ABSTRACT
The STAR File (J. Chem. Inf Comput. Sci. 1994, 34, 505-508) is used widely in structural chemistry for exchanging numerical and text information with scientific journals and databases. These exchanges are increasingly dependent on data dictionaries to facilitate automatic data validation and checking. Definitions in data dictionaries are constructed using attribute descriptors, and this paper describes a method attribute for specifying the relationships between data items as an executable script written in a new relational expression language called dREL. The addition of this attribute improves the precision and the semantic content of dictionaries by providing relational representations of data, as well as facilitating the direct evaluation of derivable data items. The capacity to evaluate derivative data directly from the combination of primitive data and dictionary expressions is expected to change future archival approaches. The design concepts of the relational expression language dREL parser, which are applicable to any discipline, are described.