Lexical categorical disambiguation using a multi-agent systems architecture
João L. T. da Silva, Vera Lúcia Strube de Lima · 2002
We present a distributed architecture model for lexical categorical disambiguation based on multi agent systems. This proposal aims to minimize some shortcomings from sequential approaches to natural language processing systems, through cooperation among intelligent and distributed modules. We have proposed a model for a linguistic society of agents which includes linguistic knowledge as well as a set of procedures to manipulate it, and mechanisms for social reasoning. The disambiguation is solved by a subsociety of acquaintance agents that use a communication protocol based on cooperative learning in order to reach a common goal.