Natural Language Interface to Knowledge Management Systems

In this paper, we present linguistic techniques required for natural language driven knowledge management interfaces. We describe two significant aspects of such an interface: First, how the user input is handled to provide an unrestricted natural language user interface, and second, how the gathered knowledge should be preprocessed, classified and thus prepared for a natural language interactive retrieval. A framework of grammatical structures (supertags) is associated with the elements of an ontology (of the respective domain). This combination of ontologies and supertagging describes a novel and very robust way in parsing of nontrivial user utterances and allows natural language feedback generation.