OpenURL Concepts and Implementation

1
OpenURL Concepts and Implementation
Presentation for Texas Library Association Net
Fair, March 2004

• Kerry BouchardAssistant University Librarian for
  Automated Systems
• Mary Couts Burnett Library, TCU
• k.bouchard_at_tcu.edu

presentation available online at
http//lib.tcu.edu/staff/bouchard/OpenURL/TLA2004.
htm
2
Problem linking related information from
different sources
Citation in A I database that does not have
full text for cited article
Full text in database / ejournal aggregator
collection from another vendor
Print full text from library collection
Author info from citation index, biographical
sources
??? ... let your imagination run wild
3
Low tech solution bibliographic instruction

• Pros
• Encourages information literacy shouldnt
  college students learn some research skills?
• Accurate if user is persistent automated
  linking systems may fail to find resources that a
  manual search would uncover

• Cons
• Google doesnt make me print a citation and go
  to a different web site for fulltext why the
  library?
• Checking for all possible sources of an
  article might require user to go to several
  different databases with different interfaces.

4
Non-standards based tech solution links galore

• A I database vendors provide proprietary
  linking mechanism from their citations to other
  sources. Typical approaches
• Library staff upload list of all online/print
  holdings, and must keep list up to date. Format
  of the list may vary from vendor to vendor. Worst
  case may not be possible to upload list at all
  instead must manually click off every title
  from a list provided on vendor site
• Vendor has set up automated links to a selected
  list of partners, e.g., citations in Mega
  Abstracts link to your backfile holdings in
  JSTOR holdings, but not your current holdings in
  Project MUSE.

5
...Non-standards based tech solution links galore
For a library with several databases / sources of
online full text, this approach leads to a
combinatorial explosion...
Mega Abstracts
Mega Fulltext
EBSCO
JSTOR
Gale
Project MUSE
CSA
Emerald Press
Silver Platter
Royal Society
Ovid
EBSCO
etc...
Online Catalog
6
...Non-standards based tech solution links
galore
Other problems with this approach...

• Authentication typically links supplied by a
  vendor will only work if your users are
  on-campus, since they bypass your local system.
• Local holdings If vendor automatically generates
  links, linking to full text from a given source
  is all or nothing even though you may only have
  access to a portion of the titles.
• Other local functions e.g., scripting youve
  done to collect usage statistics is bypassed.

7
April 1999 Herbert Van de Sompel and Patrick
Hochstenbach publish first part of Reference
Linking in a Hybrid Library Environment in D-LIB
Magazine
http//www.dlib.org/dlib/april99/van_de_sompel/04v
an_de_sompel-pt1.html Proposes the SFX model
dynamic linking based on passing metadata about a
resource to a resolver program.
8
OpenURL Dynamic Linking Model
OpenURL metadata
Links to target may or may not be OpenURL
Mega Abstracts
Mega Fulltext

• OpenURL Resolver
• Database of resources just the titles your
  library has access to
• Authentication mechanism your users
• Parsers to construct links into systems serving
  as OpenURL targets

EBSCO
JSTOR
Gale
Project MUSE
CSA
Emerald Press
Silver Platter
Royal Society
Ovid
EBSCO
etc...
Online Catalog
9
Nomenclature
Original OpenURL model was called SFX. SFX
was later licensed by Ex Libris, and is now the
name of their implementation of OpenURL.
OpenURL is a draft NISO standard.
SFX is to OpenURL As Kleenex is to facial
tissue
10
OpenURL Standard Resources
NISO Committee AXhttp//www.niso.org/committee
s/committee_ax.html The draft standard has been
completed and has been released for ballot and
review January 26, 2004-March 10, 2004.
NISO Committee AX (committee web site at Cal
Tech)http//library.caltech.edu/openurl/
11
OpenURL Standard Resources
OpenURL listserv (general discussion)openurl_at_calt
ech.edu
OpenURL software development listLIB-OPENURL-DEV-
L_at_listserv.uiuc.edu
12
OpenURL Components
Resolver Queries local database of resources
using metadata in OpenURL link. If matches found,
constructs links to services available for that
resource (e.g., full text).
Source (e.g., article citation) Takes metadata
about the cited resource and puts it in format
defined by OpenURL standard
Target(online journal, vendor 1)
Target(online journal, vendor 2)
Target(online catalog)
Target(citation index)
13
Digression OpenURL, DOI CrossRef
Thanks to Amy Brand (Directory of Business
Development for CrossRef), Miriam Blake (Los
Alamos library), and Jenny Walker (Ex Libris),
who presented a session on OpenURL, DOI, and
CrossRef at the October 2003 LITA Forum info on
next two slides is a highly condensed summary of
that session.
DOI Document Object Identifier DOIs are
somewhat analogous to ISSNs in that they provide
metadata about online objects (e.g., a journal
article) that is not subject to variations in
metadata (e.g., A. B. Smith versus Arnold
Smith). In the case of journal articles, it is
an article-level identifier. DOIs are somewhat
analogous to PURLs in that the metadata is sent
(via hypertext link) to a registration
authority that resolves it into an actual URL
for the object.
14
...Digression OpenURL, DOI CrossRef

