Supporting Inconsistency Resolution through Predictive Change Impact Analysis

Today, model-driven software processes rely on consistency management approaches to deal with the multitude of inconsistencies that occur in large systems. To resolve a detected inconsistency, the software designer applies one resolution out of a set of resolution options to each inconsistency. To do so, the designer needs to understand the extent of changes posed by each resolution. In this paper we propose change impact analysis to support the designer in this task. We present a simple algorithm for predicting the impact of inconsistency resolutions by checking the instantiation of different meta-model relationships. Based on one small case study, we demonstrate that our algorithm provides a reasonable estimate for the number of changes that actually will be applied. We demonstrate the usage of impact analysis for inconsistency resolution and make a first step towards a decision support tool to help a software designer resolve inconsistencies.