Rapid Prototyping for Spoken Dialogue Systems

We implemented a spoken dialogue system architecture for rapid prototyping. The features that support rapid prototyping include a clear separation of generic dialogue processing algorithms from domain and language speci c knowledge sources. In an experiment, it could be shown that six individuals could specify these domain and language speci c knowledge sources within 8 to 12 hours to come up with a prototypical implementation of a spoken dialogue system. To that end, no dialogue strategy had to be speci ed. Rather, it was suÆcient to provide an ontology, a description of the services o ered by the system, parsing grammars, database conversion rules and generation templates. Furthermore, the experiment shows that it is possible to formulate dialogue strategies in a domain and language independent manner; thus not requiring a system designer to be knowledgeable about dialogue processing.