Causal Intervention-based Prompt Debiasing for Event Argument Extraction

TL;DR

This paper compares prompt formulation strategies for event argument extraction, revealing ontology-based prompts outperform name-based prompts, and introduces a causal intervention method to debias prompts, improving robustness and effectiveness.

Contribution

It is the first to analyze prompt biases in event extraction using causal intervention and proposes a debiasing method that enhances model robustness and performance.

Findings

Ontology-based prompts outperform name-based prompts in zero-shot EAE.

The causal intervention debiasing method improves model robustness against adversarial attacks.

Modified models show significant performance gains on benchmark datasets.

Abstract

Prompt-based methods have become increasingly popular among information extraction tasks, especially in low-data scenarios. By formatting a finetune task into a pre-training objective, prompt-based methods resolve the data scarce problem effectively. However, seldom do previous research investigate the discrepancy among different prompt formulating strategies. In this work, we compare two kinds of prompts, name-based prompt and ontology-base prompt, and reveal how ontology-base prompt methods exceed its counterpart in zero-shot event argument extraction (EAE) . Furthermore, we analyse the potential risk in ontology-base prompts via a causal view and propose a debias method by causal intervention. Experiments on two benchmarks demonstrate that modified by our debias method, the baseline model becomes both more effective and robust, with significant improvement in the resistance to adversarial attacks.