A Framework for Recommending Resource Allocation based on Process Mining

Abstract

Dynamically allocating the most appropriate resource to execute the different activities of a business process is an important challenge in business process management. An ineffective allocation may lead to an inadequate resources usage, higher costs, or a poor process performance. Different approaches have been used to solve this challenge: data mining techniques, probabilistic allocation, or even manual allocation. However, there is a need for methods that support resource allocation based on multi-factor criteria. We propose a framework for recommending resource allocation based on Process Mining, that does the recommendation at sub-process level, instead of activity-level. We introduce a resource process cube that provides a flexible, extensible and fine-grained mechanism to abstract historical information about past process executions from process event logs. Then, several metrics are computed over the cube, considering different criteria: fitting between resources expertise and the expertise required to perform an activity, past performance (frequency, duration, quality and cost), and resources workload. These metrics are combined to obtain a final recommendation ranking based on the BPA algorithm. The approach is applied to a help desk scenario to demonstrate its usefulness.Keywords: resource allocation, process mining, business processes, recommendation systems, organizational perspective, time perspective