Prediction of Group and Individual Behaviours on Social Web Sites

Siriplakota Balaji, Jhansi Lakshmi Vaddelle, B. Lakshmi

Abstract: Today every user is using social media such as twitter, YouTube, Flicker, twitter for communication and whats going on around the world. In this paper we analyse the data regarding users of different social networks and predict their interesting areas and collective behaviour of each user. To predict the behaviour we collect the corpus from various social networking sites. Here we use the concept of edge centric clustering scheme to retrieve social dimensions. The behaviour is predicted based on social dimensions considered. This algorithm is used to address the scalability issue. The proposed approach can effectively maintain enormous amount of actors in demonstrating comparable prediction performance to other non-scalable methods.

Keywords: Collective Behaviour, Edge centric clustering, Social Dimensions, Non-Scalable methonds