A systematic literature review of multicriteria recommender systems

Since the first years of the 90s, recommender systems have emerged as effective tools for automatically selecting items according to user preferences. Traditional recommenders rely on the relevance assessments that users express using a single rating for each item. However, some authors started to s...

Full description

Saved in:
Bibliographic Details
Published in:The Artificial intelligence review 2021, Vol.54 (1), p.427-468
Main Authors: Monti, Diego, Rizzo, Giuseppe, Morisio, Maurizio
Format: Article
Language:English
Subjects:
Algorithms
Analysis
Artificial Intelligence
Computer Science
Data mining
Literature reviews
Machine learning
Multiple criterion
Recommender systems
System effectiveness
Systematic review
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Since the first years of the 90s, recommender systems have emerged as effective tools for automatically selecting items according to user preferences. Traditional recommenders rely on the relevance assessments that users express using a single rating for each item. However, some authors started to suggest that this approach could be limited, as we naturally tend to formulate different judgments according to multiple criteria. During the last decade, several studies introduced novel recommender systems capable of exploiting user preferences expressed over multiple criteria. This work proposes a systematic literature review in the field of multicriteria recommender systems. Following a replicable protocol, we selected a total number of 93 studies dealing with this topic. We subsequently analyzed them to provide an answer to five different research questions. We considered what are the most common research problems, recommendation approaches, data mining and machine learning algorithms mentioned in these studies. Furthermore, we investigated the domains of application, the exploited evaluation protocols, metrics and datasets, and the most promising suggestions for future works.
ISSN:0269-2821
1573-7462
DOI:10.1007/s10462-020-09851-4