Future Friend Recommendation System based on User Similarities in Large-Scale on Social Network

Friendship is one of the most important issues in online social networks (OSN). Researchers analyze the OSN to determine how people are connected to a network and how new connections are developed. Most of the existing methods cannot efficiently evaluate a friendship graphs internal connectivity and decline to render a proper recommendation. This paper presented three proposed algorithms that can apply in OSN to predict future friends recommendations for the users. Using network and profile similarity proposed approach can measure the similarity among the users. To predict the user similarity, we calculated an average weight that indicates the probability of two users being similar by considering every precise subset of some profile attributes such as age, profession, location, and interest rather than taking the only average of the superset profile attributes. The suggested algorithms perform a significant enhancement in prediction accuracy 97% and precision 96.566%. Furthermore, the proposed recommendation frameworks can handle any profile attribute's missing value by assuming the value based on friends' profile attributes.