ACME-RE: Adaptive Context Memory and Evidence-guided Relation Extraction for Document-Level Relation Extraction

ACL ARR 2025 February Submission6657 Authors

Abstract:

Document-level relation extraction (DocRE) is the task of identifying all relations between each entity pair in a document. Current methods still have room for improvement in handling implicit relationships, which are relations not explicitly stated in the text but can be inferred from the context.

To address this limitation, we introduce the concept of context informativeness for entity pairs and propose ACME-RE (Adaptive Contextual Memory-Enhanced Relation Extraction), a novel framework for document-level relation extraction (DocRE). By introducing Evidence-guided context informativeness and an adaptive category memory module, ACME-RE significantly improves the performance of implicit relationship extraction. Experimental results demonstrate that our method achieves state-of-the-art (SOTA) performance on the Re-DocRED dataset. This research provides a more comprehensive solution for document-level relation extraction and offers valuable insights for future studies.

Paper Type: Long

Research Area: Information Extraction

Research Area Keywords: Information Extraction - Relation extraction

Contribution Types: Model analysis & interpretability

Languages Studied: English

Submission Number: 6657