Abstract: Most of the existing domestic institutional research data repository only provide keyword search based literal matching, which can not meet the users' requirements for knowledge retrieval. Based on the literature review of research data ontology description specification and the application of BIBFRAME, combined with the metadata organization of major university research data platforms, this paper proposes the rationality and necessity of research data ontology based on BIBFRAME extension. By means of ontology modeling, metadata of research data sharing platform and BIBFRAME 2.0 ontology mapping are realized manually, and RDF formal encoding and semantic retrieval are carried out with instance data. The conclusion is that institutional research data repository consortium applies the mode of research data ontology, It can not only deeply reveal the semantic attributes and characteristics of data, but also effectively realize knowledge discovery and data sharing of research data alliance by establishing semantic recombination association among various entities.

Key words: Research data, Ontology, Bibliographic framework, Information search, Metadata, Semantic association, Knowledge discovery