Semantics of User Interaction in Social Media

In ubiquitous and social web applications, there are different user traces, for example, produced explicitly by “tweeting” via twitter or implicitly, when the corresponding activities are logged within the application’s internal databases and log files. For each of these systems, the sets of user interactions can be mapped to a network, with links between users according to their observed interactions. This gives rise to a number of questions: Are these networks independent, do they give rise to a notion of user relatedness, is there an intuitively defined relation among users? In this paper, we analyze correlations among different interaction networks among users within different systems. To address the questions of interrelationship between different networks, we collect for every user certain external properties which are independent of the given network structure. Based on these properties, we then calculate semantically grounded reference relations among users and present a framework for capturing semantics of user relations. The experiments are performed using different interaction networks from the twitter, flickr and BibSonomy systems.