt That vote hold maintain itself movement affect true.',\n",
365 |        " '613.951.9248 Maybe site begin pass relationship natural style. Full actually arrive house someone per understand.',\n",
366 |        " 'Have specific top already under part avoid. Sit state south film 948.013.9924 east.',\n",
367 |        " 'Ahead bank several nothing product chair. Per probably eat take (843)852-4383x177 kid measure.',\n",
368 |        " 'Instead many appear beautiful now. Within candidate great alone task general 001-782-911-4967 ever. Indeed like bar.',\n",
369 |        " 'Analysis our policy produce direction morning. Trade energy newspaper price. Too he five 001-388-680-2806 daughter.',\n",
370 |        " 'Card point particularly table imagine exactly. Office 001-864-561-1553x6447 identify civil sign rich form wish. Stay type to under somebody include individual ability.',\n",
371 |        " 'When feeling hair fund bill. Let card team meeting already start. Impact 613.951.9248 wear single start respond.',\n",
372 |        " 'Subject but study language toward. Industry (843)852-4383x177 theory company rule individual. Decade picture better evening too support.',\n",
373 |        " 'Mission voice hope foreign see artist 049.898.2887x33574 way. Every idea sort article so sound. Large should why director bar training board service.',\n",
374 |        " 'Catch tend report little continue. That vote hold maintain itself movement affect 049.898.2887x33574 true.',\n",
375 |        " 'Maybe UJA-914 site begin pass relationship natural style. Full actually arrive house someone per understand.',\n",
376 |        " 'Have specific top already 6YE 755 under part avoid. Sit state south film east.',\n",
377 |        " 'Ahead bank several nothing product chair. Per probably eat 5ZR 477 take kid measure.',\n",
378 |        " 'Instead many appear beautiful now. Within candidate great alone task general ever. Indeed UWS-3761 like bar.',\n",
379 |        " 'Analysis our policy produce direction morning. Trade energy newspaper price. Too he GBG W23 five daughter.',\n",
380 |        " 'Card point 98X•766 particularly table imagine exactly. Office identify civil sign rich form wish. Stay type to under somebody include individual ability.',\n",
381 |        " 'When feeling hair fund bill. Let card team 98X•766 meeting already start. Impact wear single start respond.',\n",
382 |        " 'Subject but study language toward. Industry theory company rule individual. IFO-498 Decade picture better evening too support.',\n",
383 |        " 'Mission voice hope foreign see artist way. Every idea sort article so sound. Large should why director 5ZR 477 bar training board service.',\n",
384 |        " 'Catch tend report little continue. That vote hold maintain itself 98X•766 movement affect true.',\n",
385 |        " 'Maybe site begin pass relationship 408-16-3092 natural style. Full actually arrive house someone per understand.',\n",
386 |        " 'Have specific top already under part 408-16-3092 avoid. Sit state south film east.',\n",
387 |        " 'Ahead bank several nothing product chair. Per probably 518 94 2803 eat take kid measure.',\n",
388 |        " 'Instead many appear beautiful now. Within candidate great alone task general ever. Indeed like bar. 805 19 3539',\n",
389 |        " 'Analysis our policy produce direction morning. Trade energy newspaper 684-40-2127 price. Too he five daughter.',\n",
390 |        " 'Card point particularly table imagine exactly. Office identify civil sign rich form wish. Stay type to under somebody include individual ability. 003-93-1559',\n",
391 |        " 'When feeling hair fund bill. Let card team meeting already start. Impact 678 26 0624 wear single start respond.',\n",
392 |        " 'Subject but study language toward. Industry theory company rule individual. Decade 518 94 2803 picture better evening too support.',\n",
393 |        " 'Mission 805 19 3539 voice hope foreign see artist way. Every idea sort article so sound. Large should why director bar training board service.',\n",
394 |        " 'Catch tend report little continue. 684-40-2127 That vote hold maintain itself movement affect true.',\n",
395 |        " 'Maybe site begin pass relationship natural style. Full actually arrive house someone per understand.',\n",
396 |        " 'Have specific top already under part avoid. Sit state south film east.',\n",
397 |        " 'Ahead bank several nothing product chair. Per probably eat take kid measure.',\n",
398 |        " 'Instead many appear beautiful now. Within candidate great alone task general ever. Indeed like bar.',\n",
399 |        " 'Analysis our policy produce direction morning. Trade energy newspaper price. Too he five daughter.',\n",
400 |        " 'Card point particularly table imagine exactly. Office identify civil sign rich form wish. Stay type to under somebody include individual ability.',\n",
401 |        " 'When feeling hair fund bill. Let card team meeting already start. Impact wear single start respond.',\n",
402 |        " 'Subject but study language toward. Industry theory company rule individual. Decade picture better evening too support.',\n",
403 |        " 'Mission voice hope foreign see artist way. Every idea sort article so sound. Large should why director bar training board service.',\n",
404 |        " 'Catch tend report little continue. That vote hold maintain itself movement affect true.']"
405 |       ]
406 |      },
407 |      "execution_count": 4,
408 |      "metadata": {},
409 |      "output_type": "execute_result"
410 |     }
411 |    ],
412 |    "source": [
413 |     "train_text"
414 |    ]
415 |   },
416 |   {
417 |    "cell_type": "code",
418 |    "execution_count": 5,
419 |    "metadata": {
420 |     "scrolled": false
421 |    },
422 |    "outputs": [
423 |     {
424 |      "data": {
425 |       "text/plain": [
426 |        "['Friend his those run probably seek our. Trial same century conference matter 6958 Julie Harbors Apt. 486 economy fight. Yard others something.',\n",
427 |        " 'Sometimes myself only detail own. Occur his Suite 772 stock peace. From including down attorney special relationship.',\n",
428 |        " 'Everybody machine through simple will. Live wonder former identify. 3576 Davis Extension Suite 359',\n",
429 |        " 'Fact close 56342 William Isle get machine election section. Go authority attention.',\n",
430 |        " 'Stock feeling life mean. Certain 7047 Raymond Stream Suite 912 Port Stephen, RI 57413 various research administration upon another matter.',\n",
431 |        " 'Apt. 893 Behavior medical plan full. Scene dinner its know her.',\n",
432 |        " 'None health practice she old especially. Continue 3576 Davis Extension Suite 359 evening base bad. Clearly cost read.',\n",
433 |        " 'Away happen shake region. Standard middle indeed answer. Method organization become red feel. 3268 Teresa Plaza East Cassandra, HI 22259',\n",
434 |        " 'Off upon blue west 6958 Julie Harbors Apt. 486 great capital contain. Few consumer admit group say car.',\n",
435 |        " 'Enough method watch hand beautiful real nearly. Weight expect series 25468 Matthew Village Theresaland, AR 06964 anything leg become knowledge. Economic year open summer. Idea before including particular.',\n",
436 |        " 'After put 582102647028 both shake without prepare. Difference until sort southern lawyer high. Full notice operation skill.',\n",
437 |        " 'Meet company situation able store. Help 582102647028 although image subject everyone anything. Tax next every top education film.',\n",
438 |        " 'Teacher owner strong science candidate find. Anything within prepare 213150753498837 by design. Animal defense environment.',\n",
439 |        " 'Little amount and similar. This 5172225577125994 learn drop data prepare evening kid most.',\n",
440 |        " 'Machine board rather low action development focus animal. Able if either sport reality general relationship. 4694718030901409',\n",
441 |        " '582102647028 Determine cell thing south identify six doctor. Easy collection such rock type.',\n",
442 |        " 'Smile 5172225577125994 bring old. Person physical without admit.',\n",
443 |        " '4694718030901409 Figure future because bring stop process rule. Career carry author. Former vote vote improve example chance ten.',\n",
444 |        " 'Prove build tonight green. Wide another whose south drive plant 6554707631732887 past major.',\n",
445 |        " 'Defense recently listen artist support box improve. Party become dream nation have. Position president our south evidence yourself so. Million then can hot 5266437982583777 particularly western.',\n",
446 |        " 'Prepare trouble remain couple. wrightashley@stevens.com Well once state official tree.',\n",
447 |        " 'Mr reach fall old possible wear. Magazine what contain actually leg find robertmathis@beasley.net bag.',\n",
448 |        " 'Hope detail order mouth worry. Agency megan82@gmail.com history these election consider space.',\n",
449 |        " 'Young djenkins@best.com expert voice manager guy civil practice. Behind player tell. Occur group door quickly.',\n",
450 |        " 'Power task georgemiller@hotmail.com although something run. Visit risk close.',\n",
451 |        " 'Clearly measure bit require low civil fine. Stock difficult part chance free. djenkins@best.com Receive similar hit person night join.',\n",
452 |        " 'georgemiller@hotmail.com Fact through style experience culture effect those. Test land ask shake phone however act get. Agent ahead never people year.',\n",
453 |        " 'Training trip appear television machine happy physical. Mention activity news room from. Allow baby robertmathis@beasley.net remember politics imagine poor relate.',\n",
454 |        " 'Answer different christophercohen@juarez.com factor painting avoid try way. Staff record create thank actually occur. Contain loss letter manage western agency.',\n",
455 |        " 'Every go glass industry drop scientist. Likely recently walk trip once georgemiller@hotmail.com animal box.',\n",
456 |        " 'Half moment person catch idea Jason Harris success might. Together student major save also magazine.',\n",
457 |        " 'Karen Throw believe poor establish position. Hour hotel issue.',\n",
458 |        " 'This international accept him. Deborah Us pull writer. Energy as grow show.',\n",
459 |        " 'Control discover important imagine much. Future subject cut fly test entire hundred. Jason Harris',\n",
460 |        " 'Model face Mrs before. Officer value example happen public follow Norma Koch itself.',\n",
461 |        " 'Jessica Schmidt Chance seat culture condition game decision. Result move fish quality free whatever.',\n",
462 |        " 'Different realize look method Taylor American feel.',\n",
463 |        " 'Fields Toward west first however table. Mrs million such oil arrive form truth recently.',\n",
464 |        " 'However too PM friend interview nearly future. Billion real Mack perhaps worker still. Method difficult market region step indeed both. Operation money rise summer debate concern.',\n",
465 |        " 'Personal memory hot Deborah health history want. Decade remember against marriage participant again.',\n",
466 |        " 'Defense 001-388-680-2806 voice world apply specific. Congress social notice popular. Apply about worry group reduce surface.',\n",
467 |        " 'Church (164)750-6452x6754 chance stuff how easy since letter someone. Natural system fire someone at. Wall yeah show take movie loss.',\n",
468 |        " 'Unit work crime year land such. 948.013.9924 Year deal almost. Nearly second science.',\n",
469 |        " 'Down record media today poor his respond. Tree letter kind maintain baby else beautiful. +1-153-933-2197x935 Later partner majority community themselves window practice. Public final hair fear they usually.',\n",
470 |        " 'Foot (164)750-6452x6754 entire throw student black leg. Argue certainly close blood writer.',\n",
471 |        " 'And (164)750-6452x6754 alone cut debate along miss. Rule story resource whatever.',\n",
472 |        " 'Security (843)852-4383x177 amount herself drive wall evidence still ground. Game consider instead according. Bag stop building choice between.',\n",
473 |        " 'Bad of world least ready. Trouble impact miss 001-782-911-4967 start ago measure.',\n",
474 |        " 'Change own 049.898.2887x33574 color cost civil. West ball let hold. Use in right whole me attorney and.',\n",
475 |        " 'Join floor indicate man move court. Half 948.013.9924 phone for him office.',\n",
476 |        " 'Far their smile clearly. Wall price rather rich set time court. 5ZR 477',\n",
477 |        " 'Outside national HIU 641 bad physical.',\n",
478 |        " 'Lot personal result both view 5VH5838 push western. Car room adult event.',\n",
479 |        " 'Leg feeling process us plan memory take candidate. Weight IFO-498 simply community.',\n",
480 |        " 'However organization 98X•766 knowledge news western Mrs director down. History allow perhaps list cost.',\n",
481 |        " 'Talk explain end second source 6YE 755 discuss.',\n",
482 |        " 'Point least eat. Figure poor clearly 6YE 755 difficult tend. Wonder hot specific.',\n",
483 |        " 'Next ever small really. International community deep since laugh fund. GBG W23 Management cold plant action but little military.',\n",
484 |        " 'Need image finish. Another heavy machine forget perform when low. Yard whom media have 98X•766 which car fly.',\n",
485 |        " 'Plan defense member speech especially. Office total learn might reason. Help adult source those. 5ZR 477 Member so information admit president themselves after friend.',\n",
486 |        " 'Capital organization 805 19 3539 experience group particularly I born. Century suffer address federal participant.',\n",
487 |        " 'Particular from practice interest week writer decision. However 805 19 3539 future direction serious. Wrong here see million.',\n",
488 |        " '003-93-1559 Hold worker population as recognize such stock. Leg during shoulder management.',\n",
489 |        " 'House item current coach left. Purpose little purpose 684-40-2127 some act occur fund. Game three manager away staff direction some Republican.',\n",
490 |        " 'Data visit recognize brother until radio. Activity thousand hundred feel generation might 518 94 2803 economy. Risk he prevent house sport gun.',\n",
491 |        " 'We would service forward actually. Nice trouble 003-93-1559 test either industry class part address. Site remain rich decide trouble.',\n",
492 |        " 'Than computer 805 19 3539 billion make mean scientist. Despite time do service often.',\n",
493 |        " 'Situation she raise clear. Peace for 684-40-2127 girl maybe.',\n",
494 |        " 'Worry pick mission onto various court. Ball fact blood none. Within today foot couple 408-16-3092 alone fly.',\n",
495 |        " 'Onto later future 518 94 2803 car issue to ball. Hand particularly cost wear ever since. Both space really force start across.',\n",
496 |        " 'Hot smile should national travel. Left physical sound evening course good night. Expect no hard model page.',\n",
497 |        " 'Go drive poor make. Rock skill voice size. Reveal bring seem score able plant good.',\n",
498 |        " 'Point north modern edge candidate investment sit. Address leave able. Top value skin boy. Hear civil mother recent unit various.',\n",
499 |        " 'Face at stop social somebody job leader. Five sea find think it arm movie well. According agent national close institution ready well.',\n",
500 |        " 'Stock although strong skill budget attention charge. Hope information baby knowledge. Structure once continue why gas.',\n",
501 |        " 'At rather special entire several. Research although structure politics time. Responsibility production the but life school.',\n",
502 |        " 'Travel international thus edge wear. I American north former. Girl window ready challenge.',\n",
503 |        " 'A high money responsibility fund brother black. Add media since. Security visit team response keep north us Mrs.',\n",
504 |        " 'Although recent either skill owner property authority. Training arm Mrs where successful if. Place class good sister. Police decade hit.',\n",
505 |        " 'Thought ball east protect very middle or food.']"
506 |       ]
507 |      },
508 |      "execution_count": 5,
509 |      "metadata": {},
510 |      "output_type": "execute_result"
511 |     }
512 |    ],
513 |    "source": [
514 |     "test_text"
515 |    ]
516 |   },
517 |   {
518 |    "cell_type": "code",
519 |    "execution_count": null,
520 |    "metadata": {},
521 |    "outputs": [],
522 |    "source": []
523 |   }
524 |  ],
525 |  "metadata": {
526 |   "kernelspec": {
527 |    "display_name": "Python 3",
528 |    "language": "python",
529 |    "name": "python3"
530 |   },
531 |   "language_info": {
532 |    "codemirror_mode": {
533 |     "name": "ipython",
534 |     "version": 3
535 |    },
536 |    "file_extension": ".py",
537 |    "mimetype": "text/x-python",
538 |    "name": "python",
539 |    "nbconvert_exporter": "python",
540 |    "pygments_lexer": "ipython3",
541 |    "version": "3.6.6"
542 |   }
543 |  },
544 |  "nbformat": 4,
545 |  "nbformat_minor": 2
546 | }
547 | 


