Applying Statistical Approach to Topic Analysis for more Comprehensive and Appropriate Modeling

Topic analysis is useful to construct probabilistic modeling and interpret real data. It is applied to data analysis in various fields. Topic analysis is expected to construct modeling appropriately and obtain interpretable results to assist our decision making about the field. More and more data of more and more fields give us many problems to solve with topic analysis. In this paper, we focus on treating complex data, constructing an appropriate model and interpreting data structure more meaningfully with topic analysis. To solve these problems, we suggest combinning topic analysis methods and statistical approaches. Finally, some results of experiments to study utility of our suggesting methods are given.