Real time mining of ego networks for exploring social associations

Ego-networks represent social circles and associations within social graphs. The interconnection between an individual (ego) in a social network and other users (alters) that exist in various groups or clusters are community driven and continuously evolving. Manual identification of social circles and associations are time consuming for large networks due to its busty nature and unpredictable growth. In this research work, we propose clustering based real time mining of ego network to explore social associations. Real time ego network data has been acquired with the help of Twitter API. By applying Page Rank algorithm on the obtained datasets, we determine top influencers in an ego network and recommend a user to follow them. Alters of an ego are clustered into various groups using clustering techniques; K-means, spectral and affinity propagation clustering. These algorithms have been applied on small, medium and large size Twitter data for better understanding of dynamic nature of ego network. Various performance evaluation coefficients of clusters are considered to compare the performance of clustering algorithms, especially in the light of ego network, which highlighted effectiveness of approach in real time environment.