Hi Karen, On 3 January 2012 22:17, Karen Coyle <[email protected]> wrote: > Ben, thanks for taking a close look at this. It is unfortunately true > that there are no archives for the Archive.org lists, although they > would be handy in this case. (There are social issues for not keeping > them, not technical ones.)
I can think of possible social issues. People getting verbally abused for not reading the archives before posting ;-) ? > > First, we might want to pick a record that has a more full > description, since this one doesn't have subject headings or notes. > Alternatively we could edit this one to have what we need, and then > revert it later. (I'm thinking along the lines of "subject 1" "subject > 2" kind of thing.) How about "IT Architectures and Middleware": http://openlibrary.org/works/OL8021697W.rdf and e.g. http://openlibrary.org/books/OL21701098M.rdf It seems that the original work has one author, but later editions have two authors on the cover. There are subjects and classifications. > Second, we need to decide the usual "FRBR question" -- should we use > FRBR relationships even though they are not a perfect match? (The OL > "Edition" is a FRBR Manifestation+Expression.) I've come across this and liked the fact that the RDA Relationships for Works, Expressions, Manifestations, Items at metadataregistry.org includes the workManifested and manifestationOfWork relationships. Expression is skipped - but that made me think: isn't what is called Work in Open Library actually Work/Expression or maybe even just Expression: Works that are expressed in the form of books? I guess this list is not meant to elaborate on that... (I'm not on the librarianship list, I'm not a librarian, I'm just a computer science student.) Talis's RDF asserts both <Edition> <dcterms:isVersionOf> <Work> and <Edition> <http://open.vocab.org/terms/commonManifestation> <Work>, although I'm not sure I understand the latter (and the term is unstable). > > Some other details: > > 3 - Works can also have multiple authors. We need to decide whether to > repeat dcterms:creator or use the bibo:authorList wrapper. I believe > that authorList was used with the assumption that keeping the order > of authors is important. However, I'm not at all sure that OL does > maintain the order of authors. My vague recollection is that the > translation from MARC data to OL puts the MARC main author (100) into > OL/author and the remainder of authors (700's) into dc:contributor or > bibo:contributorList. But since the records can be edited there could > be multiple authors that are added later. The Bibo documentation indeed says order is important within authorList. This example Work/Edition was imported from http://openlibrary.org/show-records/marc_oregon_summit_records/catalog_files/washs02192008.mrc_revrev.mrc:1808839739:1085 The Edition's authorList contains the Work's author, now Raj changed the template. Peter Bye is listed as dcterms:contributor (as "Bye, Peter."), but not in a bibo:contributorList. Sidenote: if I click Edit on the Edition page, Peter Bye is not in the Edition's list of contributors nor is he an author of the Work. > > 4 - I'm not sure what the right way is to include the edition > information along with the work data. I'm thinking that the Work > should have a URL link to the "manifestation" and the rdf:Description > for each edition should be a separate description, outside of the > <frbr:Work></frbr:Work> I think it doesn't matter. Unless I misunderstood the RDF/XML syntax document, both forms are serializations of the same RDF graph. The only difference between my version and the current output is the direction of the relationship between Work and Edition: current output asserts <Edition> <rdrel:workManifested> <Work>, I asserted <Work> <rdrel:manifestationOfWork> <Edition>. I think these properties are (or should be) each other's inverse, making the assertions equivalent. > > I'm sure there's more, but that's a quick set of thoughts. > > kc > Thanks for your comments! Ben > Quoting Ben Companjen <[email protected]>: > >> Hi Anand, >> >> Using an example work/edition by a greatuncle of mine, I attached two >> edited RDF files, one for the work and the other for the edition. >> >> Work like it is now on OL.org: http://openlibrary.org/works/OL16419933W.rdf >> Edition: http://openlibrary.org/books/OL25154702M.rdf >> >> Some perhaps less noticeable changes I made: >> - all edition URIs now end with '/' (why do all work and edition URIs >> end with '/', whereas author URIs don't, by the way? I'd say lose >> them, like they appear in Talis's search results.) >> - added FOAF-namespace to both files; >> - added rdrel-namespace to the edition file; >> - copied the work's author to the edition's bibo:authorList; >> - added the names of contributors to the edition's (new) >> bibo:contributorList - although they lost their specific roles. >> >> Regards, >> >> Ben >> >> P.S. Perhaps I should say I'm pretty new to the developers' side of >> Open Library, so I only noticed the Launchpad and RDF related issues >> after Karen's mail and Github after visiting the OL Launchpad >> homepage. I thought I'd look through the archives before sending >> anything, but there appears to be no archive? >> >> >> On 3 January 2012 09:52, Anand Chitipothu <[email protected]> wrote: >>> >> [snip] >>> >>> Hi Ben Companjen, >>> >>> Can you please make a RDF response with your suggestions for a >>> sample edition record? >>> >>> I can convert that into a template for generating RDF in OL. >>> >>> Anand >>> _______________________________________________ >>> Ol-tech mailing list >>> [email protected] >>> http://mail.archive.org/cgi-bin/mailman/listinfo/ol-tech >>> To unsubscribe from this mailing list, send email to >>> [email protected] >> > > > > -- > Karen Coyle > [email protected] http://kcoyle.net > ph: 1-510-540-7596 > m: 1-510-435-8234 > skype: kcoylenet > > _______________________________________________ > Ol-tech mailing list > [email protected] > http://mail.archive.org/cgi-bin/mailman/listinfo/ol-tech > To unsubscribe from this mailing list, send email to > [email protected] _______________________________________________ Ol-tech mailing list [email protected] http://mail.archive.org/cgi-bin/mailman/listinfo/ol-tech To unsubscribe from this mailing list, send email to [email protected]
