Understanding chemical terminology

This paper describes the formal properties of a system that understands organic chemical terminology. The system recognises terms in texts, analyses these terms, reconstructs their molecular structure (to the extent it is made explicit in the term), and classifies these terms according to functional and structural properties. The system is able to deal with fully specified chemical terms, underspecified terms such as deoxypentose, class names and terms that are built up by subterms from any of these. Not only the depth of analysis but also its ability to cope with underspecification and class names distinguishes it from other existing systems.