The Research Of Corn Disease And Pest Prevention And Cure Semantic Searching System Based On Ontology

The definition of ontology is "the explicit and formal specification of share conceptualization". It has been an important issue since it was proposed, because it can afford the semantic information of the resource that can be understood by the machines. The semantic search based on ontology is a valuable issue in the many applications of ontology. Semantic search contains many techniques. Examples are information retrieval, artificial intelligence and natural language processing. Its kernel technique is semantic matching mechanism based on concepts. Semantic search is different from the traditional information retrieval methods, it does the semantic matching on the information, and improves precision and recall a lot.Our work is to do some research on the important techniques of semantic search. It contains building the domain ontology, semantic similarity calculating model and semantic pertinency calculating model, building the semantic searching system and so on. Our work is a part of the 863 item which is took in hand by the Key Laboratory of Symbolic Computing and Knowledge Engineering of Ministry of Education.Firstly, we construct an ontology named CIPO. CIPO is built to describe the knowledge of how to prevent and cure plant disease and insect pests of corn. To achieve this goal, CIPO builds some concepts, relations between the concepts and attributes restrictions. We studied "Agriculture Thesaurus" and "Chinese Library Classification-Agriculture Classification", and constructed CIPO based on the W3C standard. It realizes the highly reuse of domain knowledge.We choose OWL to describe CIPO, and choose Protege 3.3 as the building tool. Protege 3.3 is developed by Stanford Medical Informatics, and is easy to use. It has friendly user interface, and supports many ontology constructing language. The ontology constructing method is a pop research domain, and there isn't a standard method, so we use many methods synthetically. Secondly, we study the semantic similarity calculating model and the semantic pertinency calculating model, and propose a new semantic similarity calculating model PHSS.The semantic similarity and semantic pertinency of concepts are hard to be differentiated. The prior one represents the matching degree of two concepts, and the hind one represents the dependent relation of concepts. They also have affinity. In other words, if two concepts are similar, they are pertinent, but if two concepts are pertinent, maybe they are not similar.The calculating of semantic similarity and pertinency are two important techniques of semantic search. Good model can improve the precision and recall. The semantic similarity calculating model can solve the problem that the similarity between qualitative concepts can't be calculated quantificationally. We proposed an improved semantic calculating model PHSS after studying the existing ones fully. It considers both the hierarchy and the property restrictions of the concepts, and describes the semantic distance between concepts more accurately. Semantic pertinency calculating is the base of improving search pertinency and realizing semantic reasoning. In this paper, we analyze the model proposed by Lijun Zhu, and apply it to the CDPPIRS.At last, we design a semantic searching system CDPPIRS, which explores the process of semantic search and validates the describing ability of CIPO. It also shows the validity of PHSS.CDPPIRS can do the semantic reasoning based on query information and the repository, then provides the users with accurate bibliographies depending on the user's different kinds of requirements, and satisfies them well. CDPPIRS consists of five modules: the user interface, the query pretreatment module, the semantic reasoning module, the literature database and the knowledge database.CIPO is stored in the knowledge database. The semantic pertinency calculating model and PHSS are used in the semantic reasoning module. There are 1027 literatures in the literature database. Some the literatures are tooted in "Corn Science". Others are got from network, they are pertinent with some concepts in CIPO. We choose 80 query information to do the experiment, and the results show that: when the amount of information is small and the description is strict, semantic reasoning can achieve our anticipative goal, and improves precision and recall a lot.The meaning of this paper is as follows: do some useful research on constructing ontology; propose a semantic similarity calculating model PHSS which describes the semantic distance between concepts more accurately; do research on designing the structure of the semantic system. The experiences can be refer-red by other systems. Our work is primary, there are still a lot of work to be do-ne in the future.