Enhancing Recommendation Systems Through Contextual Bandit Models

Zhaoxin Chen

2024

Abstract

The paper introduces the widespread application of personalized recommendation systems across various platforms, which provide tailored online services to users based on their historical movements and personal profiles, thereby satisfying individualized needs and enhancing user experience and satisfaction. It identifies the core objectives of recommendation systems as recommending items most likely to bring maximum utility to users while exploring users' potential interest points to balance the "Exploration vs. Exploitation" problem. The paper proposes a recommendation system based on the Multi-Armed Bandit (MAB) model, which combines the Upper Confidence Bound (UCB) algorithm and the Linear Upper Confidence Bound (LinUCB) algorithm into the mixed algorithm, incorporating context-aware selection, dynamic adjustment, and weighted averaging to optimize the effectiveness of the hybrid algorithm, ensuring robust and reliable decision-making in diverse scenarios. It dynamically adjusts weights to balance the recommendation needs for new and existing users, thereby improving overall system performance and user satisfaction. The system design addresses the challenges of limited interaction records for new users and limited interaction information and user features within the system.

Paper Citation

in Harvard Style

Chen Z. (2024). Enhancing Recommendation Systems Through Contextual Bandit Models. In Proceedings of the 1st International Conference on Engineering Management, Information Technology and Intelligence - Volume 1: EMITI; ISBN 978-989-758-713-9, SciTePress, pages 622-627. DOI: 10.5220/0012960800004508

in Bibtex Style

@conference{emiti24,
author={Zhaoxin Chen},
title={Enhancing Recommendation Systems Through Contextual Bandit Models},
booktitle={Proceedings of the 1st International Conference on Engineering Management, Information Technology and Intelligence - Volume 1: EMITI},
year={2024},
pages={622-627},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012960800004508},
isbn={978-989-758-713-9},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on Engineering Management, Information Technology and Intelligence - Volume 1: EMITI
TI - Enhancing Recommendation Systems Through Contextual Bandit Models
SN - 978-989-758-713-9
AU - Chen Z.
PY - 2024
SP - 622
EP - 627
DO - 10.5220/0012960800004508
PB - SciTePress