Multi-Item-Query Attention for Stable Sequential Recommendation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Xu, Mingshi, Zhu, Haoren, Ng, Wilfred Siu Hung
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908565293236224
author Xu, Mingshi
Zhu, Haoren
Ng, Wilfred Siu Hung
author_facet Xu, Mingshi
Zhu, Haoren
Ng, Wilfred Siu Hung
contents The inherent instability and noise in user interaction data challenge sequential recommendation systems. Prevailing masked attention models, relying on a single query from the most recent item, are sensitive to this noise, reducing prediction reliability. We propose the Multi-Item-Query attention mechanism (MIQ-Attn) to enhance model stability and accuracy. MIQ-Attn constructs multiple diverse query vectors from user interactions, effectively mitigating noise and improving consistency. It is designed for easy adoption as a drop-in replacement for existing single-query attention. Experiments show MIQ-Attn significantly improves performance on benchmark datasets.
format Preprint
id arxiv_https___arxiv_org_abs_2509_24424
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Multi-Item-Query Attention for Stable Sequential Recommendation
Xu, Mingshi
Zhu, Haoren
Ng, Wilfred Siu Hung
Information Retrieval
Artificial Intelligence
Machine Learning
The inherent instability and noise in user interaction data challenge sequential recommendation systems. Prevailing masked attention models, relying on a single query from the most recent item, are sensitive to this noise, reducing prediction reliability. We propose the Multi-Item-Query attention mechanism (MIQ-Attn) to enhance model stability and accuracy. MIQ-Attn constructs multiple diverse query vectors from user interactions, effectively mitigating noise and improving consistency. It is designed for easy adoption as a drop-in replacement for existing single-query attention. Experiments show MIQ-Attn significantly improves performance on benchmark datasets.
title Multi-Item-Query Attention for Stable Sequential Recommendation
topic Information Retrieval
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2509.24424