Title :
Formal neuron system for the natural language analysis
Author :
Borzenko, Alexander
Author_Institution :
Independent Res. Group, Kharkiv, Ukraine
Abstract :
The basic operation set of any natural language does not depend on one particular language but on their universal core-methods of data transformation in natural neural networks. The mathematical model of an artificial neural network is designed. The deductions from this model form the basis for protolanguage definition as a formal logic system for the analysis of natural languages. The experimental test of protolanguage principles is provided by building the neural computer VNC as a facility of neural network structure unification. VNC was realized through its complete C-program model and was tested on the unlimited Russian language. The main distinctive feature of VNC is its ability for accumulating, automatic systematizing and comparing natural language messages in the current context. There are numerous applications where this ability can be especially effective
Keywords :
formal logic; formal specification; natural languages; neural nets; Russian language; formal logic; formal neuron system; mathematical model; natural language analysis; neural networks; protolanguage; structure unification; Application software; Buildings; Computer networks; Humans; Logic; Mathematical model; Natural languages; Neural networks; Neurons; Testing;
Conference_Titel :
Neural Networks Proceedings, 1998. IEEE World Congress on Computational Intelligence. The 1998 IEEE International Joint Conference on
Conference_Location :
Anchorage, AK
Print_ISBN :
0-7803-4859-1
DOI :
10.1109/IJCNN.1998.687262