POPEL-HOW. A distributed parallel model for incremental natural language production with feedback
Wolfgang Finkler, Günter Neumann, Saarbruecken (Germany). Sonderforschungsbereich 314 Kuenstliche Intelligenz - Wissensbasierte Systeme Saarland Univ. · 1989
This paper 1 presents a new model for the produc tion of natural language. The novel idea is to com bine incremental and bidirectional generation with parallelism. The operational basis of our model is a distributed parallel system at every level of repre sentation. Starting point of the production are seg ments of the conceptual level. These segments are related to active objects which have to map them selves across the linguistic levels (i.e. functional-semantic, syntactic and morphologic level) as fast and as independently as possible. Therefore the in cremental behavior caused by a successive input is propagated on all levels. Linguistic requirements de tected by objects which are related to already pro duced segments at any level influence and restrict the decision of what to say next. 1