Personalized News Recommendation under Cold-Start and Data-Sparsity Conditions: An Empirical Investigation
IPS Journal of Physical Sciences · 2026
DOI: 10.54117/ijps.v3i2.46 ↗Kurz gefasst
Fachartikel in der Zeitschrift IPS Journal of Physical Sciences, 2026. Verfasst von I. J. Mgbeafulike und 3 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Personalized news recommendation systems help reduce information overload by delivering content aligned with users’ interests. However, their effectiveness is constrained by cold-start and data-sparsity problems, which limit reliable user–item relationship learning. This study aims to investigate the effectiveness of a hybrid personalized news recommendation approach under varying cold-start and data-sparsity conditions. An empirical experimental research design is adopted using the Microsoft News Dataset (MIND) as the principal benchmark. Collaborative Filtering (CF), Content-Based Filtering (CBF), a Deep Neural Network (DNN), and a Hybrid model integrating behavioral, content, and neural representations are comparatively evaluated using Precision@10, Recall@10, MAP, Coverage, nDCG@10, and MRR. Experiments examine overall performance, user cold-start, item cold-start, progressive sparsity, component ablation, and statistical differences among models. The results show that the Hybrid model achieved Precision@10 of 0.85, Recall@10 of 0.79, MAP of 0.81, and Coverage of 0.89, compared with 0.68, 0.59, 0.61, and 0.73, respectively, for CF. The findings suggest that integrating complementary recommendation signals can improve robustness when interaction information is limited.
Zitieren
APA
Mgbeafulike, I. J., Okeke, O. C., Mmaduakonam, N. E. G., & Nwaoha, S. (2026). Personalized News Recommendation under Cold-Start and Data-Sparsity Conditions: An Empirical Investigation. IPS Journal of Physical Sciences. https://doi.org/10.54117/ijps.v3i2.46
Vancouver
Mgbeafulike IJ, Okeke OC, Mmaduakonam NEG, Nwaoha S. Personalized News Recommendation under Cold-Start and Data-Sparsity Conditions: An Empirical Investigation. IPS Journal of Physical Sciences. 2026. doi:10.54117/ijps.v3i2.46
BibTeX
@article{Mgbeafulike2026,
author = {Mgbeafulike, I. J. and Okeke, O. C. and Mmaduakonam, N. E. G. and Nwaoha, S.O.},
title = {Personalized News Recommendation under Cold-Start and Data-Sparsity Conditions: An Empirical Investigation},
journal = {IPS Journal of Physical Sciences},
year = {2026},
doi = {10.54117/ijps.v3i2.46}
}
EUVIMED