A Japanese semantic network built on a pulsed neural network with encoding associative concept dictionaries

Abstract

A semantic network with the dynamics such as spreading activity is available for various systems as well as the structure model of knowledge. In this study we suggest a practical dynamic semantic network available for NLP, which has the structure from associative concept dictionaries and the dynamics from a pulsed neural network. We built the semantic network by means of constructing the platform called Brain Memory Model based on a pulsed neural network first, then encoding data of associative concept dictionaries into it. We also constructed the module as one of the applications of the semantic network built on BMM to NLP, especially to simile understanding. The outputs of the module were understandable in general, indicating the possibility of our semantic network. We are now considering to expand the scale of a semantic network and to introduce the learning algorithm for its self-organization.