Research On Mapping Relational Database Schema To Ontology

With the rapid development of bioengineering technology, the biological data grows exponential. How to integrate these distributed, heterogeneous, autonomous biological databases effectively and provide query services is becoming a hot topic. In order to address the current problems encountered in the integrated query of biological data, a data integration program based on semantic metadata is proposed by the research group of which the author is a member. Firstly, the metadata of the distributed database to be queried is integrated into a metastore in accordance with the uniform metadata standards. Then the metadata is annotated with domain ontology to generate semantic metadata. The program tries to exploit structural metadata and semantic metadata to address heterogeneous database integration issues, enabling the query on various biological databases.On the basis of the meta store and ontology repository established by the research group, the design and implementation of a data integration system based on CWM Metastore and ontology annotation technique is proposed in this paper. An algorithm of mapping relational database schema to ontology for the important module of building semantic metadata in the system is studied and implemented. And a semi-automatic ontology annotation tool is developed. The main research points of this subject put forward as following:(1)A data integration system and the complete process of data integrating based on CWM Metastore and ontology annotation technique are proposed. In this paper discussion has been made about how to resolve the semantic conflicts among database schemas with ontology as the semantic mediator, after the database schemas (SQL DDL script) have been imported to CWM Metastore and annotated by ontology.(2)Three types of schema matching problems are surveyed, which are data schema matching, ontology matching and matching between relational database schema and ontology.(3)For the needs of domain data integration, a hybrid algorithm of mapping relational database schema to ontology is proposed. In the process of element-level matching, not only string-based similarity but also the sense similarity based on WordNet is taken into account in calculating the name similarity. During structure-level matching, based on the similarity characteristics exhibited by the elements of different types in the mapping pair, the structure similarity is calculated.(4)An ontology annotation tool is designed and implemented base on the mapping algorithm proposed. It can help domain experts annotating the metadata with ontology in a way that they can confirm or revise those mappings output by the algorithm. This tool can relive the domain experts from the burdensome work of comparison. The experiment indicates that the mapping algorithm proposed by this paper gives better results than the previous works. And the mapping algorithm based ontology annotation tool can help the domain experts annotating the metadata with ontology with more convenience, and this tool can be integrated as the semantic metadata building tool into the schema of proteomics data resource integration based on metadata proposed by the research group, therefore promotes further study.