Clusters in Semantic Space Analysing natural language texts to model word meaning as a procedural representation
Burghard B. Rieger · 1983
This paper will report on one of the objectives of a project in Computational Semantics currently being worked on by the MESY-group at the Technical University of Aachen. Among others, the project is concerned with the development of automatic frame construction from natural language discourse. Both, in linguistic semantics and in artificial intelligence, most of the language data processed is still obtained introspectively or by way of questioning test-persons. Based mainly on the investigator's or the systemdesigner 's own linguistic competence and/or world knowledge, the relevant data for the modelling of semantic and/or conceptual structures has a more or less ad hoc character and often lacks intersubjective control. Therefore, we have been and are trying to circumvent this deficiency by developing an algorithmic procedure that takes natural language texts on a certain subject domain as input and produces as output a distance-like datastructure of linguistically labeled space points whose positions represent (connotative) meanings.