[WEB4LIB] Re: Overlap analysis of online journals

Thu Nov 25 17:22:45 EST 2004

I'd like to follow up on Tim's comments regarding overlap analysis tools,
particularly Serials Solutions' offering. Yes, these tools must be used with
care: in my opinion it's unfair to penalize one vendor for providing the
same title as another vendor, but if you've got multiple databases from a
single vendor, comparisons among those databases can quickly identify
database dollars that may not be put to their most efficient use. In those
cases, coverage dates for any given title are almost always identical among
any individual vendor's multiple databases. 

Regarding variations in dates among titles, Tim makes some valuable points.
As he points out, most comparison tools, including Serials Solutions'
original Overlap Analysis tool, included title level comparisons only. Using
what we learned from that version, and feedback from many of our customers,
the second version of our overlap tool included full date discrimination.
That means our clients know when they have full overlap for a journal (that
is, coverage dates overlap completely), when they have partial overlap (only
portions of the journal coverage overlap), when they have overlapping title
coverage but no overlap in coverage dates, and when a journal is completely
unique to a database. The tool lets you easily rank your databases by either
unique content percentage or full overlap percentage, and then drill down to
see which specific journals fall into each set.

A few other key points:

1. The feature uses data from our title authority database in order to make
title level comparisons. This means you've always got an "apples to apples"
comparison at the title level, even when competing vendors send different
title variations.

2. Coverage dates are fully normalized and embargo dates accounted for
within the comparison process. That means even dates such as "Spring 1997"
as a start date or "to 6 months ago" as an end date are accurately used in
the comparison. (This normalization is also critical for accurate date
discrimination in OpenURL results sets.)

3. You can compare any combination of databases that you want (up to a
maximum of 300 *databases* per query). Comparing one database to another
outside the context of your entire collection would be less relevant than
comparing them all at once. But this also means you can compare the coverage
of a database to your collection before buying it, not just after.

Feedback from clients tells us that this tool has been extremely helpful in
making collection development decisions. Is overlapping coverage the only
criteria that you should use? Of course not; we would never argue otherwise.

But not knowing this information means a library would not have all relevant
information about their subscriptions before making an acquisition decision
to purchase or cancel a resource. Serials budgets are under constant
pressure, and libraries have already told us they've used this tool to
identify and cancel those databases that were not giving them enough value,
confident in the knowledge that they knew exactly which percentage of unique
content they were giving up, and how much overlapping content remained in
the collection. Having that knowledge is a critical part of an intelligent
decision-making process.

Peter McCracken, MLS
Serials Solutions
444 NE Ravenna Blvd, Suite 211
Seattle, WA  98115
(206) 545-9056 ext.17  //  (866) SERIALS ext. 17
fax: (206) 525-9066  //  peter at serialssolutions.com

-----Original Message-----
From: web4lib at webjunction.org [mailto:web4lib at webjunction.org]
On Behalf Of Tim Bucknall BUCKNALL
Sent: Wednesday, November 24, 2004 9:44 AM
To: Multiple recipients of list
Subject: [WEB4LIB] Re: Overlap analysis of online journals

At the University of North Carolina at Greensboro we have a very nice home
grown OpenURL link resolver and knowledge base that is now used by 25
institutions. One feature of Journal Finder is an overlap analysis tool.

I don't personally think our overlap tool (or anyone else's) generates
terribly useful data. It does tell you what percentage of journals appear in
multiple databases, but that isn't really meaningful in and of itself. 
After all, a title may appear in two databases - but have only a year of
coverage in one and 10 years of coverage in the other. And that difference
in coverage may of may not really matter to you, depending on what the title
is. Or a title may appear in two databases, but have a significant embargo
in one, but not the other. Again, that might or might not matter, depending
on the particular title. 

So, knowing that there is a 45% overlap in title coverage between 2
databases doesn't really give you enough info to make a decision. Which
titles overlap? How do their years of coverage overlap? Are the unique
titles the ones that you really care about, or are those the titles that the
aggregator has "thrown in" to inflate their total number of titles covered?

Point being - Overlap analysis figures by themselves can be very misleading.
And if you have to go in and really look at the individual titles anyway,
there really isn't much value to knowing what percentage of overlap exists
between two databases.

"VanderHart, Robert" <Robert.VanderHart at umassmed.edu> Sent by:
web4lib at webjunction.org
11/24/2004 10:30 AM
Please respond to
Robert.VanderHart at umassmed.edu

To
Multiple recipients of list <web4lib at webjunction.org>
cc

Subject
[WEB4LIB] Overlap analysis of online journals

Serials Solutions ejournal management system offers a feature called
Overlap Analysis, which allows you to compare databases to see where
they overlap in coverage.

I'm interested in hearing opinions from Serials Solutions customers on
this feature.  How useful have you found it?  Does it really aid you in
collection development? 

I'm also interested in hearing from anyone who's implemented a similar
feature in-house (i.e. "homegrown").  What system are you using (PHP,
ColdFusion, Perl)?  How difficult was it to set up?

Thanks in advance for any feedback.  Have a nice holiday!

Robert Vander Hart
Electronic Resources Librarian
Lamar Soutter Library
University of Massachusetts Medical School
Worcester  MA  01655

Voice: 508-856-3290
Email: Robert.VanderHart at umassmed.edu
Web: http://library.umassmed.edu 

*********************************************************************
Due to deletion of content types excluded from this list by policy,
this multipart message was reduced to a single part, and from there
to a plain text message.
*********************************************************************