├── .gitignore ├── LICENSE ├── README.md ├── jupyter ├── 00_DataPrep_German_Political_Speeches.ipynb ├── 00_DataPrep_Stopwords.ipynb ├── 00_InstallMallet.ipynb ├── 01_CreateCorpus_German_Political_Speeches.ipynb ├── 02_LearnModel_LDA_GPS.ipynb ├── 03_AnalyseModel_LDA_GPS.ipynb ├── 05_SimilaritySearch.ipynb ├── 50_topics_vis.html └── 99_test_bigrams.ipynb ├── requirements.txt └── start-jupyter.sh /.gitignore: -------------------------------------------------------------------------------- 1 | # Byte-compiled / optimized / DLL files 2 | __pycache__/ 3 | *.py[cod] 4 | *$py.class 5 | 6 | # C extensions 7 | *.so 8 | 9 | # Distribution / packaging 10 | .Python 11 | build/ 12 | develop-eggs/ 13 | dist/ 14 | downloads/ 15 | eggs/ 16 | .eggs/ 17 | lib/ 18 | lib64/ 19 | parts/ 20 | sdist/ 21 | var/ 22 | wheels/ 23 | *.egg-info/ 24 | .installed.cfg 25 | *.egg 26 | MANIFEST 27 | 28 | # PyInstaller 29 | # Usually these files are written by a python script from a template 30 | # before PyInstaller builds the exe, so as to inject date/other infos into it. 31 | *.manifest 32 | *.spec 33 | 34 | # Installer logs 35 | pip-log.txt 36 | pip-delete-this-directory.txt 37 | 38 | # Unit test / coverage reports 39 | htmlcov/ 40 | .tox/ 41 | .coverage 42 | .coverage.* 43 | .cache 44 | nosetests.xml 45 | coverage.xml 46 | *.cover 47 | .hypothesis/ 48 | .pytest_cache/ 49 | 50 | # Translations 51 | *.mo 52 | *.pot 53 | 54 | # Django stuff: 55 | *.log 56 | local_settings.py 57 | db.sqlite3 58 | 59 | # Flask stuff: 60 | instance/ 61 | .webassets-cache 62 | 63 | # Scrapy stuff: 64 | .scrapy 65 | 66 | # Sphinx documentation 67 | docs/_build/ 68 | 69 | # PyBuilder 70 | target/ 71 | 72 | # Jupyter Notebook 73 | .ipynb_checkpoints 74 | 75 | # pyenv 76 | .python-version 77 | 78 | # celery beat schedule file 79 | celerybeat-schedule 80 | 81 | # SageMath parsed files 82 | *.sage.py 83 | 84 | # Environments 85 | .env 86 | .venv 87 | env/ 88 | venv/ 89 | ENV/ 90 | env.bak/ 91 | venv.bak/ 92 | 93 | # Spyder project settings 94 | .spyderproject 95 | .spyproject 96 | 97 | # Rope project settings 98 | .ropeproject 99 | 100 | # mkdocs documentation 101 | /site 102 | 103 | # mypy 104 | .mypy_cache/ 105 | -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- 1 | GNU GENERAL PUBLIC LICENSE 2 | Version 3, 29 June 2007 3 | 4 | Copyright (C) 2007 Free Software Foundation, Inc. 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 | 635 | Copyright (C) 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 . 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 | Copyright (C) 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 | . 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 | . 675 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # Natural Language Processing (NLP) using Topic Modeling 2 | 3 | Application of topic model with special focus on German texts. 4 | 5 | **Datasets:** 6 | 7 | * [German Political Speeches](http://purl.org/corpus/german-speeches) 8 | * **`TODO`** [Offenes Parlament](https://github.com/Datenschule/offenesparlament-data) 9 | * **`TODO`** [Project Gutenberg](https://www.gutenberg.org/) 10 | * **`TODO`** German news articles 11 | * **`TODO`** German Wikipedia articles 12 | 13 | **Algorithms:** 14 | 15 | * **`TODO`** LSI - Latent Semantic Indexing (SVD) 16 | * LDA - Latent Dirichlet Allocation 17 | * **`TODO`** NMF - Non-negative Matrix Factorization 18 | 19 | **Tools:** 20 | 21 | * Gensim 22 | * Mallet 23 | * **`TODO`** [lda](https://pythonhosted.org/lda/) 24 | * **`TODO`** NLTK 25 | * **`TODO`** sklearn 26 | * **`TODO`** [BigARTM](http://bigartm.org/) 27 | * **`TODO`** [Vowpal Wabbit](https://github.com/VowpalWabbit/vowpal_wabbit) (Online LDA) 28 | * **`TODO`** [tmtoolkit](https://github.com/WZBSocialScienceCenter/tmtoolkit) 29 | * **`TODO`** [tcma](http://ilcm.informatik.uni-leipzig.de/software/download/) 30 | 31 | ## Useful and inspirational resources 32 | 33 | ### Topic Modeling Tutorials 34 | 35 | About: Building, Evaluating, Visualizing Topic Models 36 | 37 | * [Gensim Tutorials](https://radimrehurek.com/gensim/tutorial.html) 38 | * [Topics and Transformations](https://radimrehurek.com/gensim/tut2.html) 39 | * [Tutorial on Mallet in Python](https://rare-technologies.com/tutorial-on-mallet-in-python/) (2014-03-20) 40 | * [Mallet](http://mallet.cs.umass.edu/) 41 | * [pyLDAvis Library](https://github.com/bmabey/pyldavis) 42 | * http://nbviewer.jupyter.org/github/bmabey/pyLDAvis/blob/master/notebooks/pyLDAvis_overview.ipynb 43 | * [Machine Learning Plus Tutorials](https://www.machinelearningplus.com/blog/) ([Topic Modeling](https://www.machinelearningplus.com/tag/topic-modeling/), [NLP](https://www.machinelearningplus.com/tag/nlp/)) 44 | * [Topic modeling visualization – How to present the results of LDA models?](https://www.machinelearningplus.com/nlp/topic-modeling-visualization-how-to-present-results-lda-models/) (2018-12-04) 45 | * [LDA in Python – How to grid search best topic models?](https://www.machinelearningplus.com/nlp/topic-modeling-python-sklearn-examples/) (2018-04-04) 46 | * [Topic Modeling with Gensim](https://www.machinelearningplus.com/nlp/topic-modeling-gensim-python/) (2018-03-26) 47 | * [Lemmatization Approaches with Examples in Python](https://www.machinelearningplus.com/nlp/lemmatization-examples-python/) (2018-10-02) 48 | * [Gensim