Recent progress in large-vocabulary continuous speech recognition (LVCSR) has raised the possibility of applying information retrieval techniques to the resulting text. This paper presents a novel unsupervised text segmentation method. Assuming a generative model of a text stream as a left-to-right hidden Markov model (HMM), text segmentation can be formulated as model parameter estimation and model selection using the text stream. The formulation is derived based on the variational Bayes framework, which is expected to work well with highly sparse data such as text. The effectiveness of the proposed method is demonstrated through a series of experiments, where broadcast news programs are automatically transcribed and segmented into separate news stories. © 2007 Wiley Periodicals, Inc. Electron Comm Jpn Pt 2, 90(12): 1–11, 2007; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjb.20421

REFERENCES

1Attias H. Inferring parameters and structure of latent variable models by variational Bayes. Proc 15th Conf on Uncertainty in Artificial Intelligence, p 21–30, 1999.
Google Scholar
2Beeferman D, Berger A, Lafferty J. Statistical models for text segmentation. Machine Learning 1999; 34: 177–210.
10.1023/A:1007506220214
Web of Science® Google Scholar
3Blei DM, Moreno PJ. Topic segmentation with an aspect hidden Markov model. COMPAQ Cambridge Res Labs Tech Rep, CRL-2001-7, 2001.
Google Scholar
4Blei DM, Ng AY, Jordan MI. Latent Dirichlet allocation. J Mach Learning Res 2003; 3: 993–1022.
10.1162/jmlr.2003.3.4-5.993
Web of Science® Google Scholar
5Furui S. Recent progress in corpus-based spontaneous speech recognition. IEICE Trans Inf Syst 2005; E88-D: 366–375.
10.1093/ietisy/e88-d.3.366
Web of Science® Google Scholar
6Hayashi Y, Ohtsuki K, Bessho K, Mizuno O, Matsuo Y, Matsunaga S, Hayashi M, Hasegawa T, Ikeda N. Speech-based and video-supported indexing of multimedia broadcast news. Proc ACM SIGIR, 2003.
Google Scholar
7Hearst MA. Multi-paragraph segmentation of expository text. 32nd Annual Meeting of the Association for Computational Linguistics, p 9–16, 1994.
Google Scholar
8Hofmann T. Probabilistic latent semantic indexing. Proc 22nd Int Conf on R&D in Information Retrieval (SIGIR'99), p 50–57.
Google Scholar
9Jitsuhiro T, Nakamura S. Variational Bayesian based topology training and mixture component splitting for acoustic modeling. Tech Rep IEICE 2004; SP204-91: 61–66.
Google Scholar
10Shinoda K, Lee CH. A structural Bayes approach to speaker adaptation. IEEE Trans Speech Audio Process 2001; 9: 276–287.
10.1109/89.906001
Web of Science® Google Scholar
11Stokes N, Carthy J, Smeaton AF. Segmenting broadcast news streams using lexical chains. STarting AI Researchers Symposium (STAIRS2002), p 145–154.
Google Scholar
12Ueda N, Ghahramani Z. Bayesian model search for mixture models based on optimizing variational bounds. Neural Networks 2002; 15: 1223–1241.
10.1016/S0893-6080(02)00040-0
PubMed Web of Science® Google Scholar
13Yamron J, Carp I, Gillick L, Lowe S, van Mulbregt P. Hidden Markov model approach to text segmentation and event tracking. Proc ICASSP98, p 333–336.
Google Scholar
14Bessho K. Text segmentation using word conceptual vectors. IPSJ J 2001; 42: 2650–2662. (in Japanese)
Google Scholar
15Isotani R, Hatazaki K, Hattori H, Okumura A, Watanabe T. Basic technologies for spontaneous speech recognition and its applications. IPSJ Tech Rep, NL-169-16, p 109–116, 2005. (in Japanese)
Google Scholar
16Mishina T, Yamamoto M. Context adaptation using variational Bayesian learning for ngram models based on probabilistic LSA. IEICE Trans Inf Syst Pt 2 2004; J87-D-II: 1409–1417. (in Japanese)
Google Scholar
17Ueda N. Bayesian learning [III]: Basics of variational Bayesian learning. J IEICE 2002; 85: 504–509. (in Japanese)
Google Scholar
18Watanabe S, Minami Y, Namamura A, Ueda N. Selection of shared-states hidden Markov model structure using Bayesian criterion. IEICE Trans Inf Syst Pt 2 2003; J86-D-II: 776–786. (in Japanese)
Google Scholar
19ChaSen: A Japanese morphological analysis system, http://chasen.naist.jp/ (in Japanese)
Google Scholar

Volume90, Issue12

December 2007

Pages 1-11

HMM-based text segmentation using variational Bayes learning and its application to audio-visual indexing

Abstract

REFERENCES

References

Information

About Wiley Online Library

Help & Support

Opportunities

Connect with Wiley

HMM-based text segmentation using variational Bayes learning and its application to audio-visual indexing

Abstract

REFERENCES

References

Related

Information