On 5/19/2012 3:07 PM, Karen Coyle wrote:
Joe, this is the thrust of my blog post, which started this thread:
http://kcoyle.blogspot.com/2012/05/rda-dbms-rdf.html
and I say:
" Where the goal in relational database design is to identify and
isolate data elements that are the same, the goal in library cataloging
data is exactly the opposite: headings are developed primarily to
differentiate at the data creation point rather than allow combination
within the database management system. The goal is to have each data
point be as unique as possible and to be assigned to as few records as
possible. "
I think this shows a fundamental misunderstanding of entity relational
modelling as well as relational databases (two things that actually are
somewhat autonomous).
Many aspects of well normalized data in an rdbms are in fact designed to
'differentiate at the data creation point', and _certainly_ to "have
each data point be as unique as possible". (That latter is, basically,
essentially, a 'foreign key' -- and no, I am _not_ saying this means
that individual headings should be used as 'foreign keys' in an rdbms.
Our data is complex (with or without rdbms, with or without linked data,
our data is fundamentally complex), and a simple 'hello world' example
of naive approach to modelling it in rdbms is unlikely to be productive.
Thinking that rdbms can not accomodate such cases is a fundamental
misunderstanding of rdbms and how they are used.
Seriously, this is getting weirder and weirder. If you want to
understand how rdbms are used and what they can do or can not do, if you
want to make pronouncements on that to the internet and get the library
community to take action based on your pronouncements -- taking a course
in databases might be a really good idea, to make sure you aren't
getting everyone to follow you down a path based on misunderstanding.
Again, I'd like to make clear that I agree that the current ways we
create and store metadata are highly problematic for flexible software
usage -- but this has got almost nothing to do with rdbms's. Thinking
that what rdbms can or can't do somehow requires or forbids certain ways
of modelling our data is, IMO, a fundamental misapprehension.
