Automated Evaluation of Novice Programs Written in Assembly Language

We consider the main educational objective of programming courses for beginners is to get students to understand important concepts in programming and be able to write programs using the concepts. For this purpose, teachers give various programs, and read their programs to see whether they obtain the concepts or not. However, teachers’ loads of evaluation tend to be heavy. Furthermore, students have to wait quite a long time to get teachers’ advice, if we intend to respond to them during a class. To avoid these unreasonable affairs, we aim at automating the reading and checking processes of the raw program; so that teachers can concentrate themselves on higher level work. There are several research projects on educational systems which detect bugs and advise about them based on program recognition [1,2,3,5]. Konishi and Itoh[4] indicated that these automated debuggers require a huge amount of knowledge on bugs and it would be difficult to constitute the systems practically. We also consider that the system would become practical if we restrict its roles to mere testing the acceptability and leave the whole judgment on programs to human teachers. For the target of automated evaluation, we selected a simple assembly language CASL that is adopted in examinations for information-technology engineers certified by the Japanese ministry of international trade and industry.