Empirical Investigation of UML Models Matching through Different Weight Calibration

Abstract

UML model matching and retrieval is widely known as optimization problem. This is because of the inconsistencies between software properties. Matching is a fundamental operation for UML model reuse, as such accurate matching between models' elements results in better reuse of such models. UML models consist of number of properties such as functional properties, structural properties, and behavioral properties. Such properties are source of numerous errors during software matching, because each property represents software system from different views. In this paper we empirically investigate the use of different weight values when computing the similarity of software system from multiple views. The paper investigates the improvement of similarity values through the calibration of aggregated metrics. The result reported shows the superiority of structural properties if assign higher metric value compared to other properties.