Tutorial](https://www.machinelearningplus.com/nlp/gensim-tutorial/) 49 | * [Data Science Plus Tutorials](https://datascienceplus.com/) 50 | * [Topic Modeling in Python with NLTK and Gensim](https://datascienceplus.com/topic-modeling-in-python-with-nltk-and-gensim/) (2018-04-26) 51 | * [Evaluation of Topic Modeling: Topic Coherence](https://datascienceplus.com/evaluation-of-topic-modeling-topic-coherence/) (2018-05-03) 52 | * [Towards Data Science](https://towardsdatascience.com/) 53 | * [The complete guide for topics extraction with LDA (Latent Dirichlet Allocation) in Python](https://towardsdatascience.com/the-complete-guide-for-topics-extraction-with-lda-latent-dirichlet-allocation-in-python-4d0200d0be98) (2018-12-14) 54 | * [WZB Data Science Blog (NLP)](https://datascience.blog.wzb.eu/category/nlp/) 55 | * [Topic Modeling – Background, Hyperparameters and Common pitfalls](https://datascience.blog.wzb.eu/2018/01/26/slides-on-topic-modeling-background-hyperparameters-and-common-pitfalls/) (2018-01-26) 56 | * [Practical Topic Modeling: Preparation, Evaluation, Visualization](https://datascience.blog.wzb.eu/2018/05/17/slides-on-practical-topic-modeling-preparation-evaluation-visualization/) (2018-05-17) 57 | * [Topic Model Evaluation in Python with tmtoolkit](https://datascience.blog.wzb.eu/2017/11/09/topic-modeling-evaluation-in-python-with-tmtoolkit/) (2017-11-09) 58 | * [A topic model for the debates of the 18th German Bundestag](https://github.com/WZBSocialScienceCenter/tm_bundestag) 59 | 60 | ### Topic Models applied on Wikipedia 61 | 62 | * https://radimrehurek.com/gensim/wiki.html 63 | * https://www.kdnuggets.com/2017/11/building-wikipedia-text-corpus-nlp.html 64 | 65 | ### Other NLP 66 | 67 | * https://github.com/adbar/German-NLP 68 | 69 | ## Research 70 | 71 | * [Abteilung Automatische Sprachverarbeitung, Universität Leipzig](http://asv.informatik.uni-leipzig.de/) 72 | * [Leipzig Corpus Miner](http://lcm.informatik.uni-leipzig.de/) 73 | 74 | ## Data Sources 75 | 76 | * [Link List - Wissenschaftszentrum Berlin für Sozialforschung](https://wzb.eu/de/literatur-daten/suchen-finden/datenquellen) 77 | * [Link List - Institut für deutsche Sprache und Linguistik (HU Berlin)](https://www.linguistik.hu-berlin.de/de/institut/professuren/korpuslinguistik/links/korpora_links) 78 | * [POLLUX - Informationsdienst Politikwissenschaft](https://www.pollux-fid.de/) 79 | * [German Microdata Lab (gesis)](https://www.gesis.org/institut/forschungsdatenzentren/fdz-german-microdata-lab/) 80 | * [Leipzig Corpora Collection](https://wortschatz.uni-leipzig.de/en/download) 81 | * [DWDS Corpora](https://www.dwds.de/d/korpora) 82 | 83 | ## Bibliography 84 | 85 | **LDA** 86 | 87 | * David M. Blei, Andrew Y. Ng, Michael I. Jordan. *Latent Dirichlet Allocation*. In: Journal of Machine Learning Research, 2003 88 | 89 | **Sentiment** 90 | 91 | * R. Remus, U. Quasthoff & G. Heyer: *SentiWS - a Publicly Available German-language Resource for Sentiment Analysis*. In: Proceedings of the 7th International Language Ressources and Evaluation (LREC'10), pp. 1168-1171, 2010 92 | -------------------------------------------------------------------------------- /jupyter/00_DataPrep_German_Political_Speeches.ipynb: -------------------------------------------------------------------------------- 1 | { 2 | "cells": [ 3 | { 4 | "cell_type": "markdown", 5 | "metadata": {}, 6 | "source": [ 7 | "# German Political Speeches\n", 8 | "\n", 9 | "Text Corpus from http://purl.org/corpus/german-speeches\n", 10 | "\n", 11 | "Barbaresi, Adrien (2018). \"A corpus of German political speeches from the 21st century\"." 12 | ] 13 | }, 14 | { 15 | "cell_type": "code", 16 | "execution_count": null, 17 | "metadata": {}, 18 | "outputs": [], 19 | "source": [ 20 | "import urllib\n", 21 | "\n", 22 | "url = \"http://adrien.barbaresi.eu/corpora/speeches/German-political-speeches-LREC2018-legacy-release.zip\"\n", 23 | "filename = url.rsplit('/', 1)[-1]\n", 24 | "datadir = '../data/'" 25 | ] 26 | }, 27 | { 28 | "cell_type": "markdown", 29 | "metadata": {}, 30 | "source": [ 31 | "## 1. Download Dataset" 32 | ] 33 | }, 34 | { 35 | "cell_type": "code", 36 | "execution_count": null, 37 | "metadata": {}, 38 | "outputs": [], 39 | "source": [ 40 | "urllib.request.urlretrieve(url, datadir + filename)" 41 | ] 42 | }, 43 | { 44 | "cell_type": "markdown", 45 | "metadata": {}, 46 | "source": [ 47 | "## 2. Extract and Convert XML documents" 48 | ] 49 | }, 50 | { 51 | "cell_type": "code", 52 | "execution_count": null, 53 | "metadata": {}, 54 | "outputs": [], 55 | "source": [ 56 | "from zipfile import ZipFile\n", 57 | "import xml.etree.ElementTree\n", 58 | "import pandas as pd\n", 59 | "import re, html\n", 60 | "\n", 61 | "def clean_str(str):\n", 62 | " return(html.unescape(re.sub(r\"[\\r\\n]\", \" \", str.strip())))\n", 63 | "\n", 64 | "def xml2csv(xmlfile, csvfile):\n", 65 | " root = xml.etree.ElementTree.parse(xmlfile).getroot()\n", 66 | " df = pd.DataFrame(([text.get('datum'),\n", 67 | " text.get('person'),\n", 68 | " clean_str(text.find('rohtext').text)] for text in root.iter('text')),\n", 69 | " columns=('date', 'speaker', 'text'))\n", 70 | " df.to_csv(csvfile, index=False)\n", 71 | "\n", 72 | "with ZipFile(datadir + filename, 'r') as zf:\n", 73 | " for info in zf.infolist():\n", 74 | " if (info.filename.endswith('.xml')):\n", 75 | " print('extracting', info.filename)\n", 