---------------------- Envoyée par Patrick LE BOEUF/811/DDSR/BnF le 06/02/2002 16:39 --------------------------- martin <[email protected]> le 06/02/2002 16:26:01 Pour : Patrick LE BOEUF/811/DDSR/BnF@BnF cc : Objet : Re: R
ιf. Re: Rιf. Re: [crm-sig]Mapping FRBR to CRM Dear Patrick, I think your answers below are very important for crm-sig. Please send them to the list. I think we completely agree. best wishes, martin [email protected] wrote: > > (My comments are integrated hereafter in your text, and introduced with a "<<") > > martin <[email protected]> le 06/02/2002 12:01:36 > > Pour : Patrick LE BOEUF/811/DDSR/BnF@BnF > cc : Panos Constantopoulos <[email protected]>, Anne Hume <[email protected]> > Objet : Re: Rιf. Re: [crm-sig] Mapping FRBR to CRM > > ------------------------------------------------------------------------------------------------------------------------------------------------- > > Dear Patrick, > > I fully agree with what you write below. You stress below a definition more > based on the observation of > the instances than on the plan. This duality is common to the ontological notion > of a class. Classes can be defined by "intention", i.e. the ideal by which we > measure > that something is instance of a class, or by "extension", i.e. the factual > instances. > A definition by "intention" is open to the future. The definition allows more > instances > to appear in the world at any time. The definition by "extension" closes the set > once > forever. I was not sure, to which degree the latter holds or not in reality. > The term "intention of a class" does not only pertain to a plan, but also to a > (not necessarily explicit) set of characteristics. > > Question: Once a thousand of books is produced, under an ISBN, and sold. A year > later, > the publisher decides to resume production. Can that be under the same ISBN? Can > that still be the same manifestation? If yes, the class "Manifestation" is open > to > the future, and intentionally defined. > > <<Typically, ISBN will remain the same in that case. ISBN will change only if > the publication shows a statement like "New edition" or "2nd edition" or > "Revised edition" etc. ISBN may be the same for two items produced at different > dates, provided it is not a "new edition" but just a reprint. On the other hand, > two items with "same" content and produced at the same date may have different > ISBNs, as is the case for simultaneous production of the "same" book in > hardcover and in paperback. At the beginning of e-publication there was some > hesitation but it is now admitted that the e-version and the printed version of > the "same" book should have different ISBNs. The ISBN standard is about to get > revised however. So it is tricky to speculate on the nature of ISBN in this > moment. > > The fact, that a "Manifestation" pertains to physical attributes, brings the > concept very close to the notion of "Type", as usual in archeology and museum > documentation (often referred to with the misleading term "Object Name"). > Therefore I spoke in my comment also about metaclasses. Now, a formalism of > metaclasses > is definitely a construct too mathematical for the CRM. But systematically > we have bypassed such formalisms by introducing Types and Conceptual Objects, > were a class is simultaneously a thing to group real world objects and an object > of human activity and discourse. I think we have so far not encountered any > logical > problems in doing so. I expect our discussion about Natural History requirements > will be > full of processes managing Types and their relationship to instances. > > <<Actually, there is some similarity between the manifestation-to-item > relationship in FRBR and the frog_as_species-to-one_individual_frog relationship > in Natural History. > > In this light, we can regard an Item as "has type: Manifestation". In Man-Made > Objects, > Types are directly related (but not necessarily identical) to the plan by which > they are made. I would like to stress, that the notion you refer below: > "all of their physical attributes are instantiated in the same way at the time > of their production" is relative in the strict sense. If that would be true in > an > absolute sense, one could not distinguish items. I have proposed to see "all > physical attributes" as relative to the functionality of the product, which > includes > of course all aesthetic aspects. It does not include minor damage, wrinkles, > wholes and stains in the paper below a threshold, changes in smell, finger > prints > etc. I even assume, that creating a set pertaining to one Manifestation with > a running sequence number would not be regarded as a feature breaking the > homogeneity > of the set. A museum may even gather characteristic "drop-outs" or > mis-productions > of such a set ("B-quality"). In that sense, the Manifestation is not purely > observationally > defined. > > <<I had first written: "all of their physical attributes as listed under > "manifestation"", but before sending my e-mail I dropped "as listed under > manifestation" because it sounded like some tautology, something like: "a > manifestation is defined as gathering all those items that show features that > define them as belonging to the same manifestation": this kind of definition > does not bring anything, it is a vicious circle. Perhaps I should have written > "a number of their physical attributes" instead of "all of their physical > attributes": as a matter of fact, nothing prevents a publisher from reissuing > the "same" book with a red cover instead of a blue one: it will be essentially > the same manifestation, provided there is no "new edition" statement nor shift > from "hardcover" to "paperback" edition. And features like "minor damage, > wrinkles, wholes and stains in the paper below a threshold, changes in smell, > fingerprints etc." are listed as attributes of "item" in FRBR. And as you say, > in those cases where each item has a running sequence number, these numbers do > not affect the existence of "manifestation" as such, they only pertain to the > items. Occasional printing defects on a given item do not prevent this item from > belonging to a manifestation, nor do they alter the essential characteristics of > the manifestation as a whole. > > Considering the fact, that the manifestation implies the information from the > Expression in a > clearly recognizable form, and the Type aspects you have stressed below, one > could > see Manifestation as subclass of Information Object and Type. > > The comparison between FRBR and the CRM is not a mapping like those of simple > data structures. > FRBR and the CRM are parallel in the attempt to interpret data structures. This > makes > the comparison intellectually demanding, but on the other side it is an > important measure > for the validity of the CRM concepts. It will be important to discuss the scope > of the > CRM with respect to the view the FRBR offers and the kind of interoperability > we want to achieve in practical terms to library data. Four scenaria are > possible: > > 1) Show a way to interpret FRBR compatible data as CRM compatible data, as a > "museum view". > 2) If necessary, extent the CRM by notions that allow to map back to FRBR > compatible data structures. > 3) Propose a CRM-compatible, but discrete extension for including FRBR semantics > consistently in the CRM, > preserving FRBR concepts, but without making it part of the CRM. > 4) Propose such extensions as integral part of the CRM. > > <<This deserves of course more thinking but at first glance I would say scenario > #3 sounds best. I don't think we should change CRM to have it fit to FRBR, but > since it is scalable it should be possible to define an extension, not intended > to stay permanently, but, according to the needs, occasionally allowing for > intercommunication between CRM and FRBR. > > best, > > Martin -- -------------------------------------------------------------- Dr. Martin Doerr | Vox:+30(810)391625 | Principle Researcher | Fax:+30(810)391609 | Project Leader SIS | Email: [email protected] | | Information Systems Laboratory | Institute of Computer Science | Foundation for Research and Technology - Hellas (FORTH) | | Vassilika Vouton,P.O.Box1385,GR71110 Heraklion,Crete,Greece | | Web-site: http://www.ics.forth.gr/proj/isst | --------------------------------------------------------------
