In-context Contrastive Learning for Event Causality Identification

TL;DR

This paper introduces In-Context Contrastive Learning (ICCL), a novel approach that improves event causality identification by effectively leveraging both positive and negative demonstrations through contrastive learning, leading to significant performance gains.

Contribution

The paper proposes ICCL, a new contrastive learning framework for event causality detection that enhances in-context learning by distinguishing positive and negative examples.

Findings

ICCL outperforms state-of-the-art methods on EventStoryLine and Causal-TimeBank datasets.

Contrastive learning improves the discrimination of causal vs. non-causal event pairs.

The approach effectively utilizes both positive and negative demonstrations for better event causality identification.

Abstract

Event Causality Identification (ECI) aims at determining the existence of a causal relation between two events. Although recent prompt learning-based approaches have shown promising improvements on the ECI task, their performance are often subject to the delicate design of multiple prompts and the positive correlations between the main task and derivate tasks. The in-context learning paradigm provides explicit guidance for label prediction in the prompt learning paradigm, alleviating its reliance on complex prompts and derivative tasks. However, it does not distinguish between positive and negative demonstrations for analogy learning. Motivated from such considerations, this paper proposes an In-Context Contrastive Learning (ICCL) model that utilizes contrastive learning to enhance the effectiveness of both positive and negative demonstrations. Additionally, we apply contrastive learning to event pairs to better facilitate event causality identification. Our ICCL is evaluated on the widely used corpora, including the EventStoryLine and Causal-TimeBank, and results show significant performance improvements over the state-of-the-art algorithms.