A COLLABORATIVE CONTENT-BASED MOVIE RECOMMENDER SYSTEM

Abstract

The introduction of the Internet has brought us into a world that portrays a huge amount of information items such as web pages, music, movies, and books, with varying quality. As a result of this huge amount of items, people get confused in choosing a particular item at any point in time. In this thesis, concepts of collaborative filtering, and content-based filtering techniques are studied in making personalized recommendation of movie items to users. The main idea behind making predictions using content data is the assumption that people with similar characteristics enjoy similar movies; the movie item for the new user is predicted based on the profile of existing similar users. Pearson’s Correlation Coefficient is employed in the collaborative filtering aspect due to its ability to manipulate numerical data as well as determine linear relationship among existing users. Its steps involve a user-user representation, similarity generation and prediction generation with a goal to produce a predicted opinion of the active user about a specific item. Concept of parental control is also incorporated for enhancement. Evaluation of the system was done using Precision, Recall, F-measure, Discounted Cumulative Gain (DCG), Idealized Discounted Gain (IDCG), normalized Discounted Gain (nDCG), and Mean Absolute Error (MAE). A total of 346 datasets were used, out of which 126 were gathered from video shops, and local, the remaining 220 were extracted from Internet Movie Database (IMDb); these were used for the experiments, and the results generated through mining of data obtained from profiles and ratings of system users prove the system average ranking quality of the collaborative filtering algorithm is 95.9%.