Henson RW & Tjahjadi T

Proc IASTED Int. Symp. A.I. Applications and Neural Networks, 1990, 275-278

Abstract

An X-ray rocking curve is an angular reflectivity plot used for analysing the structure and composition of crystal lattices. In order to assess the integrity of complex crystals, it is necessary to simulate the experimental rocking curve by selecting the most likely configuration for the lattice parameters. This paper shows that expertise is partly a function of the initial selection procedure used when building the lattice parameters and proposes that experts have a prototypical representation of rocking curves from which they make an initial assumption about the material composition. Experimental evidence suggests that prototype construction tends to operate with graphical data and helps an expert's performance. By transferring such prototype configuration to an expert system database, it is envisaged that novices will be better able to select lattice parameters.