Sábháilte in:
| Príomhchruthaitheoirí: | , , , , , |
|---|---|
| Formáid: | Preprint |
| Foilsithe / Cruthaithe: |
2024
|
| Ábhair: | |
| Rochtain ar líne: | https://arxiv.org/abs/2404.01921 |
| Clibeanna: |
Cuir clib leis
Níl clibeanna ann, Bí ar an gcéad duine le clib a chur leis an taifead seo!
|
| _version_ | 1866909194704125952 |
|---|---|
| author | Ding, Bowen Min, Qingkai Ma, Shengkun Li, Yingjie Yang, Linyi Zhang, Yue |
| author_facet | Ding, Bowen Min, Qingkai Ma, Shengkun Li, Yingjie Yang, Linyi Zhang, Yue |
| contents | Based on Pre-trained Language Models (PLMs), event coreference resolution (ECR) systems have demonstrated outstanding performance in clustering coreferential events across documents. However, the existing system exhibits an excessive reliance on the `triggers lexical matching' spurious pattern in the input mention pair text. We formalize the decision-making process of the baseline ECR system using a Structural Causal Model (SCM), aiming to identify spurious and causal associations (i.e., rationales) within the ECR task. Leveraging the debiasing capability of counterfactual data augmentation, we develop a rationale-centric counterfactual data augmentation method with LLM-in-the-loop. This method is specialized for pairwise input in the ECR system, where we conduct direct interventions on triggers and context to mitigate the spurious association while emphasizing the causation. Our approach achieves state-of-the-art performance on three popular cross-document ECR benchmarks and demonstrates robustness in out-of-domain scenarios. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2404_01921 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | A Rationale-centric Counterfactual Data Augmentation Method for Cross-Document Event Coreference Resolution Ding, Bowen Min, Qingkai Ma, Shengkun Li, Yingjie Yang, Linyi Zhang, Yue Computation and Language Based on Pre-trained Language Models (PLMs), event coreference resolution (ECR) systems have demonstrated outstanding performance in clustering coreferential events across documents. However, the existing system exhibits an excessive reliance on the `triggers lexical matching' spurious pattern in the input mention pair text. We formalize the decision-making process of the baseline ECR system using a Structural Causal Model (SCM), aiming to identify spurious and causal associations (i.e., rationales) within the ECR task. Leveraging the debiasing capability of counterfactual data augmentation, we develop a rationale-centric counterfactual data augmentation method with LLM-in-the-loop. This method is specialized for pairwise input in the ECR system, where we conduct direct interventions on triggers and context to mitigate the spurious association while emphasizing the causation. Our approach achieves state-of-the-art performance on three popular cross-document ECR benchmarks and demonstrates robustness in out-of-domain scenarios. |
| title | A Rationale-centric Counterfactual Data Augmentation Method for Cross-Document Event Coreference Resolution |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2404.01921 |