Neural Collaborative Filtering for Steam Dataset

Neural Collaborative Filtering for Steam_Dataset

It is estimated that nearly 60 percent of Americans play video games, bringing in annual revenue of over $25 billion for PC gaming alone 1 The Steam digital distribution service is the largest digital stores for video games, thus providing a great place to start to look into users’ behavior of purchasing video games. This paper aims to build a recommendation system for predicting user’s preference towards video games based on the purchasing history obtained from Steam API using neural collaborative filtering. The paper also compare the performance between Generalized Matrix Factorization(GMF), one of the neural collaborative filtering model, and traditional Matrix Factorization(MF).