LAUSR

This article considers the main models of the architecture of agency systems of project-oriented management as stages of their development. The agent technology allows us to decentralize problem solving and create complex systems of project-oriented management, combining various processing methods such as modeling, reasoning, and machine learning, and also allows us to distribute knowledge. One of these models is an aggregated architecture for systems of project-oriented management, based on agents of a marked deductive system. This approach allows us to divide algorithms into separate modules and distribute the knowledge base into parts. The focus is on existing multi-agent data mining architectures and the roles of agents in them. An architecture is described to support the decision-making process in conjunction with the use of event-driven and task-driven data mining agents, as well as helpers and knowledge management agents. The article then considers a mathematical model of the proposed decision-making system, identifies key parameters, and suggests improvements to the model based on the proposed integrated software solution. The practical significance of this study is determined by the fact that not only was the software architecture developed and presented for the first time, but also a fully extended mathematical model of a project-oriented management system.