Lexicon for multiword expression identification

Anonymous

Abstract: Following the idea that lexicons are needed in order for automatic identification of multiword expressions(MWE) to handle the unpredictable nature of MWEs, this paper proposes a lexicon formalism, itself declined in multitudes of possible sub-formalisms depending on the linguistic features considered , along with an evaluation method which could be used to compare lexicon formalisms to each other.An exploration of the powerset of features is done in order to find the bests of such subset of features to be used. The impact of the proposed lexicon formalism on MWE identification is investigated, leading us to conjecture that lexicon indeed have the potential to help MWE identification.

Paper Type: long