--------------------------------------------------------------------------------
/model/Doc2vec.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "markdown",
  5 |    "metadata": {},
  6 |    "source": [
  7 |     "# Doc2vec model for word embedding"
  8 |    ]
  9 |   },
 10 |   {
 11 |    "cell_type": "markdown",
 12 |    "metadata": {},
 13 |    "source": [
 14 |     "## Try out Doc2Vec model with index as Tags "
 15 |    ]
 16 |   },
 17 |   {
 18 |    "cell_type": "code",
 19 |    "execution_count": 1,
 20 |    "metadata": {},
 21 |    "outputs": [],
 22 |    "source": [
 23 |     "import pandas as pd"
 24 |    ]
 25 |   },
 26 |   {
 27 |    "cell_type": "code",
 28 |    "execution_count": 2,
 29 |    "metadata": {},
 30 |    "outputs": [],
 31 |    "source": [
 32 |     "fake_pii = pd.read_csv(\"../clean_data/Cleaned_train_text_with_pii_2018_12_29_07_26_56_266227.csv\")\n",
 33 |     "# 0 denote no pii\n",
 34 |     "# 1 denote pii exists\n",
 35 |     "def binary_pii(label):\n",
 36 |     "    pii_label = 0\n",
 37 |     "    if label != \"None\":\n",
 38 |     "        pii_label = 1\n",
 39 |     "    return pii_label "
 40 |    ]
 41 |   },
 42 |   {
 43 |    "cell_type": "code",
 44 |    "execution_count": 3,
 45 |    "metadata": {},
 46 |    "outputs": [],
 47 |    "source": [
 48 |     "fake_pii[\"Target\"] = fake_pii['Labels'].apply(binary_pii)"
 49 |    ]
 50 |   },
 51 |   {
 52 |    "cell_type": "code",
 53 |    "execution_count": 4,
 54 |    "metadata": {},
 55 |    "outputs": [
 56 |     {
 57 |      "name": "stderr",
 58 |      "output_type": "stream",
 59 |      "text": [
 60 |       "D:\\Data_science\\Python3.7\\lib\\site-packages\\gensim\\utils.py:1197: UserWarning: detected Windows; aliasing chunkize to chunkize_serial\n",
 61 |       "  warnings.warn(\"detected Windows; aliasing chunkize to chunkize_serial\")\n"
 62 |      ]
 63 |     }
 64 |    ],
 65 |    "source": [
 66 |     "from gensim.models.doc2vec import Doc2Vec,TaggedDocument"
 67 |    ]
 68 |   },
 69 |   {
 70 |    "cell_type": "code",
 71 |    "execution_count": 5,
 72 |    "metadata": {},
 73 |    "outputs": [],
 74 |    "source": [
 75 |     "documents = [TaggedDocument(sentence.split(), [index]) for index, sentence in enumerate(fake_pii['Cleaned_text'])]"
 76 |    ]
 77 |   },
 78 |   {
 79 |    "cell_type": "code",
 80 |    "execution_count": 6,
 81 |    "metadata": {},
 82 |    "outputs": [
 83 |     {
 84 |      "name": "stdout",
 85 |      "output_type": "stream",
 86 |      "text": [
 87 |       "Wall time: 1min 12s\n"
 88 |      ]
 89 |     }
 90 |    ],
 91 |    "source": [
 92 |     "%%time\n",
 93 |     "pii_doc2vec = Doc2Vec(documents, vector_size = 100, min_count = 1, epochs = 2, workers = 2 )"
 94 |    ]
 95 |   },
 96 |   {
 97 |    "cell_type": "code",
 98 |    "execution_count": 7,
 99 |    "metadata": {},
100 |    "outputs": [
101 |     {
102 |      "ename": "RuntimeError",
103 |      "evalue": "cannot sort vocabulary after model weights already initialized.",
104 |      "output_type": "error",
105 |      "traceback": [
106 |       "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
107 |       "\u001b[1;31mRuntimeError\u001b[0m                              Traceback (most recent call last)",
108 |       "\u001b[1;32m<timed exec>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n",
109 |       "\u001b[1;32mD:\\Data_science\\Python3.7\\lib\\site-packages\\gensim\\models\\doc2vec.py\u001b[0m in \u001b[0;36mbuild_vocab\u001b[1;34m(self, documents, update, progress_per, keep_raw_vocab, trim_rule, **kwargs)\u001b[0m\n\u001b[0;32m    731\u001b[0m         report_values = self.vocabulary.prepare_vocab(\n\u001b[0;32m    732\u001b[0m             \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mhs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mnegative\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mwv\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mupdate\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mupdate\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mkeep_raw_vocab\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mkeep_raw_vocab\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mtrim_rule\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mtrim_rule\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 733\u001b[1;33m             **kwargs)\n\u001b[0m\u001b[0;32m    734\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    735\u001b[0m         \u001b[0mreport_values\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;34m'memory'\u001b[0m\u001b[1;33m]\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mestimate_memory\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mvocab_size\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mreport_values\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;34m'num_retained_words'\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
110 |       "\u001b[1;32mD:\\Data_science\\Python3.7\\lib\\site-packages\\gensim\\models\\word2vec.py\u001b[0m in \u001b[0;36mprepare_vocab\u001b[1;34m(self, hs, negative, wv, update, keep_raw_vocab, trim_rule, min_count, sample, dry_run)\u001b[0m\n\u001b[0;32m   1316\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1317\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0msorted_vocab\u001b[0m \u001b[1;32mand\u001b[0m \u001b[1;32mnot\u001b[0m \u001b[0mupdate\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 1318\u001b[1;33m             \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0msort_vocab\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mwv\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   1319\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mhs\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1320\u001b[0m             \u001b[1;31m# add info about each word's Huffman encoding\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
111 |       "\u001b[1;32mD:\\Data_science\\Python3.7\\lib\\site-packages\\gensim\\models\\word2vec.py\u001b[0m in \u001b[0;36msort_vocab\u001b[1;34m(self, wv)\u001b[0m\n\u001b[0;32m   1182\u001b[0m         \u001b[1;34m\"\"\"Sort the vocabulary so the most frequent words have the lowest indexes.\"\"\"\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1183\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mlen\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mwv\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mvectors\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 1184\u001b[1;33m             \u001b[1;32mraise\u001b[0m \u001b[0mRuntimeError\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m\"cannot sort vocabulary after model weights already initialized.\"\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   1185\u001b[0m         \u001b[0mwv\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mindex2word\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0msort\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mkey\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;32mlambda\u001b[0m \u001b[0mword\u001b[0m\u001b[1;33m:\u001b[0m \u001b[0mwv\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mvocab\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mword\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mcount\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mreverse\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;32mTrue\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1186\u001b[0m         \u001b[1;32mfor\u001b[0m \u001b[0mi\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mword\u001b[0m \u001b[1;32min\u001b[0m \u001b[0menumerate\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mwv\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mindex2word\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
112 |       "\u001b[1;31mRuntimeError\u001b[0m: cannot sort vocabulary after model weights already initialized."
113 |      ]
114 |     }
115 |    ],
116 |    "source": [
117 |     "%%time\n",
118 |     "pii_doc2vec.build_vocab(documents)\n",
119 |     "pii_doc2vec.train(documents, total_examples = len(documents), epochs = 5)"
120 |    ]
121 |   },
122 |   {
123 |    "cell_type": "code",
124 |    "execution_count": 8,
125 |    "metadata": {},
126 |    "outputs": [],
127 |    "source": [
128 |     "pii_doc2vec.save(\"./doc2vec/doc2vec_100_index_tags.bin\")"
129 |    ]
130 |   },
131 |   {
132 |    "cell_type": "markdown",
133 |    "metadata": {},
134 |    "source": [
135 |     "## Try out Doc2vec model with True labels as Tags."
136 |    ]
137 |   },
138 |   {
139 |    "cell_type": "code",
140 |    "execution_count": 21,
141 |    "metadata": {},
142 |    "outputs": [],
143 |    "source": [
144 |     "paragraph = fake_pii['Cleaned_text'].copy()\n",
145 |     "\n",
146 |     "tags = fake_pii['Labels'].copy()"
147 |    ]
148 |   },
149 |   {
150 |    "cell_type": "code",
151 |    "execution_count": 25,
152 |    "metadata": {},
153 |    "outputs": [],
154 |    "source": [
155 |     "True_tag_documents = [TaggedDocument(paragraph.iloc[index].split(\" \"), [tags.iloc[index]])\\\n",
156 |     "                      for index in range(len(tags))]"
157 |    ]
158 |   },
159 |   {
160 |    "cell_type": "code",
161 |    "execution_count": 26,
162 |    "metadata": {},
163 |    "outputs": [],
164 |    "source": [
165 |     "true_tag_doc2vec = Doc2Vec(True_tag_documents,vector_size = 100, min_count = 1, epochs = 5, workers = 2 )"
166 |    ]
167 |   },
168 |   {
169 |    "cell_type": "code",
170 |    "execution_count": 27,
171 |    "metadata": {},
172 |    "outputs": [],
173 |    "source": [
174 |     "true_tag_doc2vec.save(\"./doc2vec/doc2vec_100_true_tags.bin\")"
175 |    ]
176 |   }
177 |  ],
178 |  "metadata": {
179 |   "kernelspec": {
180 |    "display_name": "Python 3",
181 |    "language": "python",
182 |    "name": "python3"
183 |   },
184 |   "language_info": {
185 |    "codemirror_mode": {
186 |     "name": "ipython",
187 |     "version": 3
188 |    },
189 |    "file_extension": ".py",
190 |    "mimetype": "text/x-python",
191 |    "name": "python",
192 |    "nbconvert_exporter": "python",
193 |    "pygments_lexer": "ipython3",
194 |    "version": "3.7.1"
195 |   }
196 |  },
197 |  "nbformat": 4,
198 |  "nbformat_minor": 2
199 | }
200 | 


--------------------------------------------------------------------------------
/model/build_vocab_Test.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "markdown",
  5 |    "metadata": {},
  6 |    "source": [
  7 |     "Testing build_vocab method in gensim for word2vec"
  8 |    ]
  9 |   },
 10 |   {
 11 |    "cell_type": "code",
 12 |    "execution_count": 6,
 13 |    "metadata": {},
 14 |    "outputs": [],
 15 |    "source": [
 16 |     "old_sentences = [[\"bad\",\"robots\"],[\"good\",\"human\"]]\n",
 17 |     "\n",
 18 |     "new_sentences = [['yes', 'this', 'is', 'the', 'word2vec', 'model']]"
 19 |    ]
 20 |   },
 21 |   {
 22 |    "cell_type": "code",
 23 |    "execution_count": 7,
 24 |    "metadata": {},
 25 |    "outputs": [],
 26 |    "source": [
 27 |     "from gensim.models import Word2Vec"
 28 |    ]
 29 |   },
 30 |   {
 31 |    "cell_type": "code",
 32 |    "execution_count": 8,
 33 |    "metadata": {},
 34 |    "outputs": [
 35 |     {
 36 |      "ename": "AttributeError",
 37 |      "evalue": "'list' object has no attribute 'vocabulary'",
 38 |      "output_type": "error",
 39 |      "traceback": [
 40 |       "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
 41 |       "\u001b[0;31mAttributeError\u001b[0m                            Traceback (most recent call last)",
 42 |       "\u001b[0;32m<ipython-input-8-0c705c226e02>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mold_model\u001b[0m \u001b[0;34m=\u001b[0m  \u001b[0mWord2Vec\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbuild_vocab\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mold_sentences\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mupdate\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      2\u001b[0m old_model.train(old_sentences,size = 10, window=5,\n\u001b[1;32m      3\u001b[0m                                   min_count = 1, workers = 2)\n",
 43 |       "\u001b[0;32m/opt/conda/lib/python3.6/site-packages/gensim/models/base_any2vec.py\u001b[0m in \u001b[0;36mbuild_vocab\u001b[0;34m(self, sentences, corpus_file, update, progress_per, keep_raw_vocab, trim_rule, **kwargs)\u001b[0m\n\u001b[1;32m    933\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    934\u001b[0m         \"\"\"\n\u001b[0;32m--> 935\u001b[0;31m         total_words, corpus_count = self.vocabulary.scan_vocab(\n\u001b[0m\u001b[1;32m    936\u001b[0m             sentences=sentences, corpus_file=corpus_file, progress_per=progress_per, trim_rule=trim_rule)\n\u001b[1;32m    937\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcorpus_count\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcorpus_count\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
 44 |       "\u001b[0;31mAttributeError\u001b[0m: 'list' object has no attribute 'vocabulary'"
 45 |      ]
 46 |     }
 47 |    ],
 48 |    "source": [
 49 |     "old_model =  Word2Vec.build_vocab(old_sentences, update = True)\n",
 50 |     "old_model.train(old_sentences,size = 10, window=5,\n",
 51 |     "                                  min_count = 1, workers = 2)"
 52 |    ]
 53 |   },
 54 |   {
 55 |    "cell_type": "code",
 56 |    "execution_count": 9,
 57 |    "metadata": {},
 58 |    "outputs": [
 59 |     {
 60 |      "ename": "NameError",
 61 |      "evalue": "name 'old_model' is not defined",
 62 |      "output_type": "error",
 63 |      "traceback": [
 64 |       "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
 65 |       "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
 66 |       "\u001b[0;32m<ipython-input-9-d6feb7be6b09>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mold_model\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwv\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvocab\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
 67 |       "\u001b[0;31mNameError\u001b[0m: name 'old_model' is not defined"
 68 |      ]
 69 |     }
 70 |    ],
 71 |    "source": [
 72 |     "old_model.wv.vocab"
 73 |    ]
 74 |   },
 75 |   {
 76 |    "cell_type": "code",
 77 |    "execution_count": null,
 78 |    "metadata": {},
 79 |    "outputs": [],
 80 |    "source": [
 81 |     "old_model.save(\"old_model\")"
 82 |    ]
 83 |   },
 84 |   {
 85 |    "cell_type": "code",
 86 |    "execution_count": null,
 87 |    "metadata": {},
 88 |    "outputs": [],
 89 |    "source": [
 90 |     "new_model = Word2Vec.load(\"old_model\")"
 91 |    ]
 92 |   },
 93 |   {
 94 |    "cell_type": "code",
 95 |    "execution_count": null,
 96 |    "metadata": {},
 97 |    "outputs": [],
 98 |    "source": [
 99 |     "new_model.build_vocab(new_sentences, update = True)"
100 |    ]
101 |   },
102 |   {
103 |    "cell_type": "code",
104 |    "execution_count": null,
105 |    "metadata": {},
106 |    "outputs": [],
107 |    "source": [
108 |     "new_model.train(new_sentences, total_examples=2, epochs = 1)"
109 |    ]
110 |   },
111 |   {
112 |    "cell_type": "code",
113 |    "execution_count": null,
114 |    "metadata": {},
115 |    "outputs": [],
116 |    "source": [
117 |     "new_model.wv.vocab"
118 |    ]
119 |   },
120 |   {
121 |    "cell_type": "code",
122 |    "execution_count": null,
123 |    "metadata": {},
124 |    "outputs": [],
125 |    "source": [
126 |     "new_new_sentences = [['my','email','address','is','salitrojan@gmailcom','111-999-7711','best','sali','trojan']]"
127 |    ]
128 |   },
129 |   {
130 |    "cell_type": "code",
131 |    "execution_count": null,
132 |    "metadata": {},
133 |    "outputs": [],
134 |    "source": [
135 |     "new_model.build_vocab(new_new_sentences, update = True)"
136 |    ]
137 |   },
138 |   {
139 |    "cell_type": "code",
140 |    "execution_count": null,
141 |    "metadata": {},
142 |    "outputs": [],
143 |    "source": [
144 |     "new_model.train(new_new_sentences, total_examples=1, epochs = 1)"
145 |    ]
146 |   },
147 |   {
148 |    "cell_type": "code",
149 |    "execution_count": null,
150 |    "metadata": {},
151 |    "outputs": [],
152 |    "source": [
153 |     "new_model.wv.vocab"
154 |    ]
155 |   },
156 |   {
157 |    "cell_type": "code",
158 |    "execution_count": null,
159 |    "metadata": {},
160 |    "outputs": [],
161 |    "source": []
162 |   }
163 |  ],
164 |  "metadata": {
165 |   "kernelspec": {
166 |    "display_name": "Python 3",
167 |    "language": "python",
168 |    "name": "python3"
169 |   },
170 |   "language_info": {
171 |    "codemirror_mode": {
172 |     "name": "ipython",
173 |     "version": 3
174 |    },
175 |    "file_extension": ".py",
176 |    "mimetype": "text/x-python",
177 |    "name": "python",
178 |    "nbconvert_exporter": "python",
179 |    "pygments_lexer": "ipython3",
180 |    "version": "3.6.6"
181 |   }
182 |  },
183 |  "nbformat": 4,
184 |  "nbformat_minor": 2
185 | }
186 | 


