Skip to content

Commit

Permalink
02122023
Browse files Browse the repository at this point in the history
  • Loading branch information
@pablobernabeu
pablobernabeu committed Dec 2, 2023
1 parent bf5f2cf commit 75bdc62
Show file tree
Hide file tree
Showing 58 changed files with 5,032 additions and 871 deletions.
47 changes: 22 additions & 25 deletions ...the-louisiana-minnesota-texas-crisis-across-media-and-time-a-big-data-exercise/index.html
View file
Original file line number Diff line number Diff line change
Expand Up @@ -641,7 +641,6 @@ <h1>The Louisiana-Minnesota-Texas crisis across media and time: A big data exerc

<div class="article-style">

<script src="https://pablobernabeu.github.io/2016/the-louisiana-minnesota-texas-crisis-across-media-and-time-a-big-data-exercise/index_files/header-attrs/header-attrs.js"></script>


<div class="alert alert-success" role="alert" style="padding-bottom:0px; margin-bottom:30px;">
Expand Down Expand Up @@ -725,34 +724,34 @@ <h6><em>Tweets</em></h6>
## 10 cops baton violence</code></pre>
<p>To start, it may stand out that different topics appear across sources, all the while some are indeed shared. This is perfectly normal for topic modeling on different sources, even when the same topic is being studied. Indeed, it is very relevant for us to remark on the inclusion of foreign affairs and election matters within the NYT articles, but not within people’s tweets and Facebook comments. This makes sense for several reasons. To start, the space a journalist counts on in a NYT article is considerable, compared to tweets, and also compared to ruling conventions of Facebook posts (users may write further, but the average simply will not). Second, the breadth of relation in NYT articles likely responds to the expectations from renown journalists to enrich the news with a broader contextualization. Furthermore, this extension of topics might correspond to the tacit but doubtless alignment of journals to concrete political agendas. While people commenting on Twitter or Facebook are plausibly characterized by just the same virtues and vices, their online reactions could be driven by more emotion and immediacy of focus than those of mass media journalists.</p>
<p>For greater visualization, we also provide some captions from the interactive LDAvis tool below. Please click on the figure titles to enjoy the full visualization.</p>
<div class="figure">
<img src="https://github.com/RRisto/Summer-school-course-project/raw/master/BLM/Untitled3.png" alt="" />
<p class="caption"><br> <a href="http://rristo.github.io/NYT/index.html">↑ LDAvis visualization of NYT articles (click to explore in detail)</a></p>
<div class="float">
<img src="https://github.com/RRisto/Summer-school-course-project/raw/master/BLM/Untitled3.png" alt=" ↑ LDAvis visualization of NYT articles (click to explore in detail)" />
<div class="figcaption"><br> <a href="http://rristo.github.io/NYT/index.html">↑ LDAvis visualization of NYT articles (click to explore in detail)</a></div>
</div>
<p><br></p>
<div class="figure">
<img src="https://github.com/RRisto/Summer-school-course-project/raw/master/BLM/Untitled1.png" alt="" />
<p class="caption"><br> <a href="http://rristo.github.io/Facebook/index.html#topic=0&amp;lambda=1&amp;term=">↑ LDAvis visualization of Facebook comments (click to explore in detail)</a></p>
<div class="float">
<img src="https://github.com/RRisto/Summer-school-course-project/raw/master/BLM/Untitled1.png" alt=" ↑ LDAvis visualization of Facebook comments (click to explore in detail)" />
<div class="figcaption"><br> <a href="http://rristo.github.io/Facebook/index.html#topic=0&amp;lambda=1&amp;term=">↑ LDAvis visualization of Facebook comments (click to explore in detail)</a></div>
</div>
<p><br></p>
<div class="figure">
<img src="https://github.com/RRisto/Summer-school-course-project/raw/master/BLM/Untitled.png" alt="" />
<p class="caption"><br> <a href="http://rristo.github.io/Twitter/index.html#topic=0&amp;lambda=1&amp;term=">↑ LDAvis visualization of tweets (click to explore in detail)</a></p>
<div class="float">
<img src="https://github.com/RRisto/Summer-school-course-project/raw/master/BLM/Untitled.png" alt=" ↑ LDAvis visualization of tweets (click to explore in detail)" />
<div class="figcaption"><br> <a href="http://rristo.github.io/Twitter/index.html#topic=0&amp;lambda=1&amp;term=">↑ LDAvis visualization of tweets (click to explore in detail)</a></div>
</div>
<p><br>
<br></p>
<p>In order to specifically compare different content sources, we plotted the major language from two sources on the same plot, with an axis spanning from one source to the other, as shown below. The size of the words indicates the frequency of use, and the colour is essentially parallel with the axis, with specific different colours for different corpora, and darker hues for greater association.</p>
