A Technique for Human-readable Representation and Evaluation of Media-based Social Interactions in Social Networks

• Alan Keller Gomes USP
• Maria da Graça Campos Pimentel U

Resumo

Social networks are increasingly present on the Web, especially those supported by multimedia platforms, allowing social interaction, communication, sharing and collaboration among users. In social network analysis, some models for the representation of user's interactions have been proposed in the literature. However, those models do not explain what actions were taken by users during social interactions. In particular, this also occurs when social interactions involve media. We present a technique for a human-readable representation of social interactions in the form of if-then rules, and for evaluation of the rules using data mining procedures. Our technique allows the representation and the evaluation of media-based social interactions by making explicit both the actions performed by users, and the media used in the interaction. We present the results of applying our technique to describe interaction among a group of Facebook users.