Structures for controlling synthesis design and reaction modelling via knowledge acquisition

Abstract

The performance of programs for computer-assisted organic synthesis (CAOS) is directly related to the knowledge base and to the procedures for deciding how the knowledge base should be used. The architecture of a knowledge base (text base, data base, rule base, model base and commonsense base) is discussed. The problem of knowledge acquisition is described with regard to pruning the productivity of CAOS systems. The commercial program 1stCLASS (for symbolic processing) is used with the SAAO (devised here for numeric processing) to form a commonsense base. For this purpose, knowledge was acquired by induction from the solutions provided by human experts to given problems. The general procedure is applied to the recognition of reaction sites which are crucial in the simulation of organic reactions.