Are you being addressed? - Real-time addressee detection to support remote participants in hybrid meetings

Abstract

In this paper, we describe the development of a meeting assistant agent that helps remote meeting participants by notifying them when they are being addressed. We present experiments that have been conducted to develop machine classifiers to decide whether "you are being addressed" where "you" refers to a fixed (remote) participant in a meeting. The experimental results back up the choices made regarding the selection of data, features, and classification methods. We discuss variations of the addressee classification problem that have been considered in the literature and how suitable they are for addressee detection in a system that plays a role in a live meeting. © 2009 Association for Computational Linguistics.