It applies not to Bibframe, but to a broader range. Collation ordering is not part of RDF, but of retrieval languages.

There was a discussion about a new keyword "collate" in SPARQL's "order by" but I do not know the current state. As long as there is no collation support in SPARQL, it is no possible to sort Bibframe literals e.g. in german "telephone book" order (W3C language tag de-DE-u-attr-co-phonebk) I have implemented phonebook order in Elasticsearch with ICU collations and JSON-LD literals, so I don't care too much about SPARQL.

I would appreciate it to see the Bibframe project more like a platform - implementors need a platform approach, elements/vocabulary and discussions about fields and values and how they are linked is not enough. This is too MARC-like.

Such a platform project would require to collect "best practice" approaches, in a kind of a guide, useful for being referenced by implementors. E.g.

- how to read and write Bibframe
- how to retrieve Bibframe
- how to sort Bibframe literals
- how to update Bibframe resources
- how to consolidate (match/merge) Bibframe resources
- how to ...<insert your problem here> ... with Bibframe

The question is, will this guide demonstrate how much easier it is to implement Biframe than MARC, and how easy will it be to find out about the"Bibframe compatibility" of a platform?


On Sun, Jul 27, 2014 at 11:22 PM, Stuart Yeates <[log in to unmask]> wrote:
I'm somewhat worried by the informal use of the term "alphabetical order" in this thread. In the pure-UTF-8 world of RDF, there exists a thing called the 'Unicode Collation Algorithm' and it's much, much, much more complex than things usually implied by the term "alphabetical order."

There are very good reasons for this complexity (outlined in document above); if there's anyone who thinks that some of those reasons don't apply to BIBFRAME, I'd love to hear them.