Zero-shot Triplet Extraction by Template Infilling
This enables zero-shot learning for relation extraction, addressing a bottleneck in adapting to new relations without noisy synthetic data, which is incremental but practical for NLP applications.
The paper tackled the problem of extracting unseen relations in triplet extraction without additional training data by reducing it to a template infilling task with a pre-trained language model, achieving state-of-the-art performance on FewRel and Wiki-ZSL datasets.
The task of triplet extraction aims to extract pairs of entities and their corresponding relations from unstructured text. Most existing methods train an extraction model on training data involving specific target relations, and are incapable of extracting new relations that were not observed at training time. Generalizing the model to unseen relations typically requires fine-tuning on synthetic training data which is often noisy and unreliable. We show that by reducing triplet extraction to a template infilling task over a pre-trained language model (LM), we can equip the extraction model with zero-shot learning capabilities and eliminate the need for additional training data. We propose a novel framework, ZETT (ZEro-shot Triplet extraction by Template infilling), that aligns the task objective to the pre-training objective of generative transformers to generalize to unseen relations. Experiments on FewRel and Wiki-ZSL datasets demonstrate that ZETT shows consistent and stable performance, outperforming previous state-of-the-art methods, even when using automatically generated templates. https://github.com/megagonlabs/zett/