Data Enhancement and Named Entity Recognition Method for Water Intake Permit Management Information of Yangtze River

The Yangtze River is the longest river in Asia, with many abstraction users, making it difficult to manage abstraction permits. Water Intake Permits (WIPs) are mainly text-based, highly specialized and contain a large amount of content, which makes effective information extraction complicated. Named Entity Recognition (NER) is essential for identifying key entities in WIP texts. Aiming at the limited annotated corpus, nested entities and lengthy expressions for named entity recognition in water abstraction license materials, a data enhancement and named entity recognition method is proposed for WIP Management. According to the structure and linguistic characteristics of the Yangtze River WIP text, the data augmentation method is improved to solve the problem of annotated corpus scarcity. Comparison experimental results with other methods show that the proposed WIP model has accuracy and better generalization ability, and the F1 value on the WIP dataset reaches 89%, which provides an effective method for the recognition of named entities of water abstraction permits in the Yangtze River basin.