中央研究院 資訊科學研究所

Natural language dialogue is an important and intuitive way for people access information and services. However, current dialogue systems are limited in scope, brittle to the richness of natural language, and expensive to produce. In this talk, we will introduce Chorus and Guardian, two crowd-powered dialogue systems. When using Chorus, end users converse continuously with what appears to be a single conversational partner. Behind the scenes, Chorus leverages multiple crowd workers to propose and vote on responses. A shared memory space helps the dynamic crowd workforce maintain consistency, and a game-theoretic incentive mechanism helps to balance their efforts between proposing and voting. Chorus demonstrates a new future in which conversational assistants are made usable in the real world by combining human and machine intelligence, and may enable a useful new way of interacting with the crowds powering other systems; Guardian is a system that wraps existing Web APIs into immediately usable dialogue systems. Guardian takes as input the Web API and desired task, and the crowd determines the parameters necessary to complete it, determines how to ask for those parameters, and interprets the responses from the API. The systems is structured so that over time, computation could learn to take over from the crowd. Guardian represents a hybrid system that may make dialogue systems both more general and robust going forward.

 

Ting-Hao (Kenneth) Huang is currently a PhD student in Language Technologies Institute (LTI) of Carnegie Mellon University (CMU). His main research interests are in crowd-powered systems and natural language processing. Kenneth obtained his B.S. degree in computer science and B.A. in Chinese literature from National Taiwan University (NTU) in 2007. He then worked with Prof. Hsin-Hsi Chen on sentiment analysis and received his M.S. in computer science from NTU. Later in 2011, he started his study in LTI, CMU with Prof. Teruko Mitamura on metaphor detection and interpretation. After receiving his second M.S. in computer science, Kenneth developed his research interests toward the collaboration between language technologies and crowd intelligence. He is now working with Prof. Jeffery Bigham in Human-Computer Interaction Institute to explore the possibility of combining language technology into crowd-powered systems.