MESS: Moshiri Exam Similarity Score

MESS (Moshiri Exam Similarity Score; pun on "MOSS") is a scalable Python tool for detecting exam similarity from student responses. For information about the mathematical methods behind the tool, see the Methods section of the MESS Wiki as well as the published paper.

Installation

MESS is written in Python 3 and depends on NumPy, SciPy, and seaborn, which can be installed using pip3:

pip3 install numpy scipy seaborn

You can simply download MESS.py to your machine and make it executable:

wget "https://raw.githubusercontent.com/niemasd/MESS/main/MESS.py"
chmod a+x MESS.py
sudo mv MESS.py /usr/local/bin/MESS.py # optional step to install globally

Usage

A comprehensive list of MESS arguments can be found below, but we highly recommend following the Tutorial section of the MESS Wiki.

usage: MESS.py [-h] -i INPUT -ot OUTPUT_TSV -op OUTPUT_PDF
               [--ignore_case] [-c CORRECTION]
               [-rm REG_MIN] [-rM REG_MAX] [-rd REG_XDELTA]
               [-kc KDE_COLOR] [-kl KDE_LINESTYLE] [-kw KDE_LINEWIDTH]
               [-rc REG_COLOR] [-rl REG_LINESTYLE] [-rw REG_LINEWIDTH]
               [-t TITLE] [-xl XLABEL] [-xm XMIN] [-xM XMAX] [-yl YLABEL] [-ym YMIN] [-yM YMAX] [--no_ylog]

optional arguments:
  -h, --help                                         show this help message and exit
  -i INPUT, --input INPUT                            Input Exam Responses (TSV) (default: None)
  -ot OUTPUT_TSV, --output_tsv OUTPUT_TSV            Output MESS Spreadsheet (TSV) (default: None)
  -op OUTPUT_PDF, --output_pdf OUTPUT_PDF            Output MESS Distribution (PDF) (default: None)
  --ignore_case                                      Ignore Case in Student Responses (default: False)
  -nt NUM_TESTS, --num_tests NUM_TESTS               Number of Significance Tests to Perform (default: None)
  -c CORRECTION, --correction CORRECTION             Multiple Hypothesis Test Correction (options: benjamini_hochberg, bonferroni, none) (default: benjamini_hochberg)
  -rm REG_MIN, --reg_min REG_MIN                     Minimum MESS for Regression (default: None)
  -rM REG_MAX, --reg_max REG_MAX                     Maximum MESS for Regression (default: None)
  -rd REG_XDELTA, --reg_xdelta REG_XDELTA            X Delta for Regression (default: 0.0001)
  -kc KDE_COLOR, --kde_color KDE_COLOR               KDE Color (default: black)
  -kl KDE_LINESTYLE, --kde_linestyle KDE_LINESTYLE   KDE Linestyle (default: --)
  -kw KDE_LINEWIDTH, --kde_linewidth KDE_LINEWIDTH   KDE Line Width (default: 0.75)
  -rc REG_COLOR, --reg_color REG_COLOR               Regression Color (default: black)
  -rl REG_LINESTYLE, --reg_linestyle REG_LINESTYLE   Regression Linestyle (default: -)
  -rw REG_LINEWIDTH, --reg_linewidth REG_LINEWIDTH   Regression Line Width (default: None)
  -sh, --show_hist                                   Show Histogram (default: False)
  -t TITLE, --title TITLE                            Figure Title (default: MESS Distribution)
  -xl XLABEL, --xlabel XLABEL                        Figure X-Axis Label (default: MESS Score)
  -xm XMIN, --xmin XMIN                              Figure Minimum X (default: 0)
  -xM XMAX, --xmax XMAX                              Figure Maximum X (default: None)
  -yl YLABEL, --ylabel YLABEL                        Figure Y-Axis Label (default: Frequency)
  -ym YMIN, --ymin YMIN                              Figure Minimum Y (default: None)
  -yM YMAX, --ymax YMAX                              Figure Maximum Y (default: None)
  --no_ylog                                          Don't Plot Y-Axis in Log-Scale (default: False)

Citing MESS

If you use MESS in your work, please cite:

Moshiri N (2022). "A scalable approach for detecting exam similarity in CS courses." Journal of Computing Sciences in Colleges. 37(10):8–16. doi:10.5555/3533760.3533761