TL;DR本文设计了一个层次结构和支持多标签的对话行为注释方案 MIDAS,针对开放领域的人机对话。为了验证该方案的适用性,我们利用迁移学习方法训练多标签对话行为预测模型,并达到了 0.79 的 F1 分数。
Abstract
dialog act prediction is an essential language comprehension task for both
dialog system building and discourse analysis. Previous dialog act schemes,
such as SWBD-DAMSL, are designed for human-human conversations, in which
conversation partners have perfect language understanding abil