CARDS: a large knowledge-base engine

The objective of the CARDS project (Alvey Project IKBS 140) has been to focus a wide range of expertise on to the construction of a viable very large knowledge-based system. GEC were the principal investigators and were responsible for the overall management of the project. GEC also designed and implemented the transputer pipeline, modified the operating system of the SUN workstation into a special device server and incorporated other functions in response to demands from the theory and the application (viz. pattern matching of text and the BANG search area indexing). Reading University provided much of the theory and supporting software for the functional database. In particular, a method for knowledge-base design had been developed that integrated hardware considerations with data and knowledge-base demands. The functional database language FAITH was also developed at Reading University. The application chosen was the protein folding problem of the Imperial Cancer Research Fund (ICRF). This knowledge-base described what was then known about the three dimensional folding behaviour of some proteins and most of the original work was written in PROLOG. The CARDS accelerator is an example of a family of front end processors based upon the transputer. Three transputers are formed into a pipeline that supports a content addressable relational mass storage system (1.6 Gbytes)