Studying Community Dynamics with an Incremental Graph Mining Algorithm

The widespread usage of the Web and later of the Web 2.0 for social interactions has stimulated scholars of different disciplines in studying electronic communities. Traditionally, communities are observed as a static phenomenon. However, they are evolving constellations, which emerge, lose members and obtain new ones and potentially, grow, coerce, split or decline. Such dynamic phenomena require the study of social networks across the time axis. We propose the graph mining algorithm DENGRAPH for the discovery and monitoring of evolving communities. Data mining methods are successfully used for community discovery but are mostly limited to the static perspective. Taking a dynamic perspective implies the study of a stream of interactions among community members. Accordingly, our DENGRAPH is an incremental graph mining algorithm, which detects and adapts communities over time. We report on our first results in applying DENGRAPH on the social network of mail interactions of ENRON