| With the development of study of Semantic Web, ontology has been applied in many domains. The problem of information islands is still an open issule in the domain of information management. Many researches focus on how to solve the heterogeneity of information, such as federated database system, middleware, data warehouse and distributed databases. But all these methods face the same question: how to handle the semantic conflict among so many information sources.Ontology is proved to be a better idea to solve this problem. It can clearly express the relationship between concepts and understood by machine. Usually, all kinds of information sources are transformed into ontologies and then these ontologies are put together to eliminate semantic inconsistency among information sources. In this process, ontology integration appears to describe the phenomenon of the collections of ontologies.Under the background of information integration, it is a natural way to think of ontology integration as a whole to interpret its semantics. But this method encounters the problem of complexity in reasoning and reading when the amount of ontologies is increasing.In this thesis, we propose a revised distributed interpretation based on distributed description logics to describe its semantics. In our proposal, ontology integration is taken as global ontology and local ontologies connected by ontology mapping. The main work and innovation can be summarized as follow:(1) Semantics based on distributed interpretation is given for ontology integration. It comprises a set of local interpretations which respectively interpret semantics of ontology in ontology interpretation and mappings between them. For ontology integration, information is passed from local ontologies to global ontology. Due to this situation, coherence is more important than consistency to some extent. In this thesis, we propose a set of rules to reduce ontology integration to DL ontology and an algorithm is designed to check coherence of ontology integration. By these means, coherence checking of ontology integration can be performed by regular DL reasoners such as Pellet, Race pro, etc.In this way, our method can facilitate understanding and maintenance of ontology integration.(2) Under the situation of coherence of ontology integration, a special phenomenon appears when distributed query is handled on ontology integration. It is caused by ontology mapping from local ontology to global ontology which changes the relations of some mapped concepts. In this thesis, conflict is used to describe this situation. It can make dirty data exchanged between local ontologies and global ontology.In this thesis, an algorithm using ontology closure is proposed to detect and find conflict. Through ontology closure, all relations including implicit ones between mapped concepts can be found and then these relations is sent into global ontology to check whether they will cause incoherence or not. We test our ideas by some experiments and prove its correctness.(3) When local ontologies are mapped to global ontology, some mappings may cause incoherence of ontology integration. When this situation happens, it needs to locate which mapping in response to incoherence.For ontology integration, we propose a method based on belief revision to debug incoherence happened to ontology integration. In this method, all mappings are viewed as new axioms for global ontology and some postulates are defined to ensure the truth of debugging. To detect the location of incoherence, this method adopts a semi-kernel revision operation and some experiments prove its feasibility.(4) We study the application of semantics of ontology integration in an information integration model for e-government. We also show how to put the aforementioned theory and algorithm into practice by an example.Ontology-based information integration has a very broad application in many domains. Our study provides a more facilitative way to think of semantics of ontology integration. |
PDF Full Text Request