Nowadays, various services are available on the Internet, and a vast amount of website browsing history data is being accumulated. In recent years, even services with few purchase actions per user, such as booking a wedding venue or purchasing insurance, have become available on the Internet. For these services, it is assumed that the interests of users gradually change and narrow down during browsing. Then, when the users decide the product to purchase, it is considered that their interests converge on a specific subject. Therefore, it is important to implement appropriate marketing strategies depending on the degree of convergence of user interests to increase effectiveness. Therefore, a method that can analyze changing user interests over time from browsing history data is desired. In this study, we propose a Time Window Topic Model that can analyze changes in user interests by considering the interests as latent topics. The proposed method can reveal the changes in interests of users even in a real problem where it is difficult to apply conventional topic models. Finally, we verify the usefulness of the proposed method by analyzing an artificial dataset.