W0182

Smart CIF Dictionaries for Crystallographic Knowledge Bases. S.R. Hall, N. Spadaccini, D. du Boulay, I.R. Castleden, Departments of Crystallography and Computer Science & Software Engineering, The University of Western Australia, Perth, WA 6009, Australia.

The CIF dictionaries (core, pd, mm) [1] are used widely for validating data submitted to structural science journals and databases. These applications are, however, restricted by the inability to specify within the dictionaries the precise relationship between a derivative data item and other defined items. This paper will describe a new prototype dictionary language, StarDDL, which uses text method attributes to specify precise relational expressions for linking defined data items. The expressions are written in a customised symbolic scripting language called dREL [2].

A StarDDL dictionary may be compiled into Java and used as a knowledge base attached to any CIF. Data may be then requested from the CIF, and will be returned either as those instantiated values on the file, or as values derived from the dictionary methods. This knowledge base approach may be applied to either the derivation of new values, or the validation of existing values, for any scientific discipline.

A prototype core CIF dictionary written in StarDDL will be used to demonstrate the automatic evaluation of crystallographic quantities such as molecular geometry and structure factors.

[1] www.iucr.org/iucr-top/cif/#dics

[2] Spadaccini, N., Hall, S.R., and Castleden, I.R. (2000) J. Chem Inf. Comp. Sc. 40, 1289-1301.