Automatic extractive summarization for Japanese documents by LDA

The demand for automatic summarization of newspaper headlines and article sum- maries has increasing with various studies on being currently conducted. However, there are only a few Japanese documents as compared English documents. In this paper, wheter existing methods can be effective academic pa- pers written in is verified. First, we demonstrate the effectiveness topic-based extractive Latent Semantic Analysis (LSA). Then, more effec- tive possible by using Dirichlet Allocation (LDA) demonstrated.