A Comprehensive Study and Evaluation of Recommender Systems

Abstract

This paper presents a brief study within the field of recommender systems and describes the current generation of recommender system tools and evaluation metrics. Recommender system comprises of three methods, namely content-based filtering, collaborative filtering, and hybrid filtering algorithms. It addresses two common scenarios in collaborative filtering: rating prediction and item recommendation. There are many well-known accuracy metrics which replicate evaluation goals. This paper describes a few framework and libraries of recommender system that implements a state-of-the-art algorithmic rule furthermore as series of evaluation metric. We tend to find which recommender system tool performs quicker than different, whereas achieving competitive evaluating performance with steering for the comprehensive evaluation and choice of recommender algorithm.