Image content based topological analysis for friend recommendation on twitter

Abstract

Recently, there has been increase in usage of social media platform i.e., Twitter for sharing information, personal interests and breaking news during emergencies. One of the main challenges in twitter application is friend recommendation. In this paper, we propose a Novel Image Content based Topology Analysis for Friend Recommendation (IICTA-FR) for overcoming the challenges of finding the similar people. In IICTA-FR, we construct topology based tweet analysis and Image content analysis to find relevant friend for Twitter users. In this work, we provide a framework to compute relationship strength for ties based on directed interactions between users. The proposed ICTA-FR framework produces a directed and weighted graph where the nodes and edges represent Twitter users, and user interactions respectively. Further, each weight in the directed edge represents about the probability of any interaction going from the edge source to the edge destination in the future. This weight is based on both tweet analysis and image analysis. We used hierarchical generative model for understanding the images posted in twitter through a visual model. We used logistic regression based model for calculating the edge scores in the graph. The proposed methodology has been validated on real Twitter data and found to give better results than the existing state of art algorithms in terms success rate.