A Method of Discovering Genre Similarity using Aspect Based Approach

In the movie industry, movie recommendations are a manner of advertisement or promotion that target customers. In this study, we proposed a method called "TDFIDF" which extracts genre specific keywords from movie reviews using an aspect based approach model. We also proposed a method called "genre score" which indicates a degree of correlation between the movies for discovering a genre similarity. Then, the system recommends movies based on the K-means clustering result. Through this research, we verified that movie reviews contain sufficient information to find characteristics of movie genres in terms of the user perspectives. In addition, our system can suggest relevant movies using our proposed methods.