Intellectual artefacts of expert systems meta-epistemology

摘要

Essential to intelligent programs is computational epistemology. Expert systems, derive their computational epistemology from domain expertise. Construction of an expert system shell and an application requires a metaepistemology that transforms domain epistemology through a sequence of computational epistemology, shell epistemology; programming epistemology and application epistemology (as distinct from application ontology) into an expert system. This paper explains the metaepistemological method by reference to prototypical development of the eGanges shell (2002--4) that is suited to domains (such as law) with rule systems, systemic procedures and/or strategic paths (RPS). The computational epistemology of 3d logic is used as its logic reification suits object-oriented programming. Retroduction (Peirce, 1931. p.28), commonly known as abduction, used according to common expert sense, effects the transformation. The metamorphosis is achieved by a sequence of selections of what fits next, and whatneeds to be repositioned for better fit. In the eGanges design, a central two dimensional tributary structure called a river or rule map is selected out of the 3d logic model, to optimize cognitive value of the computational epistemology for the user interface. River ideographs are streamlined flowcharts that resemble fishbone diagrams. Extensive, dense river ideographs, may be variously nested as sub-maps, and also variously glossed to incorporate annotations of the RPS system. Gloss options include links between nodes in the same set of sub-maps and between parallel river systems. Sub-epistemologies may be required for the glosses. Functionality of the shell facilitates navigation and interrogation of the maps, as well as processing interrogation input.