The problem we address is how to find interesting but previously unknown implicit information within the scientific literature. Useful information can go unnoticed if it is not explicit within any single article, but can be inferred by considering together two (or more) separate articles. We have conducted numerous link analyses of title words and phrases within the MEDLINE database in order to detect complementary relationships that reveal or suggest new information. We have used the computer to identify and construct suggestive juxtapositions of medical article titles in an attempt to enhance the ability of biomedical researchers to detect new and useful relationships that otherwise would be very difficult to uncover. This software is also a tool for investigating the process and problems of link analysis in natural language text. It is available for free public use on the web.