| 1 |
2025/12/03 20:52 |
靳*营 |
Paper title：An automatic hydrological service chain generation method by integrating LLM and embedding model
Paper abstract：Hydrological models are essential for water resources management and soil and water conservation, yet traditional models are often locally deployed and limited in complex hydrological processes. Hydrological service chains can address these limitations, but existing automatic generation methods rely on formalized requirement expressions, lacking natural language interaction and thus offering insufficient support for ambiguous requirements. This study proposes an automatic hydrological service chain generation method that integrates a large language model (LLM) and an embedding model to transform natural language requirements into hydrological analysis results. Aiming at the diversity of natural language description of requirements, structured requirement templates are designed for automatic parsing of requirements using a LLM. To further address potential generation errors caused by LLM hallucinations, the embedding model maps both parsed requirements and candidate services into a unified vector space, with optimal processing services or composite services identified via vector similarity computation. Data services are then selected based on processing service or composite service constraints, and the complete service chain is executed through a service chain execution engine. An experiment and two case studies demonstrate the effectiveness of the proposed method in automating hydrological service chain generation, significantly enhancing the automation of hydrological analysis.
Paper type：research paper
该数据用于论文的案例分析部分
|
本作品采用
知识共享署名
4.0 国际许可协议进行许可。
CSTR:11738.11.ncdc.Westdc.2020.355
10.12072/ncdc.westdc.db3703.2023