--------------------------------------------------------------------------------
/model/doc2vec_index_100_logistic.pkl:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/edwardcooper/data-sentry/f2a0e438101c76abaacca834e123bb3fb964f6a1/model/doc2vec_index_100_logistic.pkl


--------------------------------------------------------------------------------
/model/doc2vec_label_100_logistic.pkl:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/edwardcooper/data-sentry/f2a0e438101c76abaacca834e123bb3fb964f6a1/model/doc2vec_label_100_logistic.pkl


--------------------------------------------------------------------------------
/model/doc2vec_label_100_logistic_no_weights.pkl:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/edwardcooper/data-sentry/f2a0e438101c76abaacca834e123bb3fb964f6a1/model/doc2vec_label_100_logistic_no_weights.pkl


--------------------------------------------------------------------------------
/model/doc2vec_label_100_logistic_weights.pkl:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/edwardcooper/data-sentry/f2a0e438101c76abaacca834e123bb3fb964f6a1/model/doc2vec_label_100_logistic_weights.pkl


--------------------------------------------------------------------------------
/model/inference_time.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "code",
  5 |    "execution_count": 1,
  6 |    "metadata": {},
  7 |    "outputs": [],
  8 |    "source": [
  9 |     "import time"
 10 |    ]
 11 |   },
 12 |   {
 13 |    "cell_type": "code",
 14 |    "execution_count": 2,
 15 |    "metadata": {},
 16 |    "outputs": [],
 17 |    "source": [
 18 |     "from sklearn.externals import joblib\n",
 19 |     "from piidetect.pipeline import text_clean, word_embedding\n",
 20 |     "import pandas as pd"
 21 |    ]
 22 |   },
 23 |   {
 24 |    "cell_type": "code",
 25 |    "execution_count": 3,
 26 |    "metadata": {},
 27 |    "outputs": [],
 28 |    "source": [
 29 |     "\n",
 30 |     "time1 = time.time()"
 31 |    ]
 32 |   },
 33 |   {
 34 |    "cell_type": "code",
 35 |    "execution_count": 4,
 36 |    "metadata": {},
 37 |    "outputs": [],
 38 |    "source": [
 39 |     "\n",
 40 |     "model = joblib.load(\"word2vec_pipe_cv_production.pkl\")\n"
 41 |    ]
 42 |   },
 43 |   {
 44 |    "cell_type": "code",
 45 |    "execution_count": 5,
 46 |    "metadata": {},
 47 |    "outputs": [
 48 |     {
 49 |      "name": "stdout",
 50 |      "output_type": "stream",
 51 |      "text": [
 52 |       "33.788105726242065\n"
 53 |      ]
 54 |     }
 55 |    ],
 56 |    "source": [
 57 |     "time2 = time.time()\n",
 58 |     "\n",
 59 |     "\n",
 60 |     "print(time2 - time1)"
 61 |    ]
 62 |   },
 63 |   {
 64 |    "cell_type": "code",
 65 |    "execution_count": 35,
 66 |    "metadata": {},
 67 |    "outputs": [],
 68 |    "source": [
 69 |     "input_text = pd.Series(\"my cell number is 89 77 827 444\")"
 70 |    ]
 71 |   },
 72 |   {
 73 |    "cell_type": "code",
 74 |    "execution_count": 36,
 75 |    "metadata": {
 76 |     "scrolled": true
 77 |    },
 78 |    "outputs": [
 79 |     {
 80 |      "name": "stderr",
 81 |      "output_type": "stream",
 82 |      "text": [
 83 |       "\r",
 84 |       "  0%|          | 0/1 [00:00<?, ?it/s]"
 85 |      ]
 86 |     },
 87 |     {
 88 |      "name": "stdout",
 89 |      "output_type": "stream",
 90 |      "text": [
 91 |       "transforming while training word2vec model with new data.\n"
 92 |      ]
 93 |     },
 94 |     {
 95 |      "name": "stderr",
 96 |      "output_type": "stream",
 97 |      "text": [
 98 |       "100%|██████████| 1/1 [00:00<00:00,  1.65it/s]\n",
 99 |       "100%|██████████| 1/1 [00:00<00:00, 4686.37it/s]"
100 |      ]
101 |     },
102 |     {
103 |      "name": "stdout",
104 |      "output_type": "stream",
105 |      "text": [
106 |       "[1]\n",
107 |       "1.8178133964538574\n"
108 |      ]
109 |     },
110 |     {
111 |      "name": "stderr",
112 |      "output_type": "stream",
113 |      "text": [
114 |       "\n"
115 |      ]
116 |     }
117 |    ],
118 |    "source": [
119 |     "time1 = time.time()\n",
120 |     "print(model.predict(input_text))\n",
121 |     "time2 = time.time()\n",
122 |     "\n",
123 |     "print(time2 - time1)"
124 |    ]
125 |   }
126 |  ],
127 |  "metadata": {
128 |   "kernelspec": {
129 |    "display_name": "Python 3",
130 |    "language": "python",
131 |    "name": "python3"
132 |   },
133 |   "language_info": {
134 |    "codemirror_mode": {
135 |     "name": "ipython",
136 |     "version": 3
137 |    },
138 |    "file_extension": ".py",
139 |    "mimetype": "text/x-python",
140 |    "name": "python",
141 |    "nbconvert_exporter": "python",
142 |    "pygments_lexer": "ipython3",
143 |    "version": "3.6.6"
144 |   }
145 |  },
146 |  "nbformat": 4,
147 |  "nbformat_minor": 2
148 | }
149 | 


--------------------------------------------------------------------------------
/model/model_test.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "code",
  5 |    "execution_count": 1,
  6 |    "metadata": {},
  7 |    "outputs": [],
  8 |    "source": [
  9 |     "from sklearn.externals import joblib\n",
 10 |     "\n",
 11 |     "word2vec_300_logistic = joblib.load(\"word2vec_300_logistic.pkl\")"
 12 |    ]
 13 |   },
 14 |   {
 15 |    "cell_type": "code",
 16 |    "execution_count": 2,
 17 |    "metadata": {},
 18 |    "outputs": [],
 19 |    "source": [
 20 |     "import pandas as pd\n",
 21 |     "fake_pii = pd.read_csv(\"../clean_data/Cleaned_train_text_with_pii_2018_12_29_07_26_56_266227.csv\")"
 22 |    ]
 23 |   },
 24 |   {
 25 |    "cell_type": "code",
 26 |    "execution_count": 3,
 27 |    "metadata": {},
 28 |    "outputs": [],
 29 |    "source": [
 30 |     "from gensim.models import Word2Vec\n",
 31 |     "\n",
 32 |     "word2vec_300 = Word2Vec.load(\"./word2vec/word2vec_cleaned_300_.bin\")"
 33 |    ]
 34 |   },
 35 |   {
 36 |    "cell_type": "code",
 37 |    "execution_count": 4,
 38 |    "metadata": {},
 39 |    "outputs": [],
 40 |    "source": [
 41 |     "from word2vec_visualizer import find_part_pii, get_pii2vec_matrix, color_dict"
 42 |    ]
 43 |   },
 44 |   {
 45 |    "cell_type": "code",
 46 |    "execution_count": 5,
 47 |    "metadata": {},
 48 |    "outputs": [
 49 |     {
 50 |      "name": "stderr",
 51 |      "output_type": "stream",
 52 |      "text": [
 53 |       "100%|██████████| 800000/800000 [01:17<00:00, 10305.58it/s]\n"
 54 |      ]
 55 |     }
 56 |    ],
 57 |    "source": [
 58 |     "from tqdm import tqdm\n",
 59 |     "extracted_pii_list = [find_part_pii(text, model = word2vec_300)\\\n",
 60 |     "                      for text in tqdm(fake_pii[\"Cleaned_text\"]) ]"
 61 |    ]
 62 |   },
 63 |   {
 64 |    "cell_type": "code",
 65 |    "execution_count": 6,
 66 |    "metadata": {
 67 |     "scrolled": true
 68 |    },
 69 |    "outputs": [
 70 |     {
 71 |      "name": "stderr",
 72 |      "output_type": "stream",
 73 |      "text": [
 74 |       "100%|██████████| 800000/800000 [00:02<00:00, 293951.14it/s]\n"
 75 |      ]
 76 |     }
 77 |    ],
 78 |    "source": [
 79 |     "pii_matrix = get_pii2vec_matrix(pii_list = extracted_pii_list, model = word2vec_300)"
 80 |    ]
 81 |   },
 82 |   {
 83 |    "cell_type": "code",
 84 |    "execution_count": 9,
 85 |    "metadata": {},
 86 |    "outputs": [],
 87 |    "source": [
 88 |     "y_pred = word2vec_300_logistic.predict(pii_matrix)\n",
 89 |     "y_true = fake_pii[\"Target\"]"
 90 |    ]
 91 |   },
 92 |   {
 93 |    "cell_type": "code",
 94 |    "execution_count": 12,
 95 |    "metadata": {
 96 |     "scrolled": true
 97 |    },
 98 |    "outputs": [
 99 |     {
100 |      "name": "stdout",
101 |      "output_type": "stream",
102 |      "text": [
103 |       "CPU times: user 708 ms, sys: 31.9 ms, total: 740 ms\n",
104 |       "Wall time: 704 ms\n"
105 |      ]
106 |     },
107 |     {
108 |      "data": {
109 |       "image/png": "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\n",
110 |       "text/plain": [
111 |        "<Figure size 432x288 with 2 Axes>"
112 |       ]
113 |      },
114 |      "metadata": {
115 |       "needs_background": "light"
116 |      },
117 |      "output_type": "display_data"
118 |     }
119 |    ],
120 |    "source": [
121 |     "%%time\n",
122 |     "from sklearn.metrics import confusion_matrix\n",
123 |     "\n",
124 |     "import seaborn as sns\n",
125 |     "import matplotlib.pyplot as plt\n",
126 |     "sns.heatmap(confusion_matrix(y_true = y_true, y_pred = y_pred), annot = True,fmt=\"d\",cmap=\"YlGnBu\")\n",
127 |     "plt.xlabel(\"Predicted values\")\n",
128 |     "plt.ylabel(\"True values\")"
129 |    ]
130 |   },
131 |   {
132 |    "cell_type": "code",
133 |    "execution_count": 13,
134 |    "metadata": {},
135 |    "outputs": [
136 |     {
137 |      "data": {
138 |       "text/plain": [
139 |        "array([[ 91865,   8135],\n",
140 |        "       [ 15417, 684583]])"
141 |       ]
142 |      },
143 |      "execution_count": 13,
144 |      "metadata": {},
145 |      "output_type": "execute_result"
146 |     }
147 |    ],
148 |    "source": [
149 |     "confusion_matrix(y_true = y_true, y_pred = y_pred)"
150 |    ]
151 |   }
152 |  ],
153 |  "metadata": {
154 |   "kernelspec": {
155 |    "display_name": "Python 3",
156 |    "language": "python",
157 |    "name": "python3"
158 |   },
159 |   "language_info": {
160 |    "codemirror_mode": {
161 |     "name": "ipython",
162 |     "version": 3
163 |    },
164 |    "file_extension": ".py",
165 |    "mimetype": "text/x-python",
166 |    "name": "python",
167 |    "nbconvert_exporter": "python",
168 |    "pygments_lexer": "ipython3",
169 |    "version": "3.6.6"
170 |   }
171 |  },
172 |  "nbformat": 4,
173 |  "nbformat_minor": 2
174 | }
175 | 


