Ontology-based Data Integration

Nowadays, people focus on the problem of heterogeneity, more specifically on semantic heterogeneity in data integration. An ontology give the name and the descriptions of the entities of specific domains using predicates that represent relationship between these entities. It provides a vocabulary to represent and communicate knowledge about the domain and a set of relationship containing the term of the vocabulary at a conceptual level. Therefore, ontology might be used for data integration tasks because of its potential to describe the semantic of information sources and to solve heterogeneity problems. In this paper, we apply the idea of ontologies as a tool for data integration.In this paper, we have made a detailed introduction to how to build the structure of the ontology. We choose Description Logic as the method of representation and reasoning, and in turn provide new description logic with default reasoning, discuss how to use description logic represent resource at last.In this paper, we extract concept from information based on Formal Concept Analysis, an Algorithm named CLCA. We use ontology annotate assistantly information source, adopt VSM extracting character vector, use FCM clustering and calculate distance based on variance.In this paper, we propose a method of ontology query based on compositive similarity computing, which involves concept definition similarity, concept structure similarity and concept instance similarity. We discuss these similarities computing and the attributes weights determination method based on the Immunity Algorithm.In this paper, we apply the theories of relevance feedback to extend query. At last, a combined algorithm is proposed. We adopt VSM extend query keywords, and adjust weight based on analyzing row vector and calculating variance. The experiment proves that this way improve recall ratio and precision ratio.Lastly we provide a simple experimentation. We build an ontology on virus, and design a retrieval interface based on web. With our retrieval system, user could exactly find out the solve to the virus, and look for more information about it from checking the attribute similarity and figuring out the relationship between different virus.