For reasons that are hard to explain in an email, bibliographic data and
relational db models are not compatible. Although bibliographic databases
often use normal relational DBMSs (Sybase, Oracle), they do not use them in
a truly relational way. In general, software must be written to create
heading and keyword indexes, and the bibliographic data is stored as a
"blob". Putting bibliographic data into a DBMS is somewhat like putting
full text data into a DBMS -- you end up using specialized indexing
software (like Inktomi or Google) rather than the DBMS capabilities.
There are DBMSs that claim to be XML-enabled. That would mean that the
input to the DBMS can be in XML and the DBMS can be instructed which fields
to index. The big names in database technology all make this claim, but I
don't believe that the two you mention, which are open source, have this
capability. One system that does take in XML and was designed for
bibliographic data is Cheshire, developed at UC Berkeley
(http://cheshire.berkeley.edu/). It looks like you can download the source
code, with the usual "buyer beware" kind of caveats.
At 12:19 PM 7/11/2003 -0800, Bruce D'Arcus wrote:
>As I recall, there are a few IT people on this list. Has anyone looked at
>or worked on implementing MODS in a relationship database (oh, something
>like PostgreSQL or MySQL)?
>If not, then all MODS storage is in flat files now?
>I'm trying hard to convince the people involved in the Open Office
>bibliographic module project to adopt a MODS-like database model, but I
>really know nothing about DB design, so I don't feel qualified to offer
>specific suggestions on how to do this. If anyone has any suggestions,
>please let me know.
>http://www.fastmail.fm - A fast, anti-spam email service.