Neural network based lexicon representation to support MT & ML

Abstract EN

This paper suggests a method for neural network-based Lexicon representation to built up a machine translation system.

The method adopts a Bi-Associative memory (BAM) for Lexicon representation that provides an easy access to the meaning of words in two different languages.

Appropriate techniques are used for database representation including wording file and its relation to classification file, meanings and roots file, suffixes and prefixes file and finally the morphological one.

Such method provides an appropriate means for data representation and coding in a way that makes the process of analyzing and understanding the given texts accessible which ultimately might be utilized for machine learning.

This would be done by finding appropriate analysis of the linguistic potentials of each word which finally leads to an accurate architecture of bi-direction associative memory.

The outcome would be a minimized storage capacity needs gained through a retrieving process that is both handy and easy.

A more optimistic speculation is to transform a written text into a visual animation by using multimedia techniques.