Study On Domain Ontology-based Legal Text Mining

In recent years, with the continued expansion of local grass-roots police contingent, some of the police are lack of experience in law enforcement, and they aren't familiar with the legal provisions, this led to the situation that some of the public security cases get improper punishment often happen. In order to solve this problem, this paper designs a domain ontology based framework for legal text mining, we use legal ontology as the background knowledge for text mining, then we improve the traditional text mining methods based on the characteristics of the legal text. finally this system can analysis and process the unstructured legal text, and automatically match the relevant legal terms. So this system can provide legal support for the police in their law enforcement process.The main content of this paper can be divided into two aspects:on the one hand, the research on the building process of legal ontology; on the other hand, the study on the key technologies of the legal text mining. First, we give a brief introduction to the domain ontology construction, containing principles, methods and tools. Then we construct a legal ontology model based on this application, according to the requirement of the legal text mining, and drawing on the seven-step method which is proposed by Stanford University.In addition, the key technologies of the legal text mining include the formal representation of the legal text and the classification of the legal text. We improve the effect of the word segmentation by adding custom keywords library in the text preprocessing. Then we introduce the traditional model of text representation, we find that it lack of the support of domain knowledge and ignore the semantics of the text, so we improve this text vector model using the legal ontology as the background knowledge of legal text mining, then we use the concepts of the domain ontology instead of characteristics of the legal text so that it can reduce the dimension of the text vector..Moreover, this paper use the term mutual information matrix which is calculated by the structure tree of the legal ontology to adjust the weight of the concept, then it can make the method of the text similarity measure better, and improve the accuracy of legal text classification.