Polarized Topic Modeling for User Characteristics in Online Discussion Community

Topic modeling methods, such as latent Dirichlet allocation (LDA), are successfully applied in a number of computational linguistics applications. This paper presents a new approach to topic modeling within a new domain other than linguistic analysis. We present a pilot study where an LDA model is applied to an online community rather than the textual contents they produced using the idea that a user in an article is analogous to a word in a document within the context of the LDA model. We also propose a method for determining polarity using positive (+) and negative (-) signs regarding topics. As a result, each user has a topic score whose absolute value is equal to the topic distribution learned from topic modeling, and its sign indicates the polarity on that specific subject. We demonstrate the effectiveness of our proposed approach with experimental results, which provide opportunities to apply the LDA model to targets other than lexical elements.