├── README.md
└── TurkishQuestionPairs.ipynb


/README.md:
--------------------------------------------------------------------------------
 1 | # Question Paraphrases Dataset For Turkish Language
 2 | This data set contains 1377 positive question paraphrases for Turkish Languages and 2187 negative examples. 
 3 | The original questions are taken from Istanbul Bilgi Universitiy Frequently Asking Questions resource 
 4 | https://www.bilgi.edu.tr/tr/yasam/ogrenci/sss/. Later,we manually created two additional truly semantically equivalent questions. These quesitons pairs are considered postivie example of the quesiton paraphrasing. 
 5 | 
 6 | ## Negative Sampling:
 7 | We also supplemented 2187 negative examples labeled as 0. The source of negative examples are be pairs of "related questions" which, although pertaining to similar topics, are not truly semantically equivalent. It would be very much like Quora Question Pairs
 8 | https://www.quora.com/q/quoradata/First-Quora-Dataset-Release-Question-Pairs
 9 | 
10 | 
11 | ## Paraphrasing Detection
12 | To do classify wheter two questions are duplicate, we trained a simple classification algorithm based on BERT. It would be very helpfull especially for ChatBot framework in Turkish Language.
13 | 
14 | Following code shows how to run the detection model 
15 | 
16 | 
17 | ```
18 | from transformers import *
19 | import torch
20 | from transformers import AutoTokenizer, AutoModelForSequenceClassification
21 | tokenizer = AutoTokenizer.from_pretrained("savasy/TurkQP")
22 | model = AutoModelForSequenceClassification.from_pretrained("savasy/TurkQP")
23 | 
24 | 
25 | s0="Bugün yağmur yağacak mı ?"
26 | s1="Yağmur yağabilir mi bugün ?"
27 | inputs= tokenizer(s0, s1, add_special_tokens=True, return_tensors='pt')
28 | output= pytorch_model(inputs['input_ids'], token_type_ids=inputs['token_type_ids'])[0]
29 | pred = output.argmax().item()
30 | 
31 | pred==1
32 | 
33 | ```
34 | 
35 | 


