WASPBENCH: a lexicographer's workbench incorporating state-of-the-art word sense disambiguation

Human Language Technologies (HLT) need dictionaries, to tell them what words mean and how they behave. People making dictionaries (lexicographers) need HLT, to help them identify how words behave so they can make better dictionaries. Thus a potential for synergy exists across the range of lexical data in the construction of headword lists, for spelling correction, phonetics, morphology and syntax, but nowhere more than for semantics, and in particular the vexed question of how a word's meaning should be analysed into distinct senses. HLT needs all the help it can get from dictionaries, because it is a very hard problem to identify which meaning of a word applies. Lexicographers need all the help they can get because the analysis of meaning is the second hardest part of their job (Kilgarriff, 1998), it occupies a large share of their working hours, and it is one where, currently, they have very little to go on beyond intuition and other dictionaries. Thus HLT system developers and corpus lexicographers can both benefit from a tool for finding and organizing the distinctive patterns of use of words in texts. Such a tool would be an asset for both language research and lexicon development, particularly for lexicons for Machine Translation. We have developed the WAS PB EN CH, a tool that (1) presents a "word sketch", a summary of the corpus evidence for a word, to the lexicographer; (2) supports the lexicographer in analysing the word into its distinct meanings and (3) uses the lexicographer's analysis as the input to a stateof-the-art word sense disambiguation (WSD) algorithm, the output of which is a "word expert" which can then disambiguate new instances of the word.