On 8/26/15 10:20 AM, Robert Sanderson wrote:
> If the authority *describes* the person, why is that information 
> associated with the authority and not just with the person?  On the 
> other hand, if the authority is an existing identifier for the person, 
> rather than a description of them, then identifiedBy makes sense to 
> me.  The bf:Authority would play the same role as the bf:Identifier, 
> in that it's a resource which carries a string that has been used to 
> identify the subject of the triple.

Here's how I look at name authorities: name authorities (as their name 
implies) exist to establish the preferred label for a person, corporate 
body, or family. Before automation, the name authority entry was just 
another card, and was used during cataloging to determine the form of 
name (label) to include in the bibliographic description. Only the label 
existed to distinguish that entity from another. After automation, 
unfortunately, the label continued to be used to represent the entity; 
although authority records have identifiers, these are essentially 
invisible, unused except for record update. The down side of using the 
label to represent the entity is that labels can change, such that 
"Smith, John, 1932- " later becomes "Smith, John, 1932-2010", or "Boyle, 
T. Coraghessan" becomes "Boyle, T. C.". As an identifier, a label sucks. 
But conceptually, behind that label was indeed the identity for an 
entity of great importance for bibliographic control.

It makes sense to use the authority record ID as an identifier for the 
person(a) because it already exists, and it connects you to the 
authority record. However, authority records have very little 
information about the RWO because that never was their focus. For 
example, the birth and death dates function as disambiguation strings in 
the label display, and entities for which disambiguation was not needed 
do not have those dates (although lately that has not strictly been the 
case, which makes it all even murkier). Birth and death dates haven't 
been included as information about the person but as information added 
to the preferred label *where needed*. Other information, such as where 
the person was born, or when the company was founded, family 
relationships - these are logical bits of information that were never 
part of authority data because AC existed solely to create an 
unambiguous label string.

What I'm getting at is that the reason that we struggle with traditional 
authorities in RDF is that they do not have the qualities that an RDF 
graph about a person might have. We might use the existing name 
authorities record identifiers as URIs, but the thing that is today an 
authority record will not be adequate as a graph describing a person or 
corporate body. The other option is to leave name authorities alone and 
create an entire new set of identifiers (parallel, perhaps) that are 
representative of the entity, not just of the label.

I personally would like to see the concept of authorities progress from 
the determination of labels to actual information resources, of which 
labels are a part but not the only focus. Key, however, will be a 
cultural change that accepts the shift from label creation to entity 
description. This is the part that I think we are not yet agreed on, and 
IMO it is holding us back in our work to create a richer library data 


Karen Coyle
[log in to unmask]
m: +1-510-435-8234
skype: kcoylenet/+1-510-984-3600