CS132Team7Portfolio/communication.html at main · Error404Coder/CS132Team7Portfolio · GitHub

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
<!DOCTYPE HTML>
<!--
	Miniport by HTML5 UP
	html5up.net | @ajlkn
	Free for personal and commercial use under the CCA 3.0 license (html5up.net/license)
-->
<html>
	<head>
		<title>PH Twitter Fake News Analysis</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" />
		<style>
			#meth2, #meth1 {
				max-width: 80%;
			}
			td {
				border: 1px solid black;
				color: black;
				text-align: center;
				vertical-align: middle;
			}
			th {
				border: 1px solid black;
				color: black;
				text-align: center;
				width: 200px;
				font-weight: 700;
			}
			.chatter-graph {
				max-width: 100%;
			}

		</style>
	</head>
	<body class="is-preload">

		<!-- Nav -->
		<nav id="nav">
			<ul class="container">
				<li><a href="index.html">Home</a></li>
				<li><a href="overview.html">Motivation</a></li>
				<li><a href="exploration.html">Data Exploration</a></li>
				<li><a href="modeling.html">Modeling</a></li>
				<li><a href="communication.html">Communication</a></li>
			</ul>
		</nav>

		<!-- Home -->
			<article id="top" class="wrapper style1">
				<div class="container">
					<div class="row">
						<div>
							<h1>Introduction</h1>
									<p>
										With the recent results of the 2022 Philippine Presidential Elections, fake
									news became a widespread topic due to the misinformation propagated in
									social media platforms promoting certain candidates
									(
									<a href="https://fulcrum.sg/fact-checking-in-the-philippines-the-quest-to-end-disinformation-in-elections/">Labiste, 2022;</a>
									<a href="https://verafiles.org/articles/vera-files-fact-check-yearender-filipinos-face-rapid-fire-falsehoods-on-social-media-in-2022-elections">Samson, 2022;</a>
									<a href="https://www.gmanetwork.com/news/topstories/nation/849657/disinformation-influenced-operations-affected-2022-elections-says-study/story/">Locus, 2023;</a>
									 ). One of said
									candidates is Bongbong Marcos (BBM), which based on a Rappler report
									showed majority of Twitter users which used the hashtag #LabanMarcos
									were only created near BBM’s declaration of candidacy
									<a href="https://www.rappler.com/nation/elections/ferdinand-bongbong-marcos-jr-tries-take-over-twitter-freshly-made-accounts/">(Baizas, Macaraeg, & Salcedo, 2022)</a>

									. This report by
									Rappler implies that these Twitter users may have been created for the
									purpose of promoting BBM’s presidential campaign. Given these findings,
									the objective of this study is to determine if the number of credit grabbing
									tweets associated with BBM may have significantly changed during the
									presidential campaign period.
									</p>
							<h1>Materials and Methods</h1>
										<p>The methodology can be summarized in the following steps:</p>
										<p>1. Collecting credit-grabbing tweets spanning the campaign period
											using the Python library snscrape</p>
										<img src="images/meth1.png" class="center-inline-img" id="meth1">

										<p>2. Manual misinformation labeling with VERA Files criteria as basis
											<a href="https://verafiles.org/articles/what-you-want-know-about-vera-files-fact-check">[5]</a>
										</p>

										<img src="images/meth2.png" class="center-inline-img" id="meth2" >


										<p>3. Data preprocessing and visualization using Pandas, Matplotlib, Seaborn and Plotly</p>
										<img src="images/meth3.1.png" class="center-inline-img" >
										<img src="images/meth3.png" class="center-inline-img" >

										<p>4. Computing statistical significance with Kruskall Mann-Whitney U Tests</p>
										<img src="images/meth4.png" class="center-inline-img" >

										<p>5. Determining specific spikes in misinformation using Event Detection Modeling</p>
										<img src="images/meth5.2.png" class="center-inline-img"  >
										<img src="images/meth5.1.png" class="center-inline-img"  >

							<h1>Results and Discussion</h1>
							<table>
								<tr>
									<th colspan="5">Kruskall-Wallis Test</th>
								</tr>
								<tr>
									<th>
										Datasets compared
									</th>
									<th>
										Test Statistic
									</th>
									<th>
										x2 critical <br> value (df =2 )
									</th>
									<th>
										p-value
									</th>
									<th>
										Significance Level
									</th>
								</tr>

