Knowledge Base System in Logic Programming Paradigm

Computer architectures for logic-oriented data/knowledge bases

“…But a more general approach to providing the necessary data rate is based on the use of multiple disc systems with a page oriented large shared buffer memory between processing (Yokota and Itoh, 1986) elements and the disc. A contention free multi-port page memory (MPPM) has been explored as a central component of various knowledge-based machines developed in the Fifth Generation Computer System project (Itoh et al, 1988).…”

“…Figure 5 A Physical KB Organization Scheme Based on Hashing and Trie (Itoh et al, 1988) Physical data organization schemes based on transformed hash coding such as superimposed code words (SCW) or concatenated code words (CCW) have been widely investigated with the objective of enhancing retrieval from very large knowledge bases (Berra et al, 1987;Ramamohanarao and Shepherd, 1986;Shin and Berra, 1989;Wada et al, 1987). In these schemes, compact images of the knowledge bases are first generated by transformed binary codes.…”

“…A knowledge base indexing scheme based on tries and hashing has been proposed for the retrieval by unification operation as shown in Figure 5 (Itoh et al, 1988). Complex terms are stored in such a way that terms with similar structures share the common left-most subterms with each other so that the number of comparisons required for unification can be reduced.…”

“…However, it was soon recognized that the hardware unification engine could provide only limited functionality and small performance enhancement. In a more recent proposal (Itoh et al, 1988), the knowledge base machine called Mu-X uses eight general purpose microprocessors each of which has 2 Mb of main memory, instead of the unification engines. Mu-X still uses the MPPM designed by Tanaka (1984) which serves as a large shared buffer for storing pages as a basic access unit.…”

Reconciling Systems and Deductive Capabilities in Knowledge Based Systems Using Logic Programming

“…In its last phase of development, its research strands are being brought together into a prototype of a fifth generation machine. The parallel inference sub-system [24] and the parallel knowledge base sub-system [16] are to be merged to form an integrated parallel machine which executes a variant of flat GHC called KL1. The design of the parallel knowledge base sub-system will be mainly based upon enhancements of the Mu-X system and work on a parallel interface to knowledge bases from GHC [16].…”