An Approach to Determine Important Attributes for Engineering Change Evaluation

Abstract

Enterprises plan detailed evaluation of only those engineering change (EC) effects that might have a significant impact. Using past EC knowledge can prove effective in determining whether a proposed EC effect has significant impact. In order to utilize past EC knowledge, it is essential to identify important attributes that should be compared to compute similarity between ECs. This paper presents a knowledgebased approach for determining important EC attributes that should be compared to retrieve similar past ECs so that the impact of proposed EC effect can be evaluated. The problem of determining important EC attributes is formulated as the multiobjective optimization problem. Measures are defined to quantify importance of an attribute set. The knowledge in change database and the domain rules among attribute values are combined for computing the measures. An ant colony optimization (ACO)based search approach is used for efficiently locating the set of important attributes. An example EC knowledgebase is created and used for evaluating the measures and the overall approach. The evaluation results show that our measures perform better than stateoftheart evaluation criteria. Our overall approach is evaluated based on manual observations. The results show that our approach correctly evaluates the value of proposed change impact with a success rate of 83.33%.