Discussion
A question for people into metadata.
I want to mark a book as Fiction or Non-Fiction using [Schema.org](http://Schema.org)
The only applicable property I can see is https://schema.org/genre
That says the value can be text or a URl.
So should I use the plain text
"genre": "fiction"
or point to a definition like Wikidata
"genre": "https://www.wikidata.org/wiki/Q8253"
Vote in the poll… now!
@Edent other, the cursed route, a serialized json document it the text field, with a name and url key, solving your problem.
i'll see myself out.
@Edent I'd go for URI for parseability (and hopefully error correction/unification), but it sucks that it's not human readable.
Time to spin up a service with human-readable URLs that 302 to WikiData?
Of course, this would only work if clients resolve the URL, rather than just compare strings.
Why can't we have nice things?
@Edent I would not consider fiction or non-fiction to be the correct words for genre. This is too broad-grained, in my view. It is, of course, a type of genre, but more akin to category in the generic sense than how we are used to using the term genre.
Genre, on the fictional side, might be historical fiction, science fiction, or fantasy for example. On the non-fiction side, you might have military history, literary criticism, and so on.
@mpe that's fair. But I can't find any other suitable property within Schema.org :-(
@Edent I voted for URL (because I'm an RDF geek).
However, I'd expect a deeper taxonomy for the range of the "genre" property than simply "fiction"/"non-fiction"(*). Those should be the top two values of the taxonomy, immediately below the root.
Does the US Library of Congress publish a suitable taxonomy that you could use the heads of? (Dewey Decimal is, apparently, non-free).
(*) I get that this is your primary concern. Go for it! But still...
@Edent Plain text is immediately useful and doesn't rely on a 3rd party so zero risk that the site goes away and being left with something that's useless.
The element dc:subject is for this, inside an ePub, and is standardised to follow the Library of Congress Subject Headings.
https://id.loc.gov/authorities/subjects.html
LCSH generally include "fiction" or "non-fiction", inside the heading itself.
e.g.
<dc:subject id="subject-0">School sports--Fiction</dc:subject>
@shaknais I'm talking about using Schema.org metadata for a webpage, not within the XHTML of an ePub.
I understood that - I'm saying don't reinvent the wheel. Tie what you're building, to the thing it is examining. Which removes the work for you. You'll be able to copy and paste from the source.
HTML is a flexible standard. It won't mind using these tags.
@shaknais sorry, I'm obviously not making myself clear.
I'm talking about a Schema.org JSON LD document to be embedded in a page.
This needs to be something which can go in a https://schema.org/CreativeWork
Install bonfire.cafe
Get the full app experience
@Edent knowledge classification via linked data. 🥰
The tension is neither the wikdata uri nor its predicates are human parseable and the label isn’t machine readable.
So provide both and give it the concept treatment with a json object. Its id is the wiki data link with a label of “fiction”.
In skos it would look like `”genre”: {“id”: “urihere”, “prefLabel”: “fiction”}`, which gives human and machine text.
I’m sure there’s a schema.org analogue. Looking forward seeing what you chose.