mitodl / mit-learn

BSD 3-Clause "New" or "Revised" License
0 stars 2 forks source link

Include a creation date field for learning resources (particularly OCW courses) in API responses #423

Open mbertrand opened 10 months ago

mbertrand commented 10 months ago

User Story

It would be useful to include a "creation_date" field in the learning_resources API response for courses, particularly for OCW, that approximates the date that the course was created on. This is not the same as the currently existing LearningResource.created_on field which is omitted from the API and is based on the date that the resource was created in the mit-open database.

It's not a field that's required to be included for a record to be created, but given the range of dates that OCW courses have been published over, it would be helpful for a user to have when evaluating content.

Description/Context

For a course creation field, one thing that has to be considered is how to calculate this for non-OCW courses too. OCW is a bit unique in that "Quantum Physics I Spring 2013" is considered a different course than "Quantum Physics I Spring 2016", whereas for all other sources, those would be two different runs of the same course.

Acceptance Criteria

Plan/Design

Add a creation_date field to the LearningResource serializer (or alternately, the Course serializer).

Approximate the value for this field based on semester and year of the OCW course's one LearningResourceRun: Since courses are broken out into semester/year, the easiest way to list those would be ballpark dates, e.g. fall 2017 could be 2017/08/01, winter 2009 could be 2019/01/01, etc... If the semester isn't clear, defaulting to the first of the year would be a good proxy.

For non-ocw courses, use the earliest start_date value from all the course's runs. Or do we want the latest start_date instead as an indicator of how current it is?

Add this field to the search index schema as well.

ChristopherChudzicki commented 10 months ago

In light of the fact that not all OCW courses have years/semester info (and that is expected for "evergreen" courses) maybe OCW courses could expose their first_published_to_production value?

That might yield better results, too. It seems to me that the "Fall 2024" inferred date might be significantly behind the actual publication date, which could bump courses downwards in a "new courses" listing. (If we used the field for such a thing.)

pdpinch commented 10 months ago

maybe OCW courses could expose their first_published_to_production value

Sorry, I had assumed this is what we would do for this issue.

Do all (published) courses in OCW has a first_published_to_production value?

ChristopherChudzicki commented 10 months ago

Do all (published) courses in OCW has a first_published_to_production value?

The ETL pipeline draws from the s3 bucket, so we'd probably want to do some work on the Studio side to include the value in github, and on hugo-themes to include the value in output.