XML 구조적 특징을 이용한 온톨리지 기반의 지식 탐사 모델
(An) ontology based knowledge discovery model using XML, hierarchical structure
XML 온톨리지 지식탐사;
- 원문 URL
This thesis presents an ontology based knowledge discovery model to manage knowledge information for XML documents efficiently. This model use the structural characteristic of XML document and textmining techniques to create domain ontology. This model also presents an ontology learning method using feedback information to extend domain ontology continuously. The domain ontology is stored to XML database to classify and search knowledge information. This domain ontology can be used for category service for knowledge discovery in the knowledge discovery model. Presented model is composed of preprocessing stage, domain ontology creation and ontology learning in the main. In the preprocessing stage, the element weight and TWIDF technique is proposed to extract representative terms using the structural information in the XML documents. In the domain ontology creation, relationship between the domain and representative terms are analyzed by statistical value. representative term set for each domain is constructed using domain weights. A hierarchical relationship is represented by DAG analyzing the subsumption among domain representative terms. A method to establish clear-cut lines of among inter-domain terms is proposed. This model also presents an ontology learning method using the structural information in the XML documents. In the ontology learning stage, domain ontology instances are created to map a new XML document and the domain. The feedback information is collected to evaluate the degree of association between domain ontology instances and the domain. To verify the efficiency of the proposed ontology-based knowledge discovery model, classification test are performed and compared the results by F-measure for 295 XML papers in the computer science. The results of experiment show that the proposed TWIDF method using the structural information in the XML documents is more efficient than the existing TFIDF. Hence, the proposed model can be thus expected to contribute to the development of an efficient ontology-based knowledge discovery system. Also, this model can be applied to the areas, such as E-commerce, Digital Library and KDD systems.