Some natural language processing systems, e.g. machine translation, require to have a knowledge base with conceptual representations reflecting the structure of human beings' cognitive system. In some other systems, e.g. automatic indexing or information extraction, surface semantics could be sufficient, but the construction of a robust knowledge base guarantees its use in most natural language processing tasks, consolidating thus the concept of resource reuse. This paper describes how meaning postulates can be built through FunGramKB, resulting in a knowledge base which provides deep semantics.