76 | " with zf.open(info) as xmlfile:\n", 77 | " xml2csv(xmlfile, datadir + info.filename.replace('.xml','.csv')) " 78 | ] 79 | } 80 | ], 81 | "metadata": { 82 | "kernelspec": { 83 | "display_name": "Python 3", 84 | "language": "python", 85 | "name": "python3" 86 | }, 87 | "language_info": { 88 | "codemirror_mode": { 89 | "name": "ipython", 90 | "version": 3 91 | }, 92 | "file_extension": ".py", 93 | "mimetype": "text/x-python", 94 | "name": "python", 95 | "nbconvert_exporter": "python", 96 | "pygments_lexer": "ipython3", 97 | "version": "3.6.7" 98 | } 99 | }, 100 | "nbformat": 4, 101 | "nbformat_minor": 2 102 | } 103 | -------------------------------------------------------------------------------- /jupyter/00_DataPrep_Stopwords.ipynb: -------------------------------------------------------------------------------- 1 | { 2 | "cells": [ 3 | { 4 | "cell_type": "markdown", 5 | "metadata": {}, 6 | "source": [ 7 | "# Aquire Stopwords for Data Pre-processing\n", 8 | "\n", 9 | "Useful stopword sources:\n", 10 | "\n", 11 | " * nltk (231)\n", 12 | " * https://github.com/stopwords-iso/stopwords-de (621)\n", 13 | " * https://github.com/solariz/german_stopwords (1855)" 14 | ] 15 | }, 16 | { 17 | "cell_type": "code", 18 | "execution_count": null, 19 | "metadata": {}, 20 | "outputs": [], 21 | "source": [ 22 | "import urllib\n", 23 | "\n", 24 | "#url = 'https://raw.githubusercontent.com/stopwords-iso/stopwords-de/master/stopwords-de.txt'\n", 25 | "url = 'https://raw.githubusercontent.com/solariz/german_stopwords/master/german_stopwords_full.txt'\n", 26 | "filename = url.rsplit('/', 1)[-1]\n", 27 | "datadir = '../data/'" 28 | ] 29 | }, 30 | { 31 | "cell_type": "code", 32 | "execution_count": null, 33 | "metadata": {}, 34 | "outputs": [], 35 | "source": [ 36 | "urllib.request.urlretrieve(url, datadir + 'stopwords-de.txt')" 37 | ] 38 | } 39 | ], 40 | "metadata": { 41 | "kernelspec": { 42 | "display_name": "Python 3", 43 | "language": "python", 44 | "name": "python3" 45 | }, 46 | "language_info": { 47 | "codemirror_mode": { 48 | "name": "ipython", 49 | "version": 3 50 | }, 51 | "file_extension": ".py", 52 | "mimetype": "text/x-python", 53 | "name": "python", 54 | "nbconvert_exporter": "python", 55 | "pygments_lexer": "ipython3", 56 | "version": "3.6.7" 57 | } 58 | }, 59 | "nbformat": 4, 60 | "nbformat_minor": 2 61 | } 62 | -------------------------------------------------------------------------------- /jupyter/00_InstallMallet.ipynb: -------------------------------------------------------------------------------- 1 | { 2 | "cells": [ 3 | { 4 | "cell_type": "markdown", 5 | "metadata": {}, 6 | "source": [ 7 | "# Download and Extract [Mallet NLP Library](http://mallet.cs.umass.edu/download.php)" 8 | ] 9 | }, 10 | { 11 | "cell_type": "code", 12 | "execution_count": null, 13 | "metadata": {}, 14 | "outputs": [], 15 | "source": [ 16 | "import urllib, os, time, tarfile\n", 17 | "\n", 18 | "url='http://mallet.cs.umass.edu/dist/mallet-2.0.8.tar.gz'\n", 19 | "filename = url.rsplit('/', 1)[-1]\n", 20 | "dest_dir = '../lib/'\n", 21 | "\n", 22 | "# ensure destination directory exists\n", 23 | "if not os.path.exists(dest_dir):\n", 24 | " os.makedirs(dest_dir)" 25 | ] 26 | }, 27 | { 28 | "cell_type": "code", 29 | "execution_count": null, 30 | "metadata": {}, 31 | "outputs": [], 32 | "source": [ 33 | "start_time = time.time()\n", 34 | "urllib.request.urlretrieve(url, dest_dir + filename)\n", 35 | "print(\"--- took %d:%.2d minutes ---\" % divmod(time.time() - start_time, 60))" 36 | ] 37 | }, 38 | { 39 | "cell_type": "code", 40 | "execution_count": null, 41 | "metadata": {}, 42 | "outputs": [], 43 | "source": [ 44 | "with tarfile.open(dest_dir + filename, \"r:gz\") as tar:\n", 45 | " tar.extractall(path=dest_dir)" 46 | ] 47 | } 48 | ], 49 | "metadata": { 50 | "kernelspec": { 51 | "display_name": "Python 3", 52 | "language": "python", 53 | "name": "python3" 54 | }, 55 | "language_info": { 56 | "codemirror_mode": { 57 | "name": "ipython", 58 | "version": 3 59 | }, 60 | "file_extension": ".py", 61 | "mimetype": "text/x-python", 62 | "name": "python", 63 | "nbconvert_exporter": "python", 64 | "pygments_lexer": "ipython3", 65 | "version": "3.6.7" 66 | } 67 | }, 68 | "nbformat": 4, 69 | "nbformat_minor": 2 70 | } 71 | -------------------------------------------------------------------------------- /jupyter/01_CreateCorpus_German_Political_Speeches.ipynb: -------------------------------------------------------------------------------- 1 | { 2 | "cells": [ 3 | { 4 | "cell_type": "markdown", 5 | "metadata": {}, 6 | "source": [ 7 | "# Creating a Corpus for German Political Speeches\n", 8 | "\n", 9 | "The vector space model (VSM) is a common representation for documents in order to perfom clustering, topic modeling, classification, similarity search etc.\n", 10 | "\n", 11 | "In this case, we want to represent documents as bag-of-words. Therefore, the documents in the input text files (CSV, one document per line) are tokenized and converted into a corpus of indexed terms/tokens.