social media news communities: gatekeeping, coverage, and statement bias
DESCRIPTION
We examine biases in online news sources and social media communities around them. To that end, we introduce unsupervised methods considering three types of biases: selection or "gatekeeping" bias, coverage bias, and statement bias, characterizing each one through a series of metrics. Our results, obtained by analyzing 80 international news sources during a two-week period, show that biases are subtle but observable, and follow geographical boundaries more closely than political ones. We also demonstrate how these biases are to some extent amplied by social media.TRANSCRIPT
Social Media News Communities:Gatekeeping, Coverage, and Statement
Bias
Diego Saez-Trumper∗1 Carlos Castillo† Mounia Lalmas‡
∗Universitat Pompeu Fabra, Barcelona†Qatar Computing Research Institute, Doha
‡Yahoo Labs London
San Francisco, October, 2013
1This work was done while visiting the Qatar Computing Research Institute
”Media bias refers to (...) the selection of which stories arereported and how they are reported”. S. Rivolta
Selection
Coverage
Statement
Selection
Coverage
Statement
Selection
Coverage
Statement
Selection
Coverage
Statement
Goal: quantify biases present in onlinenews
Challenges
I Consider a large set of news sources.I Compare news sources with social media (Twitter).I Use unsupervised methods.
Data set - News Sources
I Use the top-100 news websites from Alexa.com.I Download all the news they publish trough RSS and
Twitter.
Data set - Twitter
Download all tweets containing a URL pointing to a newssource.
Community 6= Followers
People who have tweeted at least K1 articles from a given newssource within K2 days.
Selection Bias (Gatekeeping)
I Compute similarity among news sources using the Jaccardcoefficient.
I Project it in two dimensions using PCA.
Selection Bias (Gatekeeping)
News Sources
Geographical pattern
No clear pattern.
Coverage
I Compute similarity among news sources using theJensen-Shannon divergence (JS) .
Coverage Bias(s1, s2) = 1− JS(s1, s2)
Coverage Bias
News Sources
Stronger geographical pattern.
Coverage Bias
Geographical pattern.
Political leaning
News Sources Twitter
Stronger political leaning signal in Twitter.
Statement
I Use sentiment analyses to find positive/negativesentiments associated to a person.
Statement
Obama Thatcher
Sentiments are more extreme in Twitter.
Conclusions
I Strong geographical patterns.I Political leaning signal is stronger in Social Media.I Feelings are more extreme in Social Media.