福建电脑2025，Vol.41Issue(9)：37-42,6.DOI:10.16707/j.cnki.fjpc.2025.09.007
电力物资文档混合检索方法研究
Research on Hybrid Retrieval Method for Power Material Documents
葛星 1柏能1
作者信息
- 1. 国网江苏省电力有限公司物资分公司合同结算部 南京 210000
- 折叠
摘要
Abstract
In response to the low efficiency and insufficient accuracy of unstructured and semi-structured material document retrieval in the digital transformation of the power industry,this study proposes a hybrid retrieval method based on dynamic weight fusion.By combining the advantages of keyword retrieval,structured field retrieval,and semantic vector retrieval,a pre trained language model for the field of electric power materials is constructed,and entity recognition and text classification algorithms are developed to enhance query intent understanding and document feature extraction capabilities.Tests based on power material documents from a certain province have shown that this method significantly improves accuracy compared to traditional retrieval methods,effectively meets practical business needs,and has high application and promotion value.关键词
电力物资/混合检索/预训练模型/语义检索/文档分类Key words
Electricity Materials/Mixed Search/Pre Trained Model/Semantic Retrieval/Document Classification分类
通用工业技术引用本文复制引用
葛星,柏能..电力物资文档混合检索方法研究[J].福建电脑,2025,41(9):37-42,6.
