- From: Dan Brickley <danbri@w3.org>
- Date: Wed, 29 Jun 2005 00:08:21 +0100
- To: Dan Connolly <connolly@w3.org>
- Cc: Jonathan Borden <jonathan@openhealth.org>, "'Henry S. Thompson'" <ht@inf.ed.ac.uk>, 'www-tag' <www-tag@w3.org>
(sorry this is a bit long; I've been meaning to dig up these references for a while...) Dan Connolly wrote: >On Tue, 2005-06-28 at 17:53 -0400, Jonathan Borden wrote: > > >>Dan, >> >> >> >>>I think it's plain that Mark is not an information resource, >>>so there's something of a contradiction, or at least a potential >>> >>> >>contradiction, here. >> >>I expect that if we outfitted Mark with a heads up display and keyboard that >>allowed him to view HTTP GET requests to http://www.markbaker.ca/ that he >>would be perfectly capable as acting as a standards compliant, if not a tad >>slow, HTTP 1.1 server. In that case he *would* in fact be an information >>resource, no? *** >> >> > >No. He's made of atoms, not (just) bits. > >I think the definition is pretty clear >http://www.w3.org/TR/2004/REC-webarch-20041215/#def-information-resource > >Perhaps some of the points I made leading up to that definition >are helpful... > > > yes, thanks. >[[ >Dan suggested that a a textual work can be consumed over the web in a >way that a table cannot; if you see a table and a movie in a product >catalog, while you can learn about the table using HTTP, you can never >consume it to the point where you owe the vendor the price in the >catalog, while with information resources, you can consume them to the >point where you owe the price just by observing representations. >]] > -- http://www.w3.org/2001/tag/2004/10/05-07-tag#infores2 > > "Consume" is a metaphor that draws on our knowledge of eating. I guess the core idea is of using up something that can't be replaced with equivalent without expense? While looking at pictures of a table doesn't use up the table owner's resources, there are (controversially but undeniably) a lot of sex industry Web services where "real world" resources are "consumed" via HTTP interactions (pay-for live video streams, etc.). Is a (webcast) 1:1 strip-show an "information resource"? Or, up a level, what do we stand to gain by coming up with a definition that decides this one way or the other? While I'm *delighted* that http-range-14 has been defused, I'm really not yet sure that the class "information resource" can be uncontroversially defined without a fair bit of hard work. There's a big literature around this distinction, eg. see in the digital library world, the debates that spun out of the interaction between Dublin Core (library) and INDECS (rights holder / publisher) metadata efforts. The D-Lib paper at http://www.dlib.org/dlib/january99/bearman/01bearman.html was an effort at a hybrid view, as was the Harmony/ABC work I was involved in with Jane Hunter and Carl Lagoze. Different metadata communities carve these distinctions in different ways. If the Web architecture itself is to embody just one such conceptualisation, we should tread warily. A lot of these discussions appealed to variations on a 4-level "Work", "Expression", "Manifestation", "Item" distinction, based on Tom Delsey and other's work for IFLA on Functional Requirements for Bibliographic Retrieval *(IFLA=International* *Federation* *of* *Library* *Associations* <http://www.ifla.org/>). see http://www.ifla.org/VII/s13/frbr/frbr.htm and nearby in google('ifla','frbr') space... http://www.ifla.org/VII/s13/frbr/frbr.pdf If the TAG decide to pursue this task, I do recommend that FRBR gets some serious attention, as it has a lot of mind-share in the library and digital library world. My understanding is that FRBR is best thought of as an attempt to come up with a conceptual model that allows information systems to be clear about distinctions such as between different versions of Hamlet, different editions, different physical books and their location in library or who they've been lent to, as well as the larger challenge of engaging with complex, composite, mixed-media works. http://www.ifla.org/VII/s13/wgfrbr/wgfrbr.htm suggests that FRBR is still an active concern, with a recent workshop, efforts towards an entity-relationship and/or an OO model, clarification of core concepts based on deployment and implementation experience, and relationship to the CIDOC CRM work from the museum world. CIDOC CRM is another, related, take on this problem space, has been around for