Multiple Channels Filter Max Pooling Full Connection Output ( Dropout + Softmax ) Conversation Utterance Labeled LDA Word 2 Vec Topic Matrix Word Embedding Matrix Domain Trigger Distribution

semanticscholar(2017)

引用 0|浏览10
暂无评分
摘要
Recently, conversational robots are widely used in mobile terminals as the virtual assistant or companion. The goals of prevalent conversational robots mainly focus on four categories, namely chitchat, task completion, question answering and recommendation. In this paper, we present a Chinese intelligent conversational robot, Benben, which is designed to achieve these goals in a unified architecture. Moreover, it also has some featured functions such as diet map, implicit feedback based conversation, interactive machine reading, news recommendation, etc. Since the release of Benben at June 6, 2016, there are 2,505 users (till Feb 22, 2017) and 11,107 complete humanrobot conversations, which totally contain 198,998 single turn conversation pairs.
更多
查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络