Modeling User Interests Using Topic Model

In recommender systems, modeling user interest is a basic step to understand user's personal features. Traditional methods mostly just use the items that the target users navigated as their interests, which makes the inherent information unclear to the system and thus the recommendations are not intelligent enough. In this paper, we investigate the utility of topic model called LDA for the task of modeling user interests which are assumed as the latent variables behind user activities in the systems. By using such a probabilistic model on the generation of user profiles, the relationships amongst user、item and interest are constructed. Each user can be considered as a generation of the model. Based on this user interest model, we propose three item ranking methods for personalized recommendation. In order to get a better model for each algorithm, we use a simple automatic parameter turning way to select the model parameters. After carefully selecting the parameters, our methods can all receive encouraging results for recommender systems.