Hi, We had this problem too. It's my understanding that you can opt for
brief records with no attempt to match if there is no ISSN, or they will
try to match based on title, and as you pointed out, we got a lot of
false hits.
We changed our profile to brief records if no ISSN, and at least don't
get the wrong records now.
Ann

-----Original Message-----
From: SERIALST: Serials in Libraries Discussion Forum
[mailto:SERIALST@LIST.UVM.EDU] On Behalf Of Kay Teel
Sent: Thursday, February 19, 2009 2:10 PM
To: SERIALST@LIST.UVM.EDU
Subject: [SERIALST] SFX MARCit records for Newsbank Early American
Newspapers?

For a little over a year, we have been loading SFX's MARCit
bibliographic records for our online resources in large
packages/aggregations. For the moment, we are only loading MARCit
records that contain an ISSN (CONSER records and MARCit brief
records).

We have noticed a distressing number of problems with records supplied
by MARCit with links to newspapers, and the Newsbank Early American
Newspapers packages in particular. What frequently happens is that
MARCit supplies a CONSER record (with ISSN) for a completely unrelated
serial that happens to have the same title proper -- a common
situation with newspapers. Unsurprisingly, the early newspapers don't
actually have ISSNs, so it appears that someone simply grabbed a
CONSER record with an ISSN so they could give us a record to load. The
SFX link goes to the right place -- the Newsbank Early American
Newspaper page. Only the bibliographic record is wrong. For some
titles, they have supplied us with multiple records for multiple
unrelated serials with the same title (i.e., 2 wrong records linking
to the same resource) and/or duplicate records for different early
newspapers with the same title (i.e., 2 copies of 1 wrong record
linking to 2 different resources).

I know that bad data from MARCit is definitely not limited to the
Newsbank titles, but we're concerned about the early newspapers in
particular because of our faculty's needs.

Has anyone else encountered these problems with the MARCit service?
Have you had any success in getting cleaner data from them? Is the
problem with Newsbank -- did they pinch a bunch of ISSNs from
unrelated titles to give to their unISSN-ed newspapers? When we begin
loading the MARCit records without ISSNs, is the problem solved
because MARCit supplies correct CONSER records for the early
newspapers? :) I'd love to believe this is so! (I don't have a lot of
hope.) Or: does the problem become worse once the records without
ISSNs are loaded, because the "match" is only made on title proper,
and the universe of incorrect bibliographic records they can supply
expands dramatically? That's my big worry, and I'd love some advice or
cautionary tales if you have any.
--
Kay Teel
Serials and Arts Resources
Metadata Development Unit
Stanford University Libraries
kteel@stanford.edu