Abstract:
This paper presents a new semantic description method by building a Semantic Concepts System (SCS), in which Semantic Patterns are used to describe and recognize the speech acts expressed by spoken utterances. By this means, users’ intentions can be understood and elicited more directly and precisely. Moreover, Based on the domain knowledge related SCS, we can build high efficient lan- guage understanding models and generation models for dif- ferent domains. This method provides solutions in building universal information service SDS platform, and has been applied to develop Beijing Railway Station Ticket Informa- tion SDS (BEST) and the Capital Airport Flights Information SDS (BAFI) successfully.