I would like to start by thanking Peter for inviting me to write this
post on the Swarm Creativity blog. I have been in Peter’s COINs class for over
a month now and I am thoroughly fascinated by what social network analysis can
teach us.
For the COINs class’ midterm exam that involves gathering, analyzing and
inferring from social media data, I was assigned the topic “Joe Biden against
Paul Ryan”. It was a great topic for me since I was planning on doing
some coolhunting on this topic after the Vice Presidential debate that evening.
Conducting the analysis before the debate made the results even more
interesting. I have used Condor’s Twitter Collector to collect Twitter feeds
and Condor Web Collector’s Google Blog Search to collect website data. Here are
the results, pre and post-debate.
Pre-Debate Analysis
Twitter
Following image shows the network of users talking
about Paul Ryan and Joe Biden on Twitter. The blue dots are users talking about
Joe Biden, the green are talking about Paul Ryan and the red are talking about
both of them.
We can see that there are more people talking about
Paul Ryan compared to those talking about Joe Biden. The following image shows
that the betweenness centrality (or popularity) of Paul Ryan is much greater
than that of Joe Biden.
Now
let us see what people are actually talking about. Following image shows the
terms widely used by users while talking about Joe Biden.
We can see that “Obama”, “Ryan”
and “tonight” are quite important (or have higher “betweenness centralities” in
SNA terminology). This is expected. However, the interesting terms that appear
with a decent amount of betweenness centrality are “win” and “good”.
Now let us see what terms are being used by people
talking about Paul Ryan.
Here also, we can see that “Joe”,
“Biden” and “tonight” have high betweenness centralities. Interestingly,
although the term “good” appears with high betweenness centrality, just as it
did in Joe Biden’s network, the term “win” appears with a very low betweenness.
Could this mean that although fewer people are talking about Joe Biden, they
are talking more optimistically about him?
To go a little deeper, I used Condor’s
Taxo tool to further analyze the twitter feeds. I extracted terms from 3
wikipedia categories
1) Emotion
2) Economics
3) Employment
Following image shows the terms from these
taxonomies that appear in tweets about Joe Biden.
The interesting terms that appear
with high frequency are “hope”, “shame” and “economics”.
Following image shows the terms from these
taxonomies that appear in tweets about Paul Ryan
The interesting terms that appear
with high frequency are “economics”, “employment”, “hope”, “motivation” and
“victory”.
Although Joe Biden’s taxonomy
view doesn’t look too conclusive, Paul Ryan’s view seems very positive.
Moreover his betweeness centrality is also high on twitter. Thus I can say that
Paul Ryan is clearly more popular on twitter and positively so.
Google Blog Search
Now let us look at the google
blog search results.
Following image shows the network of websites
talking about Joe Biden and Paul Ryan. The blue dots are websites talking about
Biden, the green are talking about Ryan and red are talking about both.
Following
image shows the betweenness centralities for Paul Ryan and Joe Biden.
Clearly Joe Biden is more central
in blogs. This is different from twitter, where Ryan was more central.
Now let us see which website is more
influential. I have filtered “Joe Biden” and “Paul Ryan” from the websites.
Following image shows the betweenness centralities of the top websites.
We can see that
huffingtonpost.com is the most influential website in the blog search, followed
by foxnewsinsider.com.
Now let us look at the sentiment analysis of the
content on the websites. The dots above the zero represent positive sentiment
and the ones below the zero represent negative sentiment.
Joe Biden’s sentiment view:
We see a lot of positive
sentiment for Joe Biden.
Paul
Ryan’s sentiment view:
Ryan’s sentiment view looks quite
balanced and is not very skewed towards positive, unlike Biden’s, which is
clearly positive.
Conclusions:
1) Twitter
analysis shows Paul Ryan to be clearly more popular and positive
2) Blog
analysis shows that Joe Biden is more popular and positive
3) Since
twitter is more dynamic than blogs, I would conclude that Joe Biden has a
historical positive sentiment going for him, but Ryan is the new and popular
kid on the block.
