UniEX: An Effective and Efficient Framework for Unified Information Extraction via a Span-extractive Perspective
This addresses the need for efficient and effective universal information extraction systems that work across different tasks and schemas, representing a novel method rather than an incremental improvement.
The authors tackled the problem of unified information extraction across multiple tasks by proposing UniEX, a span-extractive framework that converts text-based IE tasks into token-pair problems. The results show that UniEX outperforms generative universal IE models on 14 benchmark datasets in both performance and inference speed, with state-of-the-art performance in low-resource scenarios.
We propose a new paradigm for universal information extraction (IE) that is compatible with any schema format and applicable to a list of IE tasks, such as named entity recognition, relation extraction, event extraction and sentiment analysis. Our approach converts the text-based IE tasks as the token-pair problem, which uniformly disassembles all extraction targets into joint span detection, classification and association problems with a unified extractive framework, namely UniEX. UniEX can synchronously encode schema-based prompt and textual information, and collaboratively learn the generalized knowledge from pre-defined information using the auto-encoder language models. We develop a traffine attention mechanism to integrate heterogeneous factors including tasks, labels and inside tokens, and obtain the extraction target via a scoring matrix. Experiment results show that UniEX can outperform generative universal IE models in terms of performance and inference-speed on $14$ benchmarks IE datasets with the supervised setting. The state-of-the-art performance in low-resource scenarios also verifies the transferability and effectiveness of UniEX.