Abstract
One function of the Semantics component of SPEECH-LIS, the Bolt Beranek and Newman (BBN) Speech Understanding System, is to gather evidence for hypotheses it has made regarding the content of an utterance, as well as to evaluate the hypotheses made by other components. Another is to produce a representation of the utterance's meaning. Specifically, this involves forming consistent, meaningful collections of words which match regions of the speech waveform, and evaluating and interpreting the possible syntactic structures built of them. This paper discusses the data structures and organization of SPEECHLIS semantics and how they are directed to the above two tasks.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
