Inferring user interests on tumblr

Abstract

Inferring user interests is one of the core tasks for online social media services. It has direct impacts on personalization, recommendation and many other features for enhanced user experience. In this work, we proposed a novel bi-relational graph model to discover individual users’ topics of interest from Tumblr, which is one of the most popular microblogging services. The proposed graph model contains two sub-graphs: one corresponds to users and the other corresponds to topics. Such a representation allows for effective exploitation of both user homophily relation and topic correlation simultaneously. This is in contrast with previous work where these two factors are considered in isolation. Subsequently, the user interest discovery problem is formulated as a multi-label learning problem on the bi-relational graph, with the goal to estimate the optimized associations between user nodes and topic nodes across the two sub-graphs. Our experiment is carried out with the complete data collected from Tumblr for a full month. To our knowledge, this work is the first attempt to conduct large-scale user interest inference on the platform.