This paper overviews and reports current progress on a research project that develops methods that enable spoken language systems to automatically extend themselves to incorporate new words. The system attempts to detect out-of-vocabulary words in spoken dialogs using multiple stochastic and symbolic knowledge sources. It combines acoustic confidence measures with the semantic, pragmatic and discourse structure knowledge embodied in the MINDS-II system. The results indicate that the conjoined usage of acoustic confidence measures of accuracy and higher-level constraints increased ability to detect misrecognitions by 36% and enabled the larger system to overcomes the weaknesses of the individual techniques. The techniques detect complementary phenomena. Current work focuses upon development of more sophisticated techniques for conjoining these two methods and techniques to use acoustic confidence measures during decoding.
Cite as: Young, S.R., Ward, W.H. (1995) The role of higher-level semantic, pragmatic and discourse knowledge in recognizing and understanding new spoken words and phrases. Proc. ESCA Workshop on Spoken Dialogue Systems, 29-32
@inproceedings{young95_sds,
author={Sheryl R. Young and Wayne H. Ward},
title={{The role of higher-level semantic, pragmatic and discourse knowledge in recognizing and understanding new spoken words and phrases}},
year=1995,
booktitle={Proc. ESCA Workshop on Spoken Dialogue Systems},
pages={29--32}
}