Extraction of key information from Chinese text of gas station safety accident cases

Keyword matching-based text structuring of gas station safety accident cases can automatically perform keyword extraction and store them in the database, and then the structured data can be subjected to multi-dimensional statistics, visual analysis, knowledge graph management and law mining. 516 cases of gas station safety accidents since 1981, which belong to unstructured electronic text data, were collected in four ways. Through the establishment of a keyword dictionary, 516 case texts were extracted and stored in a structured way using a progressive model of exact matching, phrase matching and intelligent matching+manual verification. The actual evaluation results show that the success rate of exact matching is 30% - 87%, the cumulative success rate of phrase matching is 70% - 100%, the cumulative matching rate of intelligent matching+manual verification is 91% - 100%, the overall matching effect is good, and the method is practical and useful.