Research On Information Semantic Retrieval Supported By Rough Ontology

Information semantic retrieval is a kind of retrieval method based on the semantic understanding. Compared with retrieval method based on keywords, the main difference is that:on the one hand, annotated information may get a well representation on semantic level; on the other hand, semantic information also added to the users’ retrieval request to make it unequivocal. The key to realize these is ontology in the retrieval process playing a role of knowledge base.Rough ontology is an extension of ontology, and it adds some rough concepts on the basis of precise concepts into ontology, then links them through the lower approximation extent and the upper approximation extent, so it can realize expression of precise information and imprecise information. In information retrieval process the rough ontology may not only inherit the ability of expression and inference for the precise information in a form of knowledge base, but also can be used as an information carrier with higher robustness and fault-tolerance.On the basis of studies on information semantic retrieval supported by precise ontology, rough ontology is introduced, and a retrieval model is built to realize imprecise information semantic representation and semantic retrieval.First of all, rough ontology is introduced. Studies the theory related to rough ontology, and studies the principles and methods of ontology building according to the nature of the rough ontology, and discusses the feasibility of an application of rough ontology, and analyzes the problem of imprecise information semantic representation after the rough ontology replacing the precise ontology.Secondly, the model of information semantic retrieval supported by rough ontology is given. The model has several major steps:First, build a domain rough ontology; Second, extract the retrieval request to get an initial concept set; Third, find precise concepts and rough concepts through the rough ontology, by the support of the semantic similarity calculation, get the concepts meeting the threshold value, and put them into extended concept set; Fourth, by matching the extended concept set and the annotated information document, get the retrieval results and the sorted results will be returned to users. Finally, an information semantic retrieval system supported by rough ontology is developed. After building a maritime accident rough ontology, add instance to the rough ontology, and annotate the extracted maritime accident information, and complete semantic retrieval of maritime accident information. Through testing the recall ratio and precision ratio of the system, it proved that the method of semantic information retrieval supported by rough ontology can inherit the advantages of semantic information representation and inference of the precise ontology, and through the attributes of maritime accidents, and can expand the initial concepts to similar precise concepts and rough concepts, and can make it more effective in retrieving some imprecise information which cannot be achieved before.