Mining user interest based on personality-aware hybrid filtering in social networks

With the emergence of online social networks and microblogging websites, user interest mining has been an active research topic for the past few years. However, most of the existing works suffer from two significant drawbacks, firstly, they focus on the user's explicit content and social network structure to predicate the user's interests, neglecting the fact that the user's personality might be a rich source to infer the topical interests. Secondly, they represent the user's content using the bag-of-words model that ignores the chronological order of the posted content, hence the predicted interests might contain outdated topics that the user does not interest anymore. In this paper, we propose a novel user interest mining system based on Big Five personality traits and dynamic interests. To prove the effectiveness of incorporating the user's personality traits in the interest mining process, we have implemented a social network for news sharing and conducted different experiments on the collected data. The experiment results show that considering personality traits can increase the precision and recall of interest mining systems, as well as can help to tackle the cold start problem. (C) 2020 Elsevier B.V. All rights reserved.