面向特定任务的对话主题建模
首发时间:2015-12-07
摘要:提出用分类模型和序列标注模型来解决对话主题建模问题。分析和研究了面向特定任务的对话的特点,并在此基础上考察了采用特征降维和引入对话历史特征在两种模型上对主题建模的影响。实验结果表明,在两种模型上特征降维和引入对话历史特征都可以提高对话主题建模的效果。
For information in English, please click here
Topic modeling for task-specific dialogue
Abstract:In this paper, classification model and sequence labeling model are proposed to solve dialogue topic modeling. Based on the characteristic of task-specific dialogue, feature dimension reduction and dialogue history are applied to both of the above models to improve topic modeling. Results of experiments shows, feature dimension reduction and dialogue history works in both classification model and label sequence labeling model.
Keywords: computer application technology topic modeling dialogue history feature dimension reduction
基金:
论文图表:
引用
No.4667362111911214****
同行评议
勘误表
面向特定任务的对话主题建模
评论
全部评论0/1000