• CrossRef - an independent, non-profit membership
  association - is currently one of seven official
  DOI registration agencies worldwide.
• DOIs dont provide information about
• Sources of the document other than the
  publisher. (At TCU probably less than 5 of our
  online journals are hosted directly by the
  publisher)
• Licensing and rights management - does your
  library have access to this article?
• ... so DOI is not a substitute for OpenURL.

15
...Digression OpenURL, DOI CrossRef
Diagram below illustrates how DOI might
complement OpenURL by serving as an alternate
(hopefully more complete and accurate) source of
metadata for linking.
Target(online journal, vendor 1)

• Reg. Auth.
• Sends full set of metadata about article

• Source
• Sends DOI to Resolver

Target(online journal, vendor 2)

• Resolver
• Sends DOI to Registration Authority

• Resolver
• Constructs links using better metadata

Target(online catalog)
Target(citation index)
16
OpenURL Source
Mega Fulltext
Mega Abstracts
JSTOR
EBSCO

• OpenURL Resolver
• Database of resources just the titles your
  library has access to
• Authentication mechanism your users
• Parsers to construct links into systems serving
  as OpenURL targets

Project MUSE
Gale
Emerald Press
CSA
Royal Society
Silver Platter
EBSCO
Ovid
Online Catalog
etc...
17
...OpenURL Source
URL for the link above http//lib.tcu.edu/PURL/O
penURL.asp?genrejournalISSN0009-2541DT2003061
5TICarbon20isotope20exchange20rate20of20DIC
20in20karst20groundwater2EJNChemical20Geolo
gyVI197IP1-4AUGonfiantini2C20Robertospage
319sidEBSCOaph (example from Academic Search
Premier)

18
OpenURL Source - Setup

• Setup can be as simple as sending the vendor the
  URL of your OpenURL resolver, e.g.
  http//lib.tcu.edu/PURL/OpenURL.asp
• May also include URL for a graphic to display
  next to the links, and/or the text you want to
  use (e.g. Check for Full Text Sources)

19
OpenURL Source - Setup
Local info supplied to vendor in setting up
OpenURL linking
20
OpenURL Source - Setup

• Setup may also ask you to specify which fields
  and labels to include in the parameter string
  (EBSCO is an example), e.g. DTdate1AUauth
  orISBNISBN

21
OpenURL Source - Issues

• Does source send all the necessary fields?
• Necessary may vary by target e.g., Project
  Muse requires journal name and author name for
  constructing article-level links, others use ISSN
  and date/volume/issue information.
• Does source send e-issn, print issn or both?
• Does source use labels and data formats
  specified by the standard? (e.g., are dates in
  form 20041204 or 12/25/2004?)
• Do the links display in a way that makes sense
  to the user? (If users need a bibliographic
  instruction session just to learn how to
  recognize the OpenURL links, the interface could
  use some work.)
• If you are shopping for an OpenURL solution,
  keep in mind that vendor will probably not set up
  your source links for you, nor can they guarantee
  that the resolver will work if the source
  metadata is incomplete or malformed.

22
OpenURL Source Decision Points
If you have the luxury of looking at evaluating
comparable, competing database products, make
OpenURL support part of your evaluation.

