Towards a Formal Framework for Business Process Re-Design Based on Data Mining

In today's ever changing world, business processes need to be dynamic. Data accumulated as the processes operate capture the meaning of transactions in the past, which opens a door for the dynamics of the business processes in question. Mining the operational data to explicitly represent this meaning could lead to process re-design to make the business processes more efficient. In this paper, we propose a formal framework for redesigning business processes taking data mining rules and business rules as the driver. We formally represent business processes using the artifact-centric approach put forward by the IBM Research. We devise redesigning algorithms that take classification rules extracted from data mining together with business rules and transform the business process in question by eliminating redundant tasks and/or re-ordering inefficiently placed tasks. We illustrate our algorithms and report experiments that were conducted using a proof-of-concept case-study.