--------------------------------------------------------------------------------
/TurkishQuestionPairs.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "code",
  5 |    "execution_count": 62,
  6 |    "metadata": {},
  7 |    "outputs": [],
  8 |    "source": [
  9 |     "import pandas as pd"
 10 |    ]
 11 |   },
 12 |   {
 13 |    "cell_type": "code",
 14 |    "execution_count": 63,
 15 |    "metadata": {},
 16 |    "outputs": [],
 17 |    "source": [
 18 |     "d=pd.read_csv(\"TurkQP.csv\")"
 19 |    ]
 20 |   },
 21 |   {
 22 |    "cell_type": "code",
 23 |    "execution_count": 64,
 24 |    "metadata": {},
 25 |    "outputs": [
 26 |     {
 27 |      "data": {
 28 |       "text/html": [
 29 |        "<div>\n",
 30 |        "<style scoped>\n",
 31 |        "    .dataframe tbody tr th:only-of-type {\n",
 32 |        "        vertical-align: middle;\n",
 33 |        "    }\n",
 34 |        "\n",
 35 |        "    .dataframe tbody tr th {\n",
 36 |        "        vertical-align: top;\n",
 37 |        "    }\n",
 38 |        "\n",
 39 |        "    .dataframe thead th {\n",
 40 |        "        text-align: right;\n",
 41 |        "    }\n",
 42 |        "</style>\n",
 43 |        "<table border=\"1\" class=\"dataframe\">\n",
 44 |        "  <thead>\n",
 45 |        "    <tr style=\"text-align: right;\">\n",
 46 |        "      <th></th>\n",
 47 |        "      <th>q1</th>\n",
 48 |        "      <th>q2</th>\n",
 49 |        "      <th>duplicate</th>\n",
 50 |        "    </tr>\n",
 51 |        "  </thead>\n",
 52 |        "  <tbody>\n",
 53 |        "    <tr>\n",
 54 |        "      <th>774</th>\n",
 55 |        "      <td>Öğrenci kulüplerine sağlanan imkanlar nelerdir?</td>\n",
 56 |        "      <td>öğrenci kulüplerine ne tip imkanlar sunuluyor</td>\n",
 57 |        "      <td>1</td>\n",
 58 |        "    </tr>\n",
 59 |        "    <tr>\n",
 60 |        "      <th>2370</th>\n",
 61 |        "      <td>Diplomamı kaybettim  yeni diploma alabilir miyim?</td>\n",
 62 |        "      <td>Diplomamı benim yerime bir başkası alabilir mi?</td>\n",
 63 |        "      <td>0</td>\n",
 64 |        "    </tr>\n",
 65 |        "    <tr>\n",
 66 |        "      <th>470</th>\n",
 67 |        "      <td>ders kaydımda çakışma olduğunda ne yapmalıyım</td>\n",
 68 |        "      <td>Çakışan derslerim var ders kaydım onaylanır mı?</td>\n",
 69 |        "      <td>1</td>\n",
 70 |        "    </tr>\n",
 71 |        "    <tr>\n",
 72 |        "      <th>2403</th>\n",
 73 |        "      <td>Mezuniyet başarı sıralamamı öğrenebilir miyim?</td>\n",
 74 |        "      <td>Mezuniyet başarı sıralamamı öğrenebilir miyim?</td>\n",
 75 |        "      <td>0</td>\n",
 76 |        "    </tr>\n",
 77 |        "    <tr>\n",
 78 |        "      <th>1775</th>\n",
 79 |        "      <td>Hangi üniversitelerden yatay geçiş kabul ediyo...</td>\n",
 80 |        "      <td>Ana dal yan dal çift anadal programlarında der...</td>\n",
 81 |        "      <td>0</td>\n",
 82 |        "    </tr>\n",
 83 |        "  </tbody>\n",
 84 |        "</table>\n",
 85 |        "</div>"
 86 |       ],
 87 |       "text/plain": [
 88 |        "                                                     q1  \\\n",
 89 |        "774     Öğrenci kulüplerine sağlanan imkanlar nelerdir?   \n",
 90 |        "2370  Diplomamı kaybettim  yeni diploma alabilir miyim?   \n",
 91 |        "470       ders kaydımda çakışma olduğunda ne yapmalıyım   \n",
 92 |        "2403     Mezuniyet başarı sıralamamı öğrenebilir miyim?   \n",
 93 |        "1775  Hangi üniversitelerden yatay geçiş kabul ediyo...   \n",
 94 |        "\n",
 95 |        "                                                     q2  duplicate  \n",
 96 |        "774       öğrenci kulüplerine ne tip imkanlar sunuluyor          1  \n",
 97 |        "2370    Diplomamı benim yerime bir başkası alabilir mi?          0  \n",
 98 |        "470     Çakışan derslerim var ders kaydım onaylanır mı?          1  \n",
 99 |        "2403     Mezuniyet başarı sıralamamı öğrenebilir miyim?          0  \n",
100 |        "1775  Ana dal yan dal çift anadal programlarında der...          0  "
101 |       ]
102 |      },
103 |      "execution_count": 64,
104 |      "metadata": {},
105 |      "output_type": "execute_result"
106 |     }
107 |    ],
108 |    "source": [
109 |     "d.sample(frac=1).head()"
110 |    ]
111 |   },
112 |   {
113 |    "cell_type": "code",
114 |    "execution_count": 65,
115 |    "metadata": {},
116 |    "outputs": [
117 |     {
118 |      "data": {
119 |       "text/plain": [
120 |        "0    2187\n",
121 |        "1    1377\n",
122 |        "Name: duplicate, dtype: int64"
123 |       ]
124 |      },
125 |      "execution_count": 65,
126 |      "metadata": {},
127 |      "output_type": "execute_result"
128 |     }
129 |    ],
130 |    "source": [
131 |     "d.duplicate.value_counts()"
132 |    ]
133 |   },
134 |   {
135 |    "cell_type": "markdown",
136 |    "metadata": {},
137 |    "source": [
138 |     "* 0: is not duplicate\n",
139 |     "* 1: is duplicate    "
140 |    ]
141 |   },
142 |   {
143 |    "cell_type": "code",
144 |    "execution_count": null,
145 |    "metadata": {},
146 |    "outputs": [],
147 |    "source": []
148 |   },
149 |   {
150 |    "cell_type": "code",
151 |    "execution_count": null,
152 |    "metadata": {},
153 |    "outputs": [],
154 |    "source": []
155 |   }
156 |  ],
157 |  "metadata": {
158 |   "kernelspec": {
159 |    "display_name": "Python 3",
160 |    "language": "python",
161 |    "name": "python3"
162 |   },
163 |   "language_info": {
164 |    "codemirror_mode": {
165 |     "name": "ipython",
166 |     "version": 3
167 |    },
168 |    "file_extension": ".py",
169 |    "mimetype": "text/x-python",
170 |    "name": "python",
171 |    "nbconvert_exporter": "python",
172 |    "pygments_lexer": "ipython3",
173 |    "version": "3.7.6"
174 |   }
175 |  },
176 |  "nbformat": 4,
177 |  "nbformat_minor": 4
178 | }
179 | 


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