Abstract: For the issues of heterogeneous and lack of semantics about scientific data, this paper proposes an ontology-based semantic extraction method of heterogeneous data. Firstly, we construct the semantic extraction framework of scientific data, and on this basis, respectively formulate the corresponding semantic conversion and extraction  rules for the structured, semi-structured and unstructured data, thus form the local domain ontology. Secondly, we construct the global ontology of domain data through the association mapping, and then achieve the effective organization and unified management of heterogeneous data. Finally, we apply the related method to the field of materials, and realize the effective data management and sharing service in the field of materials.

Key words:  ontology, heterogeneous data, semantic extraction, data integration, domain scientific data