Few-Shot, No Problem: Descriptive Continual Relation Extraction

TL;DR

The paper addresses Few-Shot Continual Relation Extraction (FCRE) under sequential tasks with limited samples. It proposes a retrieval-based framework that uses Large Language Model (LLM) generated relation descriptions as stable class prototypes, a bi-encoder for joint sample/class representation learning, and a Descriptive Retrieval Inference (DRI) mechanism to unify prototype proximity and description semantics. The learning objective combines sample-space losses (Supervised Contrastive and Hard Soft Margin) with description-centered losses (Hard Margin and Mutual Information) in a joint training objective, and a memory-based rehearsal strategy preserves prior knowledge. Empirical results on FewRel and TACRED with multiple backbones demonstrate state-of-the-art accuracy and robust forgetting resistance, highlighting the value of descriptive grounding and retrieval for dynamic relation extraction.

Abstract

Few-shot Continual Relation Extraction is a crucial challenge for enabling AI systems to identify and adapt to evolving relationships in dynamic real-world domains. Traditional memory-based approaches often overfit to limited samples, failing to reinforce old knowledge, with the scarcity of data in few-shot scenarios further exacerbating these issues by hindering effective data augmentation in the latent space. In this paper, we propose a novel retrieval-based solution, starting with a large language model to generate descriptions for each relation. From these descriptions, we introduce a bi-encoder retrieval training paradigm to enrich both sample and class representation learning. Leveraging these enhanced representations, we design a retrieval-based prediction method where each sample "retrieves" the best fitting relation via a reciprocal rank fusion score that integrates both relation description vectors and class prototypes. Extensive experiments on multiple datasets demonstrate that our method significantly advances the state-of-the-art by maintaining robust performance across sequential tasks, effectively addressing catastrophic forgetting.

Figures (6)

• Figure 1: Existing FCRE methods face catastrophic forgetting due to the limited and poor quality of old training samples stored in the memory buffer.
• Figure 2: Prompt to generate relation descriptions with LLMs.
• Figure 3: Our Framework.
• Figure 4: t-SNE visualization of the representations of 6 relations post-training, with and without descriptions, using our retrieval strategy.
• Figure 5: The impact of refined descriptions generated by LLMs. The green, orange, and blue bars show respectively the final accuracies of DCRE when using refined descriptions, original descriptions, and without using descriptions.