> For our preconference, “Digging into Metadata,” we’d like to get a little
> discussion going to build on once the preconference rolls around.
>
> ...
> - Dealing with free text in MARC records and how to parse them w/o too much
> heartache
>

You can find horrendous stories even with data that's fully structured.
Multiple libraries have had call numbers not migrated (or the wrong one
migrated due to the unfortunate practice of most libraries to retain
multiple call numbers) during an ILS migration -- as you can imagine, that
would make books much harder to find on the shelves. I can't remember the
names of institutions this happened to, but you could probably find someone
who can give you precise details on the autocat list.

There is the constant problem that in any migration, the data is not
structured/used the same way in the new system as in the old -- some fields
exist in one system but not the other, different numbers/types of fields
are used to represent concepts, etc.

I've personally encountered cases where the data that comes out of a system
is outright invalid or gets mangled in bizarre ways by the export routine
itself. For example, there's a system used for many digital archives that
splits a field in two anytime a field that needs to be represented by an
XML entity is encountered. Name withheld to protect the guilty.

kyle