Topical Influential User Analysis with Relationship Strength Estimation in Twitter

Xinyue Liu, Hua Shen, Fenglong Ma, Wenxin Liang

Abstract: Topical Influential User Analysis (TIUA) is an important technique in Twitter. Existing techniques neglected relationship strength between users, which is a crucial aspect for TIUA. For modeling relationship strength, interaction frequency between users has not been considered in previous works. In this paper, we firstly introduce a poisson regression-based latent variable model to estimate relationship strength by utilizing interaction frequency. We then propose a novel TIUA framework which uses not only retweeting relationship but also relationship strength. Experimental results show that the proposed TIUA algorithm can greatly improve the precision and relevance on finding topical influential users in Twitter.