Thesis paper for achievement of the scientific degree Doctor of Philosophy in the specialty 121 «Software Engineering» (12 – «Information technologies»). – Cherkasy State Technological University, Cherkasy, 2024.
The thesis paper solved the actual scientific and applied task of improving the process of modeling and development of software tools used in e-commerce through the use of an agent-oriented approach to expand the set of their properties and potential opportunities, reduce the number of used system processes and connections between components of the system, as well as increasing the speed of their development.
The thesis paper consists of an introduction, four chapters, conclusions, a list of literary sources and appendices.
In the first chapter, an overview of the features of the development and application of software agents, as well as a generalization of a number of the main disadvantages of the dropshipping approach in electronic commerce, is carried out. Features of agent-oriented programming, which became the basis for solving the problems of development and use of software in dropshipping, were revealed.
In the second chapter, the logical structure of the subject area under study is built using the representation of entities in the form of classes, as well as their connections and interactions among themselves. The method of implementing the results of the first section into the practice of software design, as well as the step-by-step sequence and features of the software implementation of agent-oriented programming using prototyping, are considered. Also carried out: analysis and modeling of the subject area, as well as its conceptualization; architectural design and modeling of system behavior.
In the course of the conducted theoretical study of the software modeling process of the automated information system for updating retail prices, the method of object-oriented programming for the development of software in electronic commerce was improved through the use of an agent-oriented approach, which allowed to reduce the number of system processes used by 4 units and also reduce the connectivity between system components by 1–3 connections to increase their autonomy.
In the third chapter, an experimental study of the process of algorithmization of pricing by a software agent was carried out, during which the efficiency of the software implementation of the developed data processing algorithm was increased in terms of calculation accuracy by improving the process of software implementation of pricing in electronic commerce according to the methodology of object-oriented programming due to the study of data processing processes and using its results at the stage of requirements formation.
In the course of the research, it was established the need to develop an additional user interface for interaction of the operator with the software and decision-making by him regarding the processing of extreme values of input data in order to further teach the software tool to make decisions independently. As a result, the need to use artificial intelligence in future versions of the software was established.
An experimental study of the software development process was conducted, during which the model of the implementation of evolutionary prototypes was improved due to the use of an agent-oriented approach to the development of software tools in e-commerce, which allowed increasing the speed of the software implementation of the user interface and asynchronous communication between the autonomous components of the system.
In the fourth chapter, when testing the performance of software agents, data collection of performance indicators was carried out in order to analyze and determine the amount of hardware resources consumed in a certain period of time when connecting software agents with different basic types of behavior, in different quantities, as well as during the performance of their main function — price updates. As a result of the analysis of the collected data, the rate of productivity growth in terms of hardware resource consumption for various types of behavior and interfaces was determined.
During the analysis of the performance of software agents and their impact on hardware resources, the following are determined: statistical indicators of performance for various types of behavior of software agents; expansion of the set of properties and potential capabilities of the software due to the application of an agent-oriented approach is not accompanied by the use of additional hardware resources.
