Regresar

A SOMAgent for identification of semantic classes and word disambiguation

Abstract:

This work describes a method that uses artificial neural networks, specially a Self- Organising Map (SOM), to determine the correct meaning of a word. By using a distributed architecture, we take advantages of the parallelism in the different levels of the Natural Language Processing system, for modeling a community of conceptually autonomous agents. Every agent has an individual representation of the environment, and they are related through the coordinating effect of communication between agents with partial autonomy. The aim of our linguistic agents is to participate in a society of entities with different skills, and to collaborate in the interpretation of natural language sentences in a prototype of an Automatic German- Spanish Translator.