Cohesion Based Personalized Community Recommendation System

Our life is totally engaged by the progressive growth of online social networking. Because, millions of users are interconnecting with each other using different social media sites like Facebook, Twitter, LinkedIn, Google+, Pinterest, Instagram etc. Most of the social sites like Facebook, Google+ allow users to join different groups or communities where people can share their common interests and express opinions around a common cause, problem or activity. However, an information overloading issue has disturbed users as thousands of communities or groups are creating each day. To resolve this problem, we have presented a community or group recommendation system centered on cohesion where cohesion represents high degree of connectedness among users in social network. In this paper, we emphasis on suggesting useful communities (or groups in term of Facebook) that users personally attracted in to join; reducing the effort to find useful information based on cohesion. Our projected framework contains of the steps like: extracting sub-network from a social networking site (SNS), computing the impact of amity(both real-life or social and SNS connected), measuring user proclivity factor, calculating threshold from existing communities or groups of a user and lastly recommending community or group based on derived threshold. In result analysis part, we consider the precision-recall values by discarding community or group one at a time from the list of communities or groups of a certain user and checking whether the removed community or group is recommended by our proposed system. We have evaluated our system with 20 users and found 76% F-1 accuracy measure.