Evolution of shared grammars for describing simulated spatial scenes with grammatical evolution

摘要

Abstract We propose a model based on an evolutionary process combined with an adapted planning process to develop a limited spatial language with a syntactical structure in a team of artificial agents. Syntax is induced by means of a grammar and the grammar itself evolves in order to reach a syntactical agreement in the team. Evolution is implemented by adapting an evolutionary algorithm where each agent in the team manages a population of chromosomes that represent possible grammars. Grammars can be used by agents to generate utterances which are subsequently applied in language games to describe spatial relations. A planning process builds the sentences, but agents select the syntactical alternatives according to their current communicative intentions. Results in two different linguistic task show how a shared grammar can be developed in the group of agents.