E-Journal List to
OpenURL Resolver
Kerry Bouchard
Asst. University Librarian for Automated Systems
Mary Couts Burnett Library
Texas Christian University
Floundering Epoch – (pre-1998)
• Publishers begin sending our Serials librarian “this
publication is now available online at…” notices. Serials
has to be involved in setting up titles.
• Systems cobbles together a URL-rewriting proxy program
to provide off-campus access to databases and e-journals
(replaced by Ezproxy in 2000) -- Systems needs to be
involved in setting up titles.
• Introduction of web-based online catalog makes linking to
online journals from bib records possible -- Cataloging
needs to be involved in setting up titles.
• We begin subscribing to services like JSTOR and Project
Muse. More titles pile up.
• Our serials librarian, Janet Douglass, founds the “E-Team”
to coordinate policies and procedures for tracking e-
journals.
Hunter-Gatherer Era (1998-2001)
The “E-Team” begins tracking
subscription e-journals.
• Relational database of titles built. We record:
– URL for the journal main contents (volumes/issues) pages
– A local PURL.
– Authentication method. With databases there is some variation, but
98% of our e-journals use IP recognition. Off-campus users are
routed through proxy.
– Subject assignments. Initially assigned automatically based on call
number. Quickly saw need to assign multiple headings to titles, so
added a subject table.
• At the end of the E-Team assembly line, cataloger adds the
PURL to the bib record (i.e., our bib record URLs are
PURLs pointing to the local relational database.)
…Hunter-Gatherer Era (1998-2001)
Example PURL:
http://lib.tcu.edu/PURL/ejournal.asp?JSTOR:00030031
• Script for processing all
e-journal links
•Vendor identifier – script • Journal identifier (usually
uses this to look up data on ISSN)
proxy method required for
off-campus access, and to
record a “click count” for the
source in our usage log table
…Hunter-Gatherer Era (1998-2001)
E-Team does not attempt to track:
• Dates of coverage for each source
• Free publications, except a handful specifically requested by
faculty.
• The thousands of full-text titles “hidden” inside full text
databases like Academic Search Premier.
Subject category screen
Sample Title List No dates of coverage, just general notes
Vocabulary Break
You say “aggregator”
I say “fulltext database”
Is OCLC ECO an “aggregation” or a database?
Maybe all we really care about are the differences
between…
User’s Perspective: Browsing…
…versus Searching
Staff Perspective: Do we get all titles from this
source?
Examples:
• Citation databases that contain full text for some / all journals cited
(“Expanded Academic Index”, “Academic Search Premier”, etc.)
• License agreements for everything available from a collection like Project
Muse
Or Only a Selection of Titles?
Examples:
• E-journal collections like OCLC ECO, Highwire Press; JSTOR; IEEE
journals, etc. Doesn’t really matter if all the titles in the collection are from
the same publisher, since we don’t necessarily subscribe to everything from a
given publisher either.
Agriculture Invented: Fulltext
Journal Locator v1.0
Our then computer services librarian James Lutz downloads lists of
journal holdings from various fulltext databases and sets up SQL
“union query” to merges all these lists together with the collections
tracked by the “E-Team”:
Agriculture Invented: Fulltext
Journal Locator v1.0
Downloaded Links
Links to E-Team Records (take user to database
(take user to list of available search screen)
volumes/issues)
…Fulltext Journal Locator v1.0
• Vendor journal lists are inconsistent – different vendors supply
different info, in different formats.
• Some citation databases list every journal they *cite* -- many
links turn out not to be full text.
• Our local “E-Team” list doesn’t have dates of coverage – these
links look odd next to the vendor records that do have dates
• Finally concluded that it was too much work to keep the lists up to
date.
Agribusiness: We pay someone to
harvest e-journal info for us.
Summer of 2001 we begin subscribing to SerialsSolutions journal
tracking system. Every two months they send an updated list of our
e-journal holdings from all our sources. This:
• Eliminates the need to manually download holdings lists from all our
fulltext database vendors – everything is now in one file. Locator data can
now be relied on to be accurate.
• Gives us dates of coverage for all the collections tracked by the “E-Team”
Fall 2002 we add a MARC record feed to our S.S. contract.
