Re: CoreData Best Practices

FROM : Scott Ellsworth
DATE : Sat Apr 30 00:43:55 2005

On Apr 29, 2005, at 3:29 PM, John Timmer wrote:

> Most sources of author information (ie - PubMed) don't track 
> individual authors, so he'll essentially have no way of identifying 
> when two authors are identical.


When we wrote such a tool, we parsed the pubmed author list, and 
noted that Andrews J A could match a number of different authors. 
Andrews J A was thus a separate author in our databases than John 
Alan Andrews.  A search for Andrews J A would return both.

> The second thing is that, although relating authors to the paper is 
> easy, retaining the order of the multiple authors for a publication 
> is absolutely essential.  There has to be some way of 
> reconstructing an ordered list out of that relationship set.


The best way I found was to store an index set.

Paper -> author_index_set ->> authors

If author_id is the unique author identifier, author_index_set rows 
consist:

author_index_set_id
author_id
author_order

For CD/WO-like systems where the pk is not user-serviceable data, we 
allow the system to add a seperate pk column.  For other tools, 
author_index_set_id and author_id act as a composite key.

Scott