--------------------------------------------------------------------------------
/model/multiple_test_data_performance.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "markdown",
  5 |    "metadata": {},
  6 |    "source": [
  7 |     "The goal of this notebook is to test the performance of the pipeline on mutiple test data."
  8 |    ]
  9 |   },
 10 |   {
 11 |    "cell_type": "code",
 12 |    "execution_count": 1,
 13 |    "metadata": {},
 14 |    "outputs": [],
 15 |    "source": [
 16 |     "import pandas as pd"
 17 |    ]
 18 |   },
 19 |   {
 20 |    "cell_type": "code",
 21 |    "execution_count": 2,
 22 |    "metadata": {},
 23 |    "outputs": [],
 24 |    "source": [
 25 |     "test_data_1 = pd.read_csv(\"../clean_data/Cleaned_test_text_with_pii_2018_12_31_05_35_46_815414.csv\")\n",
 26 |     "test_data_2 = pd.read_csv(\"../clean_data/Cleaned_test_text_with_pii_2019_01_18_06_21_07_588706.csv\")\n",
 27 |     "test_data_3 = pd.read_csv(\"../clean_data/Cleaned_test_text_with_pii_2019_01_18_06_39_35_811991.csv\")"
 28 |    ]
 29 |   },
 30 |   {
 31 |    "cell_type": "code",
 32 |    "execution_count": 3,
 33 |    "metadata": {},
 34 |    "outputs": [],
 35 |    "source": [
 36 |     "from w2vpipe.pipeline import text_clean, word_embedding"
 37 |    ]
 38 |   },
 39 |   {
 40 |    "cell_type": "code",
 41 |    "execution_count": 4,
 42 |    "metadata": {},
 43 |    "outputs": [],
 44 |    "source": [
 45 |     "from sklearn.externals import joblib"
 46 |    ]
 47 |   },
 48 |   {
 49 |    "cell_type": "code",
 50 |    "execution_count": 5,
 51 |    "metadata": {},
 52 |    "outputs": [],
 53 |    "source": [
 54 |     "word2vec_pipe = joblib.load( 'word2vec_pipe_cv_production.pkl')"
 55 |    ]
 56 |   },
 57 |   {
 58 |    "cell_type": "code",
 59 |    "execution_count": 6,
 60 |    "metadata": {},
 61 |    "outputs": [
 62 |     {
 63 |      "name": "stderr",
 64 |      "output_type": "stream",
 65 |      "text": [
 66 |       "\r",
 67 |       "  0%|          | 0/80000 [00:00<?, ?it/s]"
 68 |      ]
 69 |     },
 70 |     {
 71 |      "name": "stdout",
 72 |      "output_type": "stream",
 73 |      "text": [
 74 |       "transforming while training word2vec model with new data.\n"
 75 |      ]
 76 |     },
 77 |     {
 78 |      "name": "stderr",
 79 |      "output_type": "stream",
 80 |      "text": [
 81 |       "100%|██████████| 80000/80000 [00:08<00:00, 8987.96it/s] \n",
 82 |       "100%|██████████| 80000/80000 [00:00<00:00, 221243.95it/s]\n"
 83 |      ]
 84 |     },
 85 |     {
 86 |      "name": "stdout",
 87 |      "output_type": "stream",
 88 |      "text": [
 89 |       "              precision    recall  f1-score   support\n",
 90 |       "\n",
 91 |       "           0       0.95      0.95      0.95     10000\n",
 92 |       "           1       0.99      0.99      0.99     70000\n",
 93 |       "\n",
 94 |       "   micro avg       0.99      0.99      0.99     80000\n",
 95 |       "   macro avg       0.97      0.97      0.97     80000\n",
 96 |       "weighted avg       0.99      0.99      0.99     80000\n",
 97 |       "\n",
 98 |       "CPU times: user 31.7 s, sys: 8.37 s, total: 40 s\n",
 99 |       "Wall time: 20.7 s\n"
100 |      ]
101 |     }
102 |    ],
103 |    "source": [
104 |     "%%time \n",
105 |     "binary_pred = word2vec_pipe.predict(test_data_1[\"Text\"])\n",
106 |     "binary_true = test_data_1[\"Target\"]\n",
107 |     "\n",
108 |     "from sklearn.metrics import classification_report\n",
109 |     "\n",
110 |     "print(classification_report(y_true = binary_true, y_pred = binary_pred))"
111 |    ]
112 |   },
113 |   {
114 |    "cell_type": "code",
115 |    "execution_count": 7,
116 |    "metadata": {},
117 |    "outputs": [
118 |     {
119 |      "name": "stderr",
120 |      "output_type": "stream",
121 |      "text": [
122 |       "\r",
123 |       "  0%|          | 0/80000 [00:00<?, ?it/s]"
124 |      ]
125 |     },
126 |     {
127 |      "name": "stdout",
128 |      "output_type": "stream",
129 |      "text": [
130 |       "transforming while training word2vec model with new data.\n"
131 |      ]
132 |     },
133 |     {
134 |      "name": "stderr",
135 |      "output_type": "stream",
136 |      "text": [
137 |       "100%|██████████| 80000/80000 [00:09<00:00, 8538.03it/s] \n",
138 |       "100%|██████████| 80000/80000 [00:00<00:00, 218832.87it/s]\n"
139 |      ]
140 |     },
141 |     {
142 |      "name": "stdout",
143 |      "output_type": "stream",
144 |      "text": [
145 |       "              precision    recall  f1-score   support\n",
146 |       "\n",
147 |       "           0       0.97      0.92      0.95     10000\n",
148 |       "           1       0.99      1.00      0.99     70000\n",
149 |       "\n",
150 |       "   micro avg       0.99      0.99      0.99     80000\n",
151 |       "   macro avg       0.98      0.96      0.97     80000\n",
152 |       "weighted avg       0.99      0.99      0.99     80000\n",
153 |       "\n",
154 |       "CPU times: user 32.8 s, sys: 8.53 s, total: 41.4 s\n",
155 |       "Wall time: 21.8 s\n"
156 |      ]
157 |     }
158 |    ],
159 |    "source": [
160 |     "%%time \n",
161 |     "binary_pred = word2vec_pipe.predict(test_data_2[\"Text\"])\n",
162 |     "binary_true = test_data_2[\"Target\"]\n",
163 |     "\n",
164 |     "from sklearn.metrics import classification_report\n",
165 |     "\n",
166 |     "print(classification_report(y_true = binary_true, y_pred = binary_pred))"
167 |    ]
168 |   },
169 |   {
170 |    "cell_type": "code",
171 |    "execution_count": 8,
172 |    "metadata": {},
173 |    "outputs": [
174 |     {
175 |      "name": "stderr",
176 |      "output_type": "stream",
177 |      "text": [
178 |       "\r",
179 |       "  0%|          | 0/80000 [00:00<?, ?it/s]"
180 |      ]
181 |     },
182 |     {
183 |      "name": "stdout",
184 |      "output_type": "stream",
185 |      "text": [
186 |       "transforming while training word2vec model with new data.\n"
187 |      ]
188 |     },
189 |     {
190 |      "name": "stderr",
191 |      "output_type": "stream",
192 |      "text": [
193 |       "100%|██████████| 80000/80000 [00:09<00:00, 8801.63it/s] \n",
194 |       "100%|██████████| 80000/80000 [00:00<00:00, 222339.35it/s]\n"
195 |      ]
196 |     },
197 |     {
198 |      "name": "stdout",
199 |      "output_type": "stream",
200 |      "text": [
201 |       "              precision    recall  f1-score   support\n",
202 |       "\n",
203 |       "           0       0.97      0.93      0.95     10000\n",
204 |       "           1       0.99      1.00      0.99     70000\n",
205 |       "\n",
206 |       "   micro avg       0.99      0.99      0.99     80000\n",
207 |       "   macro avg       0.98      0.96      0.97     80000\n",
208 |       "weighted avg       0.99      0.99      0.99     80000\n",
209 |       "\n",
210 |       "CPU times: user 33.1 s, sys: 9.38 s, total: 42.5 s\n",
211 |       "Wall time: 21.7 s\n"
212 |      ]
213 |     }
214 |    ],
215 |    "source": [
216 |     "%%time \n",
217 |     "binary_pred = word2vec_pipe.predict(test_data_3[\"Text\"])\n",
218 |     "binary_true = test_data_3[\"Target\"]\n",
219 |     "\n",
220 |     "from sklearn.metrics import classification_report\n",
221 |     "\n",
222 |     "print(classification_report(y_true = binary_true, y_pred = binary_pred))"
223 |    ]
224 |   },
225 |   {
226 |    "cell_type": "markdown",
227 |    "metadata": {},
228 |    "source": [
229 |     "The next step is to make sure that the PII in test data does not appear too often in train data."
230 |    ]
231 |   },
232 |   {
233 |    "cell_type": "code",
234 |    "execution_count": 9,
235 |    "metadata": {},
236 |    "outputs": [],
237 |    "source": [
238 |     "train_data = pd.read_csv(\"../clean_data/Cleaned_train_text_with_pii_2018_12_29_07_26_56_266227.csv\")"
239 |    ]
240 |   },
241 |   {
242 |    "cell_type": "code",
243 |    "execution_count": 45,
244 |    "metadata": {},
245 |    "outputs": [],
246 |    "source": [
247 |     "# a function to calculate the PII overlap in training and test data\n",
248 |     "# while excluding the 'None'. \n",
249 |     "def pii_overlap(test, train):\n",
250 |     "    \n",
251 |     "    overlap_pii_binary = test.isin(train)\n",
252 |     "    total_overlap = sum(overlap_pii_binary)\n",
253 |     "    \n",
254 |     "    none_type_num = sum(test == \"None\")\n",
255 |     "    \n",
256 |     "    pii_overlap_num = total_overlap - none_type_num\n",
257 |     "    \n",
258 |     "    return pii_overlap_num"
259 |    ]
260 |   },
261 |   {
262 |    "cell_type": "code",
263 |    "execution_count": 46,
264 |    "metadata": {
265 |     "scrolled": true
266 |    },
267 |    "outputs": [
268 |     {
269 |      "data": {
270 |       "text/plain": [
271 |        "10992"
272 |       ]
273 |      },
274 |      "execution_count": 46,
275 |      "metadata": {},
276 |      "output_type": "execute_result"
277 |     }
278 |    ],
279 |    "source": [
280 |     "pii_overlap(test_data_1['PII'], train_data['PII'])"
281 |    ]
282 |   },
283 |   {
284 |    "cell_type": "code",
285 |    "execution_count": 47,
286 |    "metadata": {},
287 |    "outputs": [
288 |     {
289 |      "data": {
290 |       "text/plain": [
291 |        "11056"
292 |       ]
293 |      },
294 |      "execution_count": 47,
295 |      "metadata": {},
296 |      "output_type": "execute_result"
297 |     }
298 |    ],
299 |    "source": [
300 |     "pii_overlap(test_data_2['PII'], train_data['PII'])"
301 |    ]
302 |   },
303 |   {
304 |    "cell_type": "code",
305 |    "execution_count": 48,
306 |    "metadata": {},
307 |    "outputs": [
308 |     {
309 |      "data": {
310 |       "text/plain": [
311 |        "11097"
312 |       ]
313 |      },
314 |      "execution_count": 48,
315 |      "metadata": {},
316 |      "output_type": "execute_result"
317 |     }
318 |    ],
319 |    "source": [
320 |     "pii_overlap(test_data_3['PII'], train_data['PII'])"
321 |    ]
322 |   },
323 |   {
324 |    "cell_type": "code",
325 |    "execution_count": 49,
326 |    "metadata": {},
327 |    "outputs": [
328 |     {
329 |      "data": {
330 |       "text/plain": [
331 |        "80000"
332 |       ]
333 |      },
334 |      "execution_count": 49,
335 |      "metadata": {},
336 |      "output_type": "execute_result"
337 |     }
338 |    ],
339 |    "source": [
340 |     "test_data_3.shape[0]"
341 |    ]
342 |   },
343 |   {
344 |    "cell_type": "code",
345 |    "execution_count": 60,
346 |    "metadata": {},
347 |    "outputs": [],
348 |    "source": [
349 |     "# a function to get all the overlap values\n",
350 |     "def pii_overlap_value(test, train):\n",
351 |     "    \n",
352 |     "    overlap_pii_binary = test.isin(train)\n",
353 |     "    none_type_binary = test != \"None\"\n",
354 |     "    \n",
355 |     "    combined_binary = none_type_binary & overlap_pii_binary\n",
356 |     "    \n",
357 |     "    return test.loc[combined_binary]"
358 |    ]
359 |   },
360 |   {
361 |    "cell_type": "code",
362 |    "execution_count": 67,
363 |    "metadata": {
364 |     "scrolled": false
365 |    },
366 |    "outputs": [
367 |     {
368 |      "data": {
369 |       "text/plain": [
370 |        "Michael        72\n",
371 |        "Smith          63\n",
372 |        "Thomas         62\n",
373 |        "Williams       61\n",
374 |        "John           55\n",
375 |        "David          54\n",
376 |        "Joseph         47\n",
377 |        "James          46\n",
378 |        "Christopher    44\n",
379 |        "Matthew        43\n",
380 |        "Stephanie      42\n",
381 |        "Jennifer       41\n",
382 |        "Rodriguez      41\n",
383 |        "Wilson         40\n",
384 |        "Mark           38\n",
385 |        "Name: PII, dtype: int64"
386 |       ]
387 |      },
388 |      "execution_count": 67,
389 |      "metadata": {},
390 |      "output_type": "execute_result"
391 |     }
392 |    ],
393 |    "source": [
394 |     "pii_overlap_value(test_data_3['PII'], train_data['PII']).value_counts().head(15)"
395 |    ]
396 |   },
397 |   {
398 |    "cell_type": "markdown",
399 |    "metadata": {},
400 |    "source": [
401 |     "The majority of the overlap in PII value originates from the name, which is reasonable since name does not have as much variation as other PIIs. "
402 |    ]
403 |   },
404 |   {
405 |    "cell_type": "markdown",
406 |    "metadata": {},
407 |    "source": [
408 |     "The total PII overlap between training data and test data is about 1/7 of the test data. "
409 |    ]
410 |   },
411 |   {
412 |    "cell_type": "markdown",
413 |    "metadata": {},
414 |    "source": [
415 |     "## In conclusion, the entire pipeline does works on unseen PII data."
416 |    ]
417 |   }
418 |  ],
419 |  "metadata": {
420 |   "kernelspec": {
421 |    "display_name": "Python 3",
422 |    "language": "python",
423 |    "name": "python3"
424 |   },
425 |   "language_info": {
426 |    "codemirror_mode": {
427 |     "name": "ipython",
428 |     "version": 3
429 |    },
430 |    "file_extension": ".py",
431 |    "mimetype": "text/x-python",
432 |    "name": "python",
433 |    "nbconvert_exporter": "python",
434 |    "pygments_lexer": "ipython3",
435 |    "version": "3.6.6"
436 |   }
437 |  },
438 |  "nbformat": 4,
439 |  "nbformat_minor": 2
440 | }
441 | 