• Means that we have MARC records with URLs for about 25,000 titles
(35,000+ links) that were not tracked by the “E-Team”.
…Now our old list is obsolete and
incomplete
At this point we have around 25,000 e-journal titles in our catalog
and our Fulltext Journal Locator, versus the 2,000 titles entered with
subject headings by the “E-Team”.
• E-Team replaces the alphabetical-by-title and subject lists of e-journals
pages on the library web site with pages telling users to search the catalog or
fulltext journal locator, since these lists are much more complete.
…But people really liked the old
lists
Several faculty and graduate students strongly object to having
these lists removed. They don’t feel that searching the online
catalog is an adequate substitute for being able to browse the titles
in their field.
• LC subject headings don’t always work well for browsing -- no consistent
mapping of specific to broad headings.
• If our online catalog software had the capability to retrieve all titles in a
range of call numbers, that would partly address the problem.
• There would still be the problem that we want to assign more than one
broad subject category to a single title in many cases, and that’s not possible
if the subject assignments are derived from call numbers.
…So we put the lists back
• The E-Team reinstates the title and subject lists on the web site,
with modifications to the underlying relational database so that:
• We can differentiate sources that allow browsing by volume and issue from
sources that take users to a search screen. Only sources that allow browsing
by volume and issue (the same ones previously tracked by the E-Team) are
displayed on the title and subject lists.
Data from
Our local list of
SerialsSolutions
“browsable” sources.
“SSName” field links to
S.S. “Provider” field.
…So we put the lists back
• We can continue assigning non-LC subject categories to titles, so that
departments can browse “their” lists. (Subject categories are linked to ISSN
numbers, so once we assign a subject heading to one source of the title, additional
sources of the same journal automatically get mapped to the same headings.)
Subject codes Links codes to ISSNs link to SerialsSolutions
(“Geology”, etc.) ISSNs Records
…Title list, with dates of coverage
…Fulltext Journal Locator v2.0
…Fulltext Journal Locator v2.0
Citation databases begin offering linking
capabilities.
Some are explicitly OpenURL based, and others are not. Either
way it is possible to make journal-level linking work by simply
modifying the Fulltext Journal Locator script so that it can
search by ISSN as well as by words in the title. (Script will
search by two ISSN’s – print and electronic – if the citation
links provide them.) Not all the journals in the S.S. data have
ISSN’s at however.
Citation with “OpenURL like” link
URL for the link above: ISSN is all we need for journal-level link
http://lib.tcu.edu/PURL/OpenURL.asp?genre=journal&ISSN=0009-
2541&DT=20030615&TI=Carbon%20isotope%20exchange%20rate%20of%20DIC%20i
n%20karst%20groundwater%2E&JN=Chemical%20Geology&VI=197&IP=1-
4&AU=Gonfiantini%2C%20Roberto&spage=319
(example from Academic Search Premier)
Front end OpenURL resolver
At this point we provide separate links for the catalog and locator,
because MARC records aren’t yet available for all the titles in the
locator. When that changes, a separate full text journal locator will
no longer be needed, for journal level linking.
Link to search
Fulltext Journal
Locator
Link to search
online catalog
http://libnt4.lib.tcu.edu/PURL/JournalLocator.asp {issn passed in hidden form field}
OpenURL Resolver Result Screen
URL in link above:
http://lib.tcu.edu/PURL/ejournal.asp?ScienceDirect:http://www.sciencedirect.com/w
eb-editions/journal/00092541
Used to retrieve local info about TCU’s
ScienceDirect subscription
Article Level Linking
The OpenURL provided by Academic Search Premier in the
previous example contained all the information needed for building
an article-level link:
http://lib.tcu.edu/PURL/OpenURL.asp?genre=journal&ISSN=0009-
2541&DT=20030615&TI=Carbon%20isotope%20exchange%20rate%20of%20DIC%2
0in%20karst%20groundwater%2E&JN=Chemical%20Geology&VI=197&IP=1-
4&AU=Gonfiantini%2C%20Roberto&spage=319
ISSN – tells us it’s a journal and which one
DT – date the article was published, in YYYYMMDD format
TI – title of the article (may or may not be needed, depending on how the target
system builds article-level links)
VI – volume
IP – issue
AU – author (required by some target systems)
…Article Level Linking
Two enhancements to the Fulltext Journal Locator / OpenURL
Resolver script are required to make article-level linking work:
• A date normalization routine, so that we can compare the date an
article was published to dates of coverage for each potential
source of the article
• A function to convert meta-data in OpenURL format to a format
that can be used to create an article-level link for a given source.
