The problem of speech recognition error is still the major obstacle to making spoken dialog systems be widely used in our daily lives. In this paper, we proposed an error-tolerant language understanding model to improve the robustness of dialogue systems by detecting and recovering the errors arising from speech recognition. The basic idea of our approach is using example sentences to provide the clues for detecting and recovering errors. The procedure of detection and recovery is guided by a unified scoring function which consults not only the scores from the speech recognizer and concept parser but also the scores of concept-bigram and edit operations, such as deleting, inserting and substituting concepts. Experimental results show that the proposed model effectively reduces the impact of speech recognition errors. The precision and recall rates of semantic slot are improved by 40.3% and 25.4% respectively.
Cite as: Lin, Y.-C., Wang, H.-M. (2000) Error-tolerant language understanding for spoken dialogue systems. Proc. 6th International Conference on Spoken Language Processing (ICSLP 2000), vol. 1, 242-245, doi: 10.21437/ICSLP.2000-60
@inproceedings{lin00_icslp,
author={Yi-Chung Lin and Huei-Ming Wang},
title={{Error-tolerant language understanding for spoken dialogue systems}},
year=2000,
booktitle={Proc. 6th International Conference on Spoken Language Processing (ICSLP 2000)},
pages={vol. 1, 242-245},
doi={10.21437/ICSLP.2000-60}
}