Argument-Centric Causal Intervention Method for Mitigating Bias in Cross-Document Event Coreference Resolution
Research paper by Long Yao and collaborators. arXiv record dated 2025-06-02. Research area: Language technology.
Resource information
- Authors
- Long Yao, Wenzhong Yang, Yabo Yin, Fuyuan Wei, Hongzhen Lv, Jiaren Peng, Liejun Wang, Xiaoming Tao
- Research area
- Language technology
- Application domain
- General AI research
- AI methods
- Natural language processing
- Research topics
- cs.CL, cs.IR
- arXiv record date
- 2025-06-02
- Publication
- arXiv
- Upstream last modified
- Jun 3, 2025
- Last record update
- Nov 4, 2026
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.