Weak Reward Model Transforms Generative Models into Robust Causal Event Extraction Systems
Abstract: The inherent ambiguity of cause and effect boundaries poses a challenge in evaluating causal event extraction tasks. Traditional metrics like Exact Match and BertScore poorly reflect model performance, so we trained evaluation models to approximate human evaluation, achieving high agreement. We used them to perform Reinforcement Learning with extraction models to align them with human preference, prioritising semantic understanding. We successfully explored our approach through multiple datasets, including transferring an evaluator trained on one dataset to another as a way to decrease the reliance on human-annotated data. In that vein, we also propose a weak-to-strong supervision method that uses a fraction of the annotated data to train an evaluation model while still achieving high performance in training an RL model.
Paper Type: Long
Research Area: Information Extraction
Research Area Keywords: event extraction,human evaluation,automatic evaluation,data augmentation,reinforcement learning,causality,information extraction
Contribution Types: Approaches to low-resource settings, Publicly available software and/or pre-trained models, Data resources
Languages Studied: English
Submission Number: 1266
Loading