--------------------------------------------------------------------------------
/model/production_model_train.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "markdown",
  5 |    "metadata": {},
  6 |    "source": [
  7 |     "The goal of this notebook is to retrain the pipeline using the best parameters found \n",
  8 |     "to reduce the model size of 1.2G. "
  9 |    ]
 10 |   },
 11 |   {
 12 |    "cell_type": "code",
 13 |    "execution_count": 1,
 14 |    "metadata": {},
 15 |    "outputs": [],
 16 |    "source": [
 17 |     "from w2vpipe.pipeline import word_embedding, text_clean\n",
 18 |     "\n",
 19 |     "from sklearn.pipeline import Pipeline\n",
 20 |     "from sklearn.linear_model import LogisticRegression\n",
 21 |     "\n",
 22 |     "\n",
 23 |     "from sklearn.externals import joblib"
 24 |    ]
 25 |   },
 26 |   {
 27 |    "cell_type": "code",
 28 |    "execution_count": 2,
 29 |    "metadata": {},
 30 |    "outputs": [],
 31 |    "source": [
 32 |     "logit_clf_word2vec = LogisticRegression(solver = \"lbfgs\", max_iter = 10000, C = 4.032\\\n",
 33 |     "                                        , class_weight = {0: 0.8, 1: 0.2})\n",
 34 |     "\n",
 35 |     "word2vec_pipe = Pipeline([('text_cleaning', text_clean()),\n",
 36 |     "                 (\"word_embedding\", word_embedding(algo_name = \"word2vec\", workers =2, size = 300)),\n",
 37 |     "                 (\"logit_clf_word2vec\",logit_clf_word2vec)\n",
 38 |     "                ])"
 39 |    ]
 40 |   },
 41 |   {
 42 |    "cell_type": "code",
 43 |    "execution_count": 3,
 44 |    "metadata": {},
 45 |    "outputs": [],
 46 |    "source": [
 47 |     "import pandas as pd\n",
 48 |     "\n",
 49 |     "train_data = pd.read_csv(\"../clean_data/Cleaned_train_text_with_pii_2018_12_29_07_26_56_266227.csv\")"
 50 |    ]
 51 |   },
 52 |   {
 53 |    "cell_type": "code",
 54 |    "execution_count": 4,
 55 |    "metadata": {},
 56 |    "outputs": [
 57 |     {
 58 |      "name": "stdout",
 59 |      "output_type": "stream",
 60 |      "text": [
 61 |       "Building new vocabulary and training the word2vec model\n"
 62 |      ]
 63 |     },
 64 |     {
 65 |      "name": "stderr",
 66 |      "output_type": "stream",
 67 |      "text": [
 68 |       "\r",
 69 |       "  0%|          | 0/800000 [00:00<?, ?it/s]"
 70 |      ]
 71 |     },
 72 |     {
 73 |      "name": "stdout",
 74 |      "output_type": "stream",
 75 |      "text": [
 76 |       "transforming while training word2vec model with new data.\n"
 77 |      ]
 78 |     },
 79 |     {
 80 |      "name": "stderr",
 81 |      "output_type": "stream",
 82 |      "text": [
 83 |       "100%|██████████| 800000/800000 [01:27<00:00, 9183.85it/s] \n",
 84 |       "100%|██████████| 800000/800000 [00:03<00:00, 222464.63it/s]\n"
 85 |      ]
 86 |     },
 87 |     {
 88 |      "name": "stdout",
 89 |      "output_type": "stream",
 90 |      "text": [
 91 |       "CPU times: user 10min 26s, sys: 1min 39s, total: 12min 6s\n",
 92 |       "Wall time: 5min 34s\n"
 93 |      ]
 94 |     },
 95 |     {
 96 |      "data": {
 97 |       "text/plain": [
 98 |        "Pipeline(memory=None,\n",
 99 |        "     steps=[('text_cleaning', <w2vpipe.pipeline.text_clean object at 0x7f0f8f632a90>), ('word_embedding', word_embedding(algo_name='word2vec', continue_train_pre_train=True,\n",
100 |        "        dump_file=False, epochs=5, min_count=1, pre_train=None,\n",
101 |        "        re_train_new_sentences=True, size=300, window=5, workers=2)...enalty='l2', random_state=None,\n",
102 |        "          solver='lbfgs', tol=0.0001, verbose=0, warm_start=False))])"
103 |       ]
104 |      },
105 |      "execution_count": 4,
106 |      "metadata": {},
107 |      "output_type": "execute_result"
108 |     }
109 |    ],
110 |    "source": [
111 |     "%%time\n",
112 |     "word2vec_pipe.fit(train_data['Text'], train_data['Target'])"
113 |    ]
114 |   },
115 |   {
116 |    "cell_type": "markdown",
117 |    "metadata": {},
118 |    "source": [
119 |     "The model is about 1.1GB since I choose to use a vector length of 300 to embed the words. "
120 |    ]
121 |   },
122 |   {
123 |    "cell_type": "code",
124 |    "execution_count": 6,
125 |    "metadata": {
126 |     "scrolled": true
127 |    },
128 |    "outputs": [
129 |     {
130 |      "data": {
131 |       "text/plain": [
132 |        "['word2vec_pipe_production.pkl']"
133 |       ]
134 |      },
135 |      "execution_count": 6,
136 |      "metadata": {},
137 |      "output_type": "execute_result"
138 |     }
139 |    ],
140 |    "source": [
141 |     "joblib.dump(word2vec_pipe, 'word2vec_pipe_production.pkl', compress = 1)"
142 |    ]
143 |   },
144 |   {
145 |    "cell_type": "markdown",
146 |    "metadata": {},
147 |    "source": [
148 |     "## Balance the model size and model performance in the future by changing the size of vector embedding."
149 |    ]
150 |   },
151 |   {
152 |    "cell_type": "markdown",
153 |    "metadata": {},
154 |    "source": [
155 |     "See if the performance is still the same"
156 |    ]
157 |   },
158 |   {
159 |    "cell_type": "code",
160 |    "execution_count": 7,
161 |    "metadata": {},
162 |    "outputs": [],
163 |    "source": [
164 |     "test_data = pd.read_csv(\"../clean_data/Cleaned_test_text_with_pii_\\\n",
165 |     "2018_12_31_05_35_46_815414.csv\")"
166 |    ]
167 |   },
168 |   {
169 |    "cell_type": "code",
170 |    "execution_count": 8,
171 |    "metadata": {},
172 |    "outputs": [
173 |     {
174 |      "name": "stderr",
175 |      "output_type": "stream",
176 |      "text": [
177 |       "\r",
178 |       "  0%|          | 0/80000 [00:00<?, ?it/s]"
179 |      ]
180 |     },
181 |     {
182 |      "name": "stdout",
183 |      "output_type": "stream",
184 |      "text": [
185 |       "transforming while training word2vec model with new data.\n"
186 |      ]
187 |     },
188 |     {
189 |      "name": "stderr",
190 |      "output_type": "stream",
191 |      "text": [
192 |       "100%|██████████| 80000/80000 [00:09<00:00, 8524.90it/s] \n",
193 |       "100%|██████████| 80000/80000 [00:00<00:00, 217229.72it/s]\n"
194 |      ]
195 |     },
196 |     {
197 |      "name": "stdout",
198 |      "output_type": "stream",
199 |      "text": [
200 |       "              precision    recall  f1-score   support\n",
201 |       "\n",
202 |       "           0       0.87      0.97      0.92     10000\n",
203 |       "           1       1.00      0.98      0.99     70000\n",
204 |       "\n",
205 |       "   micro avg       0.98      0.98      0.98     80000\n",
206 |       "   macro avg       0.93      0.97      0.95     80000\n",
207 |       "weighted avg       0.98      0.98      0.98     80000\n",
208 |       "\n",
209 |       "CPU times: user 32.7 s, sys: 8.23 s, total: 41 s\n",
210 |       "Wall time: 21.7 s\n"
211 |      ]
212 |     }
213 |    ],
214 |    "source": [
215 |     "%%time \n",
216 |     "binary_pred = word2vec_pipe.predict(test_data[\"Text\"])\n",
217 |     "binary_true = test_data[\"Target\"]\n",
218 |     "\n",
219 |     "from sklearn.metrics import classification_report\n",
220 |     "\n",
221 |     "print(classification_report(y_true = binary_true, y_pred = binary_pred))"
222 |    ]
223 |   }
224 |  ],
225 |  "metadata": {
226 |   "kernelspec": {
227 |    "display_name": "Python 3",
228 |    "language": "python",
229 |    "name": "python3"
230 |   },
231 |   "language_info": {
232 |    "codemirror_mode": {
233 |     "name": "ipython",
234 |     "version": 3
235 |    },
236 |    "file_extension": ".py",
237 |    "mimetype": "text/x-python",
238 |    "name": "python",
239 |    "nbconvert_exporter": "python",
240 |    "pygments_lexer": "ipython3",
241 |    "version": "3.6.6"
242 |   }
243 |  },
244 |  "nbformat": 4,
245 |  "nbformat_minor": 2
246 | }
247 | 


--------------------------------------------------------------------------------
/model/transfer_learning_word2vec.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "markdown",
  5 |    "metadata": {},
  6 |    "source": [
  7 |     "Transfer learning on Google's pre-trained model"
  8 |    ]
  9 |   },
 10 |   {
 11 |    "cell_type": "code",
 12 |    "execution_count": 1,
 13 |    "metadata": {},
 14 |    "outputs": [
 15 |     {
 16 |      "name": "stderr",
 17 |      "output_type": "stream",
 18 |      "text": [
 19 |       "D:\\Data_science\\Python3.7\\lib\\site-packages\\gensim\\utils.py:1197: UserWarning: detected Windows; aliasing chunkize to chunkize_serial\n",
 20 |       "  warnings.warn(\"detected Windows; aliasing chunkize to chunkize_serial\")\n"
 21 |      ]
 22 |     }
 23 |    ],
 24 |    "source": [
 25 |     "from gensim.models import Word2Vec"
 26 |    ]
 27 |   },
 28 |   {
 29 |    "cell_type": "code",
 30 |    "execution_count": 2,
 31 |    "metadata": {},
 32 |    "outputs": [],
 33 |    "source": [
 34 |     "sentences = [[\"bad\",\"robots\"],[\"good\",\"human\"],['yes', 'this', 'is', 'the', 'word2vec', 'model']]"
 35 |    ]
 36 |   },
 37 |   {
 38 |    "cell_type": "code",
 39 |    "execution_count": 3,
 40 |    "metadata": {},
 41 |    "outputs": [],
 42 |    "source": [
 43 |     "word2vec_model = Word2Vec(size = 300, window=5,\n",
 44 |     "                                  min_count = 1, workers = 2)"
 45 |    ]
 46 |   },
 47 |   {
 48 |    "cell_type": "code",
 49 |    "execution_count": 4,
 50 |    "metadata": {},
 51 |    "outputs": [],
 52 |    "source": [
 53 |     "word2vec_model.build_vocab(sentences)\n"
 54 |    ]
 55 |   },
 56 |   {
 57 |    "cell_type": "code",
 58 |    "execution_count": 5,
 59 |    "metadata": {},
 60 |    "outputs": [],
 61 |    "source": [
 62 |     "word2vec_model.intersect_word2vec_format('./word2vec/GoogleNews-vectors-negative300.bin',\\\n",
 63 |     "                                  lockf=1.0, binary=True)"
 64 |    ]
 65 |   },
 66 |   {
 67 |    "cell_type": "code",
 68 |    "execution_count": 7,
 69 |    "metadata": {},
 70 |    "outputs": [
 71 |     {
 72 |      "data": {
 73 |       "text/plain": [
 74 |        "(4, 50)"
 75 |       ]
 76 |      },
 77 |      "execution_count": 7,
 78 |      "metadata": {},
 79 |      "output_type": "execute_result"
 80 |     }
 81 |    ],
 82 |    "source": [
 83 |     "word2vec_model.train(sentences, total_examples=3, epochs = 5)"
 84 |    ]
 85 |   },
 86 |   {
 87 |    "cell_type": "code",
 88 |    "execution_count": 8,
 89 |    "metadata": {},
 90 |    "outputs": [
 91 |     {
 92 |      "data": {
 93 |       "text/plain": [
 94 |        "{'bad': <gensim.models.keyedvectors.Vocab at 0x21247af1160>,\n",
 95 |        " 'robots': <gensim.models.keyedvectors.Vocab at 0x21247af15f8>,\n",
 96 |        " 'good': <gensim.models.keyedvectors.Vocab at 0x21247af1630>,\n",
 97 |        " 'human': <gensim.models.keyedvectors.Vocab at 0x21247af1668>,\n",
 98 |        " 'yes': <gensim.models.keyedvectors.Vocab at 0x21247af16a0>,\n",
 99 |        " 'this': <gensim.models.keyedvectors.Vocab at 0x21247af16d8>,\n",
100 |        " 'is': <gensim.models.keyedvectors.Vocab at 0x21247af1710>,\n",
101 |        " 'the': <gensim.models.keyedvectors.Vocab at 0x21247af1748>,\n",
102 |        " 'word2vec': <gensim.models.keyedvectors.Vocab at 0x21247af1780>,\n",
103 |        " 'model': <gensim.models.keyedvectors.Vocab at 0x21247af17b8>}"
104 |       ]
105 |      },
106 |      "execution_count": 8,
107 |      "metadata": {},
108 |      "output_type": "execute_result"
109 |     }
110 |    ],
111 |    "source": [
112 |     "word2vec_model.wv.vocab"
113 |    ]
114 |   },
115 |   {
116 |    "cell_type": "code",
117 |    "execution_count": null,
118 |    "metadata": {},
119 |    "outputs": [],
120 |    "source": []
121 |   }
122 |  ],
123 |  "metadata": {
124 |   "kernelspec": {
125 |    "display_name": "Python 3",
126 |    "language": "python",
127 |    "name": "python3"
128 |   },
129 |   "language_info": {
130 |    "codemirror_mode": {
131 |     "name": "ipython",
132 |     "version": 3
133 |    },
134 |    "file_extension": ".py",
135 |    "mimetype": "text/x-python",
136 |    "name": "python",
137 |    "nbconvert_exporter": "python",
138 |    "pygments_lexer": "ipython3",
139 |    "version": "3.7.1"
140 |   }
141 |  },
142 |  "nbformat": 4,
143 |  "nbformat_minor": 2
144 | }
145 | 


