Designing an Adaptive Electronic Document Management System Based on Neural Network Architecture

Abstract

The article deals with the issues of design automation of adaptive electronic document management systems. Automation of the processes of analysis, processing and transmission of information in the development of information systems will reduce the complexity of implementation, time and material costs, free up the resources of developers to solve more complex and creative problems. One of the ways to automate these processes is the use of machine learning methods, however, without a formalized methodological and mathematical apparatus, it is impossible to provide a comprehensive solution to the problem. The article describes the approbation of a neural network architecture, including a set of approaches and methods based on neural network technologies, using the example of the subject area of electronic document management systems (EDMS). The structure of an adaptive EDMS, implemented in accordance with this architecture, is presented. In the course of experimental research, two test EDMS were implemented: the classical one, developed according to the RAD methodology and the MVC pattern, and the adaptive one, the design of which was carried out within the framework of a neural network architecture. As a result, a decrease in the cost (by 24.7%) and complexity (by 32.5%) of the EDMS implementation was achieved, and the adaptability of the system was increased (by 13.6%). There is also an improvement in its quality and an increase in productivity. The results obtained confirm the effectiveness of the proposed approaches and tools.