Decision-making in partially known business process environments using Markov theory and policy graph visualisation

This paper designs and validates an innovative solution to solve the problem of lack of information available for the deciders due to business process environments that are only partially known. The solution is applied to a partially observable case study and the validation is grounded in the interpretation of results delivered by Markov theory. Firstly, the domain of interest is formalised by a set of definitions; afterwards, an instantiation in a agrofood industrial company is presented to show its applicability and usefulness. The algorithmic solution, and visualisation, is fully presented to the reader. Results reveal a control policy that forecasts the future behaviour of business processes operation. Compared with related work that analyses past executions from available data, our solution has the advantage of forecasting decision impacts from current data. Moreover, this solution supports the management decisions, providing control policy graphs that express the impacts of decisions in the organisational operation, and therefore, minimises the risk of making wrong decisions. In the end, organisation is enforced with resiliency capabilities that are triggered whenever any misalignment occurs.