Abstract of FKI-204-95
Title: Neuronale Lernverfahren zur Modellierung der Semantik
spatialer Ausdrücke - Stand der Forschung und Entwicklung
Authors: Gabriele Scheler
Category: Technical Report (Forschungsberichte Künstliche Intelligenz)
Abstract: For several years there has been a strong interest in
computational models, which use the adaptivity, learning
ability and fault tolerance of artifical neural networks.
This is also increasingly becoming applied to the modelling
of language functions, such as lexical disambiguation,
syntax-semantics mapping or syntactic parsing.
These models have reached a level, where they are applicable
to real world problems and allow to use textual corpora for
evaluation. In this article, we give an overview
on the state of the art in the application of neuronal
learning procedures to linguistic problems, specifically
problems of syntax-semantics mapping, including a motivation
for using neural nets or similar learning techniques
instead of the still prevalent rule-based systems.
Finally a detailed model for the learning of a syntax-semantics
mapping function is developed, which is directed at inter-
pretation and generation of spatial expressions, and possible
applications in the area of grammar checking and machine
translation are being discussed.
Keywords: semantics, corpus-based linguistic analysis, connectionist
linguistics, spatial expressions
Size: 20 pages
Copyright: The ``Forschungsberichte Künstliche Intelligenz''
series includes primarily preliminary publications,
specialized partial results, and supplementary
material. In the interest of a subsequent final
publication these reports should not be copied. All
rights and the responsability for the contents of the
report are with the authors, which would appreciate