Research On Urology Knowledge Base And Intelligent Information Processing Based On Ontology

In recent years, with the fast development of global information and Internet technologies, medical information science has got rapid advancement and become one of the most important research areas of information science. Knowledge system has formed completely in the medical domain after years of research. However, with the further development of global informatization and presentation of knowledge discovery concepts, the normalized problems of traditional medicine appeared gradually in the process of development. Because of lack of unified descriptions of medical resources on knowledge level, it is more difficult to achieve information sharing and further research of medical field. Therefore, it is very important to establish the medical semantic-level knowledge hierarchy and system.This paper firstly analyzes the domestic and international researches in the fields of ontology and information retrieval, and briefly describes the theory of ontology and related technologies, then puts forward the ontology in urology domain and defines the diagnosis rules. Next by integrating original meaning of semantic distance, density and depth in HowNet, a new similarity calculation method of multi-factor mixed (SCMMM) is proposed. Based on the concept of distance, depth and information capacity, a novel relevance calculation method of multi-factor mixed (RCMMM) is proposed as well. On the basis of the new calculation methods of similarity and relevance, this paper builds a semantic retrieval model of Urology ontology. The basic idea of this model is to normalize the concept and relationship to expand the scope of retrieval by the similarity calculation. Then the implicit relationship of results and retrieval content is discovered through relevance calculation. Finally, the result is sorted by the relevance and feed back to users.Experiments show that SCMMM makes rationally use of the features of the original meaning tree, and overcome the shortcomings of the existed method which the position of the original meaning does not have an effect on the similarity. The results indicate that SCMMM is more reasonable than the existed method. RCMMM is more effective than the single-factor relevance calculation method as well, because the method eliminates the shortcomings that the concept distance of relevance results did not distinguish when two pair concepts have the same number of edges.Semantic retrieval model experiment shows that the threshold of similarity is inversely proportional to the recall, and the threshold of relevance is proportional to the precision.The main contents and results of this paper are as follows:â‘ On the basis of analyzing the theory of ontology, the urology ontology is established and the diagnostic rules on the Urology ontology are analyzed and defined.â‘¡An novel multi-factor mixed calculation methods of similarity and relevance are proposed. And the experiments results demonstrate that the two methods have excellent performance.â‘¢A semantic retrieval model based on domain ontology is proposed, and it is applied to implement Urology Clinical Decision Support Diagnosis Prototype System.