New Technology for an Encyclopedic Browsing HSI

The best Human System Interface (HSI) to a large multi-domain knowledge base is a knowledgeable human librarian. Current computer based browsing technologies fall far short of this ideal because the number of possible data relationships is very large and any attempts to capture and utilize them with traditional indexing and search methods quickly becomes overwhelming. we propose to test a radical, new approach to domain-independent context-sensitive encyclopedic browsing that employs our emerging theory of "correlithm objects," or "corobs." We will show that this technology provides new ways to represent and utilize knowledge where (1) links between data are implicit and probabilistic, and are only explicitly instantiated when needed; (2) object content is learned by associations that can be supervised or unsupervised; (3) searching proceeds by automatic "freee association"; and, (4) search speeds can approach those searches one in highly structured index systems such as B-trees. In Phase I, we will execute a comprehensive project to demonstrate each of these points and the overall feasibility and advantages of this new approach, to show the appropriateness of this approach as a basis for encyclopedic browsing, and to develop the basics of an HSI which provides the required domain-independent context-sensitive encyclopedic browsing capability. Phase II work will produce and demonstrate a practical encyclopedic browsing HSI for Phase III deployment in a wide range of dual-use commercial applications.