A Model of Online Social Interactions based on Sentiment Analysis and Content Similarity

In this paper we create a model of human behavior in online communities, based on the network topology and on the communication content. The model contains eleven distinct hypotheses, which validate three intuitions. The rst intuition is that the network topology alone fails to clearly distinguish between the users who contribute to the community and the troublemakers. The second intuition is that the content of the messages exchanged in an online community can separate good and insightful contri- butions from the rest. The third intuition is that there is a delay until the network stabilizes and un- til standard measures, such as betweenness central- ity, can be used accurately. Taken together, these three intuitions are a solid case against indiscrimi- nately using network measures. They also underline the importance of the communication content. We show that the sentiment within the messages, espe- cially antagonism, can signicantly alter the commu- nity perception. We create a novel sentiment analysis technique to identify antagonistic behavior. We use real world data, taken from the Slashdot discussion forum to validate our model. All the find- ings are accompanied by extremely signicant t-test p-values.