On 6/30/15 11:37 AM, J. McRee Elrod wrote:
> Stephen said:
>> I agree, we're heading toward a model where systems are asked not to
>> compare two heading strings but to compare two sets of properties and
>> relationships for two resources to determine sameness/difference.
> I hope such programs are more accurate than that used by OCLC to merge
> e-book records according the the Provider Neutral (PN) standard.
> Records for differing manifestations are regularly combined.  I would
> hope not have have records for differing works combined.

As we found when doing merging on the U Calif union catalog, you have a 
trade-off -- to avoid mis-merging you will end up with more un=merged 
items. The algorithms cannot overcome lack of clarity in the metadata 
itself. We ended up with record sets where even a human couldn't 
determine if they should be merged. I think the only solution is going 
to be allowing some kind of crowd-sourcing of merge decisions as a way 
to share the pain.


>     __       __   J. McRee (Mac) Elrod ([log in to unmask])
>    {__  |   /     Special Libraries Cataloguing   HTTP://
>    ___} |__ \__________________________________________________________

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