Utilizing scientific literature as a potential source for new knowledge is an extremely attractive idea. Professor Don Swanson at the University of Chicago has demonstrated that complementary pairings of existing knowledge can be applied in the solution of new problems. Supercomputers can perform portions of the analysis process more quickly, but they also make possible a structurally different approach. A pair of topical literatures can be assumed to be related, and the search for intermediate literatures may then work from "both ends toward the middle". Pairs proving to not be of interest are simply discarded.
As this approach relies less upon human expertise, it can potentially be scaled up, to enable the examination of significantly more topical pairs. Not arguing for the substitution of relatively weak statistical clues for researcher understanding in the search process, it may still be possible to employ the supercomputer as a form of filter or sorter. Pairs exhibiting stronger than normal statistical ties via intermediate literatures could be identified to researchers. This may help in the organizational processes of filtering and prioritizing, to determine those pairs researchers should first examine.
This project was active from 2001-2002. The project team was: F. Wilf Lancaster (Principal Investigator), Larry S. Jackson, Carole Palmer, Mark Spasser, and David Dubin
For an overview of our work, please read our papers:
Copyright: UIUC GSLIS 2001-2002. All rights reserved. Last update: 9 December 2005.