In the real world, human knowledge and natural language have a big deal of imprecision and uncertainty. Imprecision and uncertainty play in the semantic Web context, as well as to many applications that use description logics (DLs) to capture, represent and perform reasoning with domain knowledge. In this paper, a fuzzy extension of description logic language SI is presented, which combines vague sets with SI. Its syntax, semantics and inference problems are investigated in the paper. Also the tableau algorithm is developed for reasoning in the vague SI.