Information Technology Reference
In-Depth Information
The largest area, in red, is information science, including information retrieval
and citation indexing. The second largest one, in green, includes scientometrics,
neurology, and artificial intelligence. The third largest area, in blue, contains articles
on headache and magnesium. Swanson's articles are highlighted with stronger
brightness in the scene. A dozen of his articles are located in the second area.
About a handful of his articles also appear in the first area. The strongest impact
of Swanson's work, purely according to this map, appears to be in the areas of
artificial intelligence and neurology.
Additional insights into the impact of Swanson's 15-year quest become clearer
when we study a 3-dimensional visualization, in which most highly cited articles
are displayed in the context of the underlying co-citation network. The highest cited
article in the entire landscape is Swanson's 1988 article in Perspectives in Biology
and Medicine, which identified 11 neglected connections between migraine and
magnesium. This article is almost located right on the boundary between the clinical
medicine literature and the literature of artificial intelligence and neurology. This
unique position and the fact that it has the highest citations in this data set imply
that this article is the gateway between the two disciplinary literatures. Not only has
Swanson established missing links between concepts in the literature of medical
sciences, he has also made a strong connection between information science and
medical sciences.
7.4
Pathfinder Networks' Impact
In our earlier research, we incorporated Pathfinder networks into our Generalized
Similarity Analysis (GSA) framework (Chen
1998a
,
b
,
1999b
; Chen and Paul
2001
;Chenetal.
2001
,
2002
). Traditionally a typical application of Pathfinder
networks relies on proximity data judged manually. The number of nodes in a
typical Pathfinder network ranges from 30 to 50, although Pathfinder networks
of 2,000 nodes were reported in one occasion in the 1980s (Schvaneveldt et al.
1989
). We introduced a variety of computer-generated proximity measures along
with GSA including document-document similarity computed based on information
retrieval models, state transition probabilities derived from a Web navigation, and
co-citations of authors as well as documents (Chen
1999b
)(SeeFig.
7.8
). These
proximity data have extended the scope of Pathfinder networks to a much wider
variety of phenomena beyond the amount of proximity data one can measure by
hand. This extension has not only enriched the topological properties of Pathfinder
networks but also led to valuable insights into the meaning of Pathfinder networks.
The Pathfinder case study is motivated by the question: How does this extension
fit into the general picture of Pathfinder network applications with reference to
traditional Pathfinder applications?
Search WWH ::
Custom Search