								<tr>
									<td>
										Campaign vs pre-campaign
									</td>
									<td>
										5.423
									</td>
									<td>
										5.991
									</td>
									<td>
										0.066
									</td>
									<td>
										0.05
									</td>
								</tr>
							</table>

							<p>
								Given that the p-value is less than the significance level, we fail to reject the null hypothesis,
								i.e., there is no significant difference in the number of BBM credit-grabbing tweets before,
								during and after the campaign period. This implies that the number of BBM credit-grabbing
								tweets may have been prevalent before and after the campaign period. To confirm this, a post
								hoc analysis was also performed
							</p>


							<table>
								<tr>
									<th colspan="5">One-Tailed Mann-Whitney U Test</th>
								</tr>
								<tr>
									<th>
										Datasets compared
									</th>
									<th>
										Test Statistic
									</th>
									<th>
										Critical value
									</th>
									<th>
										p-value
									</th>
									<th>
										Significance Level
									</th>
								</tr>

								<tr>
									<td>
										Campaign vs pre-campaign
									</td>
									<td>
										309
									</td>
									<td>
										331
									</td>
									<td>
										0.009
									</td>
									<td>
										0.05
									</td>
								</tr>

								<tr>
									<td>
										Campaign vs post-campaign
									</td>
									<td>
										382
									</td>
									<td>
										317
									</td>
									<td>
										0.201
									</td>
									<td>
										0.05
									</td>
								</tr>
							</table>

							<p>
								With the p-value less than the significance level for the campaign vs pre-campaign
								datasets, the null hypothesis is rejected, i.e., BBM credit-grabbing tweets are significantly
								less during the campaign period compared to the pre-campaign period. It may be the case
								that BBM credit-grabbing tweets were even more widespread before the campaign period
								has started
							</p>

							<img src="images/meth5.2.png" class="center-inline-img"  >

							<p>
								Numerous dates pop up in the event detection model. October 20 and 21, 2021 are the first
								dates of the peak and the change event, respectively. These dates are within the same
								month as the date that Bongbong Marcos filed for his certificate of candidacy (COC) on
								October 6, 2021
								<a href="https://www.pna.gov.ph/articles/1155753">[6]</a>
								. Note that after the peak and the change points, there is a noticeable
								drop in the number of misinformation tweets. It was on January 24 and January 30 of 2022
								that the number of tweets started to rise again. This drop might be explained by Twitter’s
								report on suspending hundreds of twitter accounts for violating platform manipulation and
								spam policy
								<a href="https://www.rappler.com/nation/elections/twitter-suspends-accounts-ferdinand-bongbong-marcos-jr-network-january-2022/">[7]</a>
								. The next and last change point detected was on May 29, 2022, weeks after
								the election. Little to no tweets appeared afterwards, until September 2022 where another
								peak can be seen, September 23, 2022. This peak could be the result of Joe Biden praising
								BBM for his ‘work’ on windmills, on September 22, 2022 at the UN General Assembly
								<a href="https://www.philstar.com/pilipino-star-ngayon/bansa/2022/09/23/2211712/biden-impressed-sa-ilocos-norte-windmills-na-itinayo-naman-ng-private-sector">[8]</a>
							</p>

							</p>
							<h1>Conclusion and Implications</h1>
								<p>
									From the statistical tests, it can be concluded that BBM credit-grabbing tweets were
									already prevalent before the presidential campaign. It is therefore recommended to explore
									data beyond the project’s data collection timeline, October 2021 to September 2022, in order
									to trace the wave of misinformation associated with the 2022 Philippine Elections. Moreover,
									the analysis of the dates shows that BBM’s filing for COC, the suspension of pro-BBM
									Twitter accounts, the end of elections, and Joe Biden’s mention of the windmills occurs near
									the peaks and change points of dis/misinformation tweets. These dates could be further
									studied to determine the existence/non-existence of a correlation and/or causation to the
									amount of dis/misinformation tweets.
								</p>
								<p>

									This study can help with fact checking organizations in order to know and predict when misinformation is
									most prevalent, and to allot their resources accordingly.
										These findings could also be integrated into machine learning models of misinformation detector in order to make them more accurate.
								</p>
						</div>
					</div>
				</div>
			</article>

		<!-- Scripts -->
			<script src="assets/js/jquery.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>