Etutriche is a quick experiment to automatically detect cheating students. It reads exams' responses from a bunch of CSVs (one per exam) and tries to find pairs of students with very similar answering patterns.
python etutriche.py [options] <path>
path should be a path for either a directory of .csvs or one .csv.
Supported options (see --help for more info):
--threshold: set a custom threshold for the Sørensen-Dice score.--max-score: students with a score higher or equal to this value are excluded; we assume that high scores equal good students equal no cheating. The downside of this is that it can lead to false negatives; but we try to avoid false positives.
The data format comes from the Google Form’s output of our online exams. Each CSV has one header line then one line per student’s input; only the first input is used. Each line has the following fields:
- Date of submission (
%m/%d/%Y %H:%M:%S) - Username (email address)
- Score (In the format
17 / 20in our case) - Last name (ignored)
- First name (ignored)
The N following fields are the student’s answers to the N questions. They
must be single- or multiple-choices questions; we don’t support answers typed
by the students.