Data-parallel compilation and query power extension of large knowledge bases

An associative data-parallel compilation model of logic programs capable of answering queries with unspecified relations concerning the given objects is described. The model benefits from the synergy resulting from associative search, data-parallelism during goal reduction, and the use of low-level code to invoke the subgoals and savings in data-transfers resulting from the presence of global registers. The use of associative tables extends the power of logic programming to answer a large class of queries and derive unspecified relation-names for the given objects. In contrast to the interpretation based on a pure data parallel model, this model does not suffer from data sequentiality caused by the presence of multiple-occurrence variables in the goals. The model also handles variable aliasing in the clauses efficiently using the associative data-parallel search and data-parallel assignment property