Comments on: I2: Strawman http://lackoftalent.org/michael/blog/2009/06/13/i2-strawman/ The occasional rambling of a digital library artisan Mon, 22 Feb 2010 22:08:57 +0000 http://wordpress.org/?v=2.9.1 hourly 1 By: Michael Giarlo http://lackoftalent.org/michael/blog/2009/06/13/i2-strawman/comment-page-1/#comment-93279 Michael Giarlo Sun, 14 Jun 2009 20:56:50 +0000 http://lackoftalent.org/michael/blog/?p=346#comment-93279 @Jonathan: Right. And you'll notice I didn't just say HTTP URIs because suggesting an identifier scheme is the easy part; I'm interested in linked data, which builds upon HTTP URIs but it more than just assigning identifiers. Linked data patterns already include ways to disseminate metadata. The 800-lb. gorilla is, of course, how such a decentralized (or even a centralized) system is managed. Who gets access? Are records mutable? How do you handle changes in organizations? How much metadata do you need to return about an institution? Who manages that metadata, and how? And so on. The big questions have yet to be answered, and I sure don't mean to give the impression that I've got all the answers. However, I would like to see NISO fully explore a solution to this problem that is "of the web." It may work and it may not work, but I'd like to put this idea through the paces before discarding it. @Jonathan: Right. And you'll notice I didn't just say HTTP URIs because suggesting an identifier scheme is the easy part; I'm interested in linked data, which builds upon HTTP URIs but it more than just assigning identifiers. Linked data patterns already include ways to disseminate metadata.

The 800-lb. gorilla is, of course, how such a decentralized (or even a centralized) system is managed. Who gets access? Are records mutable? How do you handle changes in organizations? How much metadata do you need to return about an institution? Who manages that metadata, and how? And so on.

The big questions have yet to be answered, and I sure don't mean to give the impression that I've got all the answers. However, I would like to see NISO fully explore a solution to this problem that is "of the web." It may work and it may not work, but I'd like to put this idea through the paces before discarding it.

]]> By: Jonathan Rochkind http://lackoftalent.org/michael/blog/2009/06/13/i2-strawman/comment-page-1/#comment-93278 Jonathan Rochkind Sun, 14 Jun 2009 18:17:05 +0000 http://lackoftalent.org/michael/blog/?p=346#comment-93278 Cool. To me, saying "http URIs" isn't an answer at all. I mean, we could assume there would be SOME string representing an institution that came out of this -- that's the point, right? So, if that string is just random chars, or an http URI, no big deal. The hard part is figuring out how they get assigned, discovered, have metadata attached to them which is also discovered, etc. Saying "http URI" might be suggestive of certain directions, but it doesn't answer the actual hard questions. I'm not sure if the answers to the actual hard questions are dependent on whether the actual token is an http URI or not. Maybe? Cool. To me, saying "http URIs" isn't an answer at all. I mean, we could assume there would be SOME string representing an institution that came out of this — that's the point, right? So, if that string is just random chars, or an http URI, no big deal. The hard part is figuring out how they get assigned, discovered, have metadata attached to them which is also discovered, etc. Saying "http URI" might be suggestive of certain directions, but it doesn't answer the actual hard questions. I'm not sure if the answers to the actual hard questions are dependent on whether the actual token is an http URI or not. Maybe?

