Avviso di Seminario - Ingmar Weber: Political Polarization in Web Search and on Twitter
In this line of work, we look at left-vs.-right polarization in web search queries issued and hashtags being used. This is joint work with Erik Borra (U. of Amsterdam), Venkata Rama Kiran Garimella (Yahoo! Research) and Asmelash Kiran (L3S).
In the first part, I'll look at queries issued to Yahoo! in the US. A
political leaning is assigned to queries in relation to the results
being clicked. E.g. a query for which predominantly The Huffington
Post (a left-leaning site) would be labeled as left-leaning and a
similar for queries landing on The Drudge Report (a right-leaning
site). This leaning information is then augmented with information
about query popularity, query sentiment and even factual accuracy of
controversies related to the query. We show that our methodology can
be used to gain insights into how issues are "framed" by either side
and present both expected and surprising results linking truth values,
leaning and popularity. A related demo can be publicly accessed at:
http://politicalsearchtrends.
political context. Starting from a set of seed users (e.g.,
@BarackObama on the left and @MittRomney on the right) we use retweet
behavior to identify users of a particular party affiliation. Looking
at who uses a particular hashtag, we then assign a leaning to it and
also track changes in this leaning over time. Focusing on sudden
changes in leaning, we identify a set of "hashtag hijackers" who
actively engage in hashtag wars to redefine the terminology being used
and to bring a hashtag "over to their side". A related demo can be
publicly accessed at:
http://politicalhashtagtrends.
not 1-dimensional. I'd also like to discuss ideas of how to extend
such an analysis to Arab Politics and will show preliminary results
for a "Islamist vs. Secular" analysis for Egypt.