Development of Neural Network Information Retrieval System from Text Documents

The aim of the paper is to describe the information retrieval model which retrieves information from text documents in natural language by neural networks. This model comes from the linguistic and conceptual approach for the analysis of text documents. The neural network model accepts the structure of conceptual, lingvistic oriented model, where the problem of document database creation and document indexing for keyword determining is solved. Query entering uses the same mechanisms as document formalization for example document database creating method. Proposed structure of neural network model loses the problem of document retrieval on the base of user question. However, learning algorithms and neural network invariancy is used by usage of neural networks, it is possible to decrease the complexity of language analysis algorithm computatation.