@Lady

RDF ontology for fanfiction tags that can be used across separated archives.

Separation of communities of (a) work hosting, (b) namespace authority(s) -- where synning/subtagging/metatagging is wrangled, (c) final determination about what a given tag is synned to (whether you want to to be categorized), which gets text-closeness suggestions by a wrangling databases's list of "observed tags" but is decided by the author (an uncategorizeable freeform being not even given the data of `<... tag-ident: "tagcomm:my-innovated-long-tag-which-tagcomm-hasn't-wrangled-yet-but-could" ... >`)

If you subscribe to RSS feeds (possibly WebSub-managed, for more scalable distribution) of various work-hosting archives you can filter on whatever tags yourself. Or someone injests the tagging metadata from a bunch of subscriptions and hosts the graph databases and a SPARQL search server for a section of the community.

@gaditb is this an intentional or unintentional rep for https://ns.1024.gdn/Tagging/

@gaditb more friendly discussion here: https://blog.ladys.computer/2023-05-26/federated_tagging/

@Lady It's definitely knowlingly downstream of it but I think putting emphsis on the nuances of the vocabulary is losing focus on the methods of using it and the specific, much more finite and "primarily like two different things, but a lot (and hopefully we can luck out and predict those two things in the design)", utility it will provide people.

And expects a larger-than-necessary amount of work for people wrangling a given tagset who just want, like, the tagging their community has around their specific kinks and fics to be useful to them, and okay if it isn't too much work also to have meaningful consistent tie-in to adjacent community/tagsets, if some of their output comes our way.

(... and then in a FEW places there may and will be some gloriously intricate nerds crocheting great tag trees, works of art in their meticulous meaning, who may or may not kink on the organizational structure itself... but that's can't be the primary design audience.)

@gaditb ah well i think part of that is just the lack of implementations

the design goal is something that can be hosted as JSON files on Github pages. i think that's the right level of necessary technicality but has the flipside of being utterly unopinionated about how you generate those JSON files and get them in Github pages. i have my own opinions here but i don't think there SHOULD be an opinion, generally, at a schema level

i don't think there is a NECESSARY lot of work around getting a basic tagset. i think it can be really simple. (i mean, i think you can do it by hand.) but i also think that there are plenty of fannish librarians who love categorizing things and i’m really frustrated in the lack of tools they have to DO that right now, so this is growing a bit out of that frustration

i also think you might be underestimating the advantages of structured data: if you ARE working with a limited tagset of known values, it is FAR easier to find the tag you want by going “Comics > Superheroes > Marvel > Spiderman” than by trying to remember if the correct tag is “Spiderman” or “Spidey” or “Peter Parker”. fuzzy text matching offers a better experience over a structured knowledge system if you don’t know what you’re looking for, but a typically worse experience if you do, because people remember a concept’s location in a hierarchy much better than they do its exact name. in both the case of people assigning tags to works as well as that of people looking for works, i think they can generally be assumed to have familiarity with the tags they are using, so i want to design for the system which works best for them. that means emphasis on structured relationships and text labels can be thrown to the wind (which is to say, left to authors without consequence)

(this is better for internationalization too)

@Lady But that's only if you're trying to /categorize/ your work -- for most people, I'm pretty sure tagging is primarily communication to their imagined audience. And if so, I wouldn't say to you "this is a comics story, a superheroes story, a marvel story, a Spiderman story" I'd say "Peter Parker is a Nice Jewish Boy".
If it wasn't too much of a hassle, sure I might toss a comma in there to help it have more legs -- "Peter Parker, is a Nice Jewish Boy" -- but even that's already secondary.

People have some sense of the place in a tag hierarchy they're posting in, but I'm pretty confident they only have an idea of two, maybe three or four places at once -- that is, using spacial reasoning to distinguish a subcategory of tags also runs into limits-of-the-functional-imagination that restrict its use as such.

@gaditb oh, i think we disagree about the purpose of tagging then

i think the primary purpose of tagging is aiding discovery systems, and i think this stance is strongly evidenced by how tags are used for search and filtering on AO3. i question why “Peter Parker is a Nice Jewish Boy” is being entered into the “tags” field and not the “description” field if the purpose isn’t associating the work with other works where Peter Parker is a Nice Jewish Boy

if that association IS desired, then it’s to everyone’s benefit if that tag is suggested to authors, for example, if they have already tagged the work with “Peter Parker”, so they don’t recreate the tag with a different spelling and require wrangling work to smooth over the differences (assuming user-suggested tags are even an option)

i don’t want to foreclose the idea of tags that aren’t discovery-oriented. i think you’re right that it’s a use case. but i also feel that users might want works tagged “Peter Parker is a Nice Jewish Boy” to appear in tag listings for “Peter Parker”, “Nice Jewish Boys”, “Jewish culture”, et cetera, even if those other tags weren’t supplied… and it’s an open question whether it is a good thing or a bad thing that this can happen even when a creator is primarily in a communicative frame of mind

@Lady I mean I definitely overstated my case and I don't mean to imply that tags are NOT discovery-oriented, just that I do think people are imagining the act of tagging a lot more finitely and conversationally than "categorizing according to a schema".

I think it is a lot more about imagining a particular audience they share context with, and describing the contexts of the work /to them/ in tag form.

(Not universally, but very heavily.)

@Lady AO3 tagging practices inherit a lot from Tumblr tagging practices, where often tags are circuitous references made knowing that the audience of followers shares the context to understand it. Similarly, many tags on AO3 are written with the context-carrying audience in mind: things like "no beta we die like Uncle Ben" are not the author saying "unbetaed", but wanting to change the text displayed for their own amusement -- that text is inherently the contebt they wish to communicate to an audience that understands it, and it is /AO3/ that translates (syns) it into its database.

Asking the author to do that translation themself is a dead-end. Your ontology is not their audience; they have no inherent allegiance to to it.

@gaditb i think you're assuming that ontologies are being made from people outside of the community and my point is the opposite, that in a federated context communities can express themselves directly to their audiences using the language they both have agreed on, instead of needing a third party to wrangle that language into a globally-recognized form

@gaditb you can't have that cake and eat it too though without those communities doing the work to, you know, define that agreed-upon ontology and put it out there in a format their members can use, which is the problem space i'm trying to tackle

@Lady I don't disagree, but then in your discussion of your proposal you're talking about using opaque hexstring identifiers, and tags objects being internationalizeable and translateable with no canonical language, and finding a tag by drilling from a universalized external-to-everything start-point down to the specifics, picking the intended meaning as it goes.

Rather than a use-case of something starting from a particular meaning and a particular context, with the job of the ontology (with hopefully a section of that ontology maintained by people living inthat context) to provide enough available LOCAL hooks to connect the known meaning into, to translate it up to more general audiences.

e.g. avoiding the needing-to-include-all-of-"#‌Gamer #‌Gamers #‌Gaming #‌Gameing" issue.

(I dunno, maybe I'm talking nonsense here.)