-
Notifications
You must be signed in to change notification settings - Fork 1
/
index.html
379 lines (346 loc) · 13.6 KB
/
index.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
<!DOCTYPE HTML>
<!--
Stellar by HTML5 UP
html5up.net | @ajlkn
Free for personal and commercial use under the CCA 3.0 license (html5up.net/license)
-->
<html>
<head>
<title>Stellar by HTML5 UP</title>
<meta charset="utf-8" />
<meta name="viewport" content="width=device-width, initial-scale=1, user-scalable=no" />
<link rel="stylesheet" href="assets/css/main.css" />
<noscript><link rel="stylesheet" href="assets/css/noscript.css" /></noscript>
</head>
<body class="is-preload">
<!-- Wrapper -->
<div id="wrapper">
<!-- Header -->
<header id="header" class="alt">
<h1>HatEval</h1>
<p>Classifying tweets as hateful/non-hateful, aggressive/non-aggressive, targeted/non-targeted <br />
along with indentifying the target as group or individual.</p>
<h5>Vaibhav Bajaj,  Sriven Reddy,  Kripa Anne,  Karthika Ramineni</h5>
</header>
<!-- Nav -->
<nav id="nav">
<ul>
<li><a href="#abstract" class="active">Abstract</a></li>
<li><a href="#methodology">Method</a></li>
<li><a href="#results">Results</a></li>
<li><a href="#video">Video Presentation</a></li>
</ul>
</nav>
<!-- Main -->
<div id="main">
<!-- Introduction -->
<section id="abstract" class="main">
<div class="spotlight">
<div class="content" style="text-align: justify;">
<header class="major">
<h2>Abstract</h2>
</header>
<p>The aim of this project is two-fold:
<ul>
<li>We initially classify hateful tweets from a corpus where hate speech against women or immigrants has been identified, as aggressive or not aggressive. </li>
<li>Additionally, we classify these hateful tweets to identify if the target being harassed is a generic group of people or a specific individual. </li>
</ul>
We employ machine learning methods like logistic regression and support vector classification with various hyperparameters to initially perform the above binary classification tasks. We have also used LSTM on word embeddings and CNN on character level.
</p>
</div>
</div>
</section>
<section id="methodology" class="main">
<div class="spotlight">
<div class="content" style="text-align: justify;">
<header class="major">
<h2>Methods Used</h2>
</header>
<p>
We have used the following methods to classigy tweets:
<ul>
<li>General preprocessing and cleaning of data through pattern matching that includes:</li>
<ul style="list-style-type:circle;">
<li>Converting to lowercase</li>
<li>Removing special characters that won’t contribute positively to the accuracy (like the ‘#’s used in hashtags and punctuation)</li>
<li>Removing stop words</li>
<li>Removing URLs</li>
<li>Removing mentions</li>
<li>Stemming words</li>
</ul>
<li>Tokenization using bag of words (sklearn vectorization).</li>
<li>Logistic Regression model</li>
<li>Support vector machines model</li>
<li>Word level embedding : 1-gram through 3-gram</li>
<li>Character level embedding : 2-gram through 6-gram</li>
<li>LSTM model on word embeddings</li>
<li>CNN model on char level embeddings</li>
</ul>
</p>
</div>
</div>
</section>
<section id="results" class="main">
<div class="spotlight">
<div class="content" style="text-align: justify;">
<header class="major">
<h2>Results</h2>
</header>
<h3>CNN on Target detection</h3>
<div class="table-wrapper">
<table class="alt">
<thead>
<tr>
<th>Metric</th>
<th>Score</th>
</tr>
</thead>
<tbody>
<tr>
