Date
Attendees
Goals
Discussion items
Time | Item | Who | Notes |
---|---|---|---|
announcements | |||
upcoming meetings, calls for papers, submission deadlines | PIDapalooza; IDCC; RDA; iPres | ||
new draft published (old one expires 12/23) proposed: link most up-to-date draft from arks.org | |||
"Relevant to the discussion of returning metadata associated with digital resources. The recommendation here is to use HTTP headers." | GJ: just and fyi; it's actually from 2017; the idea is that resources can link to each other, to a landing page, and to metadata; all done through link headers, returnable with a HEAD request; it's complimentary to the data citation practices (Fenner et al), which embeds metadata in the HTTP payload vs the HTTP headers RM: I found this stuff very interesting because you can get metadata along with a returned object (immersive experience), but it's still on my "to do" list MP: curl -I https://digital.library.unt.edu/ark:/67531/metadc12345/ and they have a number of different examples at https://signposting.org/identifier/ MP: at a minimum, a link header could give a hint that "there's more metadata to request, eg, with an inflection" (as opposed to there's not more, which could avoid frustration) | ||
proposed: minimalist strategy regarding definition of the ark?info inflection Pros:
Thank you again for bearing with me this year during the exposition of several model sketches, which were my way of thinking out loud about a fairly comprehensive and ambitious approach to generic data objects and metadata. These modeling exercises have given me confidence in the general direction even though I doubt that any of them have been particularly comprehensible to all of you, let alone persuasive. I’m at a familiar milestone in the process where my inclination is to shift away and ignore, without effacing, the long term goals – effectively just “greying them out”. Instead, with your help I’d like to look down at our feet and consider what concrete baby steps we can take now. What’s the minimum we can add to the ARK spec to achieve the RFC without precluding the long term goals or disrupting current implementations? The baby steps for the “arkinfo model” should be very achievable and move us in a good direction, even if they leave us far from the ambitious goals. A first requirement, since the RFC freezes the spec, is R1. The RFC links to a place where community-based “arkinfo” evolution occurs. A second requirement is to explicitly mention the diversity in scope of ARK application. R2. ARKs are assigned to many different things, for example,
Dealing with the landing page (guide attribute) vs “plunging” (immersive, non-landing) page issue is vital to being able to link to metadata no matter what the ARK points to. Let’s define an arkinfo response to be a package of named data elements (whether in JSON, YAML, XML, ANVL, etc) returned when an ARK is submitted with the “?info” inflection. If the ARK were K, then the inflected ARK would be “K?info”, which is an arkinfo request. As a shorthand, if we don’t say which ARK it was, we can assume the canonical name of “K”. R3. If K is an encapsulation variant with immediate encapsulation ancestor R, the arkinfo response for K should describe its encapsulation siblings. Example: K is the English language, PDF format of chapter 3 section 2, and the siblings of chapter 3 section 2 also include variants in French and Spanish, and in RTF and HTML formats. R4. If K is contained as a part with an immediate partition ancestor R, the arkinfo response for K should describe its partition siblings. Example: K is the English language, PDF format of chapter 3 section 2, and the partition siblings also include sections 1, 3, and 4 of chapter 3. R5. An arkinfo response may include arbitrary metadata. R6. An arkinfo request can be modified to request alternate formats, eg, JSON, YAML, XML, etc. Similar to the traditional ARK spec, perhaps the safest immediate way forward may be to continue to be non-committal and define these baby steps by example rather than by prescription. | MP: like decoupling and keeping the RFC simple, doing the evolution externally; have found the non-prescriptive approach to be helpful; getting everything right will take time; the RFC is a kind of parent document that could reference the community spec (that could become other RFCs) GJ: to pursue RFC for ARK, what if we simply said that "ark?info" was reserved for future use |