Measuring and visualizing interest similarity between microblog users

Microblog users share their life status and opinions via microposts, which usually reflect their interests. Measuring interest similarity between microblog users has thus received increasing attention from both academia and industry. In this paper, we design a novel framework for measuring and visualizing user interest similarity. The framework consists of four components: (1) Interest representation. We extract keywords from microposts to represent user interests. (2) Interest similarity computation. Based on the interest keywords, we design a ranking framework for measuring the interest similarity. (3) Interest similarity visualization. We propose a integrated word cloud scenario to provide a novel visual representation of user interest similarity. (4) Annotation data collection. We design an interactive game for microblog users to collect user annotations, which are used as training dataset for our similarity measuring method. We carry out experiments on Sina Weibo, the largest microblogging service in China, and get encouraging results.