<div class="figure">
<img src="https://github.com/RRisto/Summer-school-course-project/raw/master/BLM/CorporaCompareFbvsNYT.png" alt="" />
<p class="caption">↑ Facebook comments and NYT articles</p>
<div class="float">
<img src="https://github.com/RRisto/Summer-school-course-project/raw/master/BLM/CorporaCompareFbvsNYT.png" alt="↑ Facebook comments and NYT articles" />
<div class="figcaption">↑ Facebook comments and NYT articles</div>
</div>
<div class="figure">
<img src="https://github.com/RRisto/Summer-school-course-project/raw/master/BLM/CorporaCompareFbvsTw.png" alt="" />
<p class="caption">↑ Facebook comments and tweets</p>
<div class="float">
<img src="https://github.com/RRisto/Summer-school-course-project/raw/master/BLM/CorporaCompareFbvsTw.png" alt="↑ Facebook comments and tweets" />
<div class="figcaption">↑ Facebook comments and tweets</div>
</div>
<div class="figure">
<img src="https://github.com/RRisto/Summer-school-course-project/raw/master/BLM/CorporaCompareTwvsNYT.png" alt="" />
<p class="caption">↑ Tweets and NYT articles</p>
<div class="float">
<img src="https://github.com/RRisto/Summer-school-course-project/raw/master/BLM/CorporaCompareTwvsNYT.png" alt="↑ Tweets and NYT articles" />
<div class="figcaption">↑ Tweets and NYT articles</div>
</div>
<p><br></p>
<p>We went on to analyze the overlap in topics across journals, in order to quantitatively check whether some topics were indeed shared across sources, even if in different positions (for instance, topic 1 in some source and topic 3 in some other). We did this by means of cosine similarity scores. These scores represent the degree of similarity of two sources on a continuous scale from 0 to 1, where 1 would mean identical. The plots illustrate these comparisons in turn.</p>
Expand All @@ -772,9 +771,9 @@ <h6><em>Tweets</em></h6>
<br>
<br></p>
<p>Last, the interaction of Time and Media was analyzed. As expected, we found differences in the way topics fluctuated over time in the different sources, albeit in unexpected ways. NYT and tweets articles presented great variation, suggesting day-bound sensitivity to the developments. This was to be expected from Twitter, as it is famous for its immediacy. However, the immediacy of NYT articles was rather surprising, as they might have lagged behind due to the necessary investigation and editing for such kind of journalistic pieces. Unlike traditional paper-based NYT articles, this immediacy is now enabled by the publication online. Another unexpected finding was the relative stillness of Facebook posts over time. Since they are published at the minute, and nowadays mostly from mobile, we had thought they would present greater immediacy than NYT articles. We could hypothesize on this, but this would be best analyzed in further research. The plot below illustrates this interaction.</p>
<div class="figure">
<img src="https://github.com/RRisto/Summer-school-course-project/raw/master/BLM/TopicTime.png" alt="" />
<p class="caption">↑ Topic fluctuations over time for the three content sources</p>
<div class="float">
<img src="https://github.com/RRisto/Summer-school-course-project/raw/master/BLM/TopicTime.png" alt="↑ Topic fluctuations over time for the three content sources" />
<div class="figcaption">↑ Topic fluctuations over time for the three content sources</div>
</div>
</div>
</div>
Expand Down Expand Up @@ -803,7 +802,7 @@ <h2>References</h2>
<p>Wang, P., He, W., &amp; Zhao, J. (2014). A Tale of Three Social Networks: User Activity Comparisons across Facebook, Twitter, and Foursquare. <em>IEEE Internet Computing, 18</em>(2), 10-15.</p>
</div>
</div>
<div class="footnotes">
<div class="footnotes footnotes-end-of-document">
<hr />
<ol>
<li id="fn1"><p>A later update: On July 17, 2016—days after the current analysis—, the LMT crisis was extended with the killing of two policemen in the same Louisiana city where Alton Sterling had been killed.<a href="#fnref1" class="footnote-back">↩︎</a></p></li>
Expand All @@ -818,8 +817,6 @@ <h2>References</h2>

<div class="article-tags">

<a class="badge badge-light" href="/tags/s/">s</a>

<a class="badge badge-light" href="/tags/big-data/">big data</a>

<a class="badge badge-light" href="/tags/data-mining/">data mining</a>
Expand Down
2 changes: 0 additions & 2 deletions 2022/deception-in-a-survey-from-polar-insight-/-polar-intelligence/index.html
View file
Original file line number Diff line number Diff line change
Expand Up @@ -696,8 +696,6 @@ <h1>Deception in a survey from Polar Insight / Polar Intelligence</h1>

<a class="badge badge-light" href="/tags/fraud/">fraud</a>

<a class="badge badge-light" href="/tags/s/">s</a>

</div>


Expand Down
Loading

0 comments on commit 75bdc62

Please sign in to comment.