What can we recommend to game players? - Implementing a system of analyzing game reviews

With the rapid development of game industry, games take an increasing important role of entertainment in our daily life. With more and more new games coming to the market, it is difficult for players to choose suitable games. This thesis proposed a way of analyzing game user reviews to help players selecting games. A large amount of game reviews is generated by game players in text format every day, evaluating thousands of reviews one by one becomes impossible. Topic mining techniques makes it possible to extract the topics(aspects) from a large amount of textual data and summarize the reviews in a general level. A lot of research has put efforts on developing and optimizing topic mining algorithms, including developing systems or applications based on these algorithms to analyze user review data from different sources. Theses source includes social media platforms such as Twitter, mobile application stores such as Google Play, movies websites such as Netflix, etc. However, there is a lack of an application especially focus on game reviews. The aim of this thesis is to develop a system which covers all the necessary steps of topic mining of game reviews, including review data collecting, storage, processing and visualization. Topics generated by the system are supposed to reveal the content of the game and user’s game experience. Testing and evaluation of the system are conducted after implementation. After a series of experiments, it is validated that the system runs smoothly in real-world situations and produces suitable topics that can help users to understand and select games