Generating natural language interfaces for terminological knowledge representation systems

A major problem when using knowledge representation or database systems for information retrieval applications is that although the query language might be very expressive it may not support a comfortable working style. The main reason is that the user has to type long error prone strings and has to know the exact names of concepts and roles. This paper describes how systems that provide a natural language interface for querying terminological knowledge representation systems can be improved by an automatic generation process. Apart the terminological knowledge base all that is needed is a lexicon and a case table. A case grammar is automatically generated based on the naming and modelling conventions used in knowledge engineering.