基于深度学习和情绪识别的智能聊天机器人设计
Intelligent Chatbot Design Based on Deep Learning and Emotion Edentification
徐羽 1张俊杰 1王曦雨 1吴海涛1
作者信息
- 1. 南京工程学院信息与通信工程学院,江苏 南京 211167
- 折叠
摘要
目前聊天机器人在对话过程中存在单一匹配传统知识库生成回复语句,多使用情感通用性回复语句以及情感单一等问题.针对这些问题,首先对传统的编码器-解码器模型进行了研究,在原有模型的基础上使用深度LSTM结构并且加入注意力机制,提高了其适应不同长度的问句的能力.其次,借鉴CNN在图像处理方面的研究,改进CNN的一般架构,从而提高聊天机器人对中文文本的情绪识别能力.最终提出了一个新的聊天机器人模型,即能够准确识别人类情绪,以陪伴、共情的方式与人类进行自然语言交流的机器人.
Abstract
At present,chatbots have the problems of single matching traditional knowledge base to generate reply statements,multiple use of emotionally generalized reply statements,and single emotion in the conversation process.In or-der to solve these problems,this paper studies the traditional encoder-decoder model and uses the deep LSTM structure combined with the attention mechanism on the basis of the original model,which improves its ability to adapt to questions of different lengths.Secondly,by drawing on the research of CNN in image processing,the general structure of CNN is improved to enhance chatbot's ability of recognizing the emotion of Chinese text.Finally,a new chatbot model is proposed,that is,a robot that can accurately identify human emotions and communicate with humans in natural language in a com-panionable and empathetic way.
关键词
深度学习/聊天机器人/自然语言处理/情绪识别Key words
deep learning/chatbot/natural language processing/emotion recognition引用本文复制引用
出版年
2024
NETL
NSTL
万方数据