Information Technology Reference
In-Depth Information
our field of view from mainstream domain knowledge to latent domain knowledge.
A key component in this domain expansion process is the selection of a so-called
“exit” landmark from the citation landscape. This “exit” landmark will play a
pivot role in tracking latent knowledge by “pulling” highly relevant but relatively
rarely cited documents into the scene. The “exit” landmark is selected based on
both structural and topical characteristics. Structurally important documents in
the citation landscape include branching points, from which one can reach more
documents along citation paths preserved by the network. Topically important
documents are the ones that are closely related to the subject in question. Ideally,
a good “exit” landmark should be a classic work in a field of study and it can link
to a cluster of closely related documents by citation. We will explain in more detail
through case studies how we choose “exit” landmarks. Once an “exit” landmark is
chosen from the citation landscape, the four-step procedure can be applied again to
all the documents within a citation chain of up to two citation links. The resultant
citation network represents the latent domain knowledge. Finally, we embed this
local structure back into the global context by providing a reference from the “exit”
landmark in the global context to the latent knowledge structure.
In this chapter, we describe how we applied this approach to three case studies,
namely, Swanson's work, cross-domain applications of Pathfinder network scaling
techniques, and the perceived connection between BSE and vCJD in contemporary
literature. We use the Web of Science, a Web-based interface to citation databases
compiled by the Institute for Scientific Information (ISI). We start with a search in
the Web of Science using some broad search terms in order to generate a global
context for subsequent visualization. For example, in the Pathfinder case, we chose
to use search terms such as
knowledge discovery
,
knowledge acquisition
,
knowledge
modeling
,and
Pathfinder
. Once the global context is visualized, it is straightforward
to identify an “exit” landmark. In the Pathfinder case, a classic citation of Pathfinder
networks is chosen as an “exit” landmark. This “exit” landmark article serves as the
seed in a citation search within the Web of Science. The citing space of the seeding
article
s
contains articles that either cite the seeding article directly or cite an article
that in turn cites the article.
C
One
Step
.s/
D
f
c
j
c
!
s
g
C
Two
Step
.s/
D
˚
c
ˇ
ˇ
9
c
0
)
c
!
c
0
^
c
0
!
s
CitingSpace
Theme
.s/
D
C
One
Step
.s/
[
C
Two
Step
.s/
Such citing spaces may contain articles beyond the boundary of the mainstream
domain knowledge. One can repeatedly apply this method by identifying another
“exit” landmark. Articles connected to the landmark by two-step citation chains are
gathered to represent latent domain knowledge. By using different ways to select
citing articles, we can visualize latent knowledge structures with reference to highly
established and frequently cited knowledge structures. In the following two case
studies, we apply the same spiral methodology to illustrate our approach.
Search WWH ::
Custom Search