EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Structure-Aware Lightweight Document-Level Event Extraction via Code-Based Large Language Models

Xing Xu, Jianbin Zhao, Pengfei Zhang, Yaduo Liu, Bingyang Yu, Puyuan Zheng, Dingyuan Hu, Zhongchen Deng, Ping Zong, Guoxin Zhang, Zhonghong Ou, Meina Song und 1 weitere

Electronics · 12.03.2026 · 15 (6) · 1187

DOI: 10.3390/electronics15061187 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Electronics, online erschienen am 12.03.2026. Verfasst von Xing Xu und 12 weiteren Autor:innen. Die Arbeit nennt 34 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Science and Technology Project of State Grid Hebei Information and Telecommunication Branch.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Document-level Event Extraction (DEE) requires identifying complex event records and arguments dispersed across unstructured texts. However, applying general Large Language Models (LLMs) to DEE is intrinsically hindered by their lack of inductive bias for rigid structural constraints, often leading to schema violations and suboptimal performance in complex structural prediction tasks. To address this, we propose the S tructure-Aware Lightweight DEE, termed SALE, which leverages the structural reasoning potential of Code-Based LLMs (Code-LLMs) as a favorable inductive preference. We leverage the natural isomorphism between event schemas and programming object definitions, formulating event extraction as a Python 3.9 class instantiation task to bridge the gap between semantic understanding and structural adherence. Specifically, SALE employs a novel two-stage training paradigm: First, a Structure-Aware Fine-tuning stage injects general structural knowledge via diverse code-style instruction tasks derived from broad Information Extraction (IE) datasets; second, an Event Extraction Alignment stage utilizes a reward-based alignment loss—optimized via policy gradient—to adapt this capability to document-level intricacies. The effectiveness of SALE stems from the synergy between its structure-aware prompting and the specialized alignment stage built on a code-oriented backbone. Extensive experiments on established news-domain benchmarks (RAMS and WikiEvents) demonstrate that our approach significantly outperforms representative supervised and general LLM baselines in cross-task zero-shot and few-shot transfer settings (e.g., surpassing supervised baselines by over 7% in F1 score). Furthermore, SALE maintains a highly efficient inference profile and parameter-efficient footprint, offering a practical and scalable solution for vertical domain applications.

Förderung

  • Science and Technology Project of State Grid Hebei Information and Telecommunication Branch

Zitieren

APA

Xu, X., Zhao, J., Zhang, P., Liu, Y., Yu, B., Zheng, P., Hu, D., Deng, Z., Zong, P., Zhang, G., Ou, Z., Song, M., & Zhu, Y. (2026). Structure-Aware Lightweight Document-Level Event Extraction via Code-Based Large Language Models. Electronics, 15(6), 1187. https://doi.org/10.3390/electronics15061187

Vancouver

Xu X, Zhao J, Zhang P, Liu Y, Yu B, Zheng P, et al. Structure-Aware Lightweight Document-Level Event Extraction via Code-Based Large Language Models. Electronics. 2026;15(6):1187. doi:10.3390/electronics15061187

BibTeX
@article{Xu2026,
  author = {Xu, Xing and Zhao, Jianbin and Zhang, Pengfei and Liu, Yaduo and Yu, Bingyang and Zheng, Puyuan and Hu, Dingyuan and Deng, Zhongchen and Zong, Ping and Zhang, Guoxin and Ou, Zhonghong and Song, Meina and Zhu, Yifan},
  title = {Structure-Aware Lightweight Document-Level Event Extraction via Code-Based Large Language Models},
  journal = {Electronics},
  year = {2026},
  volume = {15},
  number = {6},
  pages = {1187},
  doi = {10.3390/electronics15061187}
}
Kostenlos starten →