An extracting method of movie genre similarity using aspect-based approach in social media
Youngsub Han, Yanggon Kim · ACM SIGAPP Applied Computing Review · 2017
In the movie industry, a movie recommendation is a manner of advertisement or promotion. To recommend movies, we used movie reviews and YouTube comments because user-generated data containing people's opinions. To analyze the data, we used MSP model, which is one of aspect-based approaches and it guarantees relatively higher accuracy than existing approaches. To discover a genre similarity, we proposed two methods, which are "TDF-IDF" and "Genre Score". The "TDF-IDF" is designed to extract genre specific keywords and the "Genre Score" indicates a degree of correlation between a movie and genres. Then, the system recommends movies based on results of K-means and K-NN.