Core Concepts
GEMEL is a generative framework that leverages Large Language Models to enhance Multimodal Entity Linking efficiently.
Abstract
Introduction to Multimodal Entity Linking (MEL)
Challenges in existing MEL methods and the need for GEMEL
Methodology of GEMEL, including Feature Alignment and Language Model Generation
Experimental results showcasing the effectiveness of GEMEL on two MEL datasets
Analysis of generality, scalability, demonstration selection, and popularity bias in LLMs
Case study illustrating the impact of visual information on entity linking accuracy
Stats
"With only ∼0.3% of the model parameters fine-tuned, GEMEL achieves state-of-the-art results on two well-established MEL datasets."
"GEMEL exhibits high parameter efficiency and strong scalability."
"GEMEL outperforms all other approaches and achieves state-of-the-art performance on both MEL datasets."
Quotes
"Multimodal Entity Linking has attracted increasing attention in the natural language processing community."
"GEMEL can leverage the capabilities of LLMs from large-scale pre-training to directly generate corresponding entity names."