Architecture to track feedback for specific content nodes

Supports: https://github.com/learningequality/studio/issues/4044
Related: https://github.com/learningequality/studio/issues/4133

Stories

As an admin of Studio, if I see content that violates our Community Standards or Terms of Service, I would like a way to flag the piece of content for visibility purposes. Any resources flagged for these reasons, I would like to see in the Administration section. Also within the Administration section, I would like to be able to remove the flags if they are no longer valid. I would also like to prevent publishing, and their visibility in search, of these flagged resources until the flag is removed.

As an engineer interested in improving the recommendation models utilized in the Search and folder contextualized content recommendations project, I would like to collect information on how users utilize the content recommendations. I would like to know which recommendations that they do not select when they do choose content to import. I would also like to provide a way for a user to explicitly provide feedback for a recommended resource if it isn't a good recommendation. I would like to store this information in a way that distinguishes between these two situations and is optimal for re-training our deep learning models.

Requirements

• The data model should be flexible for both use cases to minimize development time
• The data model should require a reason, the user who submitted it, a reference to the content node, and some additional context metadata which is specific to the feedback
• For search recommendations, the context will need to be able to handle the same information that is sent to the inference API

Out of Scope

• Integrating the feedback architecture into search recommendations