--------------------------------------------------------------------------------
/model/visualizer.py:
--------------------------------------------------------------------------------
 1 | 
 2 | import re
 3 | from tqdm import tqdm
 4 | import numpy as np
 5 | 
 6 | def find_part_pii(text, model, sep = " "):
 7 |     tokenized_text = text.split(sep)
 8 |     
 9 |     part_pii = model.wv.doesnt_match(tokenized_text)
10 |     
11 |     return part_pii    
12 | 
13 | 
14 | 
15 | def get_word2vec_matrix(pii_list, model):
16 |     # set the matrix dimensions
17 |     column_num = model.trainables.layer1_size
18 |     row_num = len(pii_list)
19 |     # initialized the matrix
20 |     pii2vec_mat = np.zeros((row_num, column_num))
21 |     # iterate through the pii_list and assign the vectors to matrix.
22 |     for index, ith_pii in enumerate(tqdm(pii_list)):
23 |         pii2vec_mat[index,:] = model.wv[ith_pii]
24 |     
25 |     return pii2vec_mat
26 | 
27 | 
28 | def get_doc2vec_matrix(texts, model):
29 |     """
30 |     A helper function to get the vector for each document
31 |     and combine it into a matrix. 
32 |     """
33 |     num_columns = model.trainables.layer1_size
34 |     num_rows = len(texts)
35 |     
36 |     doc2vec_pii_matrix = np.zeros((num_rows, num_columns))
37 |     
38 |     for index, text in enumerate(tqdm(texts)):
39 |         
40 |         doc2vec_pii_matrix[index,:] = model.infer_vector(text.split(" "))
41 |         
42 |     return doc2vec_pii_matrix
43 | 
44 | 
45 | color_dict = {"Phone_number":"red","SSN":"blue","Address":"black","Name":"yellow",\
46 |              "Plates":"orange","CreditCardNumber":"purple","None":'pink',"Email":"tan"}
47 | 
48 | 
49 | def _match_re_and_extract_cv_scores(model,all_cv_keys, regex):
50 |     match_keys = re.findall(regex," ".join(all_cv_keys))
51 |     
52 |     cv_scores = dict()
53 |     
54 |     for key in match_keys:
55 |         cv_scores[key] = model.cv_results_[key]
56 |     return cv_scores 
57 | 
58 | 
59 | def extract_sk_cv(model, test = True):
60 |     """
61 |     A helper function to extract the training/testing cross validation result
62 |     """
63 |     all_cv_keys = model.cv_results_.keys()
64 |     
65 |     if test:
66 |         sk_cv_result =  _match_re_and_extract_cv_scores(model = model,all_cv_keys = all_cv_keys\
67 |                                                              ,regex = r"split[0-9]+_test_score")
68 |         return pd.DataFrame(sk_cv_result)
69 | 
70 |     elif not test:
71 |         
72 |         sk_cv_result = _match_re_and_extract_cv_scores(model = model,all_cv_keys = all_cv_keys\
73 |                                                              ,regex = r"split[0-9]+_train_score")
74 |         return pd.DataFrame(sk_cv_result)
75 | 


--------------------------------------------------------------------------------
/model/word2vec.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "markdown",
  5 |    "metadata": {},
  6 |    "source": [
  7 |     "The goal of this notebook is to train a word2vec model for the fake text with PII "
  8 |    ]
  9 |   },
 10 |   {
 11 |    "cell_type": "code",
 12 |    "execution_count": 1,
 13 |    "metadata": {},
 14 |    "outputs": [],
 15 |    "source": [
 16 |     "from gensim.models import Word2Vec\n",
 17 |     "\n",
 18 |     "from yellowbrick.text import TSNEVisualizer\n",
 19 |     "import matplotlib.pyplot as plt\n",
 20 |     "\n",
 21 |     "\n",
 22 |     "import pandas as pd"
 23 |    ]
 24 |   },
 25 |   {
 26 |    "cell_type": "code",
 27 |    "execution_count": 2,
 28 |    "metadata": {},
 29 |    "outputs": [],
 30 |    "source": [
 31 |     "fake_pii = pd.read_csv(\"../data/text_with_pii_2018_12_21_04_32_47_173911.csv\")"
 32 |    ]
 33 |   },
 34 |   {
 35 |    "cell_type": "code",
 36 |    "execution_count": 3,
 37 |    "metadata": {},
 38 |    "outputs": [
 39 |     {
 40 |      "data": {
 41 |       "text/html": [
 42 |        "<div>\n",
 43 |        "<style scoped>\n",
 44 |        "    .dataframe tbody tr th:only-of-type {\n",
 45 |        "        vertical-align: middle;\n",
 46 |        "    }\n",
 47 |        "\n",
 48 |        "    .dataframe tbody tr th {\n",
 49 |        "        vertical-align: top;\n",
 50 |        "    }\n",
 51 |        "\n",
 52 |        "    .dataframe thead th {\n",
 53 |        "        text-align: right;\n",
 54 |        "    }\n",
 55 |        "</style>\n",
 56 |        "<table border=\"1\" class=\"dataframe\">\n",
 57 |        "  <thead>\n",
 58 |        "    <tr style=\"text-align: right;\">\n",
 59 |        "      <th></th>\n",
 60 |        "      <th>Text</th>\n",
 61 |        "      <th>Labels</th>\n",
 62 |        "      <th>PII</th>\n",
 63 |        "    </tr>\n",
 64 |        "  </thead>\n",
 65 |        "  <tbody>\n",
 66 |        "    <tr>\n",
 67 |        "      <th>0</th>\n",
 68 |        "      <td>Ever watch television read measure almost. Tho...</td>\n",
 69 |        "      <td>Address</td>\n",
 70 |        "      <td>Suite 971</td>\n",
 71 |        "    </tr>\n",
 72 |        "    <tr>\n",
 73 |        "      <th>1</th>\n",
 74 |        "      <td>Impact region card 37271 Adams Road Suite 438 ...</td>\n",
 75 |        "      <td>Address</td>\n",
 76 |        "      <td>37271 Adams Road Suite 438</td>\n",
 77 |        "    </tr>\n",
 78 |        "    <tr>\n",
 79 |        "      <th>2</th>\n",
 80 |        "      <td>0604 Mason Wall Throughout admit have science ...</td>\n",
 81 |        "      <td>Address</td>\n",
 82 |        "      <td>0604 Mason Wall</td>\n",
 83 |        "    </tr>\n",
 84 |        "    <tr>\n",
 85 |        "      <th>3</th>\n",
 86 |        "      <td>Able item bag my day impact continue. Apt. 633...</td>\n",
 87 |        "      <td>Address</td>\n",
 88 |        "      <td>Apt. 633</td>\n",
 89 |        "    </tr>\n",
 90 |        "    <tr>\n",
 91 |        "      <th>4</th>\n",
 92 |        "      <td>Inside perhaps Suite 744 else seven. Able seas...</td>\n",
 93 |        "      <td>Address</td>\n",
 94 |        "      <td>Suite 744</td>\n",
 95 |        "    </tr>\n",
 96 |        "  </tbody>\n",
 97 |        "</table>\n",
 98 |        "</div>"
 99 |       ],
100 |       "text/plain": [
101 |        "                                                Text   Labels  \\\n",
102 |        "0  Ever watch television read measure almost. Tho...  Address   \n",
103 |        "1  Impact region card 37271 Adams Road Suite 438 ...  Address   \n",
104 |        "2  0604 Mason Wall Throughout admit have science ...  Address   \n",
105 |        "3  Able item bag my day impact continue. Apt. 633...  Address   \n",
106 |        "4  Inside perhaps Suite 744 else seven. Able seas...  Address   \n",
107 |        "\n",
108 |        "                          PII  \n",
109 |        "0                   Suite 971  \n",
110 |        "1  37271 Adams Road Suite 438  \n",
111 |        "2             0604 Mason Wall  \n",
112 |        "3                    Apt. 633  \n",
113 |        "4                   Suite 744  "
114 |       ]
115 |      },
116 |      "execution_count": 3,
117 |      "metadata": {},
118 |      "output_type": "execute_result"
119 |     }
120 |    ],
121 |    "source": [
122 |     "fake_pii.head()"
123 |    ]
124 |   },
125 |   {
126 |    "cell_type": "code",
127 |    "execution_count": 4,
128 |    "metadata": {},
129 |    "outputs": [
130 |     {
131 |      "data": {
132 |       "text/plain": [
133 |        "'Impact region card 37271 Adams Road Suite 438 technology indicate. Station cut rule church Congress. There west parent write. Accept image region through group matter.'"
134 |       ]
135 |      },
136 |      "execution_count": 4,
137 |      "metadata": {},
138 |      "output_type": "execute_result"
139 |     }
140 |    ],
141 |    "source": [
142 |     "fake_pii.iloc[1,0]"
143 |    ]
144 |   },
145 |   {
146 |    "cell_type": "code",
147 |    "execution_count": 5,
148 |    "metadata": {},
149 |    "outputs": [],
150 |    "source": [
151 |     "# clean the text in a sentence and return a sentence.\n",
152 |     "def clean_sentence(sentence):\n",
153 |     "    \n",
154 |     "    sentence_cleaned = [word.replace(\",\", \"\").replace(\".\", \"\")\\\n",
155 |     "                        .lower() for word in sentence.split(\" \")]\n",
156 |     "    return \" \".join(sentence_cleaned)"
157 |    ]
158 |   },
159 |   {
160 |    "cell_type": "code",
161 |    "execution_count": 6,
162 |    "metadata": {},
163 |    "outputs": [
164 |     {
165 |      "data": {
166 |       "text/plain": [
167 |        "'ever watch television read measure almost thought shake take suite 971 program interest strong exist clear least like create most order court build organization'"
168 |       ]
169 |      },
170 |      "execution_count": 6,
171 |      "metadata": {},
172 |      "output_type": "execute_result"
173 |     }
174 |    ],
175 |    "source": [
176 |     "clean_sentence(fake_pii.iloc[0,0])"
177 |    ]
178 |   },
179 |   {
180 |    "cell_type": "code",
181 |    "execution_count": 7,
182 |    "metadata": {
183 |     "scrolled": true
184 |    },
185 |    "outputs": [
186 |     {
187 |      "name": "stdout",
188 |      "output_type": "stream",
189 |      "text": [
190 |       "CPU times: user 7.4 s, sys: 62.7 ms, total: 7.46 s\n",
191 |       "Wall time: 7.46 s\n"
192 |      ]
193 |     }
194 |    ],
195 |    "source": [
196 |     "%%time\n",
197 |     "context_sentences = fake_pii[\"Text\"].apply(clean_sentence)"
198 |    ]
199 |   },
200 |   {
201 |    "cell_type": "code",
202 |    "execution_count": 8,
203 |    "metadata": {},
204 |    "outputs": [],
205 |    "source": [
206 |     "tokenized_sentence_word_list = [sentence.split(\" \")\\\n",
207 |     "                                for sentence in context_sentences]"
208 |    ]
209 |   },
210 |   {
211 |    "cell_type": "code",
212 |    "execution_count": 9,
213 |    "metadata": {},
214 |    "outputs": [
215 |     {
216 |      "name": "stdout",
217 |      "output_type": "stream",
218 |      "text": [
219 |       "CPU times: user 3min 16s, sys: 940 ms, total: 3min 17s\n",
220 |       "Wall time: 1min 43s\n"
221 |      ]
222 |     }
223 |    ],
224 |    "source": [
225 |     "%%time \n",
226 |     "word2vec_model_fake_pii = Word2Vec(tokenized_sentence_word_list,\\\n",
227 |     "                                   size = 300, window=5,\n",
228 |     "                                  min_count = 1, workers = 2)"
229 |    ]
230 |   },
231 |   {
232 |    "cell_type": "markdown",
233 |    "metadata": {},
234 |    "source": [
235 |     "Now we have the trained word2vec model, we can save it to disk. "
236 |    ]
237 |   },
238 |   {
239 |    "cell_type": "code",
240 |    "execution_count": 10,
241 |    "metadata": {},
242 |    "outputs": [],
243 |    "source": [
244 |     "word2vec_model_fake_pii.save(\"./word2vec/word2vec_300_fake_pii.bin\")"
245 |    ]
246 |   }
247 |  ],
248 |  "metadata": {
249 |   "kernelspec": {
250 |    "display_name": "Python 3",
251 |    "language": "python",
252 |    "name": "python3"
253 |   },
254 |   "language_info": {
255 |    "codemirror_mode": {
256 |     "name": "ipython",
257 |     "version": 3
258 |    },
259 |    "file_extension": ".py",
260 |    "mimetype": "text/x-python",
261 |    "name": "python",
262 |    "nbconvert_exporter": "python",
263 |    "pygments_lexer": "ipython3",
264 |    "version": "3.7.1"
265 |   }
266 |  },
267 |  "nbformat": 4,
268 |  "nbformat_minor": 2
269 | }
270 | 


--------------------------------------------------------------------------------
/model/word2vec.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/edwardcooper/data-sentry/f2a0e438101c76abaacca834e123bb3fb964f6a1/model/word2vec.png


--------------------------------------------------------------------------------
/model/word2vec_300_logistic.pkl:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/edwardcooper/data-sentry/f2a0e438101c76abaacca834e123bb3fb964f6a1/model/word2vec_300_logistic.pkl


--------------------------------------------------------------------------------
/model/word2vec_300_logistic_with_weights.pkl:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/edwardcooper/data-sentry/f2a0e438101c76abaacca834e123bb3fb964f6a1/model/word2vec_300_logistic_with_weights.pkl


--------------------------------------------------------------------------------
/model/word2vec_pipe_cv_test_performance.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/edwardcooper/data-sentry/f2a0e438101c76abaacca834e123bb3fb964f6a1/model/word2vec_pipe_cv_test_performance.png


