Information Extraction Based on the Dependency Relationship of Sentences

Abstract

In order to get the information about the technical examination of vehicles in accident from the document, and to provide the foundation for the auto-generation of examination report in natural language, the information extraction techniques are studied, and an information extraction technique based on the dependency relationship of sentences is produced. It constructs the template according to the dependency relationship between the extracted words and frequently-used words. The template, as a set of dependency characteristics in the form of logic expression, is simple and direct. Based on these templates, we can select the word tallying mostly with the dependency characteristics of the template as the final extracted word, so the whole process is straight and speedy.