A New Focus Strategy for Efficient Dialog Management

来源 :第十五届全国计算语言学学术会议(CCL2016)暨第四届基于自然标注大数据的自然语言处理国际学术研讨会(NLP-NABD | 被引量 : 0次 | 上传用户:srsyzjks
下载到本地 , 更方便阅读
声明 : 本文档内容版权归属内容提供方 , 如果您对本文有版权争议 , 可与客服联系进行内容授权或下架
论文部分内容阅读
  The dialog manager is the most important component for a dialog system,in which the dialog state tracking is crucial to a real-world system.We claim that the intractability of dialog states comes from two aspects: the large slot size in user's goal and the large candidate value size for each slot.For the first time,we propose a new focus strategy to deal with the former problem,by reducing the full slots of the user's goal into a small subset focus slot.We also implement a partition-based method to deal with the latter problem.Then we combine both strategies to take advantage of their complement property.In our experiment of a real-world application in an image purchase domain,our proposed focus strategy is far faster than both the partition method and the na(i)ve algorithm with comparable quality.
其他文献
  In this paper,a novel image tag recommendation framework is developed by fusing the deep multimodal feature representation and cross-modal correlation minin
会议
  The problem of automatically labelling the appearances of characters in video with their names is challenging due to the huge variation in the appearance of
会议
  Neural network based Chinese Word Segmentation(CWS)approaches can bypass the burdensome feature engineering comparing with the conventional ones.All previou
会议
  Symptom entities are widely distributed in Chinese electronic medical records.Previous approaches on symptom entity extraction usually extract continuous st
会议
  The availability of labeled corpus is of great importance for emotion classification tasks.Because manual labeling is too time-consuming,hashtags have been
会议
  In this paper,we apply a bidirectional Long Short-Term Memory with a Conditional Random Field to the task of disfluency detection.Long-range dependencies is
会议
  The Chinese language is a character-based language,with no explicit separators between words like English.Traditionally,word segmentation is conducted to co
会议
  Large-scale annotated corpora are a prerequisite for developing high-performance age regression models.However,such annotated corpora are some-times very ex
会议
  This paper presents the state of art research progress on multilingual multi-document summarization.Our method utilizes hLDA(hierarchical Latent Dirichlet A
会议
  Previous researches on event relation classification primarily rely on lexical and syntactic features.In this paper,we use a Shallow Convolutional Neural Ne
会议