EUVIMEDEUVIMED entdecken
Crossref

Improving the Accuracy of Collaborative Filtering Recommender Systems Using a Multi-Criteria Review-Aware Approach

Aminu Zainab Abubakar, Abdullahi Sale Toro

BIMA JOURNAL OF SCIENCE AND TECHNOLOGY GOMBE · 30.03.2026 · 10 (1A) · 378-389

DOI: 10.64290/bima.v10i1a.1550 ↗

Kurz gefasst

Fachartikel in der Zeitschrift BIMA JOURNAL OF SCIENCE AND TECHNOLOGY GOMBE, online erschienen am 30.03.2026. Verfasst von Aminu Zainab Abubakar und 1 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Traditional Collaborative Filtering (CF) recommender systems primarily rely on single-criterion user–item interactions, limiting their ability to capture the multidimensional nature of user preferences and resulting in reduced predictive accuracy, especially under sparse data conditions. This study proposes a Multi-Criteria Review-Aware Recommender Model (MCR-RM) that extends classical CF by integrating structured review-derived signals—namely aspect-level features, comparative expressions, and emoticon-based sentiment—into the collaborative filtering framework using an attention-based deep learning architecture. The proposed model is evaluated on benchmark datasets (Amazon, Yelp, and TripAdvisor) and compared against baseline CF models (Matrix Factorization and Neural Collaborative Filtering) and state-of-the-art approaches (DeepCoNN and BERT4Rec). Experimental results show that MCR-RM achieves a 4.3% improvement in RMSE and MAE over BERT4Rec, and significantly outperforms traditional CF models in both prediction and ranking tasks. Specifically, the model achieves an average RMSE of 0.782 and NDCG@10 of 0.766, demonstrating superior ranking quality. These findings confirm that incorporating multi-criteria review-aware features into CF enhances recommendation accuracy, robustness, and interpretability. The study contributes a unified hybrid framework that bridges collaborative filtering and review-based deep learning models for next-generation recommender systems.

Zitieren

APA

Abubakar, A. Z., & Toro, A. S. (2026). Improving the Accuracy of Collaborative Filtering Recommender Systems Using a Multi-Criteria Review-Aware Approach. BIMA JOURNAL OF SCIENCE AND TECHNOLOGY GOMBE, 10(1A), 378-389. https://doi.org/10.64290/bima.v10i1a.1550

Vancouver

Abubakar AZ, Toro AS. Improving the Accuracy of Collaborative Filtering Recommender Systems Using a Multi-Criteria Review-Aware Approach. BIMA JOURNAL OF SCIENCE AND TECHNOLOGY GOMBE. 2026;10(1A):378-389. doi:10.64290/bima.v10i1a.1550

BibTeX
@article{Abubakar2026,
  author = {Abubakar, Aminu Zainab and Toro, Abdullahi Sale},
  title = {Improving the Accuracy of Collaborative Filtering Recommender Systems Using a Multi-Criteria Review-Aware Approach},
  journal = {BIMA JOURNAL OF SCIENCE AND TECHNOLOGY GOMBE},
  year = {2026},
  volume = {10},
  number = {1A},
  pages = {378--389},
  doi = {10.64290/bima.v10i1a.1550}
}
Kostenlos starten →