Automation of information distribution in adaptive electronic document management systems using machine learning

Abstract

Introduction. Electronic document management systems (EDMS) are used to store, process and transmit large amounts of information. Automation of these processes is a challenge that requires a comprehensive solution. Its solution willreduce the time and material costs for design and make the transition to a more advanced, adaptive EDMS. The paper is devoted to the development of new methods for automating the process of distributing information in the EDMS. The work objective is to improve the accuracy of the information distribution in the EDMS through moving from analytical or algorithmic solutions to the use of new methods based on machine learning technologies. The application of neural networks in the furtherance of this purpose will also improve the efficiency of software development through automating the analysis and processing of information. Materials and Methods. A new method of the automated information distribution based on machine learning technologies including a mathematical description of the information distribution rules is proposed. The formulated list of conditions for the information distribution provides the implementation of software based on neural networks for solving the problem of automatic data distribution in the EDMS. Results. The method of automated information distribution has been tested on the example of the EDMS subject area when solving the problem of analyzing the correctness of information entered by the user. In the course of experimental studies, it was found that the proposed method, based on machine learning technologies, provides better accuracy (8 % higher) and is more efficient (in accordance with the Jilb metrics and cyclomatic complexity). Discussion and Conclusions. The results obtained confirm the efficiency and accuracy of the method proposed. The presented results can be used to automate the processes of distribution and verification of information in adaptive EDMS, as well as in other information systems. Based on the method developed, it is also possible to solve connected problems: search for duplicates and similar documents, classification and placement by file categories.