Research On Agriculture-Oriented Ontology Learning Modeling

Ontology has become a hot research since it was proposed, which is a model used to describe the concepts and their relations. Ontology has been broadly applied in knowledge engineering, information retrieval, Web heterogeneity processing, Semantic Web and so on. As a basic task of ontology research, ontology modeling has been paid much attention in many fields, but the traditional construction ontology by handwork is a time-consuming and fussy task, so it is very necessary to research the automatic or semiautomatic ontology construction by ontology learning.This paper which focuses on the field of tea science aims at the problems existing in manual construction of agricultural ontology. Especially three key points of ontology learning, in terms of concepts extraction, taxonomy and non-taxonomy relations extraction for agricultural ontology, will be primarily resolved in ontology learning modeling of agricultural ontology. In order to solve the above problems, the theories and methods of ontology learning modeling for agriculture are proposed and an agriculture-oriented prototype system of ontology learning modeling is developed.The main content on our research as followings:①An agriculture-oriented automatic theory and method of extracting ontology concepts is proposed in this paper. Firstly, agricultural corpus is processed by Chinese word segmentation and the set of compound words are obtained by mutual information technology, then the ontology concepts of agriculture are automatically extracted by judging the domain coherence of the compound words and non-compound words.②An agriculture-oriented theory and method of extracting automatically taxonomy relations among ontology concepts is proposed in this paper. Taxonomy relations among compound concepts are automatically extracted by building a generalized suffix tree of agricultural ontology concepts; according to hierarchical clustering, taxonomy relations among non-compound concepts are automatically extracted from agricultural ontology concepts set.③An agriculture-oriented theory and method of extracting automatically non- taxonomy relations among ontology concepts is proposed in this paper. Non-taxonomy relations among concepts are automatically extracted by obtaining the concept couples based on the method of Association Rule Mining.④An agriculture-oriented prototype system of ontology learning modeling is developed in Eclipse platform, which adopts Servlet / Jsp programming technology and build a formalization description of ontology model. We can verify the validity of theories and methods in this paper by some experiments.Achievements of this paper will have a certain theoretical research value and practical significance in some aspects, such as the in-depth study of ontology learning modeling, building large-scale agricultural ontology, facilitating the establishment of agricultural Semantic Web and knowledge grid, and realizing complete share and collaboration of agricultural knowledge.