| Paper: | SLP-L3.2 | ||
| Session: | Spoken Language Dialog | ||
| Time: | Tuesday, May 16, 16:50 - 17:10 | ||
| Presentation: | Lecture | ||
| Topic: | Speech and Spoken Language Processing: Spoken Language Dialog | ||
| Title: | Towards Learning to Converse: Structuring Task-Oriented Human-Human Dialogs | ||
| Authors: | Srinivas Bangalore, Giuseppe Di Fabbrizio, AT&T Labs – Research, United States; Amanda Stent, Stony Brook University, United States | ||
| Abstract: | Data-driven techniques have influenced many aspects of speech and language processing tasks. Models derived from data are generally more robust than hand-crafted systems since they better reflect the distribution of the phenomena being modeled. Dialog management is at the threshold of reaping the beneflt of data-driven techniques with the availability of large human-human dialog corpora. In this paper, we present our view of structuring human-human dialogs in order to learn models for human-machine dialogs. We present the problem of dialog segmentation and dialog act labeling, develop a model for predicting and labeling segments and dialog acts and evaluate the models on customer-agent dialogs from a catalog service domain. | ||
©2018 Conference Management Services, Inc. -||- email: webmaster@icassp2006.org -||- Last updated Friday, August 17, 2012