internetarchive / openlibrary

One webpage for every book ever published!
https://openlibrary.org
GNU Affero General Public License v3.0
4.97k stars 1.25k forks source link

Add "is not" property to the work #9500

Open bicolino34 opened 1 week ago

bicolino34 commented 1 week ago

Problem

A clear and concise description of what you want to happen

There are a lot of works that might have the same titles, but are actually different works or the other work might be an adaptation of the first.

In case of manga, there are often two editions being published: Tankōbon and Bunkōbon. They have the same content, but different amount of chapters inside them.

So, volume 1 of Tankobon might contain 5 chapters, and Bunkobon would have 10 chapters in the 1st volume. When the series gets fully published there might be 10 Tankobon volumes, but just 5 of Bunkobon. This might cause a confusion when making merges.

Currently, there is nothing preventing these works from merging and the librarian might accidentally do a faulty merge. There is an option of adding into a title info about the format, but I think a property to prevent merge would be a more reliable solution.

I think, this was discussed in some other issues, but no separate one was created.

Proposal & Constraints

No response

Leads

Related files

Stakeholders


Instructions for Contributors

Freso commented 1 week ago

Related: https://github.com/internetarchive/openlibrary/issues/9452

mekarpeles commented 4 days ago

It's worth noting that admins already have a field that looks like this:

Screenshot 2024-07-01 at 11 17 00 AM

I'd like to discuss a few solutions.

  1. Allowing librarians to create (subject) Tags for these common things like manga, Tankōbon, and Bunkōbon, as well as other common distinctions like Textbook, Non-Fiction. A challenge here is that Tags currently operate on Work rather than Edition
  2. Extend admin section to include trained super librarians and give ability to let super librarians add keys for certain editions in areas where they are sufficiently subject-matter experts and have coordinated w/ @seabelis et al
  3. Add a few specific fields/checkboxes to the form specifically for manga or other serials
bicolino34 commented 3 days ago

Tankobon and Bunkobon editions should be in two different works. They should not be merged

mekarpeles commented 6 hours ago

This gets even more confusing with series :)

For now, let's just make the metadata field available to super librarians, small permission change, then we can figure out the rest later

Freso commented 4 hours ago

I'd like to discuss a few solutions.

  1. Allowing librarians to create (subject) Tags for these common things like manga, Tankōbon, and Bunkōbon, as well as other common distinctions like Textbook, Non-Fiction. […]

Isn’t this already the case? I’ve added subjects like "non-fiction" and "dictionary" to Works before.

  1. Extend admin section to include trained super librarians and give ability to let super librarians add keys for certain editions in areas where they are sufficiently subject-matter experts and have coordinated w/ seabelis et al

This seems to be the one you want to go with for the time being. Do you have any plan on a schema that addresses this issue? is_not:OL…W or different_from:OL…W (akin to Wikidata’s P1889) or WHATEVER_YOU_DO_IN_YOUR_LIVE_DO_NOT_MERGE_THIS_WITH:OL…W? Would these then be checked during a(n attempted) merge and block the merge if they exist and conflict? What happens if a Work mentioned with the tag is merged; will the process check the redirect table, or will the references get updated at merge time?


My own suggestion would be to implement https://github.com/internetarchive/openlibrary/issues/9452 for a free text field and show this note (if it exists) on the merge page. This would allow to not just warn about similar Works, but also provide more context for why they’re not the same. E.g., instead of just being told is_not:OL123456789W you can be told This is the 2009 bunkōbon; do not merge with the tankōbon of the same name or with the 2023 bunkōbon that contain the content of an additional 10 tankōbons.

The free text approach is also more flexible and with fewer pitfalls in cases where Works get merged or where new Works need merging (“should this newly created Work be merged with OldWork 1 or OldWork 2?”).