Human Activity Supporting by Deontological Knowledgebases

A concept of computer systems supporting human activity by recommendations for actions leading to desired intermediate and final goals is here presented. The recommendations are given in a standard form of deontological statements specifying the current states, the desired next states and operations transforming the current into the desired states. The quality of recommended actions is by parameters describing various favorable and unfavorable effects assessed. The quality parameters are presented as vectors in semi-ordered linear (Kantorovitsch) space where preferences between vectors is induced by definition of a positive vectors’ cone. It is shown how the positive cone by systems of linear inequalities (without constant terms) can be defined so as to establish a compromise between different quality parameters in order to chose the most preferable actions. Two methods of the above-mentioned methods extension on the paths of actions comparative quality assessment are proposed. The presented basic concepts by simple examples are illustrated.