Social and temporal-aware personalized recommendation for best spreaders on information sharing

Abstract

As the growth of online information sharing and online shopping is tremendous, the social networking sites and Online Shops (OSs) have become the potential information sources to the Recommender System (RS). The RS provides either services or products to the users based on their preferences. An Online Social Network (OSN) enables the users to share information with their social neighbors. An OS furnishes the products to the customers or users based on their requirements. The conventional context-aware recommender systems are inept at predicting the new user’s preferences, and user’s recent preferences. Usually, customers like to drift their preferences over time due to the evolution of the products in the OS. Hence, together considering of the key parameters of social popularity and temporal dynamics is crucial for modeling the RS. This paper presents Social and Temporal-Aware personalized Recommendation for the best Spreaders (STARS) approach which recommends the products based on the social influence and recent context information about the user. It employs the collaborative filtering and incorporates the three phases such as influence user identification using OSN, user’s preference identification using OS, and recent preference based recommendation using temporal dynamics. Initially, the STARS identifies the best spreader in OSN by applying Eigen Vector Centrality (EVC) measurement in the k-shell structure. Secondly, it analyzes the customer’s explicit as well as implicit feedback information using a user-item matrix factorization and Pearson correlation measurement. Finally, the STARS recommends the appropriate products to the users by predicting the user’s recent preferences reading it from the context-aware explicit and implicit feedback information. The experimental results show that the STARS significantly outperform the conventional context-aware recommender systems.