Hierarchical geometric constraint networks as a representation for spatial structural knowledge.

A representation is proposed for capturing generic spatial knowledge about classes of objects in a structural hierarchy of biology. The basic premise is that spatial properties such as shape and relative relationships can be expressed as networks of interacting constraints. The representation is defined, partial implementations that demonstrate practical utility are described for model based organ and protein structure determination, and research issues are presented that must be solved before the representation can be implemented in its entirety. As these issues are resolved the representation will find increasing utility as the foundation for a spatial knowledge base of structural biology.