Provenance model for user recommendation in social network

Abstract

In recent years social network is growing rapidly and become a popular medium for disseminating information and connecting concurring people. Also, this network enabled large numbers of users to communicate, create and share content and become a huge information source. Filtering the truthful content creator of such huge information source is not easy. Such valuable content creator can be filtered with the help of the online recommender system. But, in reality, society has high confidence on recommendation made from the people they trust than from untrusted sources.To build a recommender system from such trusted people, this research work proposes a Trust-Boosted Recommender System (TBRS). The aim of TBRS is to predict and recommend the truthful user (here reviewer) with reduced error rate. The TBRS is designed by extracting the data about the user, quantifying the trust score of the user, classifying the trustworthy user and finally predicting and recommending the top ranked trustworthy user. The user s data are collected from the social network using an Application Programming Interface (API) and are preprocessed to get cleaned data. As the trust in the semantic web is derived from the provenance, this concept is adopted for trustworthy user s identification. The W7 provenance model is applied to the cleaned data to choose appropriate provenance elements in the context of trust. A novel conceptual relation based trust quantification algorithm is proposed to find the trust score of the user for various provenance elements. newline