Tagging in ChRIS

This came up in today's ChRIS status meeting.

Right now, we have a tagging system in ChRIS. Some notes about how it works today:

- Tags can be added and removed on a per-feed basis.
- Each feed (specific run of plugins/workflow, not abstracted) can have many tags
- Each tag can be applied to many feeds.
- (Tags are many-to-many.)
- Tags are per user account. They aren't global right now.
- The initial intention behind tags was to allow researchers to organize their feeds, e.g. by project or type. They were kept per user because naming structures could be as personal as an individual's personal file folder structure.
- As noted by Rudolph and Jorge, users today aren't making use of the tag feature.

We should think about the tag system and how we want to use it going forward and how we might want to roadmap it. Some ideas in no particular order:

- autocomplete / hinting when user is adding tags to help with consistency (this is a problem inherent in tagging systems, inconsistency between tags)
- tag suggestions - based on ML for example - when you open a feed
- autotagging - can we automate tagging based on various things (e.g. dataset has already been tagged "dyslexia" - can future feeds using this data set be autotagged "dyslexia" ? would that be helpful?)
- tagging reminders - when user views feed, suggest to them they should add some tags
- there is some research / standards on folder hierachy / structures for various kinds of work (I can think of some for design work, particularly I think for 3D artists.) It would be interesting to review some of this and see if any of it applies in this context. If we could build a reasonable structure that would be useful to researchers using ChRIS, it would be nice to suggest it / allow it to be easily enforced in an automatic way so it's less work on them keeping their work organized.

The issue names no files, tests, or entry points. Start by locating the existing ChRIS tagging implementation, then review the listed autocomplete, suggestion, autotagging, reminder, and hierarchy ideas with maintainers; done means an agreed direction and roadmap for the tagging system.

Written by the indexing model from the issue text.