Turn Taking versus Discourse Structure
Turn Taking versus Discourse Structure
复制标题
轮流与话语结构
DOI：
--
复制
发表时间：
1999
期刊：
影响因子：
--
通讯作者：
Scott Prevost
中科院分区：
文献类型：
--
作者：
Justine Cassell;Obed E. Torres;Scott Prevost
图灵测试一直被认为是测试计算机对对话内容的贡献。也就是说，从输入的输出中，我们应该尝试判断文本是由人类还是计算机生成的。语音技术的最新进展使我们想到了通过电话进行图灵测试。那面对面的图灵测试呢一台计算机必须表现出什么样的行为才能让我们相信它不是一个灰色的盒子，而是一个活生生的、会呼吸的身体？也许我们今天还没有准备好迎接这样的竞争，但总有一天我们会准备好的。本文试图将人机对话的领域向这个方向移动，即向与计算机进行具体对话的方向移动。在其他工作中（Cassell等人，1994; Cassell and Thorisson，in press），我们集中在手势，语调，头部运动和凝视。目前的工作重新审视凝视的问题，并试图调和两个相互竞争的方法来产生非语言行为的一般问题。
The Turing test has always been conceived of as a test of the content of a computer’s contribution to a conversation. That is, from typed output, we are supposed to try to tell whether text was generated by a human or a computer. Recent advances in speech technology have led us to conceive of a Turing test taken over the phone. What about a face-to-face Turing test? What kinds of behaviors would a computer have to exhibit to convince us that it was not a grey box but a living, breathing body? We are perhaps not ready today for such a competition, but7 we may be one day. This paper attempts to move the field of human-computer conversation in that direction—in the direction of embodied dialogue with computers. In other work (Cassell et al., 1994; Cassell and Thorisson, in press), we have concentrated on hand gestures, intonation, head movement, and gaze. The current work revisits the question of gaze and attempts to reconcile two competing approaches to the general question of generating nonverbal behaviors.