Combining an LNS-based approach and organizational mining for the Resource Replacement Problem

For companies, it is crucial to promptly react to (even short-term) lack of resources, for guaranteeing the continuity of the operations in business processes. This leads to the solution of a Resource Replacement Problem (RRP) aimed at reassigning as many activities performed by resources that are no longer available to those that are available. To this purpose, several aspects are considered simultaneously, e.g., resources skills, workloads and other domain-specific constraints. In this paper, we propose an innovative hybrid approach for solving RRP, combining mathematical optimization with organizational mining. In particular, logs of past process executions are used to model a social network of resources by organizational mining techniques. Then, a similarity measure among resources is derived and exploited along with run-time resource workload and information on activities priority to formulate an Integer Linear Programming (ILP) model for reassigning the activities of unavailable resources, minimizing the total reassignment cost. To efficiently solve RRP, a Large Neighborhood Search based matheuristic is developed. Computational experiments show that the proposed matheuristic outperforms the commercial solver used to solve the ILP model. A sensitivity analysis, on possible variations of the input parameters and on the moves of the matheuristic, concludes the work.