LLM with Relation Classifier for Document-Level Relation Extraction
This addresses the problem of inefficient LLM attention in document-level relation extraction for NLP researchers, representing an incremental improvement by hybridizing classifier and LLM methods.
The paper tackled the performance gap of LLM-based methods in document-level relation extraction by identifying attention dispersion as a key issue and introduced a classifier-LLM approach that selects entity pair candidates before LLM classification. Experiments showed it significantly outperforms recent LLM-based models and narrows the gap with BERT-based state-of-the-art methods.
Large language models (LLMs) have created a new paradigm for natural language processing. Despite their advancement, LLM-based methods still lag behind traditional approaches in document-level relation extraction (DocRE), a critical task for understanding complex entity relations within long context. This paper investigates the causes of this performance gap, identifying the dispersion of attention by LLMs due to entity pairs without relations as a key factor. We then introduce a novel classifier-LLM approach to DocRE. Particularly, the proposed approach begins with a classifier designed to select entity pair candidates that exhibit potential relations and then feed them to LLM for final relation classification. This method ensures that the LLM's attention is directed at relation-expressing entity pairs instead of those without relations during inference. Experiments on DocRE benchmarks reveal that our method significantly outperforms recent LLM-based DocRE models and narrows the performance gap with state-of-the-art BERT-based models.