Argument-Centric Causal Intervention Method for Mitigating Bias in Cross-Document Event Coreference Resolution

2025-06-02

AI Summary

To address superficial lexical bias in cross-document event coreference resolution (CD-ECR) caused by over-reliance on trigger words, this paper proposes an argument-centric causal intervention framework. It constructs a structural causal model to characterize confounding paths between trigger words and coreference decisions, then applies backdoor adjustment and counterfactual perturbation to isolate and mitigate trigger-induced spurious correlations, thereby focusing modeling on the causal effect of argument semantics. The framework innovatively integrates argument-aware representation enhancement and counterfactual reasoning modules, enabling end-to-end, bias-mitigated learning without data augmentation or post-processing. Evaluated on the ECB+ and GVC benchmarks, it achieves CoNLL F1 scores of 88.4% and 85.2%, respectively—setting new state-of-the-art results at the time of publication. The code and resources are publicly available.

Technology Category

• Natural Language Processing: Discourse, Pragmatics & Argument Mining
• Knowledge Representation and Reasoning: Argumentation
• Reasoning under Uncertainty: Causality

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Cross-document Event Coreference Resolution (CD-ECR) is a fundamental task in natural language processing (NLP) that seeks to determine whether event mentions across multiple documents refer to the same real-world occurrence. However, current CD-ECR approaches predominantly rely on trigger features within input mention pairs, which induce spurious correlations between surface-level lexical features and coreference relationships, impairing the overall performance of the models. To address this issue, we propose a novel cross-document event coreference resolution method based on Argument-Centric Causal Intervention (ACCI). Specifically, we construct a structural causal graph to uncover confounding dependencies between lexical triggers and coreference labels, and introduce backdoor-adjusted interventions to isolate the true causal effect of argument semantics. To further mitigate spurious correlations, ACCI integrates a counterfactual reasoning module that quantifies the causal influence of trigger word perturbations, and an argument-aware enhancement module to promote greater sensitivity to semantically grounded information. In contrast to prior methods that depend on costly data augmentation or heuristic-based filtering, ACCI enables effective debiasing in a unified end-to-end framework without altering the underlying training procedure. Extensive experiments demonstrate that ACCI achieves CoNLL F1 of 88.4% on ECB+ and 85.2% on GVC, achieving state-of-the-art performance. The implementation and materials are available at https://github.com/era211/ACCI.

Problem

• Mitigating bias in cross-document event coreference resolution
• Reducing spurious correlations in lexical triggers and coreference labels
• Enhancing argument semantics sensitivity in coreference models

Innovation

• Argument-Centric Causal Intervention for debiasing
• Counterfactual reasoning module for trigger perturbations
• Argument-aware enhancement for semantic sensitivity