\n", 12 | "\n", 13 | "### Pre-processing Steps\n", 14 | "\n", 15 | "The input data is usually messy. But instead of extensive pre-processing, e.g. cleaning of markup, punctuation, etc., we will simply extract all alphabetic sequences as tokens and nomalize them including following steps:\n", 16 | "\n", 17 | " * convert to lower case\n", 18 | " * remove stopwords\n", 19 | " * create n-grams\n", 20 | " * (stemming - not so useful for interpreting topics)\n", 21 | " * **`TODO:`** lemmatization (more complicated for german)\n", 22 | "\n", 23 | "### References\n", 24 | "\n", 25 | "* [Tutorial on Corpora and Vector Spaces](https://radimrehurek.com/gensim/tut1.html) from [Gensim](https://radimrehurek.com/gensim/index.html).\n", 26 | "* [Tutorial on Topic Modeling with Gensim](https://www.machinelearningplus.com/nlp/topic-modeling-gensim-python/) from [Machine Learning Plus](https://www.machinelearningplus.com/)." 27 | ] 28 | }, 29 | { 30 | "cell_type": "markdown", 31 | "metadata": {}, 32 | "source": [ 33 | "## Prerequisites\n", 34 | "\n", 35 | "### Libraries and Constants" 36 | ] 37 | }, 38 | { 39 | "cell_type": "code", 40 | "execution_count": null, 41 | "metadata": {}, 42 | "outputs": [], 43 | "source": [ 44 | "from collections import defaultdict\n", 45 | "from pprint import pprint\n", 46 | "import pandas as pd\n", 47 | "import string\n", 48 | "import os\n", 49 | "import re\n", 50 | "import time\n", 51 | "\n", 52 | "# input files\n", 53 | "data_dir = '../data/'\n", 54 | "filename = data_dir + 'Bundesregierung.csv'\n", 55 | "\n", 56 | "# output files\n", 57 | "corpus_dir = '../corpus/'\n", 58 | "dict_filename = corpus_dir + 'gps_ngrams.dict'\n", 59 | "corpus_filename = corpus_dir + 'gps_ngrams_bow.mm'\n", 60 | "\n", 61 | "# ensure output directory exists\n", 62 | "if not os.path.exists(corpus_dir):\n", 63 | " os.makedirs(corpus_dir)" 64 | ] 65 | }, 66 | { 67 | "cell_type": "markdown", 68 | "metadata": {}, 69 | "source": [ 70 | "### Helper Functions" 71 | ] 72 | }, 73 | { 74 | "cell_type": "code", 75 | "execution_count": null, 76 | "metadata": {}, 77 | "outputs": [], 78 | "source": [ 79 | "def print_diff(start_time):\n", 80 | " print(\"--- took %d:%.2d minutes ---\" % divmod(time.time() - start_time, 60))\n", 81 | "\n", 82 | "def most_frequent(tokens, topn=10):\n", 83 | " frequency = defaultdict(int)\n", 84 | " for doc in tokens:\n", 85 | " for term in doc:\n", 86 | " frequency[term] += 1\n", 87 | " return sorted(frequency.items(), key=lambda t: t[1], reverse=True)[0:topn]" 88 | ] 89 | }, 90 | { 91 | "cell_type": "markdown", 92 | "metadata": {}, 93 | "source": [ 94 | "## Read the Documents" 95 | ] 96 | }, 97 | { 98 | "cell_type": "code", 99 | "execution_count": null, 100 | "metadata": {}, 101 | "outputs": [], 102 | "source": [ 103 | "start_time = time.time()\n", 104 | "df = pd.read_csv(filename)\n", 105 | "print_diff(start_time)\n", 106 | "\n", 107 | "print(len(df), 'documents imported')\n", 108 | "df.head()" 109 | ] 110 | }, 111 | { 112 | "cell_type": "markdown", 113 | "metadata": {}, 114 | "source": [ 115 | "## Analyse Compound Words\n", 116 | "\n", 117 | "We want to recognize these compound words later in the n-gram detection." 118 | ] 119 | }, 120 | { 121 | "cell_type": "code", 122 | "execution_count": null, 123 | "metadata": {}, 124 | "outputs": [], 125 | "source": [ 126 | "PAT_COMPOUND = re.compile(r'\\w+[-]\\w+')\n", 127 | "\n", 128 | "compounds = df['text'].apply(lambda doc: [match.group() for match in PAT_COMPOUND.finditer(doc)])\n", 129 | "pprint(most_frequent(compounds, 10))" 130 | ] 131 | }, 132 | { 133 | "cell_type": "markdown", 134 | "metadata": {}, 135 | "source": [ 136 | "### Tokenize" 137 | ] 138 | }, 139 | { 140 | "cell_type": "code", 141 | "execution_count": null, 142 | "metadata": {}, 143 | "outputs": [], 144 | "source": [ 145 | "from gensim.utils import tokenize\n", 146 | "\n", 147 | "def tokens(documents):\n", 148 | " \"\"\"Convert all documents into a list of lowercase tokens using Gensim's tokenize() function.\"\"\"\n", 149 | " return [tokenize(doc, lower=True) for doc in documents]\n", 150 | "\n", 151 | "# explicit tokenization\n", 152 | "start_time = time.time()\n", 153 | "tokens = [[t for t in tokenize(doc, lower=True)] for doc in df['text']]\n", 154 | "print_diff(start_time)\n", 155 | "\n", 156 | "pprint(\" \".join(tokens[0][0:100]))" 157 | ] 158 | }, 159 | { 160 | "cell_type": "markdown", 161 | "metadata": {}, 162 | "source": [ 163 | "## Create N-Gram Model" 164 | ] 165 | }, 166 | { 167 | "cell_type": "code", 168 | "execution_count": null, 169 | "metadata": {}, 170 | "outputs": [], 171 | "source": [ 172 | "from gensim.models.phrases import Phrases, Phraser\n", 173 | "\n", 174 | "print('Building bigrams...')\n", 175 | "start_time = time.time()\n", 176 | "bigram_model = Phrases(tokens, min_count=1, threshold=100)\n", 177 | "print_diff(start_time)\n", 178 | "\n", 179 | "print(bigram_model)\n", 180 | "\n", 181 | "print('Building trigrams...')\n", 182 | "start_time = time.time()\n", 183 | "bigrams = list(bigram_model[tokens])\n", 184 | "trigram_model = Phrases(bigrams, min_count=1, threshold=100)\n", 185 | "print_diff(start_time)\n", 186 | "\n", 187 | "print(trigram_model)\n", 188 | "\n", 189 | "print('Optimizing bigram/trigram models...')