From human-oriented dictonaries to computer-oriented lexical resources - trying to pin down words

Abstract

Dictionaries are designed for the human user; electronic lexical resources are often designed with computers in mind: to represent information about the form, use and meaning of words which will allow computers to process human language efficiently, to extract information from language input and to reason over the information extracted. Where a human user can be assumed to know a great deal by either common sense or common cultural knowledge, a computer knows nothing in advance. This paper will look at some of the typical information concerning lexical semantics (word meaning) that has been included in resources like WordNet (for English) and DanNet (for Danish), as well as some radically different kinds of resources, such as SALDO (for Swedish).