KGMEL: Knowledge Graph-Enhanced Multimodal Entity Linking

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Kim, Juyeon, Lee, Geon, Kim, Taeuk, Shin, Kijung
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866916699586953216
author Kim, Juyeon
Lee, Geon
Kim, Taeuk
Shin, Kijung
author_facet Kim, Juyeon
Lee, Geon
Kim, Taeuk
Shin, Kijung
contents Entity linking (EL) aligns textual mentions with their corresponding entities in a knowledge base, facilitating various applications such as semantic search and question answering. Recent advances in multimodal entity linking (MEL) have shown that combining text and images can reduce ambiguity and improve alignment accuracy. However, most existing MEL methods overlook the rich structural information available in the form of knowledge-graph (KG) triples. In this paper, we propose KGMEL, a novel framework that leverages KG triples to enhance MEL. Specifically, it operates in three stages: (1) Generation: Produces high-quality triples for each mention by employing vision-language models based on its text and images. (2) Retrieval: Learns joint mention-entity representations, via contrastive learning, that integrate text, images, and (generated or KG) triples to retrieve candidate entities for each mention. (3) Reranking: Refines the KG triples of the candidate entities and employs large language models to identify the best-matching entity for the mention. Extensive experiments on benchmark datasets demonstrate that KGMEL outperforms existing methods. Our code and datasets are available at: https://github.com/juyeonnn/KGMEL.
format Preprint
id arxiv_https___arxiv_org_abs_2504_15135
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle KGMEL: Knowledge Graph-Enhanced Multimodal Entity Linking
Kim, Juyeon
Lee, Geon
Kim, Taeuk
Shin, Kijung
Information Retrieval
Artificial Intelligence
Computation and Language
Entity linking (EL) aligns textual mentions with their corresponding entities in a knowledge base, facilitating various applications such as semantic search and question answering. Recent advances in multimodal entity linking (MEL) have shown that combining text and images can reduce ambiguity and improve alignment accuracy. However, most existing MEL methods overlook the rich structural information available in the form of knowledge-graph (KG) triples. In this paper, we propose KGMEL, a novel framework that leverages KG triples to enhance MEL. Specifically, it operates in three stages: (1) Generation: Produces high-quality triples for each mention by employing vision-language models based on its text and images. (2) Retrieval: Learns joint mention-entity representations, via contrastive learning, that integrate text, images, and (generated or KG) triples to retrieve candidate entities for each mention. (3) Reranking: Refines the KG triples of the candidate entities and employs large language models to identify the best-matching entity for the mention. Extensive experiments on benchmark datasets demonstrate that KGMEL outperforms existing methods. Our code and datasets are available at: https://github.com/juyeonnn/KGMEL.
title KGMEL: Knowledge Graph-Enhanced Multimodal Entity Linking
topic Information Retrieval
Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2504.15135