Structural Similarity Measurement using Graph Edit Distance-Greedy on State chart Diagrams

Abstract

With the emergence of the need for online learning, the automatic grading system is inevitable requirements in an e-learning system. The automatic grading system in software design courses requires a method for measuring similarity between the key-answer design and student-answer designs. There have been several efforts to develop methods for measuring the design similarity. The similarity measurement that has been developed based on semantic or structural aspects of the design. Nevertheless, the purpose of those methods is to reuse software designs. This study proposes a graph representation of the State chart diagram. The graph models the structural aspects of the State chart diagram. This study also proposes the use of Graph Edit Distance (GED) greedy for calculating the structural similarity between two graphs. Graph representation of the State chart diagram is used as input to the GED-greedy method. The results show that all parameters used can determine the structural similarity between two State chart diagrams with graph representation of the State chart diagram as input. State chart similarity results obtained were 0.83.