A stochastic approach to the understanding problem in dialog systems
Departamento de Sistemas Informáticos y Computación, Universidad Politécnica de Valencia
DATE: September 28 2009, 3.00 P.M.
LOCATION: Room 7 – Faculty of Science – Povo (TN)
We present a stochastic approach to the understanding component in the area of the spoken dialog systems. We learn a two-level semantic model from a semantic labeled corpus. This learned model transforms the transcription of the utterance into a set of segments, and assign to each one a semantic unit defined for the task. We obtain a semantic representation of the phrase based in concepts and restrictions to the concepts. Some experiments in the framework of the Spanish research projects BASURDE, DIHANA will be presented
Fernando Garcia Granada received the M.S. in computer science from the Universidad Politécnica de Valencia (UPV), Spain, in 1992, and the Ph.D. in computer science from the Universidad Politécnica de Valencia (UPV), Spain, in 2003. He joined the Departamento de Sistemas Informáticos y Computación (DSIC) of the UPV in 1996, where he is now an Associate Professor of computer science. He is an active member of the research of the Natural Language Engineering and Pattern Recognition group ELiRF (old RFIA) at the same institution. His research is mainly focused on the application of grammatical inference techniques and automatic learning applied to language understanding and dialog systems.
CONTACT: Giuseppe Riccardi