Work

The Multi-Modal Recommendation System Based on Behavior Benchmark Adjustment and Modality Balance

Year: 2025

Type: conference-paper

Abstract: This work presents a multi-modal recommendation system based on behavior benchmark adjustment and modality balance (BAMB), designed to improve recommendation accuracy by addressing noise during feature extraction and integration. The model dynamically adjusts feature weights from each modality based on user behavior, reducing noise interference. A modality balancer further optimizes the weights during feature integration to ensure consistency and accuracy. Experimental results on three Amazon datasets (Baby, Sports, Clothes) demonstrate the model's superiority in multiple evaluation metrics, validating its effectiveness and potential for improvement.

Source: International Conference on Data Management and Computer Science (ICDMCS)

DOI: 10.1109/icdmcs69443.2025.00019

Authors: Qinghong Jin, Weiqian Li, Jinguang Chen

Institution: Xi'an Polytechnic University

Language: English

Topic: Recommender Systems and Techniques

Subfield: Information Systems

Field: Computer Science

Domain: Physical Sciences

Open Access status: hybrid

OpenAlex ID: W7202398024

FWCI:

Cites: 14

Cited by:

Related to: 10