Low Cost Recommendation System for Online Users based on Intelligent Social Networking Clusters

Abstract

Social networks are a key part of our life in today's time. Social media is growing at a very great pace as everyone is getting connected to the internet. The Internet is filled with all kinds of stuff, but all these objects have different kinds of features associated with it. Hence instead of showcasing different kinds of recommendations, we proposed a low cost hybrid recommendation methodology to showcase posts that are related to a person's past history which are related to the person. The experimentation and results involves the clustering of different topics based on collaborative filtering and content-based filtering, making a hybrid of these filtering methods for making a recommendation system. The research focuses on getting data posted currently by different kinds of people and based on the content in segregate different kinds of people or the topics. The proposed low cost recommendation system with lower training cost and higher accuracy of 89% is observed.