TGSim: Framework of the Tree-Graph Similarity for Automatic Assessment of Java Exercises

Abstract—Technology has continuously modernized educational models, As well as the college-level computer programming courses. The instant- assess technologies have been widely adopted to stimulate the learners and enhance the learning and grading effectiveness and thus save time and reduce errors from manual grading method. The research aim to develop an automatic method for checking and grading students' programming homework/assignment in JAVA programming course, adopting tree similarity techniques created by the MS Visual Studio code editor. The code similarity checking tool called Mayat in Python has been adopted. It is the modules related to tree generation and similarity calculation. Mayat would compare the inter-files Binary Search Tree for similarity regarding the shape and topology of the tree and the subtree. Matching of the two trees was checked by comparing the Abstract Syntax Tree between the homework code written by the student and a similarity value for the student code written by the teacher's code. The results have shown that the assessment of students' homework with the automatic assessment system can save time, reduce instructor’s workload and students are enabling to know the grade or score immediately through the assessment results.