Big data integration process faced structural heterogeneity and semantic heterogeneity problems. Ontology is an explicit specification of a conceptualization, which is the core of semantic web technologies that can be used to describe the semantics of the metadata. This paper builds a big data model based on ontology by exploiting semantic web technology, and propose an ontology-based semantic model and ontology-based key/value storage model, meanwhile giving detailed steps for building the model. The model proposed try to solve the issue of understanding between heterogeneous data systems. Finally, the authors give an example of the semantic model application, which has been verified on HBase-based prototype system.