Adaptive Candidate Retrieval with Dynamic Knowledge Graph Construction for Cold-Start Recommendation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Yang, Wooseong, Zhang, Weizhi, Liu, Yuqing, Han, Yuwei, Wang, Yu, Lee, Junhyun, Yu, Philip S.
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915635854835712
author Yang, Wooseong
Zhang, Weizhi
Liu, Yuqing
Han, Yuwei
Wang, Yu
Lee, Junhyun
Yu, Philip S.
author_facet Yang, Wooseong
Zhang, Weizhi
Liu, Yuqing
Han, Yuwei
Wang, Yu
Lee, Junhyun
Yu, Philip S.
contents The cold-start problem remains a critical challenge in real-world recommender systems, as new items with limited interaction data or insufficient information are frequently introduced. Despite recent advances leveraging external knowledge such as knowledge graphs (KGs) and large language models (LLMs), recommender systems still face challenges in practical environments. Static KGs are expensive to construct and quickly become outdated, while LLM-based methods depend on pre-filtered candidate lists due to limited context windows. To address these limitations, we propose ColdRAG, a retrieval-augmented framework that dynamically constructs a knowledge graph from raw metadata, extracts entities and relations to construct an updatable structure, and introduces LLM-guided multi-hop reasoning at inference time to retrieve and rank candidates without relying on pre-filtered lists. Experiments across multiple benchmarks show that ColdRAG consistently outperforms strong seven baselines.
format Preprint
id arxiv_https___arxiv_org_abs_2505_20773
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Adaptive Candidate Retrieval with Dynamic Knowledge Graph Construction for Cold-Start Recommendation
Yang, Wooseong
Zhang, Weizhi
Liu, Yuqing
Han, Yuwei
Wang, Yu
Lee, Junhyun
Yu, Philip S.
Information Retrieval
68T05 68T05
The cold-start problem remains a critical challenge in real-world recommender systems, as new items with limited interaction data or insufficient information are frequently introduced. Despite recent advances leveraging external knowledge such as knowledge graphs (KGs) and large language models (LLMs), recommender systems still face challenges in practical environments. Static KGs are expensive to construct and quickly become outdated, while LLM-based methods depend on pre-filtered candidate lists due to limited context windows. To address these limitations, we propose ColdRAG, a retrieval-augmented framework that dynamically constructs a knowledge graph from raw metadata, extracts entities and relations to construct an updatable structure, and introduces LLM-guided multi-hop reasoning at inference time to retrieve and rank candidates without relying on pre-filtered lists. Experiments across multiple benchmarks show that ColdRAG consistently outperforms strong seven baselines.
title Adaptive Candidate Retrieval with Dynamic Knowledge Graph Construction for Cold-Start Recommendation
topic Information Retrieval
68T05 68T05
url https://arxiv.org/abs/2505.20773