On Sun, Nov 16, 2003 at 08:25:41AM -0600, Ed Summers wrote: > > Is the LC server the "definitive" Z39.50 database? If I suck down a record > > from there, send it to my database, add more information, etc., etc., how > > does it get back to the LC? Does it? Would the way I'd contribute be > > to simply run and promote my own Z39.50 server? > > I guess you could say LC has the definitive Z39.50 databasee, since they do > so much original cataloging. OCLC also has a huge database (WorldCAT) which > is incredible because of the holdings data it contains (records which tell > you what libraries have a particular title). > > Z39.50 is a mechanism for *discovery* only. If you want to run your own > Z39.50 server you might want to check out IndexData's SimleServer [1]. Just to clarify a bit: Z39.50 can be used for searching and retrieval of records from compliant databases, by means of Z39.50 servers and clients. One usually thinks of retrieving bibliographic records in MARC format via Z39.50, but the protocol is, strictly speaking, format-agnostic. I belive there are Z39.50 servers offering access to GILS records, for instance. When you search LC via Z39.50, you are searching the same underlying bibliographic data as when you search via the web-based OPAC. I believe one can expose XML data via Z39.50 as well.
As for who has a definitive database, that's a complex question. OCLC loads tapes of LC's cataloging, so you could consider OCLC a superset of LC. Many institutions modify catalog records acquired from OCLC before loading into a local database, so it's difficult to say if there is a definitive form of a given record. LC's cataloging copy is generally a very good starting point. Since they are the creators and maintainers of the Library of Congress Subject Headings and the Library of Congress Classification, their use of these headings and class numbers could be considered definitive, but again, many institutions adapt the class numbers to fit into local shelf-listing schemes. Cataloging rules don't stipulate procedures and outputs with the same degree of specificity as e.g. networking protocols. LC promulgates its own interpretations of AACR2 (Anglo-American Cataloging Rules) because there are many important specific cases not covered by AACR2 which must be resolved. Some institutions follow the LCRIs in toto. Many institutions choose to follow some of them. Some institutions may ignore them altogether. It is understood and accepted that local practice should play a role in understanding how to apply cataloging rules and standards in each setting. I would say that the lack of complete specificity, the reliance on local practice, and on the cataloger's judgement, are good things. First, catalogers should catalog to meet the needs of their users. Second, bibliographic records in their end use are in great measure understood and resolved by the human mind and not by other programs, which might choke on the discrepancies. You are picking things up very quickly. I hope this little rant was helpful in providing further background for understanding the cataloging world you are now exploring. Chuck Bearden
