Evolvable neuro-fuzzy system for artificial creativity in linguistics

Abstract

The paper describes an application of evolvable fuzzy neural networks for artificial creativity in linguistics, which consists of the intelligent mechanisms of affix and root morpheme creativity and analysis. The task of the creation of an English vocabulary was resolved with neural networks which have an evolvable architecture with learning capabilities as well as a fuzzy connectionist structure. The paper features a form of artificial creativity which creates words on its own using genetic algorithms, fuzzy logic methods, and multiple layer neural networks. A review of selected issues is carried out with regards to linguistic creativity as well as spontaneous language production. The paper also presents experimental results of affix and root morpheme recognition in the context of word meaning analysis. © 2008 Springer-Verlag Berlin Heidelberg.