Эвристика наиболее частого

This paper proposes benchmarks for systems of automatic sense identification. A textual corpus in which open-class words had been tagged both syntactically and semantically was used to explore three statistical …...