Unified Structure Generation for Universal Information Extraction
This addresses the challenge of adapting information extraction to diverse schemas and tasks, offering a universal solution that is not incremental but provides broad improvements.
The paper tackles the problem of varying targets and heterogeneous structures in information extraction by proposing UIE, a unified text-to-structure generation framework, which achieved state-of-the-art performance on 4 tasks and 13 datasets across supervised, low-resource, and few-shot settings.
Information extraction suffers from its varying targets, heterogeneous structures, and demand-specific schemas. In this paper, we propose a unified text-to-structure generation framework, namely UIE, which can universally model different IE tasks, adaptively generate targeted structures, and collaboratively learn general IE abilities from different knowledge sources. Specifically, UIE uniformly encodes different extraction structures via a structured extraction language, adaptively generates target extractions via a schema-based prompt mechanism - structural schema instructor, and captures the common IE abilities via a large-scale pre-trained text-to-structure model. Experiments show that UIE achieved the state-of-the-art performance on 4 IE tasks, 13 datasets, and on all supervised, low-resource, and few-shot settings for a wide range of entity, relation, event and sentiment extraction tasks and their unification. These results verified the effectiveness, universality, and transferability of UIE.