We use a combination of linear support vector machines and hidden markov models for dialog act tagging in the HCRC MapTask corpus, and obtain better results than those previously reported. Support vector machines allow easy integration of sparse high-dimensional text features and dense low-dimensional acoustic features, and produce posterior probabilities usable by sequence labelling algorithms. The relative contribution of text and acoustic features for each class of dialog act is analyzed.
Cite as: Surendran, D., Levow, G.-A. (2006) Dialog act tagging with support vector machines and hidden Markov models. Proc. Interspeech 2006, paper 1831-Wed2FoP.12, doi: 10.21437/Interspeech.2006-535
@inproceedings{surendran06_interspeech,
author={Dinoj Surendran and Gina-Anne Levow},
title={{Dialog act tagging with support vector machines and hidden Markov models}},
year=2006,
booktitle={Proc. Interspeech 2006},
pages={paper 1831-Wed2FoP.12},
doi={10.21437/Interspeech.2006-535}
}