PROG: A Complementary Model to the Social Networks for Mining Forums

Online discussion systems in the form of forums have been represented by graphs and analyzed through social network techniques. Each forum is regarded as a social network and it is modeled by a graph whose vertices represent forum participants. Here, we focus on the structure and the opinion content of the forum postings and we are looking at the social network that is developed from a semantics point of view. We formally define a model whose purpose is to provide complementary information to the knowledge extracted by the social network model. We present structure, opinion, temporal and topic-oriented measures that can be defined based on the new model, and we discuss how these measures facilitate the analysis of an online discussion. Applying our model to a real forum found on the Web shows the additional information that can be extracted.