Automatic Grading of Linux Commands Using String Similarity Measures

An education sector is faced with a challenge of a manual grading process that is prone to partiality, subjective influence and other human errors. There are many grading systems in the body of knowledge, however the area of Linux commands hands-on assessments is still under investigation. To advance developments in this area, existing automatic grading systems and their limitations are discussed in this paper. Having discussed existing systems, we argue that the use of history command output provides enough details to measure the correctness of student's executed commands. We further argue that the weighted-average from the combination of token-based similarity algorithm and character-based similarity algorithm can make an ideal tool for auto grading Linux commands hands-on labs. The grading tool is named Command-line History Grader (CoLHiG) for ease of reference and have substantiated its components.