Ranking of news items in rule-stringent social media based on users' importance: A social computing approach

Abstract

In this paper an innovative social media news items ranking scheme is proposed. The proposed unsupervised architecture takes into consideration user-content interactions, since social media posts receive likes, comments and shares from friends and other users. Additionally the importance of each user is modeled, based on an innovative algorithm that borrows ideas from the PageRank algorithm. Finally, a novel content ranking component is introduced, which ranks posted news items based on a social computing method, driven by the importance of the social network users that interact with them. Initial experiments on real life social networks news items illustrate the promising performance of the proposed architecture. Additionally comparisons with three different ranking ways are provided (SUMF, RSN-CO and RSN-nCO), in terms of user satisfaction.