Abstract:

Scoring and finding plagiarisms in university programming classes are burdensome and time-consuming tasks for teachers.
The aim of the study is to help teachers to achieve fair grading and reduce their burdens while organizing university programming classes.
In this paper, we improved our proposed similarity measuring method and implemented it to the education support system aiming to provide meaningful information to help teachers’ scoring task.

We developed an education support system for university programming classes to automate some parts of burdensome tasks such as scoring and plagiarism detection.

To realize the main functions of the system: scoring and plagiarism detection, we implemented two different kinds of similarity measuring methods for programming languages.

One of them is called FRef that quantify the feature of a source code by using a number of reference source codes to extract rich information to differentiate short and similar source codes, and another one is CM Algorithm that measures similarity according to the student’s coding style.
We implemented each of the two methods as a scoring function and a plagiarism detection function of the system.
Here, we need to note that the source codes produced by students as a part of programming class have special features.
That is, the length of the source codes is short and they have similar structure and algorithm because the source codes are produced to achieve the same goal according to the teachers’ instruction. Thus, it is difficult to extract and quantify features from those source codes to achieve grading or find plagiarisms.

Also, as we working on this study, we re-considered about reasonable criteria for education support system, through a survey. The paper provides a discussion on this topic too.

Keywords:

Programming education, assessment, source code similarity, education support system.