Comparing cognitive maps using graph algorithms.

The purpose of this project is two-fold: (1) to examine researchers' knowledge structures on research topics; (2) to compare knowledge structures of experts with those of non-experts. Expert is defined as the researcher who had conducted in-depth research and published on the topic and whose vocabulary used to describe the topic for online searching was the basis for constructing maps. Non-experts are also researchers in the same field but had not done any in-depth work on the topics when they were asked to make a map using the given vocabulary. Both experts and non-experts were allowed to add new terms to or drop original terms from the given vocabulary. The finished cognitive map is a structured layout of the terms on a two-dimensional plane. During the mapping process, subjects also provided thinking-aloud protocols, which revealed additional information on how they saw the relationships of the concepts represented by the terms. Preliminary analyses of ten sets of cognitive maps for ten research topics revealed differences in final vocabulary (after adding and dropping terms), configuration (topdown, left-right, radial, etc.), and foci (focusing on problems, issues or processes). These results were reported at the 1999 ASIS Annual Meeting. Work has just been completed to advance the comparison of semantic closeness using graph theory to convert maps into matrices and to calculate similarities. The following algorithms have been developed for the conversion and calculation of cognitive maps