Study On Ontology Construction Evaluation

In Knowledge Economy era, people are no longer satisfied with the stage at which the description of thing is simple, static and superficial .The description of information has transferred from literature information description to knowledge description. Research in this field has changed from static to dynamic state. The concept of Ontology proposed by the philosopher has become an important research in computer field. Research and application of Ontology have had a rapid development in recent years. Ontology goes deeply into the essence of things, and reflects the changes in the development process, as well as the internal nicety structure and macroscopic properties. According to the analysis, reasoning and logic calculation on existing information, Ontology can promote the rediscovery of knowledge.It has become a hot topic in international about Ontology, and has a large number of methodologies of ontology construction tools, but none of them are fully mature. Generally speaking, the domestic study on ontology is still at an early stage. Research of ontology construction is now in the process of continuous improvement, and the study on Ontology evaluation has just started. Most of the study is stay at the level of shallow evaluation on the construction of ontology, and has not yet set eyes on the technologies and methods of Ontology construction. Ontology has been applied into a number of specific areas, and many experts and scholars have pay attention to the evaluation of ontology construction, mainly in the following evaluation methods: user-based approach, task-based approach, principle-based approach, application-based approach, gold standard-based approach, corpus-based approach. Although the study on ontology evaluation based on the existing ontology base (such as DAML + OIL)has made a little progress, Every respect of evaluation on Ontology is unilateralism . A complete theory of ontology evaluation has not come into being and the research content of ontology evaluation is still unclear. The present study hasn't outlined a complete analytical framework. A practicable evaluation index system and a evaluation tool hasn't been established. We can rarely find some documents about ontology evaluation. Even the well-known large-scale ontologies such as Cyc, SENSUS also have problems in ontology evaluation.Considering the above-mentioned problems, we think it is imperative to construct a scientific and rational ontology evaluation system. This paper discusses the contents of ontology evaluation on the perspective of theories, technologies and application integration. It established a set of ontology evaluation index system and developed a assessment tool. Using this tool we tested the index system so as to regulate ontology construction and promote the application of ontologies. Meanwhile, this open evaluation tool offers great convenience to the end user and Ontology developer.The main contents of this paper are structured as follows:Chapter 1 is the general part of this paper. After introducing background and present status domestic and international study on ontology evaluation, it Summarized problems in all aspects appeared in the ontology construction so as to show the significance of this paper and introduce the its main content as well as the structure of this paper.Chapter 2 compares methods and tools for ontology evaluation and Confirms the content of ontology evaluation. It analyses and compares methods and tools which are in study abroad.Chapter 3 is the construction of ontology evaluation metrics. Combining the characteristics and methods of ontology construction in domestic, we established an evaluation index system based on the life cycle of ontology and endow different weight values for each level index according to the degree of importance in life cycle of ontology.Chapter 4, we developed a semi-automatic tool for evaluating ontology on the basis of concept architecture of ontology evaluation. It is convenient for user evaluating ontologies and comparing their outcome. Moreover, taking FrameNet and WordNet as study objects and conference, we tested Onto_evaluate by evaluating their concepts and concept relations.