Tracking Latent Domain Knowledge - Mapping Scientific Frontiers: The Quest for Knowledge Visualization - page 227

connection between the two literatures. A key step in Swanson's methodology is
the identification of the two premises A ! BandB ! C. In a large knowledge
domain, identifying two such premises is like searching for needles in a haystack.
Knowledge visualization aims to capture the structure of a knowledge domain and
increase the chance of finding something useful. Before we turn to issues faced by
domain visualization, let us take a look at Swanson's approach to the discovery of
neglected knowledge.
Swanson's paradigm explores connections in the biological world that can be
represented in the following generic form. If we know two premises that can be
expressed as connections of some form A ! BandB ! C, then the question is
whether A ! C holds. In the biological world, this may not be the case. One must
establish the transitivity explicitly. If A ! C does make sense, it will be worth
considering as a hypothesis to be tested by domain experts. Swanson suggests once
information scientists identify such hypotheses they should pass the question to
domain experts, who will handle it accordingly. In particular, in his Award of Merit
acceptance speech, Swanson gave his advice to information scientists (Swanson
2001 ):
First, information scientists should aim to produce new hypotheses or sugges-
tions - not discoveries. This is the job of lab scientists to test such hypotheses. Real
discoveries should come out of the lab, not the literature.
Second, when information scientists write for publication, subject content should
be limited to reporting factually, or simply quoting, selected passages from scholarly
and reputable literatures of the subject domain. Information scientists' aim is to
highlight possible implicit links in subject literatures. It is a judgment call by
scientists with subject expertise to decide whether the links are plausible and
persuasive enough to merit testing.
After the successful detective work of identifying a link between fish oil and
Raynaud's syndrome, which was later verified by medical researchers, Swanson was
able to continue his quest and find a few more examples falling into the same pattern,
especially by collaborating with Neil Smalheiser, a neurologist since 1994. By 1998,
the number of cases increased to seven. Arrowsmith is their web-based software for
discovering such links (Swanson 1999 ). See more details at its homepage: http://
kiwi.uchicago.edu .
Swanson describes three aspects of the context and nature of knowledge frag-
mentation (Swanson 2001 ):

• There is an enormous and constantly growing gap between the entire body of

recorded knowledge and the limited human capacity to make sense of it.

• Inadequate cross-specialty communication causes knowledge fragmentation. In

response to the information explosion, specialties are increasingly divided into
more and more narrowly focused subspecialties.

• One specialty might not be aware of potentially valuable information in another

specialty. Two specialized literatures may be isolated in terms of explicit citation
links, but they may have implicit, latent connections at the text level.