Пошук академічних текстів

Інформація × Реєстраційний номер 2123U006684, Матеріали видань та локальних репозитаріїв Категорія Препринт Назва роботи Business process management system for optimisation of unstructured businesses Автор Onyshkiv Roksoliana Дата публікації 01-01-2023 Постачальник інформації Український католицький університет Першоджерело https://hdl.handle.net/20.500.14570/4815 Видання Опис This thesis presents an approach to optimizing non-structural processes found in service businesses without changing the nature of such processes. In addition to structural, data-driven processes, most organizations work with non-structural, knowledge-driven processes. Because such processes lack con- sistency in execution, they are usually more complex. However, these non-structural processes are typically high-value-added, confirming their critical importance as the main operational components of a business. Most existing approaches to optimiz- ing non-structural processes "bring structure to unstructured", providing both ad- vantages and disadvantages. Changing the nature of a non-structural process may decrease flexibility, creativity, and adaptability since the structure limits the freedom of action of the human resource within the framework of the process and makes it more challenging to make changes to workflow quickly. In this thesis, we take such non-structural processes’ informality and randomness properties for granted. However, we learn to work with their ad-hoc nature and try to partially transfer the "expertise" during their execution from the human to the machine. The following approach to optimizing non-structural processes of a service business was proposed: 1) consider such processes as a set of actions that have an impact on the current state of the consumer (modeled as Finite Probabilistic Non- deterministic State Automaton); 2) implement the process of choosing the next opti- mal action step in the process instance for the consumer using the AlphaZero model- based algorithm of reinforcement learning, based on the highest customer satisfac- tion in the shortest time to recommend the optimal decision to the resource (hu- man); 3) prioritize tasks for the resource when several instances of non-structural processes are launched simultaneously, based on the assessment of urgency, conse- quences of delay, level of negative states and general context for each instance’s next best-recommended action. The proposed approach is validated in this thesis using a service business and was evaluated by comparing the resource load and customer satisfaction metrics of the As-Is (unstructured) process against the proposed approach over the course of an entire month of operations. An improvement in the performance of the consid- ered processes was found when the proposed solution was applied. However, it is impossible to precisely answer how much the proposed approach improved the process performance due to the influence of other simultaneously introduced inno- vations, such as work logging and daily retrospectives. As the main goal of this the- sis was to develop an approach, transferability efforts were limited because of time constraints. Future work would increase the case studies to establish the validity of the proposed approach. Додано в НРАТ 2025-05-09 Закрити