ReaORE: Reasoning-Guided Progressive Open Relation Extraction Empowered by Large Reasoning Models
This work addresses the challenge of generalizing to unseen relation types in open relation extraction, a key problem for information extraction in dynamic domains.
ReaORE introduces a coarse-to-fine reasoning framework for open relation extraction that first filters candidate relations via multi-aspect reasoning and embedding similarity, then predicts the target relation through fine-grained comparative reasoning, achieving state-of-the-art results on two benchmark datasets.
Open Relation Extraction (OpenRE) requires a model to extract unseen relations between head and tail entities from unstructured text for real-world applications. The core challenge of OpenRE lies in achieving reliable generalization to unseen relation types. Current OpenRE approaches either employ clustering techniques, which cannot generate relation labels and suffer from poor generalization, or rely on direct relation label generation via Large Language Models (LLMs), which lack sufficient discriminative capacity to distinguish easily confused relations. To address these limitations, we propose Reasoning-guided progressive OpenRE (ReaORE), a framework for performing relation extraction through coarse-to-fine relation reasoning. Specifically, ReaORE consists of two key stages: (i) relation filtering, which reasons over multiple aspects to understand relations and instances, yielding an initial relation set, and further supplements and filters relations via embedding-based similarity to ensure the target relation is included; (ii) relation prediction, which aims to predict the target relations from the above set via fine-grained comparative reasoning to better distinguish easily confused relations. Extensive experiments on two widely used OpenRE datasets demonstrate that ReaORE outperforms existing baselines.