These days, numerous online reviews for restaurants are available on the Internet. People often refer to these reviews to gain insights and make decisions about which restaurants they would like to visit. However, sifting through a large number of reviews can lead to confusion due to the abundance of lengthy texts. To address this issue, our study proposed a personalized restaurant recommendation system based on a topic list derived from topic modeling of online reviews. Additionally, we developed a user-friendly web application (web app) that enables users to easily find restaurants based on their preferences, including food type, location, district, and restaurant topic. The initial step involved collecting and preprocessing online reviews of restaurants in Seoul, South Korea. We then utilized Latent Dirichlet Allocation (LDA) to group restaurants according to frequently discussed themes or topics. These generated groups serve as topic descriptions, and each restaurant is assigned to a specific group based on this categorization. However, we recognized that some of the topic descriptions may contain menu names that could be unfriendly or new for certain users. To address this, we integrated OpenAI models through its Application Programming Interface (API) and developed an additional page that allows users to inquire about the full description and view image of the menu in their preferred language. The study is expected to benefit users by offering personalized restaurant recommendations and increasing spending, thus boosting economic activity in the post-pandemic era.