Analysis of Movie Recommendation Systems
Diwyanshu, Gurharpartap Singh, Harsh Agarwal, Gull Kaur · 2023
This research paper presents a comprehensive literature review of various movie recommendation systems, including techniques such as sentiment analysis of comments, movie trailer data, facial expressions, browsing history, and view percentage. The paper also compares various recommenders based on deep learning and swarm algorithms. Furthermore, this paper provides a novel approach for movie recommender system that uses a hybrid model combining content-based and collaborative filtering approaches. Specifically, the proposed approach involves creating a matrix factorization-based model and providing certain custom-designed features, such as averages of all movies and users, top similar movies, and top similar users, to produce the final model. The proposed approach is evaluated on the MovieLens dataset and achieves a mean absolute percentage error (MAPE) of 19.868316 and a root mean squared error (RMSE) of 0.672788. These outcomes illustrate the potency of the suggested strategy and its potential to enhance movie recommendation systems.