Open vancleve opened 4 years ago
I'm not sure this is a "bug" in the common sense of the word. As you note, the RDF translator is explicitly written to prefer og:description
over dc:description
and that's what's happening.
Having looked at this briefly, it looks to me that the reason for this is terribly convoluted, related to the use of dc.description in the Zotero RDF format (for the extra field).
This will require a more thorough dive than I'll be able to do in the coming months, I'm afraid, but I'll leave this ticket open for reference. I don't think the current behavior is correct -- I'd be surprised if dc: metadata isn't almost always better than og.
Thanks @adam3smith! If it's not too complicated I can take a stand at it. Though is the issue that a ton of translators rely on RDF.js and so I'd have to make sure changes didn't create new bugs?
Running the
Embedded Metadata.js
translator on https://www.nature.com/articles/ncomms14906 returnswhich is not the Abstract, but rather the content of
og:description
.From here, https://github.com/zotero/translators/blob/5a8e310028dad8b0fd3bf26bc856cf2049d0be6f/RDF.js#L1199 it looks like the Facebook OpenGraph (
og:description
) description element takes precedence overdescription
: https://github.com/zotero/translators/blob/5a8e310028dad8b0fd3bf26bc856cf2049d0be6f/RDF.js#L1301From the debug in
Embedded Metadata.js
, it looks like the RDF is being constructed with the correct abstract as:So, is this a "bug" that should be fixed in
RDF.js
orEmbedded Metadata.js
or inNature Publishing Group.js
?