HELPR: A Multi-Disciplinary Strategy for Enhancing Speech-Dialog Platforms

Abstract

People usually interact with intelligent agents (IAs) when they have certaingoals to be accomplished. Sometimes these goals are complex and may requireinteracting with multiple applications, which may focus on different domains. CurrentIAs may be of limited use in such cases and the user needs to directly managethe task at hand. An ideal personal agent would be able to learn, over time, thesetasks spanning different resources. In this paper, we address the problem of crossdomaintask assistance in the context of spoken dialog systems, and describe ourapproach about discovering such tasks and how IAs learn to talk to users about thetask being carried out. Specifically we investigate how to learn user activity patternsin a smartphone environment that span multiple apps and how to incorporate user’sdescriptions about their high-level intents into human-agent interaction