Difference between revisions of "Nbgrader notes/tools"

From Earlham CS Department
Jump to navigation Jump to search
m (Using MOSS to detect cheating)
(Using MOSS to detect similarity in answers)
 
Line 1: Line 1:
 
=== Using MOSS to detect similarity in answers ===
 
=== Using MOSS to detect similarity in answers ===
  
1. ssh as nbgrader to tools.cs, cd to "Detection" directory.
+
1. ssh as nbgrader to tools.cs.
  
2. Run the bash script, converter.sh, by providing the paths of lab you are trying to investigate. E.g.
+
2. Run run_moss.py. Example run: $python3 run_moss.py -s s20 -l lab3
  >> ./converter.sh ~/cs128/submitted/*/Lab3-Math/
+
(Note that you do NOT have to enter the full name of the lab anymore. That should be in lab-names.txt in root nbgrader directory.)
This will convert all the python notebooks to .py files and store it in a directory with the name of your choice.
 
  
3. Use MOSS on the named directory as follows:
+
3. It will return a link that will remain live for 14 days. In that link you will find a list of "similar" student work, with similarity scores, listed in descending order of similarity. Click either on the left or right link in each row that seems like it might be a high similarity. It will pull up a page with color-coded sections that look similar. Click on the same color for both students and check similarity manually until you're satisfied.
 
 
  ./moss -l python DIR-NAME/*.py
 
 
 
Where DIR-NAME is the directory you created. It will return a link that will remain live for 14 days.
 
  
 
The documentation for MOSS is in the file called moss itself.
 
The documentation for MOSS is in the file called moss itself.
  
 
To do: Make it so we can improve results using a base file.
 
To do: Make it so we can improve results using a base file.

Latest revision as of 11:40, 17 July 2020

Using MOSS to detect similarity in answers

1. ssh as nbgrader to tools.cs.

2. Run run_moss.py. Example run: $python3 run_moss.py -s s20 -l lab3 (Note that you do NOT have to enter the full name of the lab anymore. That should be in lab-names.txt in root nbgrader directory.)

3. It will return a link that will remain live for 14 days. In that link you will find a list of "similar" student work, with similarity scores, listed in descending order of similarity. Click either on the left or right link in each row that seems like it might be a high similarity. It will pull up a page with color-coded sections that look similar. Click on the same color for both students and check similarity manually until you're satisfied.

The documentation for MOSS is in the file called moss itself.

To do: Make it so we can improve results using a base file.