A MOVIE DATA RECOMMENDATION SYSTEM USING A HYBRID MODEL

Computer Science

In recent years, the rate of expansion in the amount of information available online has led to both a surge in Internet users and massive amounts of data. The use of the recommendation system has been made possible by the enormous amount of information available to users. This study, proposed a hybrid recommender engine that could combines recommendations from content-based and collaborative filtering. This aims to investigate how existing collaborative filtering frameworks can improve prediction accuracy. We examine whether a recommendation system that combines content-based and collaborative filtering, employing a Mahout Structure and developed on Hadoop, will enhance accuracy of the recommendation and as well resolve adaptability problems presently encountered in handling huge data sizes for users recommendation of items. The enhancement of features was used. The hybrid feature augmentation technique was used, in which the output from collaborative filtering was used as an input to content-based recommendation. To extract user and item content features, the well-known Movie-Lens data was linked with the Internet Movie Database (IMDB). The text files created as a result of integrating the two databases were then used as inputs for Mahout's collaborative filtering architecture. The ideal Mahout Components parameter optimization for our model was found through a number of tests. By comparing our model's Root Mean Square Error to that of the most recent model, we further investigated these models. When compared to the pure collaborative model, the study demonstrated a considerable improvement. Our investigation showed that the derived user and item content attributes may result in more accurate prediction

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.