Each vendor may require it’s own function (example to follow).
Date normalization
Dates supplied in the OpenURL links are already normalized (in YYYYMMDD
format). However, dates we get in our S.S. data feed appear in a variety of ways:
Winter 2002
12/17/1998
1 month ago {as in, “2001 to 1 month ago”}
[blank] {a blank End Date field means “to present”}
1999 {no month supplied}
Fortunately, there are only a small number of variations like this, so it was not too
difficult to write a date normalization routine that converts the SerialsSolutions
supplied starting and ending dates for a publication to YYYYMMDD format;
these can then be compared to the date of publication supplied in an OpenURL to
see if we have full text for the issue in which the article was published.
Example metadata parser - JSTOR
The JSTOR web site contains documentation for building article-level
links to JSTOR content using SICI codes. To construct a SICI, we
take the OpenURL metadata elements we ignored earlier and convert
them into a SICI. So the OpenURL data:
genre=journal&ISSN=0002-9475&DT=19950901&TI=Ancient%20anagrams%2E&
JN=American%20Journal%20of%20Philology&VI=116&IP=3&AU=Cameron%2C
%20Alan&spage=477
Becomes:
sici=0002%2D9475%2819950901%29116%3A3%3C477%3A%3E2%2E0%2ECO%3B2%2
D%23&origin=tcu
…Example metadata parser - JSTOR
And the user sees:
Article-level link using SICI
Journal-level link using old method
Project Muse also supports article-level linking, but in this case
the date parser saw that Muse would not have full text for an
article published in 1995.
Article-level linking issues
• Journal source must not only support some kind of article-level
linking, they need to use links that can be derived from metadata. For
example, an accession numbering system (e.g.,
“http://www.somevendor.com/journals/articles/23090238762.html”) won’t work.
• Would be difficult use MARC records for building article level
links:
• Much more difficult to extract and normalize dates as they’re presented in the
MARC 856 fields (MARC fixed date fields can’t be used, they’re the date
range for the publication itself, not the dates of coverage for a particular
source):
• Scripting tools (PHP, ASP, Perl, etc) make it much easier to pull data from a
relational database table than from a MARC record (at least with our current
ILS software).
Modern era? (Agribusiness crushes the small
farmer)
At least two vendors we
work with have announced
article-level linking solutions
that may eliminate the need
for local work on this.
So the Systems Librarian has to
shift back from writing cool
metadata parsers to attending
committee meetings.
…But maybe there’s still some work before we
send the farmer off to the knackers…
• Link from the OpenURL resolver to our Interlibrary Loan system
when we don’t have access to online or print.
• Still a need to track some titles and sources locally and integrate
these into the hitlists for the OpenURL resolver, since there are still
some sources that aren’t tracked by our e-journal info vendor.
• Need to keep up with “fulltext database” vendors as they add
journal-level/article-level linking capability, so that we can change
our URLs to take advantage of that instead of sending users to a
search screen.
In Conclusion…
If you have a list of your e-journals in a relational database
Either with data you entered yourself
Or data you get from a vendor tracking your holdings for you
And if the info on each e-journal includes an ISSN
And if someone on your staff knows how to write web server scripts
that retrieve info from the e-journal database…
Then you can turn your e-journal list into an OpenURL resolver with
about 20 minutes work. (Journal level links – article-level linking
takes longer.)
Kerry Bouchard
Asst. University Librarian for Automated Systems
Mary Couts Burnett Library
Texas Christian University
Email: k.bouchard@tcu.edu
This presentation online:
http://lib.tcu.edu/staff/bouchard/OpenURL/Amigos2003.htm