Research On Effective Evaluation Method Of Agricultural Ontology

As a concept model which can describe information and data on the semantic level,agricultural ontology can provide the basis for ontology services such as intelligent agricultural information retrieval,pest control and prevention and decision-making support for agricultural production.The significance of ontology evaluation aims to ensure ontology building quality,enable ontology to be effectively managed and maintained and supply better service to users.With the increasing number of agricultural ontology,it is an urgent problem to evaluate the quality of agricultural Ontology.To date,it has not achieved perfect theoretical system and methodology.This paper which focuses on the problems existing in evaluation of agricultural ontology.Especially three important parts of agricultural ontology evaluation,such as concepts evaluation,taxonomy relations evaluation and attribute relations evaluation.For purpose of solving the three problems,the methods of ontology evaluation are proposed and an prototype system of ontology evaluation is developed.This paper consists of seven parts.The main views and content are as follows:(1)The domain-dictionary based theory and method for the effectively evaluation of concepts is proposed.Firstly,constructing the tea science domain dictionary manually and calculating the total number of words;Secondly,evaluating the accuracy,recall rate and F measure by calculating ratios;Finally,analyzing the assessment results.The evaluation of concepts is completed.(2)The similarity-based theory and method for the effectively evaluation of taxonomy relations among concepts is put forward.By calculating the semantic similarity between the parent node and child node in the taxonomy,the consistency,accuracy and simplicity of the taxonomy are evaluated according to the similarity value.The methods can estimate the error types of ontology taxonomy effectively.(3)The graph-based theory and method for the effectively evaluation of nontaxonomy relations among concepts is given.Based on the graph theory,the characteristics of attribute relation are gained and the determination conditions for assessment indexes are proposed.By analyzing the assessment results,the evaluation of non-taxonomy relations among concepts is completed.(4)A prototype system of tea pest ontology evaluation is developed in Eclipseplatform by using the Java language and calling ontology development kit Jena.