Calculation of Client Similarities in Large-Scale on Social Network Using Recommendation Framework

Analysis of OSN to find out how people are being connected with each other into a network and existing methods can’t evaluate efficiently internal connectivity of a friendship graph and fail to provide proper recommendation. In this paper, we propose an efficient user similarity measurement between two users using profile and network similarity. With a view to estimating the user similarity we compute an average weight which denotes the possibility of two persons being alike. After calculating profile similarity, we also compute network similarity using friendship edge and mutual friends edge and later to determine user similarity, we apply conditional probability by considering profile similarity and network similarity as evidence. To handle missing valued of any profile attributes we provide a technique that assume the valued based on friends profile attribute. Finally, we conduct a feature analysis on our proposed scheme in a large network to prove its batter outcome.