Quantifying Controversy on Social Media

“We represent a topic of discussion with a conversation graph. In such a graph, vertices represent users, and edges represent conversation activity and interactions, such as posts, comments, mentions, or endorsements. Our working hypothesis is that it is possible to analyze the conversation graph of a topic to reveal how controversial the topic is. In particular, we expect the conversation graph of a controversial topic to have a clustered structure. This hypothesis is based on the fact that a controversial topic entails different sides with opposing points of view, and individuals on the same side tend to endorse and amplify each other’s arguments [1, 2, 10].

Our main contribution is to test this hypothesis. We achieve this result by studying a large number of candidate features, based on the following aspects of activity: (i) structure of endorsements, i.e., who agrees with whom on the topic, (ii) structure of the social network, i.e., who is connected with whom among the participants in the conversation, (iii) content, i.e., the keywords used in the topic, (iv) sentiment, i.e., the tone (positive or negative) used to discuss the topic. Our study shows that all features, except from content-based ones, are useful in detecting controversial topics, to different extents. Particularly for Twitter, we find endorsement features (i.e., retweets) to be the most useful.”