« What people skip | Main | FRBR and uniform titles »

Comments

Bryan

Awesome. I happened to hear about this project yesterday. Brenda Block and Kathy Kie gave a web talk about WorldCat database quality for some Independent libraries. Will you attempt to control the names that appear on the thousands of article records entering WorldCat.org? Do you have a sense for the percentage of names in article records that might match up with existing name authority records? Do you ever use other details like classification numbers or subject headings to suggest possible matches for the more difficult names that appear in bib records? I am just curious about all this. Thanks.

Bryan

Response:
We are interested in controlling names in articles, but there are no plans in place and we don't even have a good estimate of the overlaps. Right now most of our matching is based on author and title. We do use some of those other clues for matching in VIAF, though.

--Th

Adam Schiff

How will you be dealing with undifferentiated names? Or names that match an authority but are the incorrect heading for the particular entity?

Response:
Right now all we are doing are the 'easy' headings (only 2% change their text); there is very little correction being done. In other processing (e.g. WorldCat Identities) we try to 'fix' headings by transforming them into a standard form. Undifferentiated names, however, just get lumped together for now.

--Th

Stephen Hearn

Very impressive. Are you doing anything different with matches on non-roman bib name forms, since these can now appear as 400s in LCNAF authorities?

Response:
I don't think I've ever seen non-Latin forms in a X00 field in WorldCat, but if they did, and if the authority record 400 matched, and it had multiple subfields, we should match them up (and probably eliminate the non-Latin, since the preferred form of the name in LCNAF is still going to be in Latin script).

--Th

Deborah J. Leslie

1. Will this process be touching the Institutional records?

2. I'm also interested to hear what your plans are for unique names that have no dates are any other qualifying data. We have had some wild flips made through automated authority control in the past.

Response:
1. No, we are not touching the institutional records.
2. We expect to use a variant of how we match names for VIAF, which depends on having a title or other information match beyond a simple string match on the name.

--Th

Diana Brooking

"I don't think I've ever seen non-Latin forms in a X00 field in WorldCat, but if they did, and if the authority record 400 matched, and it had multiple subfields, we should match them up (and probably eliminate the non-Latin, since the preferred form of the name in LCNAF is still going to be in Latin script)."

There are lots of non-Roman headings in X00 fields in WorldCat, at least that's what they look like in the Cnx Client, as linked parallel fields to other X00s. Are you referring here to non-linked X00s in bibs, that is, ones that don't have hidden 880s? Given the multiple sources that OCLC is now pulling records from, I think it is only a matter of time before completely non-Roman records start appearing. Do you really want to flip those headings to Latin forms? Say the records are coming from the Russian National Library!

With article citation records, and records from other national libraries that don't follow AACR2, the environment "our" records live in is changing rapidly. "We" follow Model A for bibs, but will follow Model B for authority records. I can't put it all together in my mind, how will authority control work in this new heterogeneous environment?

Response: I'm only talking about personal name headings in English language metadata (as indicated in the 040 $b). I think we already need authorities for other cataloging languages in WorldCat, but we do try to keep records as close to MARC-21 as we can.

--Th

--Th

The comments to this entry are closed.

My Photo

February 2016

Sun Mon Tue Wed Thu Fri Sat
  1 2 3 4 5 6
7 8 9 10 11 12 13
14 15 16 17 18 19 20
21 22 23 24 25 26 27
28 29