Plot-Topic based Movie Recommendation System using WordNet
Potnuru Sirisha, Golagani Lavanya Devi, N. Ramesh · 2022
With the immense increase of content, finding the desired movie/content is a challenging task as it is being limited by genre, language filtering. Recommending the same genre of movies collectively doesn't interest users anymore. Users are interested in finding movies with particular storyline irrespective of the genre. A content-based filtering method is proposed in this paper, where recommendations will be based on a particular topic representing an aspect portrayed in the movie. Topics are assigned to movies using WordNet and topic modeling over their plot summaries. The dataset used consists of movies scraped from Internet Movie Database (IMDb) users created movie lists which are based on the movie's plot-aspect. Precision of predicted plot topics is calculated manually over movies from IMDb's most popular Telugu movies. This is a naïve approach to identify topics involved in a movie. Further research can be done to predict movie lists based on plot-aspects by identifying topic correlations.