> -----Original Message-----
> From: Resource Description and Access / Resource Description and Access
> [mailto:[email protected]] On Behalf Of Kathleen Lamantia
> Sent: April 8, 2011 1:08 PM
> To: [email protected]
> Subject: Re: [RDA-L] FRBR
>
...

>
> I dislike WEMI, but I know I am in the minority.  Every once in awhile I try
> to mention what I perceive as its shaky philosophical base.  That's what I was
> trying to do today.  It's WEMI that makes this all so difficult.  Of course a
> book and a movie are different things and they have different main entries.
> But in FRBR/RDA world, as in Wonderland, things are often not what they seem
> to be.

Some years ago I took a course in SQL database design. One assignment had the 
students design a database for a DVD rental store.

Within no time, the related tables (related entities) the students came up with 
bore a striking resemblance to FRBR. Of course an item is different from a 
manifestation -- a customer rents a particular copy, not an entity that has all 
the details about all copies of the DVD.

Of course actors and directors are covered in different tables (different 
entities). And they have different roles (relationships) to the movie, and 
these roles were covered in their own tables and were intermediaries to the 
relationships. Of course a widescreen version and a full screen version need to 
be differentiated-- these are different manifestations of the same work, and 
some customers demand this distinction. We got into a discussion about what to 
do about the different versions of Blade Runner-- a table for expression 
attributes would have to be created to account for this variation, and we 
believed it would serve the customers' interests to do so.

Designing tables in this way is based on the normal form rules-- keep it 
consistent, eliminate redundancy as much as possible. That makes functionality 
and maintenance much easier, which has to be the point of any exercise in 
modeling data to serve customers' interests. You absolutely want to keep 
entities in separate tables-- it makes it so much easier to add attributes and 
to create new relationships between entities in the future. There is nothing so 
special about library resources that would exclude them from benefitting from 
such an exercise.

That means that this novel-film dichotomy is a red herring in the discussion. 
The E-R approach neatly solves problems of designing displays that show what is 
of must use to the end-user. If the display works best by showing movies 
prominently along with novels of the original work, then that can be done with 
an E-R design simply by assigning specific relationships and grouping those 
upon generation of the display.

A relevant section in the FRBR report is here:

http://archive.ifla.org/VII/s13/frbr/frbr_current5.htm

Main quote:
"Relationships may be reflected in bibliographic records in a number of ways. 
Some relationships, especially those depicted in the entity-relationship 
diagrams in Chapter 3 (Figures 3.1 through 3.3), are often reflected simply by 
concatenating attributes of one entity with attributes of the related entity in 
a single record. For example, a record will normally couple the attributes of a 
particular manifestation with the attributes of the expression that is embodied 
in that manifestation and with the attributes of the work that is realized 
through that expression. Relationships are also frequently reflected implicitly 
by appending to the record a heading identifying a related entity."

The E-R model makes relationships explicit. The current approach, often by 
concatenating attributes, is inflexible. It's the current approach that needs 
to be rejected, because that's where the true Wonderland is-- where things are 
often not what they appear, if one can even find them in the mix of attributes.

Thomas Brenndorfer
Guelph Public Library

Reply via email to