Quoting Kelley McGrath <[log in to unmask]>:

> Three optional machine-comprehensible provenance elements attached  
> to every data element:
> 1) source of the data
> 2) the institution entering the data
> 3) date the data was input
> Perhaps something like the following for title proper:
> TitleProper: Citizen Kane
> DataSource: title frame
> DataInst: OrU
> DataDate: 2012-01-04

One possible answer is that we eliminate generalities like "title  
proper" that don't actually tell us much about title itself. "Title  
proper" is like "main entry" -- it tells you the role the title is  
playing in the record, not what the title actually is. We could have:

- title page title
- cover title
- title from jewel case insert

etc., if that information is important to capture, rather than calling  
one of them the "title proper" and essentially losing key information  
about the source of the title. In a situation where you want to  
display just one title you would have an algorithm in your display  
code that selects the one to display from a list: first title page,  
then cover, then supplied....

If you look at AACR/2 it's easy to see that the cataloger is selecting  
data elements to fit into certain display roles, and in that process  
losing the meaning of the data elements themselves. This is one of the  
difficulties in then transforming the current data into something that  
is based on the meaning of the data rather than its position in the  
record. Essentially the rule would be:

"say what it is, not where you want it to display"

And of course, notes about data elements are virtually useless in a  
data environment. The data elements themselves need to carry that  

As for the dates or versions on a data element, and the entity that  
created the data, that is being considered by the W3C provenance  
working group.

It might be interesting to offer some use cases from library  
cataloging as test cases for their work.


Karen Coyle
[log in to unmask]
ph: 1-510-540-7596
m: 1-510-435-8234
skype: kcoylenet