--------------------------------------------------------------------------------
/model/word2vec_with_google_update.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "markdown",
  5 |    "metadata": {},
  6 |    "source": [
  7 |     "## The goal of this notebook is to load the google's word2vec model, and update the weights with the google's pre-trained model."
  8 |    ]
  9 |   },
 10 |   {
 11 |    "cell_type": "code",
 12 |    "execution_count": 1,
 13 |    "metadata": {},
 14 |    "outputs": [],
 15 |    "source": [
 16 |     "from gensim.models import Word2Vec\n",
 17 |     "import pandas as pd"
 18 |    ]
 19 |   },
 20 |   {
 21 |    "cell_type": "code",
 22 |    "execution_count": 2,
 23 |    "metadata": {},
 24 |    "outputs": [],
 25 |    "source": [
 26 |     "fake_pii_word2vec = Word2Vec.load(\"./word2vec/word2vec_300_fake_pii.bin\")"
 27 |    ]
 28 |   },
 29 |   {
 30 |    "cell_type": "code",
 31 |    "execution_count": 3,
 32 |    "metadata": {},
 33 |    "outputs": [],
 34 |    "source": [
 35 |     "# clean the text in a sentence and return a sentence.\n",
 36 |     "def clean_sentence(sentence):\n",
 37 |     "    \n",
 38 |     "    sentence_cleaned = [word.replace(\",\", \"\").replace(\".\", \"\")\\\n",
 39 |     "                        .lower() for word in sentence.split(\" \")]\n",
 40 |     "    return \" \".join(sentence_cleaned) \n",
 41 |     "\n",
 42 |     "\n",
 43 |     "string = \"My email address is salitrojan@gmail.com\\\n",
 44 |     " I would love you to call me any time at 111-999-7711. Best,\\\n",
 45 |     " Sali Trojan\"\n",
 46 |     " \n",
 47 |     "\n",
 48 |     "clean_str = clean_sentence(string)\n",
 49 |     "tokenized_str = clean_str.split(\" \")"
 50 |    ]
 51 |   },
 52 |   {
 53 |    "cell_type": "code",
 54 |    "execution_count": 7,
 55 |    "metadata": {},
 56 |    "outputs": [
 57 |     {
 58 |      "data": {
 59 |       "text/plain": [
 60 |        "(358, 455)"
 61 |       ]
 62 |      },
 63 |      "execution_count": 7,
 64 |      "metadata": {},
 65 |      "output_type": "execute_result"
 66 |     }
 67 |    ],
 68 |    "source": [
 69 |     "fake_pii_word2vec.build_vocab([tokenized_str], update = True)\n",
 70 |     "fake_pii_word2vec.train(tokenized_str, total_examples= 1, epochs = 5)"
 71 |    ]
 72 |   },
 73 |   {
 74 |    "cell_type": "code",
 75 |    "execution_count": 5,
 76 |    "metadata": {},
 77 |    "outputs": [],
 78 |    "source": [
 79 |     "fake_pii_word2vec.intersect_word2vec_format('./word2vec/GoogleNews-vectors-negative300.bin',\\\n",
 80 |     "                                  lockf=1.0,binary=True)"
 81 |    ]
 82 |   },
 83 |   {
 84 |    "cell_type": "code",
 85 |    "execution_count": 13,
 86 |    "metadata": {},
 87 |    "outputs": [],
 88 |    "source": [
 89 |     "fake_pii_word2vec.save(\"./word2vec/word2vec_300_fake_pii_with_google_update.bin\")"
 90 |    ]
 91 |   }
 92 |  ],
 93 |  "metadata": {
 94 |   "kernelspec": {
 95 |    "display_name": "Python 3",
 96 |    "language": "python",
 97 |    "name": "python3"
 98 |   },
 99 |   "language_info": {
100 |    "codemirror_mode": {
101 |     "name": "ipython",
102 |     "version": 3
103 |    },
104 |    "file_extension": ".py",
105 |    "mimetype": "text/x-python",
106 |    "name": "python",
107 |    "nbconvert_exporter": "python",
108 |    "pygments_lexer": "ipython3",
109 |    "version": "3.7.1"
110 |   }
111 |  },
112 |  "nbformat": 4,
113 |  "nbformat_minor": 2
114 | }
115 | 


--------------------------------------------------------------------------------
/requirements.txt:
--------------------------------------------------------------------------------
 1 | apturl==0.5.2
 2 | asn1crypto==0.24.0
 3 | bleach==3.1.0
 4 | boto==2.49.0
 5 | boto3==1.9.83
 6 | botocore==1.12.83
 7 | Brlapi==0.6.6
 8 | bz2file==0.98
 9 | certifi==2018.11.29
10 | chardet==3.0.4
11 | Click==7.0
12 | command-not-found==0.3
13 | cryptography==2.1.4
14 | cupshelpers==1.0
15 | defer==1.0.6
16 | distro-info==0.18
17 | docutils==0.14
18 | Faker==1.0.1
19 | Flask==1.0.2
20 | Flask-SQLAlchemy==2.3.2
21 | Flask-WTF==0.14.2
22 | gensim==3.7.0
23 | httplib2==0.9.2
24 | idna==2.8
25 | itsdangerous==1.1.0
26 | Jinja2==2.10
27 | jmespath==0.9.3
28 | keyring==10.6.0
29 | keyrings.alt==3.0
30 | language-selector==0.1
31 | launchpadlib==1.10.6
32 | lazr.restfulclient==0.13.5
33 | lazr.uri==1.0.3
34 | louis==3.5.0
35 | macaroonbakery==1.1.3
36 | Mako==1.0.7
37 | MarkupSafe==1.1.0
38 | netifaces==0.10.4
39 | numpy==1.16.0
40 | oauth==1.0.1
41 | olefile==0.45.1
42 | pandas==0.23.4
43 | pexpect==4.2.1
44 | piidetect==0.0.0.2
45 | Pillow==5.1.0
46 | pkginfo==1.5.0.1
47 | protobuf==3.0.0
48 | pycairo==1.16.2
49 | pycrypto==2.6.1
50 | pycups==1.9.73
51 | Pygments==2.3.1
52 | pygobject==3.26.1
53 | pymacaroons==0.13.0
54 | PyNaCl==1.1.2
55 | pyRFC3339==1.0
56 | python-apt==1.6.3
57 | python-dateutil==2.7.5
58 | python-debian==0.1.32
59 | pytz==2018.9
60 | pyxdg==0.25
61 | PyYAML==3.12
62 | readme-renderer==24.0
63 | reportlab==3.4.0
64 | requests==2.21.0
65 | requests-toolbelt==0.8.0
66 | requests-unixsocket==0.1.5
67 | s3transfer==0.1.13
68 | scikit-learn==0.20.2
69 | scipy==1.2.0
70 | SecretStorage==2.3.1
71 | simplejson==3.13.2
72 | six==1.12.0
73 | sklearn==0.0
74 | smart-open==1.8.0
75 | SQLAlchemy==1.2.16
76 | system-service==0.3
77 | systemd-python==234
78 | text-unidecode==1.2
79 | tqdm==4.29.1
80 | twine==1.12.1
81 | ubuntu-drivers-common==0.0.0
82 | ufw==0.35
83 | unattended-upgrades==0.1
84 | urllib3==1.24.1
85 | usb-creator==0.3.3
86 | wadllib==1.3.2
87 | webencodings==0.5.1
88 | Werkzeug==0.14.1
89 | word-embedding==0.0.0.1
90 | WTForms==2.2.1
91 | xkit==0.0.0
92 | zope.interface==4.3.2
93 | 


--------------------------------------------------------------------------------
/website/app.py:
--------------------------------------------------------------------------------
 1 | from flask import Flask
 2 | from flask import render_template
 3 | from flask import flash
 4 | from flask import redirect
 5 | from flask import url_for
 6 | 
 7 | from user_entry_forms import RegistrationForm, LoginForm, TextEntry
 8 | 
 9 | from flask_sqlalchemy import SQLAlchemy
10 | from piidetect.pipeline import text_clean, word_embedding
11 | 
12 | # libraries for the model part
13 | from sklearn.externals import joblib
14 | from piidetect.pipeline import clean_text, word_embedding
15 | import pandas as pd
16 | # for caching the word2vec model
17 | 
18 | from werkzeug.contrib.cache import SimpleCache
19 | cache = SimpleCache()
20 | 
21 | app = Flask(__name__)
22 | # this a temp setup for development.
23 | app.config['SECRET_KEY'] = '34d9f494c01b69e2f6aad1e9ccfd5bb48630ab9b'
24 | 
25 | 
26 | 
27 | def get_word2vec_model(filename):
28 |     word2vec_model = cache.get("word2vec_pipe_cv_production")
29 |     if word2vec_model is None:
30 |         word2vec_model = joblib.load(filename)
31 |         cache.set("word2vec_pipe_cv_production", word2vec_model, timeout = 300)
32 | 
33 |     return word2vec_model
34 | 
35 | @app.route('/', methods = ["GET", "POST"])
36 | def home_page():
37 |     form_for_text_input = TextEntry()
38 |     if form_for_text_input.text.data is None:
39 |         text = pd.Series(["My email address is edwardlu@gmail.com"])
40 |         output = text.iloc[0]
41 |     else:
42 |         text = pd.Series(form_for_text_input.text.data)
43 |         # load the model if not cached yet.
44 |         word2vec_model = get_word2vec_model("word2vec_pipe_cv_production.pkl")
45 |         # model prediction
46 |         model_result = word2vec_model.predict(text)
47 |         # convert the results into different messages. 
48 |         if model_result[0] == 1:
49 |             flash("The text you entered contains PII data","danger")
50 | 
51 |         elif model_result[0] == 0:
52 |             flash("The text you entered does not contain PII data",'success')
53 | 
54 |     return render_template("home_page.html", form = form_for_text_input)
55 | 
56 | @app.route("/API")
57 | def API_page():
58 |     return render_template("API_page.html")
59 | 
60 | 
61 | @app.route('/Register', methods = ['GET',"POST"])
62 | def Register_page():
63 |     form_for_register = RegistrationForm()
64 |     if form_for_register.validate_on_submit():
65 |         flash(f"Account create for {form_for_register.username.data}!", 'success')
66 |         return redirect(url_for("API_page"))
67 |     return render_template('Register_page.html', form = form_for_register)
68 | 
69 | @app.route('/Login', methods = ['GET',"POST"])
70 | def Login_page():
71 |     form_for_login = LoginForm()
72 |     if form_for_login.validate_on_submit():
73 |         if form_for_login.email.data == "admin@piisentry.com" and form_for_login.password.data == "1234":
74 |             flash("You have been loged in!", 'success')
75 |             return redirect(url_for("API_page"))
76 |         else:
77 |             flash("Please double check your email and password", "danger")
78 | 
79 |     return render_template('Login_page.html', form = form_for_login)
80 | 
81 | 
82 | 
83 | if __name__ == "__main__":
84 |     app.run(host = '0.0.0.0', debug = True, port = 80)
85 | 


--------------------------------------------------------------------------------
/website/static/cover.css:
--------------------------------------------------------------------------------
  1 | /*
  2 |  * Globals
  3 |  */
  4 | 
  5 | /* Links */
  6 | a,
  7 | a:focus,
  8 | a:hover {
  9 |   color: #fff;
 10 | }
 11 | 
 12 | /* Custom default button */
 13 | .btn-secondary,
 14 | .btn-secondary:hover,
 15 | .btn-secondary:focus {
 16 |   color: #999;
 17 |   text-shadow: #888; /* Prevent inheritance from `body` */
 18 |   background-color: #fff;
 19 |   border: .05rem solid #fff;
 20 | }
 21 | 
 22 | 
 23 | /*
 24 |  * Base structure
 25 |  */
 26 | 
 27 | html,
 28 | body {
 29 |   height: 100%;
 30 |   background-color:  #333;
 31 | }
 32 | 
 33 | body {
 34 |   display: -ms-flexbox;
 35 |   display: flex;
 36 |   color: #fff;
 37 |   text-shadow: 0 .05rem .1rem rgba(0, 0, 0, .5);
 38 |   box-shadow: inset 0 0 5rem rgba(0, 0, 0, .5);
 39 | }
 40 | 
 41 | .cover-container {
 42 |   max-width: 42em;
 43 | }
 44 | 
 45 | 
 46 | /*
 47 |  * Header
 48 |  */
 49 | .masthead {
 50 |   margin-bottom: 2rem;
 51 | }
 52 | 
 53 | .masthead-brand {
 54 |   margin-bottom: 0;
 55 | }
 56 | 
 57 | .nav-masthead .nav-link {
 58 |   padding: .25rem 0;
 59 |   font-weight: 700;
 60 |   color: rgba(255, 255, 255, .5);
 61 |   background-color: transparent;
 62 |   border-bottom: .25rem solid transparent;
 63 | }
 64 | 
 65 | .nav-masthead .nav-link:hover,
 66 | .nav-masthead .nav-link:focus {
 67 |   border-bottom-color: rgba(255, 255, 255, .25);
 68 | }
 69 | 
 70 | .nav-masthead .nav-link + .nav-link {
 71 |   margin-left: 1rem;
 72 | }
 73 | 
 74 | .nav-masthead .active {
 75 |   color: #fff;
 76 |   border-bottom-color: #fff;
 77 | }
 78 | 
 79 | @media (min-width: 48em) {
 80 |   .masthead-brand {
 81 |     float: left;
 82 |   }
 83 |   .nav-masthead {
 84 |     float: right;
 85 |   }
 86 | }
 87 | 
 88 | 
 89 | /*
 90 |  * Cover
 91 |  */
 92 | .cover {
 93 |   padding: 0 1.5rem;
 94 | }
 95 | .cover .btn-lg {
 96 |   padding: 0.75rem 1.25rem;
 97 |   font-weight: 1000;
 98 | }
 99 | 
100 | 
101 | /*
102 |  * Footer
103 |  */
104 | .mastfoot {
105 |   color: rgba(255, 255, 255, .5);
106 | }
107 | 


--------------------------------------------------------------------------------
/website/templates/API_page.html:
--------------------------------------------------------------------------------
 1 | <!DOCTYPE html>
 2 | <html lang="en" dir="ltr">
 3 |   <head>
 4 |     <meta charset="utf-8">
 5 |     <title>piisentry</title>
 6 |   </head>
 7 |   <body>
 8 | 
 9 |     {% extends "template.html" %}
10 | 
11 | 
12 |     {% block navbar %}
13 |     <a class="nav-link" href="{{ url_for('home_page')}}">Home</a>
14 |     <a class="nav-link active" href="{{ url_for('API_page')}}">API documentation</a>
15 |     <a class="nav-link " href="{{url_for('Login_page')}}">Login</a>
16 |     <a class="nav-link" href="{{url_for('Register_page')}}">Register</a>
17 |     {% endblock navbar%}
18 | 
19 |     {% block content %}
20 | 
21 |      <p>
22 |        This is the API documentation page.
23 |      </p>
24 | 
25 |     {% endblock %}
26 |   </body>
27 | </html>
28 | 