<td>Precision</td>
<td>[0.68275862 0.9270073 ]</td>
</tr>
<tr>
<td>Recall</td>
<td>[0.95192308 0.57990868]</td>
</tr>
<tr>
<td>F1-Score</td>
<td>[0.79518072 0.71348315]</td>
</tr>
<tr>
<td>ROC-AUC</td>
<td>0.7659158763610818</td>
</tr>
</tbody>
</table>
</div>
<h3>CNN on Aggression detection</h3>
<div class="table-wrapper">
<table class="alt">
<thead>
<tr>
<th>Metric</th>
<th>Score</th>
</tr>
</thead>
<tbody>
<tr>
<td>Precision</td>
<td></td>
</tr>
<tr>
<td>Recall</td>
<td></td>
</tr>
<tr>
<td>F1-Score</td>
<td></td>
</tr>
<tr>
<td>ROC-AUC</td>
<td></td>
</tr>
</tbody>
</table>
</div>
<h3>LSTM on Target detection</h3>
<div class="table-wrapper">
<table class="alt">
<thead>
<tr>
<th>Metric</th>
<th>Score</th>
</tr>
</thead>
<tbody>
<tr>
<td>Precision</td>
<td></td>
</tr>
<tr>
<td>Recall</td>
<td></td>
</tr>
<tr>
<td>F1-Score</td>
<td></td>
</tr>
<tr>
<td>ROC-AUC</td>
<td></td>
</tr>
</tbody>
</table>
</div>
<h3>LSTM on Aggression detection</h3>
<div class="table-wrapper">
<table class="alt">
<thead>
<tr>
<th>Metric</th>
<th>Score</th>
</tr>
</thead>
<tbody>
<tr>
<td>Precision</td>
<td></td>
</tr>
<tr>
<td>Recall</td>
<td></td>
</tr>
<tr>
<td>F1-Score</td>
<td></td>
</tr>
<tr>
<td>ROC-AUC</td>
<td></td>
</tr>
</tbody>
</table>
</div>
<p>
While using LSTMs and CNNs on the hateful tweets (Phase 2),
A few samples did not fit the trained model and were wrongly classified.
LSTM: We found 645 correctly classified tweets, 188 false negatives and 167 false positives.
(Correctly classified as NOT hateful towards anyone)
eg: 'We must reject the "globalization of indifference" toward refugees, says the president of the International Catholic Migration Commission. https://t.co/RtZRiHxw2v'
Processed tweet - 'must reject global indiffer toward refuge say presid intern cathol migrat commiss'
</p>
</div>
</div>
</section>
<section id="video" class="main">
<div class="spotlight">
<div class="content" style="text-align: justify;">
<header class="major">
<h2>Video Presentation</h2>
</header>
<p>
The presentation below shows the working of the models and the results obtained by them.
</p>
<div style="display: block; text-align: center;">
<iframe width="1024" height="576" src="https://www.youtube.com/embed/xcJtL7QggTI" frameborder="0" allow="accelerometer; autoplay; encrypted-media; gyroscope; picture-in-picture" allowfullscreen align="middle" ></iframe>
</div>
</div>
</div>
</section>
<!--
<section id="first" class="main special">
<header class="major">
<h2>Magna veroeros</h2>
</header>
<ul class="features">
<li>
<span class="icon solid major style1 fa-code"></span>
<h3>Ipsum consequat</h3>
<p>Sed lorem amet ipsum dolor et amet nullam consequat a feugiat consequat tempus veroeros sed consequat.</p>
</li>
<li>
<span class="icon major style3 fa-copy"></span>
<h3>Amed sed feugiat</h3>
<p>Sed lorem amet ipsum dolor et amet nullam consequat a feugiat consequat tempus veroeros sed consequat.</p>
</li>
<li>
<span class="icon major style5 fa-gem"></span>
<h3>Dolor nullam</h3>
<p>Sed lorem amet ipsum dolor et amet nullam consequat a feugiat consequat tempus veroeros sed consequat.</p>
</li>
</ul>
<footer class="major">
<ul class="actions special">
<li><a href="generic.html" class="button">Learn More</a></li>
</ul>
</footer>
</section>
<section id="second" class="main special">
<header class="major">
<h2>Ipsum consequat</h2>
<p>Donec imperdiet consequat consequat. Suspendisse feugiat congue<br />
