| Paper: | SP-P7.13 (ICASSP 2003 Paper) | ||
| Session: | Topics in Speech Analysis | ||
| Time: | Wednesday, May 19, 13:00 - 15:00 | ||
| Presentation: | Poster (ICASSP 2003 Presentation) | ||
| Topic: | Speech Processing: Spoken Language Systems and Dialog | ||
| Title: | TRAINING A PROSODY-BASED DIALOG ACT TAGGER FROM UNLABELED DATA | ||
| Authors: | Anand Venkataraman; SRI International | ||
| Luciana Ferrer; SRI International | |||
| Andreas Stolcke; SRI International | |||
| Liz Shriberg; SRI International | |||
| Abstract: | Dialog act tagging is an important step toward speech understanding, yet training such taggers usually requires large amounts of data labeled by linguistic experts. Here we investigate the use of "unlabeled" data for trainingHMM-based dialog act taggers. Three techniques are shown to be effective for bootstrapping a tagger from very small amounts of labeled data: iterative relabeling and retrainingon unlabeled data; a dialog grammar to model dialog act context,and a model of the prosodic correlates of dialog acts.On the SPINE dialog corpus, the combined use of prosodic information and unlabeled data reduces the tagging error between 12% and 16%, compared to baseline systems using word information and various amounts of labeled data only. | ||
| Back | |||
Home -||-
Organizing Committee -||-
Technical Committee -||-
Technical Program -||-
Plenaries
Paper Submission -||-
Special Sessions -||-
ITT -||-
Paper Review -||-
Exhibits -||-
Tutorials
Information -||-
Registration -||-
Travel Insurance -||-
Housing -||-
Workshops