On 20 August 2012 16:54, Karen Coyle <[email protected]> wrote:
>
>
> On 8/20/12 1:24 AM, Ben Companjen wrote:
>
>>
>> As long as there is a link to the work record, not having it in the
>> edition record 'only' means the work title is an extra request away.
>> I have no idea if either field is being updated when the title of a
>> work record is updated. And which of these fields would be the one to
>> use? (There are some 750k out of 15m records with these fields, if
>> they don't both appear on one record, so it seems they are not
>> actively used.)
>
>
> Ben, you're thinking in terms of the ultimate structure, but there is a
> value in retaining, for the record, what came in on the edition record. The
> "work title" may differ somewhat from the title in the work record if it
> consists of the entire "uniform title" from the MARC record. It will be
> necessary to look at the data, not just the statistics, to determine if it
> is useful. Maybe you could pull out some examples.

Yes, I have been thinking that removing bad data could help make the
data more useful.
I still think making the records more uniform makes it at least easier
to understand and possibly more interoperable, but I understand that
just removing old data won't work. Documenting it probably works
better.
It's very hard to understand what is in the data without documentation
(issue #100 on GitHub). There is a model for Edition records,
<http://openlibrary.org/type/edition> but there are all kinds of
fields (work_title looks like a typo) that don't fit the model.
Your previous mail has helped someone who wasn't around from the
beginning :) I will try to get some examples to help complete the docs
a little bit more.

Ben

> kc
>
_______________________________________________
Ol-tech mailing list
[email protected]
http://mail.archive.org/cgi-bin/mailman/listinfo/ol-tech
To unsubscribe from this mailing list, send email to 
[email protected]

Reply via email to