摘要 |
PURPOSE:To remarkably shorten the processing of conditional collation at the time of inference, by reserving the state table of the inference, that is, an inference network in a knowledge base, and maintaining and managing it always as the inference network fitted to a current knowledge base by updating a preserved inference network simultaneously with the time of updating the knowledge base. CONSTITUTION:An inference engine 3 stores the inference network 11 used for the inference at the time of completing the inference by an inference network storage means 6 in the knowledge base 1. The stored inference network 11 is updated by a knowledge update monitoring means 7 and an inference network updating means 8 in accordance with the updating of the knowledge base 1. In such a way, by reading the inference network 11 from the knowledge base 1 by an inference network reading means 5, it is possible to make the generation of an initial inference network by the inference engine 3 in the inference being performed repeatedly unnecessary.
|