Publication at Faculty of Mathematics and Physics |
2010
Abstract
Electronic dictionaries of the future will be much in demand—for computational, pedagogical, and other applications—if they can be used as resources for mapping word meaning systematically onto word use. Research in computational linguistics and artificial intelligence over the past twenty years, despite many declarations of success, has shown that existing dictionaries, designed for human users, coupled with existing linguistic theory of a top-down, speculative nature, have failed to be suitable for this goal.
Nor are results using hierarchical ontologies such as WordNet any better. Such resources are very plausible for human users, but they fail to meet the challenges of mapping meaning systematically onto words in use in ordinary text.