Heterogeneous data produced by Internet of Things increased continually. Inconsistent semantics of data became obstacle of sharing data. To handle them effectively and intelligently, a semantic model for Internet of Things was proposed, which was based on semantic technology and cloud computing. Scalable distributed file system was the basis of storing mass data. Distributed inference engine, which was based on Map/Reduce operation model, had solved the problem of reasoning on mass ontology. Hierarchical semantic provided the basis of intelligent processing, which had reached consistent understanding between heterogeneous data.