Can We Mimic Human Pragmatics Knowledge into Computational Lexicon ?