• Whether they provide OpenURL links at all
• Quality of the links based on earlier criteria
  mentioned

23
OpenURL Target
Mega Fulltext
Mega Abstracts
JSTOR
EBSCO

• OpenURL Resolver
• Database of resources just the titles your
  library has access to
• Authentication mechanism your users
• Parsers to construct links into systems serving
  as OpenURL targets

Project MUSE
Gale
Emerald Press
CSA
Royal Society
Silver Platter
EBSCO
Ovid
Online Catalog
etc...
24
OpenURL Target Characteristics to Consider

• Fulltext database (e.g., Academic Search
  Premier) vs Ejournal aggregator (e.g., Project
  Muse)
• With databases that include fulltext content,
  you typically have access to all titles for
  which they have full text
• Because the list of journals in a fulltext
  database is typically large, and changes
  frequently, locally keeping track of the list may
  not be practical
• With aggregators, content is typically based on
  subscriptions to individual titles just as you
  probably dont subscribe to all the Elsevier
  journals in print, you probably dont have access
  to all titles in ScienceDirect Web Editions
• Dates of coverage for aggregator content may
  vary from library to library because of
  subscription differences

25
...OpenURL Target Characteristics to Consider

• How deep can a link go?
• Database search screen
• Journal volumes/issue pages (all dates)
• Specific issue of journal
• Article-level linking

26
...OpenURL Target Characteristics to Consider

• Does Target support links using OpenURL syntax,
  or does the resolver need to parse the data into
  another format?

JSTOR article-level linking example JSTOR uses
SICI codes for article-level links OpenURL
metadatagenrejournalISSN0002-9475DT19950901
TIAncient20anagrams2EJNAmerican20Journal2
0of20PhilologyVI116IP3AUCameron2C20Alan
spage477 becomes sici00022D94752819950901291
163A33C4773A3E22E02ECO3B22D23origintcu
27
OpenURL Resolver
Mega Fulltext
Mega Abstracts
JSTOR

• OpenURL Resolver
• Database of resources just the titles your
  library has access to
• Authentication mechanism your users
• Parsers to construct links into systems serving
  as OpenURL targets

EBSCO
Project MUSE
Gale
Emerald Press
CSA
Royal Society
Silver Platter
EBSCO
Ovid
Online Catalog
etc...
28
OpenURL Resolver Components
Resolver

• Database (aka knowledgebase) of your librarys
  resources (targets)
• User interface for maintaining database
• Software to accept OpenURLs as input, query
  database, construct appropriate links, and
  display available resources to user

Note Many vendors are selling both the database
and software components may or may not offer
the option to unbundle the two.
29
OpenURL Resolver Database DB Interface

• Database of resources (targets) your library
  has access to
• Should include titles in full text databases,
  not just subscription aggregators (currently TCU
  has app. 8,900 ejournals through aggregators,
  versus app. 27,500 journals in full text
  databases.)
• Should include accurate dates of coverage for
  each source of each title
• Should include information for authenticating
  for off-campus use e.g., prefixing the URLs
  with the address of an EZproxy server or local
  script
• Should include information about level of
  linking supported e.g., are article-level links
  possible?

30
...OpenURL Resolver Database DB Interface

• Database record format MARC, or tabular
  (RDBMS)?
• TCU subscribes to a MARC feed that is loaded
  into our online catalog, and a tabular file that
  is loaded into a relational database same
  information in both.
• The MARC records allow users to find our print
  and online journals with a single search of our
  catalog
• However, MARC data does not lend itself to use
  in an OpenURL resolver
• Much more difficult to write software that uses
  MARC
• Dates of coverage may be embedded in free text
  notes difficult or impossible to parse
• Difficult to create links between MARC records
  and non-MARC data

31
... OpenURL Resolver Database DB Interface
Questions to Ask Yourself

• Are you starting from scratch, or have you
  already built a database of some/all your
  e-journal holdings? If youve already done work
  in-house, find out if/how you can supply that
  data to the vendor to populate the OpenURL
  resolver database without having to re-key
  everything.
• Do you want data for other purposes than the
  OpenURL resolver e.g., MARC records for your
  online catalog?
• What processes do you have in place to keep
  track of your subscription holdings for
  example, noticing when titles you get in print
  become available online for no/reasonable
  additional cost? Think through how you will
  integrate these processes with updating the
  OpenURL resolver database.

