Community detection based on graph models of data

Analyzing relations between people belonging to a social circle helps in forming potential communities on the Internet. The increasing social circles on various networking sites make this analysis all the more important. Community detection, as it is called, is closely related to clustering; however, different data models are also used in the same. The paper aims at community detection using Weighted Graph Clustering. The proposed improvement to a recent work of Liu et al in this direction is tested for performance on two real life datasets - College Football and YouTube. The desired result is reduced cost while detecting close to original communities.