Paper: Question Detection in Spoken Conversations Using Textual Conversations

ACL ID P11-2021
Title Question Detection in Spoken Conversations Using Textual Conversations
Venue Annual Meeting of the Association of Computational Linguistics
Session Main Conference
Year 2011
Authors

We investigate the use of textual Internet con- versations for detecting questions in spoken conversations. We compare the text-trained model with models trained on manually- labeled, domain-matched spoken utterances with and without prosodic features. Over- all, the text-trained model achieves over 90% of the performance (measured in Area Under the Curve) of the domain-matched model in- cluding prosodic features, but does especially poorly on declarative questions. We describe efforts to utilize unlabeled spoken utterances and prosodic features via domain adaptation.