A Study on Evaluation Analysis for Products Using Topic Model
Shingo Tsukioka, Tomohiro Yoshikawa, Takeshi Furuhashi · 2016
Recently, E-commerce (EC) market has been expanding by diffusion of Web. Accordingly, a lot of reviews are written by users every day. Reviews contain a lot of useful information about items. Thus, there are strong needs for an automatic analysis method of reviews. For a user, it is useful for decisions whether to purchase the items or not by knowing the details of items. For a company, it is useful for marketing of the items by analyzing reviews. This paper studies the factor analysis of the evaluation of hotel putting the importance on the comparison among hotels by using user-reviews. An analysis method using a topic model and a multiple regression analysis is proposed. In the proposed method, each topic proportion is the explanatory variable and evaluation rating of review is the objective variable. This paper applies the proposed method to real user-reviews and investigates the effectiveness of this method. It also shows the difference of evaluation factor among the locality of hotels.