\n", 190 | "# optimize bigram, trigram models\n", 191 | "start_time = time.time()\n", 192 | "bigram_model = Phraser(bigram_model)\n", 193 | "trigram_model = Phraser(trigram_model)\n", 194 | "print_diff(start_time)" 195 | ] 196 | }, 197 | { 198 | "cell_type": "code", 199 | "execution_count": null, 200 | "metadata": {}, 201 | "outputs": [], 202 | "source": [ 203 | "def n_grams(documents):\n", 204 | " return trigram_model[list(bigram_model[documents])]" 205 | ] 206 | }, 207 | { 208 | "cell_type": "code", 209 | "execution_count": null, 210 | "metadata": {}, 211 | "outputs": [], 212 | "source": [ 213 | "# most frequent n-grams\n", 214 | "start_time = time.time()\n", 215 | "pprint(most_frequent([[word for word in doc if '_' in word] for doc in n_grams(tokens)]))\n", 216 | "print_diff(start_time)" 217 | ] 218 | }, 219 | { 220 | "cell_type": "markdown", 221 | "metadata": {}, 222 | "source": [ 223 | "## Load Stopwords" 224 | ] 225 | }, 226 | { 227 | "cell_type": "code", 228 | "execution_count": null, 229 | "metadata": {}, 230 | "outputs": [], 231 | "source": [ 232 | "stopwords_filename = '../data/stopwords-de.txt'\n", 233 | "\n", 234 | "with open(stopwords_filename) as f:\n", 235 | " stopwords = [line for line in f.read().splitlines() if not line.startswith(';')]" 236 | ] 237 | }, 238 | { 239 | "cell_type": "code", 240 | "execution_count": null, 241 | "metadata": {}, 242 | "outputs": [], 243 | "source": [ 244 | "# add more stopwords\n", 245 | "more_stopwords = 'anbelangt dingen genauso gerne hierzu hinzu liebe nahezu nunmehr punkt relativ sodass sozusagen trotz übrigen vielfach vielfache vielmehr voraussichtlich wahrlich wahrscheinlich zuvor'\n", 246 | "stopwords.extend(more_stopwords.split())\n", 247 | "\n", 248 | "# use dictionary for better performance\n", 249 | "stopwordsdict = dict.fromkeys(stopwords, 1)\n", 250 | "\n", 251 | "print(len(stopwordsdict), \"stopwords\")" 252 | ] 253 | }, 254 | { 255 | "cell_type": "code", 256 | "execution_count": null, 257 | "metadata": {}, 258 | "outputs": [], 259 | "source": [ 260 | "def remove_stopwords(tokens):\n", 261 | " return [[word for word in doc if word not in stopwordsdict] for doc in tokens]" 262 | ] 263 | }, 264 | { 265 | "cell_type": "markdown", 266 | "metadata": {}, 267 | "source": [ 268 | "## Stemming" 269 | ] 270 | }, 271 | { 272 | "cell_type": "code", 273 | "execution_count": null, 274 | "metadata": {}, 275 | "outputs": [], 276 | "source": [ 277 | "from nltk.stem.cistem import Cistem\n", 278 | "\n", 279 | "stemmer = Cistem(True)\n", 280 | "\n", 281 | "# even do stemming on each part of the n-grams\n", 282 | "def stemming(tokens):\n", 283 | " return [['_'.join([stemmer.stem(part) for part in word.split('_')]) for word in doc] for doc in tokens]" 284 | ] 285 | }, 286 | { 287 | "cell_type": "markdown", 288 | "metadata": {}, 289 | "source": [ 290 | "## Lemmatization\n", 291 | "\n", 292 | "This is more complicated for German than English as there are fewer good algorithms available. Usually, the results are less accurate.\n", 293 | "\n", 294 | "Note: Lemmatization requires that the text has been POS tagged.\n", 295 | "\n", 296 | " * nltk\n", 297 | " * spacy\n", 298 | " * pattern.de\n", 299 | " * stanford NLP\n", 300 | "\n", 301 | "See also:\n", 302 | "\n", 303 | " * https://datascience.blog.wzb.eu/2017/05/19/lemmatization-of-german-language-text/\n", 304 | " * https://github.com/WZBSocialScienceCenter/germalemma\n", 305 | " " 306 | ] 307 | }, 308 | { 309 | "cell_type": "markdown", 310 | "metadata": {}, 311 | "source": [ 312 | "## All Pre-processing together" 313 | ] 314 | }, 315 | { 316 | "cell_type": "code", 317 | "execution_count": null, 318 | "metadata": {}, 319 | "outputs": [], 320 | "source": [ 321 | "tokens_nostop = remove_stopwords(tokens)\n", 322 | "tokens_ngram = n_grams(tokens_nostop)\n", 323 | "tokens_stem = stemming(tokens_ngram)\n", 324 | "\n", 325 | "texts = tokens_ngram" 326 | ] 327 | }, 328 | { 329 | "cell_type": "code", 330 | "execution_count": null, 331 | "metadata": {}, 332 | "outputs": [], 333 | "source": [ 334 | "print(tokens_nostop[1][0:5])\n", 335 | "print(tokens_ngram[1][0:5])\n", 336 | "print(tokens_stem[1][0:5])" 337 | ] 338 | }, 339 | { 340 | "cell_type": "code", 341 | "execution_count": null, 342 | "metadata": {}, 343 | "outputs": [], 344 | "source": [ 345 | "# most frequent compound words after pre-processing\n", 346 | "ngrams = [[word for word in doc if '_' in word] for doc in tokens_ngram]\n", 347 | "stems = [[word for word in doc if '_' in word] for doc in tokens_stem]\n", 348 | "\n", 349 | "pprint(most_frequent(ngrams, 10))\n", 350 | "pprint(most_frequent(stems, 10))" 351 | ] 352 | }, 353 | { 354 | "cell_type": "markdown", 355 | "metadata": {}, 356 | "source": [ 357 | "# Create Dictionary" 358 | ] 359 | }, 360 | { 361 | "cell_type": "code", 362 | "execution_count": null, 363 | "metadata": {}, 364 | "outputs": [], 365 | "source": [ 366 | "from gensim import corpora\n", 367 | "\n", 368 | "print('Creating Dictionary...')\n", 369 | "\n", 370 | "start_time = time.time()\n", 371 | "dictionary = corpora.Dictionary(texts)\n", 372 | "print_diff(start_time)\n", 373 | "\n", 374 | "print(dictionary)" 375 | ] 376 | }, 377 | { 378 | "cell_type": "code", 379 | "execution_count": null, 380 | "metadata": {}, 381 | "outputs": [], 382 | "source": [ 383 | "dfs_desc = sorted(dictionary.dfs.items(), key=lambda t: t[1], reverse=True)\n", 384 | "\n", 385 | "print('--- Most Frequent Tokens in X Documents', dictionary.num_docs, 'Documents ---')\n", 386 | "for (k,v) in dfs_desc[0:10]: print('{freq}: {token}'.format(token=dictionary[k], freq=v))\n", 387 | "\n", 388 | "print('--- Least Frequent Tokens in X Documents', dictionary.num_docs, 'Documents ---')\n", 389 | "for (k,v) in dfs_desc[-10:]: print('{freq}: {token}'.format(token=dictionary[k], freq=v))" 390 | ] 391 | }, 392 | { 393 | "cell_type": "markdown", 394 | "metadata": {}, 395 | "source": [ 396 | "### Filter extreme tokens\n", 397 | "\n", 398 | "* tokens which occur in more than 30% of all documents.