search

tl-its-umich-edu / my-learning-analytics

My Learning Analytics (MyLA)
Apache License 2.0
36 stars 39 forks source link

Possibility of revealing filenames before they are available. #752

Open jonespm opened 4 years ago

jonespm commented 4 years ago

Thank you for contributing to this project!

Expected behavior (A clear and concise description of what you expected to happen) :

The current cron process to get files calls

                    status += self.update_with_bq_access()
                    status += self.update_canvas_resource()

So the bigquery process pulls down all the files for a course, and the update_canvas_resource actually removes them if they are unavailable and updates the file name.

We actually should be removing them if they are not present in the UDW. This could happen if there was a long delay between the events being inserted into BQ and the cron UDW/cron process running. An instructor clicking on the file would register as a caliper event and appear on the very bottom of the list. The student would not have access to this file but could see what the names are before, potentially before they are available.

A potentially better process @jennlove-um and I talked about was to only consider files with events generated by students as part of the update_with_bq, but that would take a little more thought to implement.

jennlove-um commented 4 years ago

Follow up with Matt. Discussion raised the issue that only files that have been accessed by at least one student will show up on the list. A file that is not available to students and has only been accessed by instructors would not be included in the list.

jonespm commented 4 years ago

So maybe this similar to how we did the binning and shouldn't reveal any access stats until at least a certain number of users access the file. It still would probably be useful to filter out non-students but that might be a separate issue.

jonespm commented 4 years ago

Another issue could be if a file is deleted, the events for the deleted file will still be pulled and I believe this record would still display on our table. We should clean it up, even though they'll be a delay between when the file is deleted and when the data warehouse is updated.

jennlove-um commented 1 year ago

Test and draft requirements.