H. Small. Visualizing science by citation mapping. Journal of the American Society for Information Science, 50(9):799–813, 1999. [url]

———————-

This paper present methods for science mapping with the aim to allowing the user to spot literature not directly related to the one another. The object of this work is to create a global structure and overview of as large a sample of data as possible and then enable the user to user to explore the underlying fine structure. They call this procedure od positioning objects in space ‘ordination’.

They use the frequency of co-citation as the building block of the visualization of science.  The authors list different techniques for graphically compute the position in space of the documents. The first is the distance tranformations, achieved through a mathematical minimization. The second is the ordination by triangulation, which principle is the arbitrary poistioning of the first two objects on the coordinate systems, and the subsequent positioning of the rest of the object according to the distance to the first two objects. Others techniques are listed to reduce the number of computations required for the task: as the sampling with fractional citation counts, and the humpty-dumpty method.

The visualization method chosen is called Volvox (McCain, 1996). This display presents lower-level objects as circles, within larger high-level objects. A linkage between the clusters is also offered, allowing for the discovery of the shortest pathways through two points/documents on the map.

Tags: clustering, map algorithms, maps