\n", 399 | "* tokens which occur in less than 5 documents." 400 | ] 401 | }, 402 | { 403 | "cell_type": "code", 404 | "execution_count": null, 405 | "metadata": {}, 406 | "outputs": [], 407 | "source": [ 408 | "print('Filtering extreme tokens')\n", 409 | "freq_before = len(dictionary)\n", 410 | "dictionary.filter_extremes(no_below=5, no_above=0.3)\n", 411 | "print('{} token before -> {} after'.format(freq_before, len(dictionary)))" 412 | ] 413 | }, 414 | { 415 | "cell_type": "code", 416 | "execution_count": null, 417 | "metadata": {}, 418 | "outputs": [], 419 | "source": [ 420 | "dfs_desc = sorted(dictionary.dfs.items(), key=lambda t: t[1], reverse=True)\n", 421 | "\n", 422 | "print('--- Most Frequent Token Occurrences in', dictionary.num_docs, 'Documents ---')\n", 423 | "for (k,v) in dfs_desc[0:10]: print('{freq}: {token}'.format(token=dictionary[k], freq=v))\n", 424 | "\n", 425 | "print('--- Least Frequent Token Occurrences in', dictionary.num_docs, 'Documents ---')\n", 426 | "for (k,v) in dfs_desc[-10:]: print('{freq}: {token}'.format(token=dictionary[k], freq=v))" 427 | ] 428 | }, 429 | { 430 | "cell_type": "markdown", 431 | "metadata": {}, 432 | "source": [ 433 | "### Saving Dictionary" 434 | ] 435 | }, 436 | { 437 | "cell_type": "code", 438 | "execution_count": null, 439 | "metadata": {}, 440 | "outputs": [], 441 | "source": [ 442 | "print('Saving Dictionary to', dict_filename)\n", 443 | "start_time = time.time()\n", 444 | "dictionary.save(dict_filename)\n", 445 | "print_diff(start_time)" 446 | ] 447 | }, 448 | { 449 | "cell_type": "markdown", 450 | "metadata": {}, 451 | "source": [ 452 | "# Create Corpus\n", 453 | "\n", 454 | "Use the n-gram tokens to construct the model." 455 | ] 456 | }, 457 | { 458 | "cell_type": "code", 459 | "execution_count": null, 460 | "metadata": {}, 461 | "outputs": [], 462 | "source": [ 463 | "print('Creating Corpus')\n", 464 | "start_time = time.time()\n", 465 | "corpus_bow = [dictionary.doc2bow(doc) for doc in texts]\n", 466 | "print_diff(start_time)" 467 | ] 468 | }, 469 | { 470 | "cell_type": "code", 471 | "execution_count": null, 472 | "metadata": {}, 473 | "outputs": [], 474 | "source": [ 475 | "print('Saving Corpus to', corpus_filename)\n", 476 | "start_time = time.time()\n", 477 | "corpora.MmCorpus.serialize(corpus_filename, corpus_bow)\n", 478 | "print_diff(start_time)" 479 | ] 480 | }, 481 | { 482 | "cell_type": "markdown", 483 | "metadata": {}, 484 | "source": [ 485 | "## LDA Model" 486 | ] 487 | }, 488 | { 489 | "cell_type": "code", 490 | "execution_count": null, 491 | "metadata": {}, 492 | "outputs": [], 493 | "source": [ 494 | "from gensim.models import LdaModel, CoherenceModel\n", 495 | "\n", 496 | "num_topics = 150\n", 497 | "\n", 498 | "start_time = time.time()\n", 499 | "model_lda = LdaModel(corpus_bow, id2word=dictionary, num_topics=num_topics)\n", 500 | "print_diff(start_time)\n", 501 | "print(model_lda)\n", 502 | "\n", 503 | "model_lda.save('../model/{}_topics/'.format(num_topics) + 'topic_model.lda')" 504 | ] 505 | }, 506 | { 507 | "cell_type": "markdown", 508 | "metadata": {}, 509 | "source": [ 510 | "### Compute Coherence\n", 511 | "\n", 512 | "Remove words from texts which are not in dictionary" 513 | ] 514 | }, 515 | { 516 | "cell_type": "code", 517 | "execution_count": null, 518 | "metadata": {}, 519 | "outputs": [], 520 | "source": [ 521 | "texts = [[token for token in text if token in dictionary.token2id] for text in texts]" 522 | ] 523 | }, 524 | { 525 | "cell_type": "code", 526 | "execution_count": null, 527 | "metadata": {}, 528 | "outputs": [], 529 | "source": [ 530 | "start_time = time.time()\n", 531 | "cm = CoherenceModel(model=model_lda, corpus=corpus_bow, dictionary=dictionary, coherence='u_mass')\n", 532 | "print('u_mass: {:0.3f}'.format(cm.get_coherence()))\n", 533 | "print_diff(start_time)\n", 534 | "\n", 535 | "start_time = time.time()\n", 536 | "cm = CoherenceModel(texts=texts, model=model_lda, corpus=corpus_bow, dictionary=dictionary, coherence='c_v')\n", 537 | "print('c_v: {:0.3f}'.format(cm.get_coherence()))\n", 538 | "print_diff(start_time)" 539 | ] 540 | } 541 | ], 542 | "metadata": { 543 | "kernelspec": { 544 | "display_name": "Python 3", 545 | "language": "python", 546 | "name": "python3" 547 | }, 548 | "language_info": { 549 | "codemirror_mode": { 550 | "name": "ipython", 551 | "version": 3 552 | }, 553 | "file_extension": ".py", 554 | "mimetype": "text/x-python", 555 | "name": "python", 556 | "nbconvert_exporter": "python", 557 | "pygments_lexer": "ipython3", 558 | "version": "3.6.7" 559 | } 560 | }, 561 | "nbformat": 4, 562 | "nbformat_minor": 2 563 | } 564 | -------------------------------------------------------------------------------- /jupyter/02_LearnModel_LDA_GPS.ipynb: -------------------------------------------------------------------------------- 1 | { 2 | "cells": [ 3 | { 4 | "cell_type": "markdown", 5 | "metadata": {}, 6 | "source": [ 7 | "# Topic Modeling with LDA (Gensim)" 8 | ] 9 | }, 10 | { 11 | "cell_type": "markdown", 12 | "metadata": {}, 13 | "source": [ 14 | "## Prerequisites" 15 | ] 16 | }, 17 | { 18 | "cell_type": "code", 19 | "execution_count": null, 20 | "metadata": {}, 21 | "outputs": [], 22 | "source": [ 23 | "import pandas as pd\n", 24 | "import string\n", 25 | "import os\n", 26 | "import time\n", 27 | "\n", 28 | "mallet_path = '../lib/mallet-2.0.8/bin/mallet'\n", 29 | "use_mallet = False\n", 30 | "\n", 31 | "# input files\n", 32 | "corpus_dir = '../corpus/'\n", 33 | "dict_filename = corpus_dir + 'gps_ngrams.dict'\n", 34 | "corpus_filename = corpus_dir + 'gps_ngrams_bow.mm'\n", 35 | "\n", 36 | "# output files\n", 37 | "model_dir = '../model/{}_topics/'\n", 38 | "model_filename = 'topic_model.lda'" 39 | ] 40 | }, 41 | { 42 | "cell_type": "markdown", 43 | "metadata": {}, 44 | "source": [ 45 | "## Load Dictionary and Corpus" 46 | ] 47 | }, 48 | { 49 | "cell_type": "code", 50 | "execution_count": null, 51 | "metadata": {}, 52 | "outputs": [], 53 | "source": [ 54 | "from gensim import corpora, models\n", 55 | "\n", 56 | "dictionary = corpora.Dictionary.load(dict_filename)\n", 57 | "print(dictionary)\n", 58 | "\n", 59 | "corpus_bow = corpora.MmCorpus(corpus_filename)\n", 60 | "print(corpus_bow)" 61 | ] 62 | }, 63 | { 64 | "cell_type": "markdown", 65 | "metadata": {}, 66 | "source": [ 67 | "## Model Computations" 68 | ] 69 | }, 70 | { 71 | "cell_type": "code", 72 | "execution_count": null, 73 | "metadata": {}, 74 | "outputs": [], 75 | "source": [ 76 | "num_topics = 150" 77 | ] 78 | }, 79 | { 80 | "cell_type": "code", 81 | "execution_count": null, 82 | "metadata": {}, 83 | "outputs": [], 84 | "source": [ 85 | "def compute_lda(corpus, id2word, num_topics):\n", 86 | " \n", 87 | " print('Computing LDA for', num_topics, 'topics...')\n", 88 | " \n", 89 | " start_time = time.time()\n", 90 | " model_lda = models.LdaModel(corpus, id2word=id2word, num_topics=num_topics)\n", 91 | " print(\"--- took %d:%.2d minutes ---\" % divmod(time.time() - start_time, 60))\n", 92 | " \n", 93 | " return model_lda\n", 94 | "\n", 95 | "def compute_lda_mallet(corpus, id2word, num_topics):\n", 96 | " \n", 97 | " print('Computing LDA (Mallet) for', num_topics, 'topics...')\n", 98 | " \n", 99 | " start_time = time.time()\n", 100 | " model_lda_mallet = models.wrappers.LdaMallet(mallet_path, corpus=corpus_bow, num_topics=num_topics, id2word=dictionary)\n", 101 | " print(\"--- took %d:%.2d minutes ---\" % divmod(time.time() - start_time, 60))\n", 102 | " \n", 103 | " return model_lda_mallet\n", 104 | "\n", 105 | "def save_model(model_lda, num_topics):\n", 106 | " pathname = model_dir.format(num_topics)\n", 107 | " \n", 108 | " # ensure directories exist\n", 109 | " if not os.path.exists(pathname):\n", 110 | " os.makedirs(pathname)\n", 111 | " \n", 112 | " filename = pathname + model_filename\n", 113 | " \n", 114 | " print('Saving LDA model to ', filename)\n", 115 | " model_lda.save(filename)\n", 116 | " return\n", 117 | "\n", 118 | "def load_model(num_topics):\n", 119 | " filename = model_dir.format(num_topics) + model_filename\n", 120 | " print('Loading LDA model from', filename)\n", 121 | " return models.ldamodel.LdaModel.load(filename)\n", 122 | "\n", 123 | "def get_model(corpus, id2word, num_topics):\n", 124 | " try:\n", 125 | " model_lda = load_model(num_topics)\n", 126 | " except IOError as e:\n", 127 | " errno, strerror = e.args\n", 128 | " print(\"I/O error({0}): {1}\".format(errno,strerror))\n", 129 | " model_lda = compute_lda(corpus, id2word, num_topics)\n", 130 | " save_model(model_lda, num_topics)\n", 131 | " \n", 132 | " print(model_lda)\n", 133 | " return model_lda\n", 134 | "\n", 135 | "def compute_coherence(model, corpus, id2word, num_topics, coherence='u_mass'):\n", 136 | " cm = models.CoherenceModel(model=model, corpus=corpus, dictionary=id2word, coherence=coherence)\n", 137 | " return cm.get_coherence()\n", 138 | "\n", 139 | "def create_models(corpus, dictionary, num_topics_list):\n", 140 | " \n", 141 | " coherence_vals = []\n", 142 | " \n", 143 | " for num_topics in num_topics_list:\n", 144 | " \n", 145 | " model_lda = get_model(corpus, dictionary, num_topics)\n", 146 | " coh = compute_coherence(model_lda, corpus, dictionary, num_topics)\n", 147 | " \n", 148 | " coherence_vals.append((num_topics, coh))\n", 149 | " \n", 150 | " return coherence_vals" 151 | ] 152 | }, 153 | { 154 | "cell_type": "markdown", 155 | "metadata": {}, 156 | "source": [ 157 | "## Compute LDA" 158 | ] 159 | }, 160 | { 161 | "cell_type": "code", 162 | "execution_count": null, 163 | "metadata": {}, 164 | "outputs": [], 165 | "source": [ 166 | "model_lda = get_model(corpus_bow, dictionary, num_topics)\n", 167 | "coherence = compute_coherence(model_lda, corpus_bow, dictionary, num_topics)\n", 168 | "print(\"coherence: {:.3f}\".format(coherence))" 169 | ] 170 | }, 171 | { 172 | "cell_type": "code", 173 | "execution_count": null, 174 | "metadata": {}, 175 | "outputs": [], 176 | "source": [ 177 | "# model_lda.print_topics(10)\n", 178 | "# model_lda.top_topics(corpus_bow)" 179 | ] 180 | }, 181 | { 182 | "cell_type": "markdown", 183 | "metadata": {}, 184 | "source": [ 185 | "### Using Mallet" 186 | ] 187 | }, 188 | { 189 | "cell_type": "code", 190 | "execution_count": null, 191 | "metadata": {}, 192 | "outputs": [], 193 | "source": [ 194 | "if use_mallet:\n", 195 | " num_topics = 10\n", 196 | " mallet = compute_lda_mallet(corpus_bow, dictionary, num_topics)\n", 197 | " compute_coherence(mallet, corpus_bow, dictionary, num_topics)" 198 | ] 