Yusuf Pisan

JollyFeedback - An Automated Feedback System

Discussion created by Yusuf Pisan on Dec 3, 2018
Latest reply on Dec 9, 2018 by Wasi Khan

[Posted to HigherEd and Instructional Designer groups]

 

Hi,

 

I always want to provide students with more feedback than I ever have time for, so I taught that some of the feedback could be automated.

 

I was not the first person to have this brilliant idea. There is a long history of automated feedback systems (and Intelligent Tutoring Systems) with various levels of success. There are free ones (such as AutoGradr - Automatically grade programming assignments, Web-CAT, Marmoset, APOGEE, Infandango, and Submit!) as well as commercial systems (such as Vocareum, Zybooks, Turings Craft, and Pearson Revel). All of these systems have their advantages and disadvantages.

 

Well, this did not stop me from building my own version, called JollyFeedback or Jolly for short.

 

If you would like to try out Jolly, the link to download is below. It is based on python3, so it should (in theory) run on any system. The main features of Jolly are:

 

  • Download all student submissions or a single submission from Canvas to instructor's computer via command-line (no logging into Canvas, clicking buttons required, just 'jolly.py --download')
  • Move each student's submission into its own directory and unzip any zip files if needed
  • Run default programs (such as unzip, cpplint, cppcheck) on each student submission and save the results into a file.
  • Email the results of the tests back to students
  • Jolly provides a test harness, as such it can work for all types of submissions. For students submitting essays, the test could be to run grammar and spell checker on the essays and send the results back (writing or integrating grammar/spell checker is left to instructor). For students submitting computer programs, it can run the programs, compare the results to the expected output and email the students the differences.

 

I have only tested it on the UW system, so I am not sure at this point how "customized" the Canvas installation is and whether the things that work on UW system will or will not work on other systems.

 

If you are interested in testing out v0.0.1-alpha-minus, have a look at the README at jollyPublic/README.md at master · pisanuw/jollyPublic · GitHub  and download the system from Releases · pisanuw/jollyPublic · GitHub 

 

I look forward to your feedback and who knows might even be able to put in "high priority" features if you were interested in using it.

 

Thanks

 

Yusuf

pisan[at]uw.edu

Outcomes