Effective Structure Matching Algorithm for Automatic Assessment of Use-Case Diagram

Introduction

Access to quality education is the key to a better quality of life for every individual on this planet in this 21^st century. Hence, there is a great demand all over the world for quality education. Providing quality education requires a very large number of highly qualified and skilled teachers in the every nook and corner of the world. There has been a tremendous shortage of such teachers in the world.

However, embracing innovative technology like information and communication technology and its application in the form of e – learning has solved this problem to a great extent. It has also helped us in overcoming the constraints of location and time. Delivery of appropriate learning material is one side of a coin for quality education. The other side of the coin for quality education is just and objective quantitative assessment and qualitative feedback, which needs to be done in both the paper based examination system and electronic examination system.

In the paper based and electronic examinations the answers can be free response, diagrammatic or selection of a correct answer for a multiple choice question (MCQ). There is a different degree of automation possible in the assessment of both the examination systems and all three types of answers. Automatic e-assessment has solved this problem to a great extent in case of objective or fixed response type of answers. It has solved this problem to some extent in case of free response answers. There is a tremendous scope for the use of imagination and hence wide variations in case of diagrammatic answers in the software engineering domain (Jayal A., Shepperd M., 2009). This makes it extremely difficult and consume lot of time and effort to do manual or semi-automatic assessment and provide qualitative feedback of such answers in case of large volumes. These types of assessments would require a very large number of skilled, qualified, willing and sincere evaluators at the specific location and at the specified time. But, there is a significant shortage of such evaluators. Hence, there is an urgent need to develop tools and technologies for fully automated assessment of various diagrams used in Software development (Vachharajani V., Vasant S., & Pareek J., 2016).

Use – case is a methodology used to recognize, explain and organize system requirements. It is used in several stages of Software Engineering life cycle like planning, designing, testing and creating user manuals. It shows functional requirements and represents interactions between different stakeholders of the system and hence one of the important UML diagrams in software engineering.

This research attempts to develop tools and algorithms for electronic submission and fully automatic assessment of use – case diagrams with quantitative and qualitative feedback.

There are various stages of this framework like graph generation of students’ diagrams and model diagram, label cleaning, syntax matching, semantic matching (Vachharajani, V. & Pareek, J., 2012a, 2019), structure matching through which the process of assessing Use – Case diagrams will be automated.

This paper proposes effective structure matching algorithm for matching two Use – Case diagrams drawn by a student and an expert respectively.

Structure matching means matching two nodes with all its properties. Apart from labels, properties include relationship between two nodes and their hierarchical association. In our case, we convert two Use – Case diagrams into m – array trees and assess them by applying tree edit distance approach where two nodes can be interrelated with each other by association, include, extend or generalization relationships. In a hierarchical association, there is a parent – child relationship. There can be a single branch (means branch of a hierarchy cannot be decomposable) or branch may be further broken into sub branches.

Different techniques are available like tree edit distance, tree alignment distance and tree inclusion problem for comparing labeled trees based on deleting, inserting and relabeling operations (Bille, 2005).