Structured Hydrological Entity and Relation Extraction From multiregion Flood‐Control Documents
International Journal of Energy Research · 2026-01 · 2026 (1)
DOI: 10.1155/er/2524372 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Energy Research, online erschienen am 19.11.2026. Verfasst von Hui Min und 4 weiteren Autor:innen. Die Arbeit nennt 32 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Reliable extraction of hydrological observations, flood risks, and control measures from multiregion flood‐control documents is challenging because regional terminology varies, entity, and relation decisions are interdependent, and unconstrained predictions may violate hydrological schemas. To address these issues, this study proposes a schema‐guided hydrological entity and relation extraction framework (SHEAR) for source‐grounded structured information extraction. The framework consists of three modules. First, the region‐aware representation module integrates regional metadata, schema descriptions, and supervised contrastive alignment to reduce lexical variation across issuing authorities. Second, the joint span extraction module identifies typed entity spans and models ordered span pairs for directed relation prediction, thereby reducing error propagation between entity and relation decisions. Third, the constraint‐guided optimization module incorporates valid head‐type–relation–tail‐type combinations into model training and decoding to improve structural consistency while retaining source offsets. On the author‐built corpus, SHEAR achieves entity and relation micro‐F1 scores of 89.95 and 83.14, exceeding the strongest supervised baseline by 2.40 and 3.20 points, respectively. It also obtains entity and strict relation F1 scores of 70.45 and 40.26 on SciERC. Ablations reduce relation micro‐F1 by 2.18–3.42 points when individual modules are removed or replaced, supporting their complementary contributions. These results support source‐grounded hydrological extraction on the evaluated datasets; applicability to live operational settings requires further validation.
Zitieren
APA
Min, H., Ye, F., Xu, D., Xu, J., & Liao, X. (2026). Structured Hydrological Entity and Relation Extraction From multiregion Flood‐Control Documents. International Journal of Energy Research, 2026(1). https://doi.org/10.1155/er/2524372
Vancouver
Min H, Ye F, Xu D, Xu J, Liao X. Structured Hydrological Entity and Relation Extraction From multiregion Flood‐Control Documents. International Journal of Energy Research. 2026;2026(1). doi:10.1155/er/2524372
BibTeX
@article{Min2026,
author = {Min, Hui and Ye, Feng and Xu, Dong and Xu, Jin and Liao, Xiaoping},
title = {Structured Hydrological Entity and Relation Extraction From multiregion Flood‐Control Documents},
journal = {International Journal of Energy Research},
year = {2026},
volume = {2026},
number = {1},
doi = {10.1155/er/2524372}
}
EUVIMED