Utilize este identificador para referenciar este registo:
http://hdl.handle.net/10071/27854
Autoria: | Medeiros, H. Batista, F. Moniz, H. Trancoso, I. Nunes, L. |
Editor: | Leal, J. P., Rocha, R., and Simões, A. |
Data: | 1-Jan-2013 |
Título próprio: | Comparing different machine learning approaches for disfluency structure detection in a corpus of university lectures |
Volume: | 29 |
Título e volume do livro: | 2nd Symposium on Languages, Applications and Technologies, SLATE 2013 |
Paginação: | 259 - 269 |
Título do evento: | 2nd Symposium on Languages, Applications and Technologies, SLATE 2013 |
Referência bibliográfica: | Medeiros, H., Batista, F., Moniz, H., Trancoso, I., & Nunes, L. (2013). Comparing different machine learning approaches for disfluency structure detection in a corpus of university lectures. In J. P. Leal, R. Rocha, & A. Simões (Eds.), 2nd Symposium on Languages, Applications and Technologies, SLATE 2013 (vol. 29, pp. 259-269). OASIcs. https://doi.org/10.4230/OASIcs.SLATE.2013.259 |
ISSN: | 2190-6807 |
ISBN: | 978-3-939897-52-1 |
DOI (Digital Object Identifier): | 10.4230/OASIcs.SLATE.2013.259 |
Palavras-chave: | Machine learning Speech processing Prosodic features Automatic detection of disfluencies |
Resumo: | This paper presents a number of experiments focusing on assessing the performance of different machine learning methods on the identification of disfluencies and their distinct structural regions over speech data. Several machine learning methods have been applied, namely Naive Bayes, Logistic Regression, Classification and Regression Trees (CARTs), J48 and Multilayer Perceptron. Our experiments show that CARTs outperform the other methods on the identification of the distinct structural disfluent regions. Reported experiments are based on audio segmentation and prosodic features, calculated from a corpus of university lectures in European Portuguese, containing about 32h of speech and about 7.7% of disfluencies. The set of features automatically extracted from the forced alignment corpus proved to be discriminant of the regions contained in the production of a disfluency. This work shows that using fully automatic prosodic features, disfluency structural regions can be reliably identified using CARTs, where the best results achieved correspond to 81.5% precision, 27.6% recall, and 41.2% F-measure. The best results concern the detection of the interregnum, followed by the detection of the interruption point. |
Arbitragem científica: | yes |
Acesso: | Acesso Aberto |
Aparece nas coleções: | ISTAR-CRI - Comunicações a conferências internacionais IT-CRI - Comunicações a conferências internacionais |
Ficheiros deste registo:
Ficheiro | Tamanho | Formato | |
---|---|---|---|
conferenceobject_12189.pdf | 413 kB | Adobe PDF | Ver/Abrir |
Todos os registos no repositório estão protegidos por leis de copyright, com todos os direitos reservados.