sgeinitz / picata Goto Github PK

Further clean up the Installation and Usage documentation with more links/screenshots/description, etc.

Additionally, there should ideally be support for both Mac/Linux and Windows (which also requires slight changes to the code).

PICATA is currently a command-line tool with only one instructor workflow taken care of.

To allow for more workflows, it would be helpful to have an improved UI. The Python module, textual, looks like a good solution for this.

The packaging and general installation for picata can be improved quite a bit. To get started, see: https://packaging.python.org/en/latest/tutorials/packaging-projects/

The peer/collaborative quizzes are done in the classroom and only students present that day are eligible to be paired. Currently, the process of checking who is present in class is manually done by the instructor (by quickly going through a csv copy of the enrollment and marks students present with a 1, and those absent with a 0). The picaTA then opens this manually populated list and will generate pairs for only those present.

Ideally, this process is simplified for the instructor. Perhaps by presenting the students physically in the classroom with a code/keyword that they send/reply to a Canvas message with. The picaTA would then read these and generate pairings for only those students that replied with code/keyword.

Currently the student does not get any feedback after the first (independent) a-quiz, nor after the second (collaborative) b-quiz.

Ideally, students would be sent a message after a quiz with some information about the question(s) that they missed. It may not make sense to do this for the a-quiz (depending on when it is due and when it is reviewed in class). But, it could certainly be done for the b-quiz. And, this could be a way to continue the collaboration/discussion between pairs of students as well.

Longer term, this would ideally be done by sending the student(s) a follow-up question that was similar to the first one that they missed. Generating new questions is likely not feasible for the instructor though, so this would be best be done using an LLM to generate a new question that is similar.

Adopt the usage of the logging module to allow for additional debugging and logging information to be easily switched on/off.

Longer-term plans are for the PICA TA to be a bot-like agent residing on a broadly used messaging platform (e.g. Discord, Slack, MS Teams, etc.). While Canvas is itself a rich LMS it is not the ideal nor preferred communication channel (this is true not just for students but instructors as well). Shifting the communication to a more natural channel will allow for more sophisticated interactions (e.g. students jointly answer a new question on the platform in a chatbot-like interface).

This will likely depend on what the final UI looks like but there should be a more (and better) structure to the program to allow for new functionality to be added.