| An ontology provides a concept sharing model for knowledge in semantic level.Thus, it plays an important role in many application domains such as data integration,peer-to-peer system, e-commerce, Semantic Web services, and social networks.Researches recently utilize the ontology driven methods to solve the domain problem as they want to reduce the semantic heterogeneity.Ontology Matching is the key technology to find the mapping correspondences between different entities such as elements, attributes and individuals in ontology.It has been seen one of the best solutions to the semantic hetergeneity problem of computer system by the researches in and out of the country.The knowledge retrieval problem in knowledge service is actually owed to solve the matching problem between user inquiries and resource descriptions in semantic level.One of the best solutions is to represent the user conditons and resource descriptions as formal knowledge which can be understood by computer automatically, and then measure the match degree by reasonsing and calculating.Thus, the sematic corespondences between the users and resources can be computed and the user can select the best candidate resources according to the ranking of semantic similarity.Because an ontology can provide a semantic expression of user inquires and resource descriptions,and ontology matching is a promising solution to the semantic relation between ontology.So the dissertation is aimed to make use of ontology matching technology as one of the ways to solve the knowledge retrieval in knowledge service domain,and attempts to slove the problem of semantic matching in knowledge service.The research of this dissertation mainly focuses on the semantic matching of the inquiries and resource descriptions and include four issues:(1) The definition of domain ontology;(2)Context based Element-level ontology matching;(3) Struture-level ontology matching based on weighted ontology; (4) The application and realization of ontology matching algorithm in knowledge retrival system for education. The work in this dissertation is supported by the National Science Foundation of China (No.60673094), the Cultivation Fund of the Key Scientific and Technical Innovation Project, Ministry of Education of China(NO705038) ,and the National Great Project of Scientific and Technical Supporting Programs Funded by Ministry of Science & Technology of China During the 11th Five-year Plan (NO. 2006BAH02A24) and is also supported by the Natural Science Foundation of Hubei Province (NO.2006ABC011) .The contributions of this dissertation include:(1) Learning resources ontology in educational domain:In term of the lack of semantics for the standard of learning object metadata, this section makes a deep survey of research for the learning object medata and its combination with ontology, and proposes the learning resource ontology based on the ontological levels of learning object metadata and domain ontologies advanced by Dragan Ga(s|ˇ)evic and Marek Hatala.The proposed ontology is an instance that combinates the ACM Computer classification system with IEEE LOM learning resources metadata in basis of ontology construction rules. Compared with the relative ontology of learning resources, the proposed Model is more comprehensive and specific, which is mainly oriented for semantic retrieval.This section paves the way for the latter research in ontology matching algorithm and its application.(2) The context based element-level ontology matching algorithm:This section summarizes the independent ontology element analysis methods and the structure ontology element analysis methods,and compares the two semantic semantic similarity measuring methods for independent ontology elements based on wordnet and corpus or ontology structure respectively.Then,it proposes the advanced independent elemental similarity method which expands the famours Hirst & St-Onge algorithm.It also proposes the context based element-level ontology matching algorithm according to the context path of independent elments.This approach takes the pre-element,attributes,and next-element of concept as the elment context,and forms the context path of element.The context based semantic similarity of element is composed by all the entities in the context path of elements.The Recall and Precision ratio as well as the F-Measure and Overall ratio for the ontology matching experiments at OAEI 2006 datasets has been designed,and the experiment results have shown that the context-based element-level ontology matching algorithm can significantly improve the quality and performance of ontology matching.(3) The structure-level ontology matching algorithm based on the weighted ontology: The section firstly makes a survey of the graph based matching algorithm, and analyzes the graph representation method of ontology structure. Then, the thesis proposes the structure-level ontology matching algorithm based on the weighted ontology. The approach adopts the top-bottom and layer-weighted idea, and deploys the different level of the ontology structure with appropriate weight coefficients.By combining all the entities in the ontology to form a whole, the approach computes the clustering semantic similarity of the binding ontology structure. The proposed method is based on the element level similarity and computes the semantic similarity of elements, attributes, and attributes value of the ontology structure of learning resources, all of which is stored with two dimension matrix formal in the database, and then the weighted average result of ontology is attained.(4) The application and realization of ontology matching algorithm in knowledge retrival system for education: this section designs and implements a semantic search system for educational knowledge, which integrates the ontology matching technology introduced in forenamed sections.The system can test the efficiency and accuracy of the proposed ontology matching algorithm.In order to the effectiveness of the semantic search system, this section designs the experiments to compare the semantic search system with the key word based search system that is also developed by us.The experiment results have proved the empirical value of the semantic search system.
|
PDF Full Text Request