Special Sessions

SS-3

Spoken language processing of human-human conversations

Introduction
Conversations provide an efficient way of human interaction, and create unique knowledge sharing opportunities between people with different areas of expertise. Although conversations are so common, there is still no globally adopted automated or semi-automated mechanism for processing these conversations, saving and analyzing their content for later use, or mining them.

The availability of human-human phone conversations, such as the Switchboard corpus, and multi-party meeting recordings such as the ICSI and AMI corpora, publicly broadcast conversations such as talk shows, and shared task evaluations such as the ones performed by NIST, have facilitated research on automatic processing of conversational speech.

There have been active research efforts on processing of human conversations, covering a broad range of tasks from speech recognition, speaker diarization, to speech understanding and social signal analysis. The purpose of this special session is to offer the opportunity to the researchers working on speech and language processing of conversations to share ideas and have constructive discussions and provide all others an opportunity to find out about the latest developments in this field, in a single, focused, special conference session.

The focus areas in the proposed special session include:
- Dialog act tagging of conversations
- Automatic summarization of conversations
- Speech analytics on human-human dialogs
- Social signal analysis in conversations
- Extraction of conversation structure
- Annotation and evaluation issues
- Information extraction from human conversations
- Multimodal analysis of human conversations

Organizers:
Dilek Hakkani-Tur, Microsoft Speech Labs, (dilek@ieee.org)
Yang Liu, University of Texas at Dallas (email:yang.liu@utdallas.edu)