Recommender systems aim to facilitate decision making for users by offering them information according to their preferences, they are now popular in several application domains. The evaluation of recommender systems is very important to have an effective application in practice. In addition, it focuses to find better algorithms and evaluate their performance. However, researchers did not give much attention to it in this field. There are various ways to evaluate a recommender system. In this paper, we will discuss the main types of evaluations in this domain, which are offline and online evaluation, we will start with an overview of recommender systems, then we will present each type of evaluation, and we will compare the offline and the online evaluation for recommender systems. We will base on ten factors which are, the reproducibility, the reliability of the results of each type of evaluation, the preparation cost, the evaluation, the stability, the possibility of extensibility that’s mean if we can add new metrics or not, the scalability, the passed time, deep analysis, and the sparsity metric. Finally, we will discuss the factors which are presented in the comparison.
[1]
Ahmed Zellou,et al.
A Comparative Study on Recommender Systems Approaches
,
2019,
BDIoT.
[2]
Yu Mao,et al.
A Bidirectional Collaborative Filtering Recommender System Based on EM Algorithm
,
2017
.
[3]
Bracha Shapira,et al.
Recommender Systems Handbook
,
2015,
Springer US.
[4]
Sean M. McNee,et al.
Being accurate is not enough: how accuracy metrics have hurt recommender systems
,
2006,
CHI Extended Abstracts.
[5]
Sam Gabrielsson,et al.
The use of Self-Organizing Maps in Recommender Systems : A survey of the Recommender Systems field and a presentation of a State of the Art Highly Interactive Visual Movie Recommender System
,
2006
.
[6]
Jonathan L. Herlocker,et al.
Evaluating collaborative filtering recommender systems
,
2004,
TOIS.