Extended Edgecluster based Technique for Social Networking Collective Behavior Learning System
Growing interest and continuous development of social network sites like Facebook, Twitter, Flicker, and YouTube etc.turn to several researchers for research study, planning and rigorous development. Exact people behavior prediction is the most important challenge of these on-line social networking websites. This research focus to learn to predict collective behavior in social media networks. Particularly provided information about some person, how can we collect the behavior of unobserved persons in the same network? These tremendous growing networks in social media are of massive size, involving large number of actors. The computational scale of these networks makes necessary scalable learning for models for collective collaborative behavior prediction. This scalability issue is solved by the proposed k-means clustering algorithm which is used to partition the edges into disjoint distinct sets, with each set is showing one separate affiliation. This edge-centric structure represents that the extracted social dimensions are definitely sparse in nature. This model idealized on the sparse natured social dimensions, shows efficient prediction performance than earlier existing approaches The proposed approach can effectively able to work for sparse social networks of any growing size. The important advantage of this method is that it easily grows upon to handle networks with large number of actors while existing methods was unable to do. This scalable approach effectively used over of online network collective behavior on a large scale.