32
...OpenURL Resolver Database DB Interface
Questions to Ask OpenURL/Database Vendors

• If the vendor is selling a bundled solution a
  database and resolver software together try to
  get as much detail as you can on the database
  where it comes from, how its updated, how you
  maintain information about your subscriptions.
• OpenURL resolvers are (so far at least) fairly
  simple applications to write. Its easy for a
  vendor to demo their software...
• Making sure the data being queried is accurate
  (e.g. reflects your holdings) is the hard part.
• Can the vendor take a list of all your fulltext
  databases and subscription journals and confirm
  which ones they can supply data for?

33
...OpenURL Resolver Database DB Interface
...Questions to Ask OpenURL/Database Vendors

• How do you input your subscription holdings
  initially, and then to keep it up to date?
  Mechanics of this could have a major impact on
  how quickly you get up and running, and how much
  labor is required to keep the information
  accurate.
• How do you input fulltext database holdings?
  (Hopefully not by clicking on each individual
  title.)
• Do they supply both e-issn and print issn
  for titles? All or some? Ability to query the
  database by either issn and find a match will
  increase the rate of successful OpenURL
  resolutions, especially if the resolver
  constructs a search of your online catalog for
  print holdings when you dont have a particular
  journal online.

34
OpenURL Resolver Software

Should be able to take the data sent by the
source, query the database, and present the user
with a list of relevant links to targets. If
source was a journal article citation, the
software

• Should take dates into account (why show a user
  eight links, if only three sources are for the
  volume/issue they need)?
• Should provide article-level links when possible
  (whether its possible is something you may want
  to verify independently)
• Should be able to check your catalog for print
  holdings when online holdings are unavailable
  ideally without the user having to click on a
  second link to search the catalog.

35
...OpenURL Resolver Software

• Should be able to take advantage of all the data
  in the database for example, if database has
  both print issns and e-issns, search both.
• If no print or online holdings found, link to
  ILL request system.
• Provide you with statistics particularly link
  failure statistics, so you can track down
  problems with Sources and Targets (which may be
  unrelated to the OpenURL resolver itself).

36
OpenURL Resolver Example Screens

Source Link
37
...OpenURL Resolver Example Screens

We currently offer a catalog link up front, in
case user actually prefers to use a print copy.
38
...OpenURL Resolver Example Screens

Other sources in case date check flaked out
Article-Level Link
Journal-Level Link
Journal-level link
Links to database search screen
39
...OpenURL Resolver Example Screens

Links from Academic Search Premier
Whats this? all links are failing!
40
OpenURL Resolver Software

Questions to Ask Yourself

• Develop in-house, buy bundled hardware/software
  solution, buy software only and rely on other
  vendor for data feed?
• Do you already have a database-driven online
  e-journal list, and is the programmer still
  available? If so, adding basic journal-level
  OpenURL functionality to it may not be much
  work.(see http//lib.tcu.edu/staff/bouchard/OpenU
  RL/OpenURL.ppt for more info)
• On the other hand, starting from scratch
  creating a database-driven list just to achieve
  OpenURL functionality may not be very practical.
• Host the server locally, or remotely at vendor
  site? May depend on the size of your staff, how
  much control (customizability) you want have over
  the software.

41
OpenURL Resolver Software
...Questions to Ask Yourself

• Usability will your users understand the
  interface?
• Are you interested in other services besides
  linking from citations to full text? e.g.,
  citation index searching, links to book reviews

42
OpenURL Resolver Software
Questions for vendor

• If hosted locally, does it require a dedicated
  server? What size? Are other licenses required
  (e.g., Oracle or SQL Server)?
• How customizable is it for example, can you
  create new services (e.g., linking to a book
  review database) that arent already canned?
• What kinds of statistics and reports are
  available?
• Can it link to your online catalog? Does user
  have to blind click to see if there are sources
  in your catalog, or can the software search the
  catalog and tell them without the extra click?
• Can it send a link on to your ILL system if no
  other sources available?

43
OpenURL - Conclusion

• No solution will be completely turnkey just as
  online catalog software is only useful if you
  have a cataloging department to maintain the
  data, OpenURL resolver software is useless
  without accurate data behind it.
• But some are more turnkey than others
  especially if you dont want/have time to
  customize the interface.
• Its not that hard to do if you approach it with
  realistic expectations, and even if you rarely
  hear from them, your users will appreciate it.