Development of Delta as a First Step to a Knowledge Base Machine

Computer architectures for logic-oriented data/knowledge bases

“…One of the issues involved in coupling a logic programming system and a database system is to resolve the mismatch between the tuple oriented access of the logic programming system and the set oriented retrieval of the database system. From the performance point of view, supporting partial match queries, join and other operations has been one of the major concerns in the back end (Gonzalez-Rubio et al, 1987;Sakai et al, 1986) since these operations are expected to occur more frequently than in conventional database systems. Consider, for example, the query <-p(a,X),q(c,b,c), and suppose that p and q are base relations (facts).…”

“…Recursively defined rules require a more complicated operation called the least fixed point (LFP) operation (Aho and Ullman, 1979;Sakai et al, 1986). When processing a query involves recursively defined subgoals, the query cannot be translated into an explicit sequence of relational operations as in the above example, since the termination point cannot be recognized in advance.…”

“…The physical data organization scheme of DELTA (Sakai et al, 1986) is a fully decomposed storage oganization, where a relation with N arguments is divided into N binary relations with a (Sakai et al, 1986) tuple identifier (TID) as the first attribute for all N binary relations (see Figure 2(a)). This organization is well suited to join operations and partial match retrieval queries, especially for those queries having a small number of arguments (attributes) specified.…”

“…Deductive database machines are back end computers designed to manage very large fact bases by efficiently supporting the relational operations. In this section, we describe the architectures of three representative deductive database machines including ILEX (Li, 1984) DELTA (Sakai et al, 1986), and MPDC (Tanaka, 1986).…”

“…The KM-1 proposed by Kellog (1986) takes a similar approach to that of ILEX in the sense that it uses existing systems rather than developing specialized hardwares (the Xerox 1100 Lisp machine as the front-end and the Britton-Lee IDM : 500 relational database machine as the back-end). Figure 12 DELTA (Sakai et al, 1986) 8.a.2 DELTA DELTA is a functionally distributed MIMD database machine supporting inference machines (PSI's) as a back-end (Sakai et al, 1986). It consists of two subsystems: the relational database management supervisory/processing (RSP) subsystem and the hierarchical memory (HM) subsystem.…”