Twitter User Modeling Based on Indirect Explicit Relationships for Personalized Recommendations

Abstract

Information overload has increased due to social network website use in recent times. Social media has increased the popularity of websites such as Twitter. It is believed that a rich environment is provided through Twitter whereby information sharing will be able to aid in recommender system research. This paper will focus upon Twitter user modeling through the utilization of indirect explicit relationships that exist amongst users. The further aim of this paper is to ensure that personal profiles are built via the use of information that will be sourced from Twitter so as to provide recommendations that are more accurate. The proposed method adopts Twitter user's indirect explicit relationships in order to get information which is vital in the process of building personal user profiles. The proposed method has been validated through the implementation of an offline evaluation using real data. Proposed user profiles' performances have been compared with each other and against the baseline profile. The performance of this has been validated using real data and is both practical and effective.

Resorting to Context-Aware Background Knowledge for Unveiling Semantically Related Social Media Posts

“…Tweet-Recommender system [20] provides tweets related to the news using topic similarities and language modeling. Several other approaches, such as [4], [18], [21], focus on hashtag-based or user interaction history for tweet recommendations. These methods primarily focused on hashtag similarities.…”