]]> By: Michael Giarlo http://lackoftalent.org/michael/blog/2009/06/13/i2-strawman/comment-page-1/#comment-93276 Michael Giarlo Sun, 14 Jun 2009 14:41:08 +0000 http://lackoftalent.org/michael/blog/?p=346#comment-93276 @Robert I'm almost scared to entertain that question. Perhaps a new metadata format stuck behind an OAI-PMH interface? I'm sure folks can think up much less webby options than that, even. @Jonathan I'm getting there, slowly. Rather than creating one mega-post about linked data and I2, I'm cutting them up into chunks and mulling them over while the I2 group's work evolves. I don't have an example yet but one is starting to form in my mind. Cool that you're interested enough to want more, though. :) At this point I want to be sure I'm not totally off the rails suggesting that linked data might be applied to this problem in light of the requirements thus far identified. I am sure as more use cases are found and as I flesh this idea out, the real issues will come out and the real problems will be revealed. It's one of the questions I wanted to explore in the next post, though I may devote the next one to examples and the final one to questions about, say, how to manage a global, decentralized inst. identifier infrastructure. But, to address your questions directly: The IR subgroup has designed a survey to elicit use cases for inst. identifiers, so I can't give a use case just yet, but we are administering the survey to repo managers around the world this week, so I hope to have more to say w/in the next month as survey results roll in. @Robert I'm almost scared to entertain that question. Perhaps a new metadata format stuck behind an OAI-PMH interface? I'm sure folks can think up much less webby options than that, even.

@Jonathan I'm getting there, slowly. Rather than creating one mega-post about linked data and I2, I'm cutting them up into chunks and mulling them over while the I2 group's work evolves. I don't have an example yet but one is starting to form in my mind. Cool that you're interested enough to want more, though. :)

At this point I want to be sure I'm not totally off the rails suggesting that linked data might be applied to this problem in light of the requirements thus far identified. I am sure as more use cases are found and as I flesh this idea out, the real issues will come out and the real problems will be revealed. It's one of the questions I wanted to explore in the next post, though I may devote the next one to examples and the final one to questions about, say, how to manage a global, decentralized inst. identifier infrastructure.

But, to address your questions directly: The IR subgroup has designed a survey to elicit use cases for inst. identifiers, so I can't give a use case just yet, but we are administering the survey to repo managers around the world this week, so I hope to have more to say w/in the next month as survey results roll in.

]]> By: Jonathan Rochkind http://lackoftalent.org/michael/blog/2009/06/13/i2-strawman/comment-page-1/#comment-93275 Jonathan Rochkind Sun, 14 Jun 2009 13:35:04 +0000 http://lackoftalent.org/michael/blog/?p=346#comment-93275 We need to see an example of what this means. Okay, HTTP URIs. How will it be decided and subsequently discovered which HTTP URI represents which insitution? Can you give an example in the context of an actual use case? I forget if the task group has some use cases identified in addition to requirements. We need to see an example of what this means. Okay, HTTP URIs. How will it be decided and subsequently discovered which HTTP URI represents which insitution? Can you give an example in the context of an actual use case?

I forget if the task group has some use cases identified in addition to requirements.

]]> By: robert http://lackoftalent.org/michael/blog/2009/06/13/i2-strawman/comment-page-1/#comment-93274 robert Sun, 14 Jun 2009 08:29:09 +0000 http://lackoftalent.org/michael/blog/?p=346#comment-93274 i think HTTP URLs as institutional identifiers may also have the beneficial effect to force institutions to think more about their URL space (and domain names in particular). i'd say some sort of conscious management of domain names within an institution is necessary anyway - it's just overlooked often. i think HTTP URLs as institutional identifiers may also have the beneficial effect to force institutions to think more about their URL space (and domain names in particular). i'd say some sort of conscious management of domain names within an institution is necessary anyway – it's just overlooked often.

]]> By: robert http://lackoftalent.org/michael/blog/2009/06/13/i2-strawman/comment-page-1/#comment-93273 robert Sun, 14 Jun 2009 08:25:46 +0000 http://lackoftalent.org/michael/blog/?p=346#comment-93273 maybe i have been brainwashed by the linkeddata crowd already - but what else than an HTTP URL and linkeddata would be an option? maybe i have been brainwashed by the linkeddata crowd already – but what else than an HTTP URL and linkeddata would be an option?

]]> By: Ross http://lackoftalent.org/michael/blog/2009/06/13/i2-strawman/comment-page-1/#comment-93272 Ross Sun, 14 Jun 2009 02:25:21 +0000 http://lackoftalent.org/michael/blog/?p=346#comment-93272 +1 You know I've got your back on this one. Assuming my scenario ever meets about anything. +1

You know I've got your back on this one.

Assuming my scenario ever meets about anything.

]]>