CMLLite: a design philosophy for CML

There is an on-going need for formal, computable representations of scientific data and documents which are also accessible to humans [1-3]. The challenge is to devise systems that people will not only use but for which they will, critically, develop additional tools and content. Our approach for chemistry is Chemical Markup Language (CML) (whose evolution and philosophy is described elsewhere in this issue [4]) which has been developed to support five main areas of chemistry (molecules, reactions, solid-state, spectroscopy and computational chemistry).The strengths and weaknesses of CML have been recently analysed by Dumontier [3] and we quote directly:Chemical Markup Language, backed by a controlled vocabulary, has been rather successful in specifying most aspects of chemistry, from small molecules and their connectivity to polymers and crystal structures.Unfortunately, while most elements of this specification can be parsed out using one of the many XML libraries, certain elements do not render themselves to facile interpretation. Consider the sample CML specification of a water molecule [...]. In order to identify the member atoms in a given bond, it is necessary to carry out string processing as an intermediate step. Further, while many of the elements of CML are defined in a controlled vocabulary, the lack of explicit, consistent, and formal axiomatization of the involved concepts gives rise to difficulties in inferring connections between chemical concepts where no such connections are stated explicitly, something that is possible in formal ontology-backed RDF-based information specifications. Although CML specifications have been increasingly evolving to incorporate elements of the Semantic Web, the lack of widespread adoption of the format, and the limited availability of large-scale CML-based chemical knowledge repositories, have somewhat limited CML-assisted federation of the world of chemical data. Furthermore, the implementation of coverage of additiona