Internet Insights
Linking on Steroids
By PÉTER JACSÓ
I wrote this column just before going to the Society
for Scholarly Publishing's (SSP) 26th Annual Meeting.
There, I spoke at a half-day seminar titled "Linking
on Steroids" with my esteemed peers from the scholarly
publishing world. We discussed the state of the art
in Web linking from the perspective of authors, publishers,
database aggregators, libraries, and end users. Clicking
on http://hypatia.slis.hawaii.edu/~jacso/ssp/linking/linking.ppt will take you to my detailed PowerPoint presentation,
which demonstrates the variety of links in different
databases on different host systems. In this column,
I'll provide some insights into the best and worst
linking practices that I encountered in preparing for
my seminar on link typology and topology.
The Power of Citation Linking
The backbone (and bane) of scholarly publishing is
the set of references cited by authors and listed at
the end of their works. These have become the primary
links in publishers' digital archives, databases of
A&I services, and aggregators' services. The greatest
advancements in linking have been the links to cited
and citing references, the technical counterparts of
the intellectual acts of referring to other works,
with possibly instant delivery of a full bibliographic
record, the abstract of the cited record, or its full
text. The last option depends on the subscription status
of the users (typically their libraries) to the cited/citing
source.
The links may be categorized according to their domain.
Some are restricted to cited/citing works within the
database. These intra-database links have been the
most widely used by aggregators. Inter-database links
within a host are more gratifying as they increase
the chances of getting access to the abstract and/or
full text of the cited/citing documents. The most powerful
are the inter-host links between, say, OCLC and EBSCO
or Elsevier and Wiley.
Not only is the access to cited/citing documents
subject to subscription status, sometimes it's also
the tracing options. For example, only subscribers
to the American Physical Society's (APS) archive can
display the cited references. (The bibliographic citations
and abstracts are free to everyone.) The Institute
of Physics' (IoP) archive displays the cited references
of articles for any user via its intelligent HyperCite
technology. The IoP archive also includes links to
its free abstracts and its sometimes free HTML/PDF
articles. In addition, IoP also provides links to those
archives/databases that offer free abstracts (APS,
ChemPort, PubMed) or free full text (PubMed Central
and most preprint archives). IoP also includes links
to subscription-based services (INSPEC, ScienceDirect,
etc.) as well as to citing articles in a variety of
databases similar to the cited articles. Many of the
links are based on the Digital Object Identifier (DOI)
and the services of CrossRef.
Among the inter-host links, one of the most powerful
combinations is when a publisher links to citing and
cited references as well as to related articles (which
share one or more cited references with the article
being consulted) in the gargantuan ISI Web of Science
databases. Typically, these links are only offered
if the user subscribes to the databases of both the
linking and linked partners. However, Annual Reviews,
Inc. offers these links (up to 10 citing and related
articles in Web of Science) to anyone, even without
registering with the Annual Reviews site. Believe me,
the Annual Reviews are top-ranked publications in many
disciplines.
The Bane of Citation Linking
Sadly, a significant proportion of the cited references
have typos in their titles and/or page numbers as well
as in the names of authors, journals, volumes, and
issues. Abstracting-and-indexing services add their
own typos and inconsistencies (and sometimes correct
the erroneous ones in the source, as H.W. Wilson does
so well). Humans can cope with most of the errors and
find the cited works (often with a little help from
librarians), but link-resolver programs (as with most
other software) are hypersensitive to accurate syntax.
They often fail to find the cited items even if they're
right under their noses in the same archive/database.
Although DOIs could be the best tools for fighting
the consequences of citation errors, only a few A&I
services add the DOI to their bibliographic records.
Even if they do add them, the host services may not
retain them. CSA, OCLC, and EBSCO deserve kudos for
retaining and displaying the DOIs. Not accidentally,
they have the best built-in link-related features—and
not just for common citation linking. Citation sloppiness
backfires in linking and occasionally defeats even
the smartest systems.
For example, in the American Psychological Association's
PsycINFO database, the beginning of the title of one
of my test articles, "The explosion of knowledge, references,
and citations," appears erroneously as "The explosion
knowledge...." The latter may be useful in the fight
against terrorism but may make the authors explode
for several reasons. One reason is the lack of links
to the full text of their article in those systems
that automatically generate links from the title combined
with one or more other data elements. The second reason
is that exact title searches in PsycINFO would not
find the record. The third is that aggregators (such
as EBSCO and CSA) that go out of their way to enrich
the records in databases by showing how many times
an article was cited (within the same database or host)
will list this article—understandably—as
uncited.
If you think that these are rare mistakes at APA
and that the organization will rush to correct them,
think again. As I reported 6 months ago in my January
2004 "Cheers and Jeers" column, at the end of 2003
PsycINFO had 11,852 records with syntactically incorrect
publisher URLs (including its own) with a backward-leaning
pair of slashes. At the end of May, there were 22,335
such records.
There are other types of power links based on citation
analysis, as in the University of Southampton's awesome
though experimental Citebase and ParaCite, as well
as other indicators of scholarly clout that I'll discuss
in a future column. As promised earlier this year,
my next column will be about metasearching, which happens
to be the subject of another session that I moderated
at the SSP conference.
Péter Jacsó is a professor of library
and information science at the University of Hawaii's
Department of Information and Computer Sciences. His
e-mail address is jacso@hawaii.edu.
|
