Study On Construction Method Of Agricultural Domain Ontology Based On Agricultural Thesaurus And Document

Agricultural development, which plays an important role in economic growth in China, is closely related with the research on agriculture. However, the biggest challenge facing the researchers now is that they could not get the data they really need, which is caused by the drawbacks of the traditional knowledge organization. Ontology can present a solution to this problem.On the basis of study on a large number of relevant literature both at home and abroad, this research put forward the idea of semi-automatic ontology construction based on the "Agricultural Thesaurus" and professional literature in agriculture. The idea is proposed due to the following two points: (1) The "Agricultural Thesaurus", compiled by a great quantity of experts, is an available knowledge system in agriculture. It provides the richest and most authoritative terminology and concepts in agriculture, guaranting the integrity and standardization of the ontological knowledge system. (2)As the other source of the ontology concepts, the agricultural literature, with huge of data and good dynamic nature, is a supplement to the thesaurus and could greatly expand the terms range for ontology construction.The paper studies the structure of professional articles and defines the framework of the agricultural domain ontology,including the identifying of main classes, properties and relationships, which ensures scientificity and rationality of the horizontal framework. The thesaurus stucture, the terms range, as well as the relationships between terms are analyzed. The main categories for the research are picked out, as the vertical delineation of the ontology scope. The combination of agricultural thesaurus and professional literature has associated external features of documents with their internal content, laying a good foundation for the ontology construction. The paper also introduces essential tools in detail, including ICTCLAS, a Chinese word segmentation tool, and PDF2WORD, a conversion tool from PDF to WORD. The key technologies are mainly the natural language processing technologies, involving Chinese word segmentation, POS tagging, extraction of sentences and phrases with specified structure, clustering and so on. The last part gives detailed descriptions of the construction procedures of all ontology elements.In the end, the semi-automatic construction method of domain ontology is summarized and problems are discussed. The following work is planned.