Challenges for computational lexicography
Résumé
There seems to exist some consensus about both the usefulness and basic form of computational dictionaries; the consensus, however, concerns only a relatively small aspect of what is in fact involved in the construction of sophisticated and adequate computational lexica. In this note, we want to address both the question of what has been achieved so far in this area, mainly as far as morphological dictionaries are concerned, and the question of what additional information should be contained in these lexica. In addition, we would like to discuss some of the requirements and issues that arise when realistic and comprehensive applications are attempted; two practical areas where we have been active ourselves are the construction of semantic search engines in the area of job search and hotel recommendations. In both cases, it turned out that without recourse to previously constructed large lexical databases for German, it would have been much more complicated to build these applications; and on the other hand, it also turned out that the dictionaries we needed to construct for these applications turned out to be still orders of magnitude larger than the sizable dictionaries we had available at the beginning of these projects.
Un consensus semble exister sur l’utilité et la forme de base des dictionnaires électroniques. Toutefois le consensus concerne uniquement un aspect restreint de ce qui est requis par la construction de lexiques électroniques sophistiqués et adéquats. Le présent article traitera de l’état des connaissances en la matière sur le plan morphologique et formulera des propositions d’amélioration. En supplément sera menée une discussion sur les prérequis et les propositions qui émergent lorsque des applications réalistes sont mises en chantier.