In this paper we present a semantic architecture that allows developers of knowledge intensive CBR systems to effectively deal with the problem of knowledge imprecision and vagueness through the utilization of fuzzy ontologies. The architecture is based on a novel knowledge intensive CBR framework, called IKARUS-CBR, which we presented in previous work and which addresses the problem of knowledge imprecision through the integration of Fuzzy Algebra in the Ontology-Based CBR paradigm. Along with the architecture we also present a comprehensive methodology that defines the necessary steps for building and using CBR systems that follow the IKARUS-CBR framework