Information Technology Reference
In-Depth Information
Chapter 7
Tracking Latent Domain Knowledge
Knowledge is power.
Francis Bacon (1561-1626)
Conventional citation analysis typically focuses on distinctive members of a
specialty - the cream of the crop. Landscape visualizations naturally emphasize
the peaks rather than the valleys. Obviously such practices remind us either the
Matthew Effect or the winner-takes-it-all phenomenon. However, scientific frontiers
are constantly changing. We cannot simply ignore the “root” of the crop or the
valleys of an intellectual landscape. Today's valleys may become tomorrow's peaks
(Fig. 7.1 ).
In this chapter, we will focus on latent domain knowledge and techniques
that may reveal latent domain knowledge. Knowledge discovery and data mining
commonly rely on finding salient patterns of association from a vast amount of
data. Traditional citation analysis of scientific literature draws insights from strong
citation patterns. Latent domain knowledge, in contrast to the mainstream domain
knowledge, often consists of highly relevant but relatively infrequently cited scien-
tific works. Visualizing latent domain knowledge presents a significant challenge to
knowledge discovery and quantitative studies of science. We will explore a citation-
based knowledge visualization procedure and develop an approach that not only
captures knowledge structures from prominent and highly cited works, but also
traces latent domain knowledge through low-frequency citation chains. This chapter
consists of three cases:
1. Swanson's undiscovered public knowledge;
2. A survey of cross-disciplinary applications of Pathfinder networks; and
3. An investigation of the current status of scientific inquiry of a possible link
between BSE, also known as mad cow disease, and vCJD, a type of brain disease
in human.
Search WWH ::

Custom Search