199 | }, 200 | { 201 | "cell_type": "markdown", 202 | "metadata": {}, 203 | "source": [ 204 | "## Calculate coherence for multiple models\n", 205 | "\n", 206 | " see also:\n", 207 | " * https://radimrehurek.com/gensim/models/coherencemodel.html\n", 208 | " * https://rare-technologies.com/what-is-topic-coherence/" 209 | ] 210 | }, 211 | { 212 | "cell_type": "code", 213 | "execution_count": null, 214 | "metadata": {}, 215 | "outputs": [], 216 | "source": [ 217 | "#coherence_vals = create_models(corpus_bow, dictionary, range(20, 501, 3))" 218 | ] 219 | }, 220 | { 221 | "cell_type": "code", 222 | "execution_count": null, 223 | "metadata": {}, 224 | "outputs": [], 225 | "source": [ 226 | "import matplotlib.pyplot as plt\n", 227 | "%matplotlib inline\n", 228 | "\n", 229 | "plt.plot(*zip(*coherence_vals))\n", 230 | "plt.xlabel(\"Num Topics\")\n", 231 | "plt.ylabel(\"Coherence Score\")\n", 232 | "#plt.legend((\"Coherence_values\"), loc='best')\n", 233 | "plt.show()" 234 | ] 235 | } 236 | ], 237 | "metadata": { 238 | "kernelspec": { 239 | "display_name": "Python 3", 240 | "language": "python", 241 | "name": "python3" 242 | }, 243 | "language_info": { 244 | "codemirror_mode": { 245 | "name": "ipython", 246 | "version": 3 247 | }, 248 | "file_extension": ".py", 249 | "mimetype": "text/x-python", 250 | "name": "python", 251 | "nbconvert_exporter": "python", 252 | "pygments_lexer": "ipython3", 253 | "version": "3.6.7" 254 | } 255 | }, 256 | "nbformat": 4, 257 | "nbformat_minor": 2 258 | } 259 | -------------------------------------------------------------------------------- /jupyter/05_SimilaritySearch.ipynb: -------------------------------------------------------------------------------- 1 | { 2 | "cells": [ 3 | { 4 | "cell_type": "markdown", 5 | "metadata": {}, 6 | "source": [ 7 | "# Similarity Search\n", 8 | "\n", 9 | "## Issues\n", 10 | "\n", 11 | " * [LDA returns only one topic probability](https://github.com/RaRe-Technologies/gensim/issues/1260)" 12 | ] 13 | } 14 | ], 15 | "metadata": { 16 | "kernelspec": { 17 | "display_name": "Python 3", 18 | "language": "python", 19 | "name": "python3" 20 | }, 21 | "language_info": { 22 | "codemirror_mode": { 23 | "name": "ipython", 24 | "version": 3 25 | }, 26 | "file_extension": ".py", 27 | "mimetype": "text/x-python", 28 | "name": "python", 29 | "nbconvert_exporter": "python", 30 | "pygments_lexer": "ipython3", 31 | "version": "3.6.7" 32 | } 33 | }, 34 | "nbformat": 4, 35 | "nbformat_minor": 2 36 | } 37 | -------------------------------------------------------------------------------- /jupyter/99_test_bigrams.ipynb: -------------------------------------------------------------------------------- 1 | { 2 | "cells": [ 3 | { 4 | "cell_type": "code", 5 | "execution_count": 40, 6 | "metadata": {}, 7 | "outputs": [ 8 | { 9 | "name": "stdout", 10 | "output_type": "stream", 11 | "text": [ 12 | "[['mayor', 'new_york'],\n", 13 | " ['machin', 'learn', 'us'],\n", 14 | " ['new_york', 'mayor', 'present']]\n" 15 | ] 16 | } 17 | ], 18 | "source": [ 19 | "documents = [\"the mayor of new york was there\",\n", 20 | " \"machine learning can be useful sometimes\",\n", 21 | " \"new york mayor was present\"]\n", 22 | "\n", 23 | "import gensim, pprint\n", 24 | "\n", 25 | "# tokenize documents with gensim's tokenize() function\n", 26 | "tokens = [list(gensim.utils.tokenize(doc, lower=True)) for doc in documents]\n", 27 | "\n", 28 | "# build bigram model\n", 29 | "bigram_mdl = gensim.models.phrases.Phrases(tokens, min_count=1, threshold=2)\n", 30 | "\n", 31 | "# do more pre-processing on tokens (remove stopwords, stemming etc.)\n", 32 | "# NOTE: this can be done better\n", 33 | "from gensim.parsing.preprocessing import preprocess_string, stem_text, remove_stopwords\n", 34 | "CUSTOM_FILTERS = [remove_stopwords, stem_text]\n", 35 | "tokens = [preprocess_string(\" \".join(doc), CUSTOM_FILTERS) for doc in tokens]\n", 36 | "\n", 37 | "# convert tokens to bigrams\n", 38 | "bigrams = bigram_mdl[tokens]\n", 39 | "\n", 40 | "pprint.pprint(list(bigrams))" 41 | ] 42 | } 43 | ], 44 | "metadata": { 45 | "kernelspec": { 46 | "display_name": "Python 3", 47 | "language": "python", 48 | "name": "python3" 49 | }, 50 | "language_info": { 51 | "codemirror_mode": { 52 | "name": "ipython", 53 | "version": 3 54 | }, 55 | "file_extension": ".py", 56 | "mimetype": "text/x-python", 57 | "name": "python", 58 | "nbconvert_exporter": "python", 59 | "pygments_lexer": "ipython3", 60 | "version": "3.6.7" 61 | } 62 | }, 63 | "nbformat": 4, 64 | "nbformat_minor": 2 65 | } 66 | -------------------------------------------------------------------------------- /requirements.txt: -------------------------------------------------------------------------------- 1 | wheel 2 | pip 3 | pandas 4 | jupyter 5 | jupyterlab 6 | gensim 7 | lda 8 | nltk 9 | spacy 10 | sklearn 11 | matplotlib 12 | bokeh 13 | pyLDAvis 14 | pattern 15 | tmtoolkit 16 | bs4 17 | 18 | -------------------------------------------------------------------------------- /start-jupyter.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | 3 | # create virtual environment 4 | # using venv instead of virtualenv 5 | python3 -m venv .venv 6 | source .venv/bin/activate 7 | 8 | pip install -U -r requirements.txt 9 | 10 | # jupyter notebook 11 | jupyter lab 12 | --------------------------------------------------------------------------------