Title:Interplay of homophily and communication in online social networks: Wikipedia-based semantic metric application on Twitter

Abstract:People are observed to assortitavely connect on a set of traits. Uncovering the reasons for people exhibiting this strong assortative mixing in social networks is of great interest to researchers and in practice. A popular case application exploiting the insights about social correlation in social networks is in marketing and product promotion. Suggested tendencies to induce observed social correlation are homophily and social influence. However, clearly identifying the causal relationship between these tendencies has proven to be a hard task. In this study we present the interplay between communication happening on Twitter (represented by user mentions) and different semantical aspects of user communication content (tweets). As a semantic relatedness metric we employ a database built from the English Wikipedia corpus according to the Explicit Semantic Analysis method. Our work, to the best of our knowledge, is the first to offer an in-depth analysis on semantic homophily on communication in social networks. Moreover, we quantify diverse levels of homophily and/or social influence, identify the semantic traits as the foci of such homophily, show insights in the temporal evolution of the homophily and influence and finally, we present their intricate interplay with the communication in Twitter.