Data base design for natural language medical data

The natural language of clinical reporting exhibits the properties of a sublanguage, having a grammar and vocabulary greatly restricted in comparison to the language as a whole. This paper presents a CODASYL-type (network) data base schema for natural language medical records based on the structures and relationships identified in such material through computerized language processing. While the schema described is special to the medical sublanguage, the methods of analysis and of schema design are applicable to natural language material in other subject areas.