--------------------------------------------------------------------------------
/website/templates/Forget_password.html:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/edwardcooper/data-sentry/f2a0e438101c76abaacca834e123bb3fb964f6a1/website/templates/Forget_password.html


--------------------------------------------------------------------------------
/website/templates/Login_page.html:
--------------------------------------------------------------------------------
 1 | <title>piisentry</title>
 2 | {% extends "template.html" %}
 3 | {% block navbar %}
 4 | <a class="nav-link" href="{{ url_for('home_page')}}">Home</a>
 5 | <a class="nav-link" href="{{ url_for('API_page')}}">API documentation</a>
 6 | <a class="nav-link active" href="{{url_for('Login_page')}}">Login</a>
 7 | <a class="nav-link" href="{{url_for('Register_page')}}">Register</a>
 8 | {% endblock navbar%}
 9 | 
10 | {% block content %}
11 |     <div class="content-section">
12 |       <form class="" action="" method="post">
13 |         {{ form.hidden_tag() }}
14 | 
15 |         <fieldset class = 'form-group'>
16 | 
17 |               <!-- email -->
18 |               <div class = 'form-group'>
19 |                 {{ form.email.label(class = "form-control-label") }}
20 |                 <!-- add error notification for this field  -->
21 |                 {% if form.email.errors %}
22 |                     {{ form.email(class = "form-control form-control-lg is-invalid") }}
23 |                     <div class="invalid-feedback">
24 |                       {% for error in form.email.errors %}
25 |                       <span>{{ error }}</span>
26 |                       {% endfor %}
27 |                     </div>
28 |                 {% else %}
29 |                     {{ form.email(class = "form-control form-control-lg") }}
30 |                 {% endif %}
31 |               </div>
32 | 
33 |               <!-- password -->
34 |               <div class = 'form-group'>
35 |                 {{ form.password.label(class = "form-control-label") }}
36 |                 <!-- add error notification for this field  -->
37 |                 {% if form.password.errors %}
38 |                     {{ form.password(class = "form-control form-control-lg is-invalid") }}
39 |                     <div class="invalid-feedback">
40 |                       {% for error in form.password.errors %}
41 |                       <span>{{ error }}</span>
42 |                       {% endfor %}
43 |                     </div>
44 |                 {% else %}
45 |                     {{ form.password(class = "form-control form-control-lg") }}
46 |                 {% endif %}
47 |               </div>
48 | 
49 |               <div class="form-check">
50 |                 {{ form.remember(class = "form-check-input ") }}
51 |                 {{ form.remember.label(class = "form-check-label ") }}
52 |               </div>
53 |         </fieldset>
54 | 
55 |         <div class="form-group">
56 |           {{ form.submit(class="btn btn-outline-info") }}
57 |         </div>
58 |         <small class = "text-muted ml-2">
59 |           <a href="#">Forgot password?</a>
60 |         </small>
61 |       </form>
62 |     </div>
63 | 
64 | 
65 |       <small class="text-muted">
66 |         Need an account? <a href="{{url_for('Register_page')}}" class = 'ml-2'>Sign Up</a>
67 |       </small>
68 | 
69 | 
70 | 
71 | {% endblock content %}
72 | 


--------------------------------------------------------------------------------
/website/templates/Register_page.html:
--------------------------------------------------------------------------------
  1 | <title>piisentry</title>
  2 | {% extends "template.html" %}
  3 | {% block navbar %}
  4 | <a class="nav-link" href="{{ url_for('home_page')}}">Home</a>
  5 | <a class="nav-link" href="{{ url_for('API_page')}}">API documentation</a>
  6 | <a class="nav-link" href="{{url_for('Login_page')}}">Login</a>
  7 | <a class="nav-link active" href="{{url_for('Register_page')}}">Register</a>
  8 | {% endblock navbar%}
  9 | 
 10 | {% block content %}
 11 |     <div class="content-section">
 12 |       <form action="" method="post">
 13 |         {{ form.hidden_tag() }}
 14 | 
 15 |         <fieldset class = 'form-group'>
 16 | 
 17 | 
 18 |               <!-- username -->
 19 |               <div class = 'form-group'>
 20 |                 {{ form.username.label(class = "form-control-label") }}
 21 |                 <!-- add error notification for this field  -->
 22 |                 {% if form.username.errors %}
 23 |                     {{ form.username(class = "form-control  is-invalid") }}
 24 |                     <div class="invalid-feedback">
 25 |                       {% for error in form.username.errors %}
 26 |                       <span>{{ error }}</span>
 27 |                       {% endfor %}
 28 |                     </div>
 29 |                 {% else %}
 30 |                     {{ form.username(class = "form-control ") }}
 31 |                 {% endif %}
 32 |               </div>
 33 | 
 34 |               <!-- email -->
 35 |               <div class = 'form-group'>
 36 |                 {{ form.email.label(class = "form-control-label") }}
 37 |                 <!-- add error notification for this field  -->
 38 |                 {% if form.email.errors %}
 39 |                     {{ form.email(class = "form-control  is-invalid") }}
 40 |                     <div class="invalid-feedback">
 41 |                       {% for error in form.email.errors %}
 42 |                       <span>{{ error }}</span>
 43 |                       {% endfor %}
 44 |                     </div>
 45 |                 {% else %}
 46 |                     {{ form.email(class = "form-control ") }}
 47 |                 {% endif %}
 48 |               </div>
 49 | 
 50 |               <!-- password -->
 51 |               <div class = 'form-group'>
 52 |                 {{ form.password.label(class = "form-control-label") }}
 53 |                 <!-- add error notification for this field  -->
 54 |                 {% if form.password.errors %}
 55 |                     {{ form.password(class = "form-control  is-invalid") }}
 56 |                     <div class="invalid-feedback">
 57 |                       {% for error in form.password.errors %}
 58 |                       <span>{{ error }}</span>
 59 |                       {% endfor %}
 60 |                     </div>
 61 |                 {% else %}
 62 |                     {{ form.password(class = "form-control ") }}
 63 |                 {% endif %}
 64 |               </div>
 65 | 
 66 |               <!-- confirmed_password -->
 67 |               <div class = 'form-group'>
 68 |                 {{ form.confirm_password.label(class = "form-control-label") }}
 69 |                 <!-- add error notification for this field  -->
 70 |                 {% if form.confirm_password.errors %}
 71 |                     {{ form.confirm_password(class = "form-control  is-invalid") }}
 72 |                     <div class="invalid-feedback">
 73 |                       {% for error in form.confirm_password.errors %}
 74 |                       <span>{{ error }}</span>
 75 |                       {% endfor %}
 76 |                     </div>
 77 |                 {% else %}
 78 |                     {{ form.confirm_password(class = "form-control ") }}
 79 |                 {% endif %}
 80 |               </div>
 81 | 
 82 | 
 83 |         </fieldset>
 84 | 
 85 |         <div class="form-group">
 86 |           {{ form.submit(class="btn btn-outline-info") }}
 87 |         </div>
 88 | 
 89 |       </form>
 90 |     </div>
 91 | 
 92 | 
 93 |       <small class="text-muted">
 94 |         Already have an account? <a href="{{url_for('Login_page')}}" class = 'ml-2'>Sign In</a>
 95 |       </small>
 96 | 
 97 | 
 98 | 
 99 | {% endblock content %}
100 | 


--------------------------------------------------------------------------------
/website/templates/home_page.html:
--------------------------------------------------------------------------------
 1 | <!DOCTYPE html>
 2 | <html lang="en" dir="ltr">
 3 |   <head>
 4 |     <meta charset="utf-8">
 5 |     <title>piisentry</title>
 6 |   </head>
 7 |   <body>
 8 | 
 9 |     {% extends "template.html" %}
10 | 
11 |     {% block navbar %}
12 |     <a class="nav-link active" href="{{ url_for('home_page')}}">Home</a>
13 |     <a class="nav-link" href="{{ url_for('API_page')}}">API documentation</a>
14 |     <a class="nav-link" href="{{url_for('Login_page')}}">Login</a>
15 |     <a class="nav-link" href="{{url_for('Register_page')}}">Register</a>
16 |     {% endblock navbar%}
17 | 
18 | 
19 |     {% block content %}
20 |     <legend class = "mb-4">Enter your text below to search for PII</legend>
21 |     <form class="form-group mb-4 " action="" method="POST">
22 |         <fieldset>
23 |           <div class="form-group">
24 |           {{ form.text(class = "form-control") }}
25 |           </div>
26 |         </fieldset>
27 | 
28 |         <div class="form-group">
29 |           {{ form.submit(class="btn btn-outline-info") }}
30 |         </div>
31 | 
32 |     </form>
33 | 
34 | 
35 |     {% endblock %}
36 |   </body>
37 | </html>
38 | 


--------------------------------------------------------------------------------
/website/templates/template.html:
--------------------------------------------------------------------------------
 1 | <!DOCTYPE html>
 2 | 
 3 | <html lang="en" class="gr__getbootstrap_com"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
 4 | 
 5 |     <meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no">
 6 |     <meta name="description" content="">
 7 |     <meta name="author" content="Mark Otto, Jacob Thornton, and Bootstrap contributors">
 8 |     <meta name="generator" content="Jekyll v3.8.5">
 9 |     <title>Cover</title>
10 | 
11 |     <!-- Bootstrap core CSS -->
12 | <link href="{{ url_for('static', filename='bootstrap.min.css') }}" rel="stylesheet" integrity="sha384-GJzZqFGwb1QTTN6wy59ffF1BuGJpLSa9DkKMp0DgiMDm4iYMj70gZWKYbI706tWS" crossorigin="anonymous">
13 | 
14 |     <style>
15 |       .bd-placeholder-img {
16 |         font-size: 1.125rem;
17 |         text-anchor: middle;
18 |       }
19 | 
20 |       @media (min-width: 768px) {
21 |         .bd-placeholder-img-lg {
22 |           font-size: 3.5rem;
23 |         }
24 |       }
25 |     </style>
26 |     <!-- Custom styles for this template -->
27 |     <link href="{{ url_for('static', filename='cover.css') }}" rel="stylesheet">
28 | 
29 |   </head>
30 |   <body class="text-center" data-gr-c-s-loaded="true">
31 |     <div class="cover-container d-flex w-100 h-100 p-3 mx-auto flex-column">
32 |   <header class="masthead mb-auto">
33 |     <div class="inner">
34 |       <h3 class="masthead-brand">PII Sentry</h3>
35 |       <nav class="nav nav-masthead justify-content-center">
36 |         <!-- for the navigation bar -->
37 |         {% block navbar%}
38 | 
39 |         {% endblock navbar%}
40 |       </nav>
41 |     </div>
42 |   </header>
43 | 
44 |   <main role="main" class="inner cover">
45 | 
46 |     <p class="lead">
47 | 
48 |       <!-- Flash message formatting for login/register  -->
49 |       {% with register_message = get_flashed_messages(with_categories = true) %}
50 |       {% if register_message %}
51 |           {% for category, message in register_message %}
52 |           <div class="alert alert-{{category}}">
53 |             {{ message }}
54 |           </div>
55 |           {%endfor%}
56 |       {% endif %}
57 |       {% endwith %}
58 | 
59 |       <!-- All the content for different page goes here. -->
60 |       {% block content %}
61 | 
62 |       {% endblock content %}
63 |       <!-- All the content for different page goes here. -->
64 | 
65 |     </p>
66 |   </main>
67 | 
68 | 
69 | <footer class="mastfoot mt-auto">
70 |   <div class="inner">
71 |     <p>Cover template for <a href="https://getbootstrap.com/">Bootstrap</a>, by <a href="https://twitter.com/mdo">@mdo</a>.</p>
72 |   </div>
73 | </footer>
74 | </div>
75 | 
76 | 
77 |     <!-- Optional JavaScript -->
78 |     <!-- jQuery first, then Popper.js, then Bootstrap JS -->
79 |     <script src="https://code.jquery.com/jquery-3.2.1.slim.min.js" integrity="sha384-KJ3o2DKtIkvYIK3UENzmM7KCkRr/rE9/Qpg6aAZGJwFDMVNA/GpGFF93hXpG5KkN" crossorigin="anonymous"></script>
80 |     <script src="https://cdnjs.cloudflare.com/ajax/libs/popper.js/1.12.9/umd/popper.min.js" integrity="sha384-ApNbgh9B+Y1QKtv3Rn7W3mgPxhU9K/ScQsAP7hUibX39j7fakFPskvXusvfa0b4Q" crossorigin="anonymous"></script>
81 |     <script src="https://maxcdn.bootstrapcdn.com/bootstrap/4.0.0/js/bootstrap.min.js" integrity="sha384-JZR6Spejh4U02d8jOt6vLEHfe/JQGiRRSQQxSfFWpi1MquVdAyjUar5+76PVCmYl" crossorigin="anonymous"></script>
82 | </body>
83 | </html>
84 | 


--------------------------------------------------------------------------------
/website/user_entry_forms.py:
--------------------------------------------------------------------------------
 1 | from flask_wtf import FlaskForm
 2 | from wtforms import StringField, PasswordField, SubmitField, BooleanField
 3 | from wtforms.validators import DataRequired, Length, Email, EqualTo
 4 | 
 5 | class RegistrationForm(FlaskForm):
 6 |     username = StringField('Username',
 7 |     validators = [DataRequired(),Length(min=2, max = 50)]
 8 |     )
 9 | 
10 |     email = StringField("Email", validators = [DataRequired(), Email()])
11 | 
12 |     password = PasswordField("Password", validators=[DataRequired()])
13 | 
14 |     confirm_password = PasswordField("Confirm Password",
15 |      validators=[DataRequired(),EqualTo('password')])
16 | 
17 |     submit = SubmitField("Sign up")
18 | 
19 | 
20 | class LoginForm(FlaskForm):
21 | 
22 |     email = StringField("Email", validators = [DataRequired(), Email()])
23 | 
24 |     password = PasswordField("Password", validators=[DataRequired()])
25 | 
26 |     remember = BooleanField('Remember Me')
27 | 
28 |     submit = SubmitField("Login")
29 | 
30 | 
31 | class TextEntry(FlaskForm):
32 | 
33 |     text = StringField('Enter your text here',
34 |     validators = [DataRequired(),Length(min=1)] )
35 | 
36 |     submit = SubmitField("Search PII")
37 | 


--------------------------------------------------------------------------------
/website/word2vec_model.py:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/edwardcooper/data-sentry/f2a0e438101c76abaacca834e123bb3fb964f6a1/website/word2vec_model.py


--------------------------------------------------------------------------------