We have proposed a method of concept-driven semantic interpretation based on general semantic knowledge of conceptual dependency. In our approach, a concept is a unit of semantic interpretation and an utterance is regarded as a sequence of concepts that convey an intention. However, a considerable number of accepted results were not syntactically meaningful. This is because the order in which linguistic features occurred in the sequence of concepts was not taken into account in constructing the whole meaning from the concepts: only semantic constraint was used to attain linguistic robustness. Therefore, we introduce a statistical language model which calculates the plausibility of a sequence of concepts from the points of view of the order in which shallow linguistic features occur. Experimental results of speech understanding for 1000-word-vocabulary spontaneous speech show that the proposed method significantly improves the system performance.
Cite as: Nagai, A., Ishikawa, Y. (1998) Concept-driven speech understanding incorporated with a statistic language model. Proc. 5th International Conference on Spoken Language Processing (ICSLP 1998), paper 1023, doi: 10.21437/ICSLP.1998-483
@inproceedings{nagai98_icslp,
author={Akito Nagai and Yasushi Ishikawa},
title={{Concept-driven speech understanding incorporated with a statistic language model}},
year=1998,
booktitle={Proc. 5th International Conference on Spoken Language Processing (ICSLP 1998)},
pages={paper 1023},
doi={10.21437/ICSLP.1998-483}
}