A Diffusion Model for Event Skeleton Generation

Fangqi Zhu, Lin Zhang, Jun Gao, Bing Qin, Ruifeng Xu, Haiqin Yang


Abstract
Event skeleton generation, aiming to induce an event schema skeleton graph with abstracted event nodes and their temporal relations from a set of event instance graphs, is a critical step in the temporal complex event schema induction task. Existing methods effectively address this task from a graph generation perspective but suffer from noise-sensitive and error accumulation, e.g., the inability to correct errors while generating schema. We, therefore, propose a novel Diffusion Event Graph Model (DEGM) to address these issues. Our DEGM is the first workable diffusion model for event skeleton generation, where the embedding and rounding techniques with a custom edge-based loss are introduced to transform a discrete event graph into learnable latent representations. Furthermore, we propose a denoising training process to maintain the model’s robustness. Consequently, DEGM derives the final schema, where error correction is guaranteed by iteratively refining the latent representations during the schema generation process. Experimental results on three IED bombing datasets demonstrate that our DEGM achieves better results than other state-of-the-art baselines. Our code and data are available at https://github.com/zhufq00/EventSkeletonGeneration.
Anthology ID:
2023.findings-acl.800
Volume:
Findings of the Association for Computational Linguistics: ACL 2023
Month:
July
Year:
2023
Address:
Toronto, Canada
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
12630–12641
Language:
URL:
https://aclanthology.org/2023.findings-acl.800
DOI:
10.18653/v1/2023.findings-acl.800
Bibkey:
Cite (ACL):
Fangqi Zhu, Lin Zhang, Jun Gao, Bing Qin, Ruifeng Xu, and Haiqin Yang. 2023. A Diffusion Model for Event Skeleton Generation. In Findings of the Association for Computational Linguistics: ACL 2023, pages 12630–12641, Toronto, Canada. Association for Computational Linguistics.
Cite (Informal):
A Diffusion Model for Event Skeleton Generation (Zhu et al., Findings 2023)
Copy Citation:
PDF:
https://preview.aclanthology.org/remove-xml-comments/2023.findings-acl.800.pdf