What are the key criteria for CML and its ontology?
(a) CML must be ontologically neutral. At present most molecular “file formats” contain complex, often implicit, ontologies which are not necessarily reconcilable between different formats. Thus many systems can describe “aromatic bonds” or “formal charges”, but these concepts are not completely portable. In other words simply mapping (say) bonds designated of order 4 (MDL-molfile) to the designation -5 (CCDC file) may be ontologically unsound since the algorithms used to create these may be different. Hydrogen counts, formal charges, atom “types”, bond orders, and valences are related in a complex manner and are not always used consistently. In some systems bond orders are related to measured lengths, in others to calculated properties and in others to electron counts. Therefore while CML must support the concept of bond order it cannot arbitrate between different uses. CML has a minimal ontology that is primarily for convenience in managing the components. Thus