EUVIMEDEUVIMED entdecken
Crossref

Improving personalization based on intrinsic differences in user preferences in streaming session-based recommendations

Vinnakota Saran Chaitanya, Aryan, P Santhi Thilagam

The Computer Journal · 08.11.2026

DOI: 10.1093/comjnl/bxag118 ↗

Kurz gefasst

Fachartikel in der Zeitschrift The Computer Journal, online erschienen am 08.11.2026. Verfasst von Vinnakota Saran Chaitanya und 2 weiteren Autor:innen. Die Arbeit nennt 46 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The streaming session-based recommendation (SSR) models have evolved to generate recommendations from continuously arriving session data in the form of clickstreams, but this continuous arrival causes them to forget previously learned user preferences. This necessitates the models to use a reservoir to store massively arriving clickstreams. Due to the reservoir’s limited capacity, existing clickstreams are replaced to accommodate newly arriving clickstreams and require sampling of existing clickstreams for model retraining to reduce retraining time. Existing SSR models adopt random replacement and sampling approaches, often overlooking clickstreams of users with unique preferences, causing the models to forget these intrinsic differences in user preferences and may not generate appropriate personalized recommendations. This work introduces a clustering-based reservoir approach that dynamically clusters incoming clickstreams and eliminates noisy interactions using dwell time analysis, thus preserving unique user preferences in the reservoir. Also, it proposes a priority-based sampling approach to selectively sample the representative clickstreams from each cluster, ensuring that the model is periodically retrained using unique preferences. Furthermore, personalized recommendations are generated using a matrix factorization-based attentive session encoder by considering both historical and ongoing session information. Experimental evaluations across three publicly available benchmark datasets show that the proposed approach achieves better performance than the state-of-the-art baseline approaches.

Zitieren

APA

Chaitanya, V. S., Aryan, & Thilagam, P. S. (2026). Improving personalization based on intrinsic differences in user preferences in streaming session-based recommendations. The Computer Journal. https://doi.org/10.1093/comjnl/bxag118

Vancouver

Chaitanya VS, Aryan, Thilagam PS. Improving personalization based on intrinsic differences in user preferences in streaming session-based recommendations. The Computer Journal. 2026. doi:10.1093/comjnl/bxag118

BibTeX
@article{Chaitanya2026,
  author = {Chaitanya, Vinnakota Saran and Aryan and Thilagam, P Santhi},
  title = {Improving personalization based on intrinsic differences in user preferences in streaming session-based recommendations},
  journal = {The Computer Journal},
  year = {2026},
  doi = {10.1093/comjnl/bxag118}
}
Kostenlos starten →