some years, has an RDF/OWL expression, and deserves some serious attention by anyone trying to model the patterns of relationship between "information resources" and the real world artifacts that they represent or describe. See http://cidoc.ics.forth.gr/ http://cidoc.ics.forth.gr/downloads.html http://jodi.ecs.soton.ac.uk/Articles/v04/i01/Doerr/ compares some ideas from Harmony/ABC with CRM. IMHO CRM was and remains vastly more mature, but the main point I want to make is that this is an active, and subtle, area of discussion and debate in the wider metadata world. And one that connects to real professional practice (re librarians see the IFLA links, re museums, look around near CIDOC). For publishers/rights-holders, Godfrey Rust and co's work in the INDECS project took the FRBR model as a starting point, but refined it from point of view of parties who are concerned to be paid for the modelled content (and hence drew some careful and fine distinctions that folk in the Dublin Core, search and resource discovery communities were happier to gloss over). I'm not sure where that work is now being developed, but it's probably not far from the MPEG scene. It's been 5 years+ since I was really involved with this scene, but from 30 minutes googling around, it is clear that no clear 'winner' has bubbled up from these debates. Lots of communities have related, but different, ways of conceptualising 'information resources' and their relationship to various abstractions such 'works', 'ideas', not to mention versioning, and models of real world artifacts (eg. in a museum collection, or a print run of books, journals etc). Its a v interesting space, but one that'd be easy to dissapear into, never to return. I keep coming back to the question of "how would we know when we've got it right?". For further reading, Google has 1000s of hits on 'rdf' and 'frbr', a sampling includes.... http://netapps.muohio.edu/blogs/darcusb/darcusb/archives/2004/01/24/libdb-bringing-rdf-and-the-frbr-to-the-masses which reminds me of Morbus' work on an RDF-backed DB inspired by FRBR, http://www.libdb.com/ [[ LibDB allows you to /smartly and easily/ catalog your movies, books, magazines, comics, etc. into your own computerized "personal library". It is a *free*, *open sourced*, library and asset management system based on and inspired by the Functional Requirements for Bibliographic Records (pdf) <http://www.ifla.org/VII/s13/frbr/frbr.pdf>, triples from the semantic web <http://www.disobey.com/d/2002/sw123/>, and "the end-user doesn't, and shouldn't, need to know this stuff". ]] http://www.libdb.com/project_goals has a nice summary that connects FRBR to the world of RDF tools: [[ The ideas behind FRBR are a core basis for the design of LibDB, with specific attention to: * *the concepts of /work/, /expression/, /manifestation/, and /item/*. This involves the ability to say "I'd like to search for /Lord Of The Rings/, but I'm not interested in every version of every book published: I'm mainly interested in when it was written, and who it was written by. When I view this info, certainly give me links to its /manifestations/ (the various editions of the book) or related /works/ (the movies, soundtrack, or artwork inspired by), but don't let me see one thousand useless queries: I care little about the Third Edition put out by A Fictional Press for Devoted Members of Abbey Square Apartment 13". This also gives the end-user the ability to add their own comments about an /item/ in their possession: "This /item/ I own, which is a /manifestation/ of the Second Edition book, has a slightly coffee-stained cover and is located in Row 3, Box 17 of my Attic." * ]] http://eprints.rclis.org/archive/00000205/ mentions another system, VisualCat, also using FRBR and RDF. I'd hope that if 'Information Resource' is to be defined, it will turn out to be obvious an uncontroversial which classes in these vocabs are subclasses of InformationResource. Unfortunately some of them are pretty abstract, so I'm not sure how that decision would be made. I don't really have a more concrete point, beyond 'this stuff is hard (but kinda interesting, and related to real use cases)". And I guess also "a lot of people have tried to make such distinctions before, and it'd be polite to take their work into account somehow, since they represent important constituencies of the Web community". cheers, Dan
Received on Tuesday, 28 June 2005 23:08:27 UTC