4) We
need to keep an eye on huffingtonpost.com and foxnewsinsider.com to see what
they are saying about the 2 candidates. What they say is clearly very
influential
Post-Debate Analysis
Twitter
The twitter feeds after the
debate were collected every hour from the time the debate ended at about
10:30PM EDT till the next morning. This was done so as to be able to see the
change in the betweenness centralities of the two candidates over the few hours
after the debate.
The following image shows the network of users
talking about Joe Biden and Paul Ryan after the debate.
The blue dots are users talking
about Joe Biden, the green are users talking about Paul Ryan and the red dots
are users talking about both the candidates. Unlike the pre-debate view, we can
see a much larger fraction of people is talking about Joe Biden after the
debate.
The following image shows the betweenness
centralities of the 2 candidates.
Joe Biden is in fact more central
post-debate than he was before the debate. This does not tell the complete
story though. Let’s analyze the content of the tweets to figure out what people
are actually saying about the 2 candidates.
The following image shows the terms that
appeared frequently in tweets about Joe Biden.
In addition to the obvious
“tonight”, “Obama” and “vice” we see the terms “rude” and “laughing” being used
a lot. These terms do indicate a bit of negativity in the tweets about Biden.
After the debate, Joe Biden was criticized quite a bit for his condescending
and rude demeanor during the debate.
The following image shows the terms that
appeared frequently in tweets about Paul Ryan.
There is nothing that really
stands out that may be attributed to a positive or negative tone in the tweets
about Paul Ryan.
Now let us look at how the betweenness
centrality of the two candidates changed over the few hours after the debate.
The blue line represents the betweenness centrality of Joe Biden and the green
line represents the betweenness centrality of Paul Ryan over time.
We can see a surge in Joe Biden’s
centrality for some time after the debate. However, Paul Ryan does catch up and
finally surpass Joe Biden in centrality in due course of time. This indicates
that more people did talk about Biden for some time after the debate, but went
back to talking about Ryan soon. Moreover, the analysis of terms shows that
quite a bit of the talk about Biden might have been negative.
Google Blog Search
The following image shows the network of
websites talking about Joe Biden and Paul Ryan after the debate.
We see that Joe Biden and Paul
Ryan have comparable betweenness centralities, unlike the view before the
debate. Paul Ryan definitely seems to have gained some ground in the blog
space. Now let us look at the sentiment analysis of the content in the
websites.
Joe Biden’s sentiment view:
Biden’s sentiment view is still
quite positive but it does seem to have lost some positive sentiment.
Paul Ryan's Sentiment View:
Paul Ryan’s sentiment view is
clearly positive now, unlike the one before the debate.
Let us now look at the websites that are most
influential. Once again I removed the search terms “Joe Biden” and “Paul Ryan”
from the websites network. Here is the picture showing the most influential
websites.
huffingtonpost.com
is once again the most influential website. However, foxnewsinsider.com seems
to have disappeared after the debate.
Conclusions:
Here
are some conclusions based on the analysis after the debate
1) People
spoke about Joe Biden a lot more on twitter after the debate than they did
before the debate
2) The
content of tweets about Joe Biden might have had some negative tone.
3) Websites
spoke more about Paul Ryan after the debate than they did before the debate
4) The
sentiment analysis of the content on blogs shows an increase in positive
sentiment for Paul Ryan and a slight decrease in positive sentiment for Joe
Biden
5) The
website to keep an eye on for news on these two candidates in
huffingtonpost.com
6) Paul
Ryan certainly seems to be the winner in terms of likability of the only Vice
Presidential debate in this race to the White House.
However, what effect the debate has had on their electability and consequently
that of the presidential candidates is a coolhunting task in itself
I
am now looking forward to the next coolhunting task that I plan to undertake.
The analysis of the second Presidential debate.
