Predictive Study of Changes in Business Process Models

Change impact analysis can play a major role in planning and establishing the feasibility of a change in forecasting the cost and complexity before its implementation. In this work, we focus in managing the change impact propagation on several levels of granularity and abstraction of business process models (BPM) systems in order to pursue the change management studies in BPM and related software units. We adopt a machine learning-based approach to study the importance of integral components of a BPM across different versions. The proposed approach allows to analyze the dependencies among the activities, data, and roles of actors in the “training” phase. The facts are collected and analyzed in the form of a matrix. Then, for the prediction phase, we use the Bayesian classification to optimize the gained experience and leverage the learning to trace the change impact propagation in similar systems. The proposed approach analyzes the dependencies among activities, data, and roles with respect to the change and related artifacts. In this regard, the current work evaluates the links, cohesion, complexity, and effectiveness to better analyze the impact of change and its propagation on the model through the dependencies among the artifacts of the model. The major objective is to reduce the unforeseen cost and the unexpected risk on the evolving business operations.