Designing a Decision-Making Process for Partially Observable Environments Using Markov Theory

0
122

Authors

Tags: 2017, conceptual modeling, Sérgio Guerreiro

This paper is motivated by the problem of deciding how to proceed in a business process when a workaround occurs. In addition to this problem, most of the times, the exact state of the business processes is not fully available to the industrial organization. Therefore, it means that something wrong happens during automatic (or manual) operation; however, the managers do not know exactly the state of the operating systems. Usually, the combination of these problems drives to management decision without enough information and thus error prone. This paper integrates Markov theory with business processes design to predict the impact of each decision in the operational environment. The solution is tested in an agro-food industrial company that transforms fresh fruit to preparations that are sold to others companies. The paper shows that anticipating the production processes changes has the benefit of minimizing lot infections or stock disrupts. The main identified limitations are the compute intensive process involved and the effort required to estimate the business processes details. In future research, this work might be (i) extended to friendly software interfaces in order to facilitate the interaction with end-users, (ii) optimized to the algorithm of informed decision-making computation, and (iii) automatic estimation of business processes details using machine learning techniques.

Read the full paper here: https://link.springer.com/chapter/10.1007/978-3-319-59466-8_16