The paper describes the project held within Russian National Corpus (http://www.ruscorpora.ru). Beside such obligatory constituents of a linguistic corpus as POS (parts of speech) and morphological tagging RNC contains semantic annotation. Six classifications are involved in the tagging: category, taxonomy, mereology, topology, evaluation and derivational classes. The operating of the context semantic rules is shown by applying them to various polysemous nouns and adjectives. Our results demonstrate semantic tags incorporated in the context to be highly effective for WSD.