Implicit text linkages between Medline records: Using Arrowsmith as an aidto scientific discovery

THE PROBLEM OF HOW TO FIND INTERESTING but previously unknown implicit info rmation within the scientific literature is addressed. Useful information c an go unnoticed by anyone, even its creators, if it can be inferred only by considering together two (or more) separate articles neither of which cite s the other and which have no authors in common. The two articles (or two s ets of articles) are in that case said to be complementary and noninteracti ve. During the past twelve years, this project has uncovered and reported n umerous complementary relationships in the biomedical literature that have led to new information of scientific interest. Several of these literature- based discoveries subsequently have been corroborated through clinical or l aboratory investigations. We describe how to use software that can create s uggestive juxtapositions of Medline records, the purpose being to help biom edical researchers detect new and useful relationships. This software, call ed Arrowsmith, has also proved valuable as a tool for investigating pattern s of complementary relationships in natural language text (Arrowsmith can b e used free of charge at http://kitvi.uchicago.edu).