See, I don't think the techniques we're talking about here are really
specific to rdbms or entity-relational databases.
The idea of using a unique identifier to build a relationship between
two records is a pretty standard way to relate two records in _any_ kind
of information system. It has one implementation in an
entity-relational rdbms, yes. But it's not really unique to that. In
the web world, using a URI of some kind as this unique identifier would
be the typical way to do it; a URI isn't the most natural way to do this
in a rdbms, but it's the same basic idea.
If you have a unique identifier for an entity (or for a record
describing an entity), you can use that unique identifier to record a
relationship between another entity/record and that one in an
information system. If you don't... then you can't record that
relationship in an efficient and effective way. It's not unique to
rdbms technology.
This is a pretty important fact of information systems that has direct
impact on how we record metadata. Those designing standards for
recording of metadata need to understand it. If those doing that
designing are catalogers are not programmers (as they probably should
be), then those catalogers need to understand at least a bit about
information systems. Because to live in information systems is the
destiny of the metadata created.
Jonathan
Weinheimer Jim wrote:
Bernhard Eversberg wrote:
What exactly do you want to say here?
Do you really mean relational databases? I see this term frequently
used erroneously instead of "entity-relationship" databases. The word
"relational" in RDBS does precisely not say that the database cares
about relations between objects or entities. The term was created by
mathematicians who developed the first models. For them, a "relation"
was just a mathematical term taken from set theory and meaning a subset
of a table.
Apologies for the shorthand. What I meant was the use of primary and foreign
keys in databases. RDF is an extreme example of this way of operating, but in
any case what I meant was the use of a primary/foreign key or a URI instead of
text strings. It is my personal view that a lot of this is highly technical and
should not be designed or decided upon by librarians or catalogers, although we
should have a lot of input and be the primary testers. Our areas of expertise
are different from those of an RDF or RDBS expert.
Will any of these projects happen or finish anytime soon? Of course not, but
look how long it took to get ISBD. In many ways, the information community is
screaming for a project that they can hook URIs into. I suspect that
instituting such a system would take less time than ISBD so long as people
remain flexible.
Jim Weinheimer
--
Jonathan Rochkind
Digital Services Software Engineer
The Sheridan Libraries
Johns Hopkins University
410.516.8886
rochkind (at) jhu.edu