posuere. Nulla massa urna, fermentum eget quam aliquet.</p>
</header>
<ul class="statistics">
<li class="style1">
<span class="icon solid fa-code-branch"></span>
<strong>5,120</strong> Etiam
</li>
<li class="style2">
<span class="icon fa-folder-open"></span>
<strong>8,192</strong> Magna
</li>
<li class="style3">
<span class="icon solid fa-signal"></span>
<strong>2,048</strong> Tempus
</li>
<li class="style4">
<span class="icon solid fa-laptop"></span>
<strong>4,096</strong> Aliquam
</li>
<li class="style5">
<span class="icon fa-gem"></span>
<strong>1,024</strong> Nullam
</li>
</ul>
<p class="content">Nam elementum nisl et mi a commodo porttitor. Morbi sit amet nisl eu arcu faucibus hendrerit vel a risus. Nam a orci mi, elementum ac arcu sit amet, fermentum pellentesque et purus. Integer maximus varius lorem, sed convallis diam accumsan sed. Etiam porttitor placerat sapien, sed eleifend a enim pulvinar faucibus semper quis ut arcu. Ut non nisl a mollis est efficitur vestibulum. Integer eget purus nec nulla mattis et accumsan ut magna libero. Morbi auctor iaculis porttitor. Sed ut magna ac risus et hendrerit scelerisque. Praesent eleifend lacus in lectus aliquam porta. Cras eu ornare dui curabitur lacinia.</p>
<footer class="major">
<ul class="actions special">
<li><a href="generic.html" class="button">Learn More</a></li>
</ul>
</footer>
</section>
<section id="cta" class="main special">
<header class="major">
<h2>Congue imperdiet</h2>
<p>Donec imperdiet consequat consequat. Suspendisse feugiat congue<br />
posuere. Nulla massa urna, fermentum eget quam aliquet.</p>
</header>
<footer class="major">
<ul class="actions special">
<li><a href="generic.html" class="button primary">Get Started</a></li>
<li><a href="generic.html" class="button">Learn More</a></li>
</ul>
</footer>
</section>
</div>
<footer id="footer">
<section>
<h2>Aliquam sed mauris</h2>
<p>Sed lorem ipsum dolor sit amet et nullam consequat feugiat consequat magna adipiscing tempus etiam dolore veroeros. eget dapibus mauris. Cras aliquet, nisl ut viverra sollicitudin, ligula erat egestas velit, vitae tincidunt odio.</p>
<ul class="actions">
<li><a href="generic.html" class="button">Learn More</a></li>
</ul>
</section>
<section>
<h2>Etiam feugiat</h2>
<dl class="alt">
<dt>Address</dt>
<dd>1234 Somewhere Road • Nashville, TN 00000 • USA</dd>
<dt>Phone</dt>
<dd>(000) 000-0000 x 0000</dd>
<dt>Email</dt>
<dd><a href="#">information@untitled.tld</a></dd>
</dl>
<ul class="icons">
<li><a href="#" class="icon brands fa-twitter alt"><span class="label">Twitter</span></a></li>
<li><a href="#" class="icon brands fa-facebook-f alt"><span class="label">Facebook</span></a></li>
<li><a href="#" class="icon brands fa-instagram alt"><span class="label">Instagram</span></a></li>
<li><a href="#" class="icon brands fa-github alt"><span class="label">GitHub</span></a></li>
<li><a href="#" class="icon brands fa-dribbble alt"><span class="label">Dribbble</span></a></li>
</ul>
</section>
<p class="copyright">© Untitled. Design: <a href="https://html5up.net">HTML5 UP</a>.</p>
</footer> -->
</div>
<!-- Scripts -->
<script src="assets/js/jquery.min.js"></script>
<script src="assets/js/jquery.scrollex.min.js"></script>
<script src="assets/js/jquery.scrolly.min.js"></script>
<script src="assets/js/browser.min.js"></script>
<script src="assets/js/breakpoints.min.js"></script>
<script src="assets/js/util.js"></script>
<script src="assets/js/main.js"></script>
</body>
</html>