Techniques for building a world-wide information infrastructure by reverse engineering existing databases to link them in a hierarchical system of subject clusters to create an integrated database are explored. The controlled vocabulary of the Library of Congress Subject Headings is used to ensure consistency and group similar items. Each database becomes a system object, and each package within the database is assigned a subject cluster based on its content. An expert system matches the user profile to the information package best suited to need and locates the appropriate database. This is supplemented by a machine-generated natural language mapping scheme to lead the user into the clusters of interest. For the prototype, an object-oriented hypermedia user interface was developed, using MARC records. Packages are grouped into subject clusters consisting of the classification number and the first subject heading/keyword assigned. Use of a hierarchical classification number (Dewey number) makes it possible to broaden or narrow a search at will. It is anticipated that the system will be useful to searchers and will also provide a basis for automated indexing. Fifteen computer prototype screens are presented as illustrations. (SLD)