The natural language component - STINA
Rolf Carlson, Sheri Hunnicutt · 1995
In this paper we will give a short overview of a dialogue system and describe the natural language and dialogue component in detail. Our work in this project is focused on a sublanguage grammar, a grammar limited to a particular subject domain- that of requesting information from a transportation database. Our parser, STINA, is knowledge based and is designed as a probabilistic language model. It contains a context-free grammar which is compiled into an augmented transition network (ATN). Probabilities are assigned to each arc after training. Characteristics of STINA are a stack-decoding search strategy and a featurepassing mechanism to implement unification. Dialogue management based on grammar rules and lexical semantic features is implemented in STINA. The notation to describe the syntactic rules has been expanded to cover some of our special needs to model the dialogue. The STINA parser is running with two different time scales corresponding to the words in each utterance and to the turns in the dialogue. Topic selection is accomplished based on probabilities calculated from user initiatives.