Implementation of an Alternating Least Square Model Based Collaborative Filtering Movie Recommendation System on Hadoop and Spark Platforms

Nowadays, consumers and businesses all face the problem of information explosion. Recommendation systems represent a powerful solution This study practices a movie recommendation system to give suggestions of films to the movie-watcher, enabling him to consume more while shortening the time interval between payments. This research implements a prototype recommendation system based on collaborative filtering with Alternating Least Squares (ALS) algorithm. Collaborative filtering has the advantage of avoiding possible violation of the Personal Information Protection Act and reducing the possibility the errors caused by poor quality of personal information. However, one of its shortcomings is the scalability. Our study attempts to improve it by adopting Spark with Hadoop Yarn platform and uses it to compute movie recommendation and to store data respectively. The result of this research shows that the proposed system offers recommendations with satisfying accuracy while keeping acceptable computation time.