DocumentCode :
3246013
Title :
Interactive grammar inference with finite state transducers
Author :
Caskey, Sasha P. ; Story, Ezra ; Pieraccini, Roberto
Author_Institution :
Speechworks Int., New York, NY, USA
fYear :
2003
fDate :
30 Nov.-3 Dec. 2003
Firstpage :
572
Lastpage :
576
Abstract :
We propose a method for improving the coverage of handcrafted context free grammars based on a set of new sentence examples. The described algorithm aims at finding the minimal set of modifications to the grammar that increase its coverage while preserving its original structure. The algorithm is based on a finite state transducer (FST) representation of context free grammars. The inference method includes an interactive component that allows developers to control the generalization of the new grammar.
Keywords :
context-free grammars; finite state machines; inference mechanisms; interactive systems; natural language interfaces; speech recognition; speech-based user interfaces; dialog systems; finite state transducers; handcrafted context free grammars; inference method; interactive grammar inference; speech applications; user interfaces; Cities and towns; Costs; Formal languages; Inference algorithms; Speech; Stochastic processes; System testing; Transducers; Usability; User interfaces;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Automatic Speech Recognition and Understanding, 2003. ASRU '03. 2003 IEEE Workshop on
Print_ISBN :
0-7803-7980-2
Type :
conf
DOI :
10.1109/ASRU.2003.1318503
Filename :
1318503
Link To Document :
بازگشت