A Comparative Study of Sentiment-Aware Collaborative Filtering Algorithms for Arabic Recommendation Systems

The rapid proliferation of online information necessitates efficient Recommendation Systems (RSs) to assist users in discovering relevant content. While English-language RSs have received significant attention, research on Arabic RSs remains limited despite the increasing demand for Arabic digital c...

Full description

Saved in:
Bibliographic Details
Published in:IEEE access Vol. 12; pp. 174441 - 174454
Main Authors: Al-Ghuribi, Sumaia Mohammed, Mohd Noah, Shahrul Azman, Mohammed, Mawal A., Tiwary, Neeraj, Saat, Nur Izyan Yasmin
Format: Journal Article
Language:English
Published: Piscataway IEEE 2024
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Subjects:
Accuracy
Algorithms
Arabic language
Classification algorithms
Collaboration
Collaborative filtering
Comparative studies
deep autoencoder network
Deep learning
Filtration
Machine learning algorithms
Natural language processing
Neural networks
Prediction algorithms
Ratings
Recommender systems
Reviews
Sentiment analysis
Support vector machines
user reviews
ISSN:2169-3536, 2169-3536
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The rapid proliferation of online information necessitates efficient Recommendation Systems (RSs) to assist users in discovering relevant content. While English-language RSs have received significant attention, research on Arabic RSs remains limited despite the increasing demand for Arabic digital content. This paper addresses the scarcity of Arabic-focused Collaborative Filtering (CF) approaches for RS. Recognizing the wealth of information embedded in user reviews, we propose novel review-based CF approaches tailored for Arabic, aiming to enhance recommendation accuracy for Arab users. Our work comprises three key stages: we first develop a comprehensive Arabic lexicon specifically for the book domain. Secondly, using this lexicon we then propose three distinct sentiment-aware ratings, leveraging sentiment analysis of Arabic reviews to enrich traditional rating predictions. Thirdly, these sentiment-aware ratings are integrated into ten diverse CF algorithms from the Surprise library and a deep autoencoder neural network, covering a spectrum of traditional and modern approaches. Extensive experiments conducted on the Large Arabic Book Reviews (LABR) dataset demonstrate the superior performance of our proposed sentiment-aware ratings compared to baseline methods across all evaluated metrics. Further analysis reveals the importance of appropriate sentiment word extraction methods and lexicon selection for accurate sentiment rating calculation. Finally, this study makes a significant contribution to the field of Arabic CF recommendation systems by providing a comprehensive framework for leveraging user review and underscores the importance of further research in this area.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:2169-3536
2169-3536
DOI:10.1109/ACCESS.2024.3489658