Identifying Social Communities by Frequent Pattern Mining

This paper presents a social network modeling technique that models the data to be analyzed to create a social network as frequent closed patterns. Frequent closed patterns have the advantage that they successfully grab the inherent information, content of the dataset and is applicable to a broader application domain. Entropies of the frequent closed patterns are used to keep the dimensionality of the feature vectors to a reasonable size. Experimental results presented in the paper shows that social network produced from these set of features successfully carries the community structure information.