Research On Ontology Based Retrieval Technology Between Heterogeneous Data Sources

According to statistics, more than 70% of the data in the Internet is stored in relational databases, while query to the database is mainly dependent on the professional interface, which can never meet the growing needs of customers. If people can retrieve the database like the search engine using natural languages, it must bring great convenience to the development of various industries. On the other hand, heterogeneous data sources usually contain rich associations, the traditional search methods force user use the results from one source to query another, which is time-consuming and laborious.In this thesis, we propose a method of ontology based retrieval technology between heterogeneous data sources, the main contents include:(1) Combining the existing technology of ontology extraction, ontology mapping, ontology integration, and research on the properties matching methods between several ontologies, to integrate the ontologies extracted from the relation schemas of heterogeneous data sources as the global ontology, which in the basis of retrieval technology of heterogeneous data sources;(2) Research on mining associations between the keywords from users’ query in natural languages using the global ontology, to find the potential sematic relations;(3) Research on transforming the query in natural language into database query language, using mapping between the global ontology and the relation schemas of heterogeneous database;(4) Design and implementation the prototype system of ontology based retrieval technology between heterogeneous data sources.Our main contributions in this thesis include:(1) Proposal of a similarity calculating method for attributes based on attributes values in ontology matching and integration;(2) Proposal of a method for retrieving heterogeneous data sources by means of the global ontology;(3) Proposal of a method for mining the relations between keywords in query by the graph data structure response to the global ontology;(4) Proposal of a method for transforming the query in natural language into database query language by the mapping between global ontology and relation schemas of heterogeneous data sources.