摘要
为了提升对话系统的实体纠错能力和意图识别正确率,提出了一种基于实体模糊识别的高校心理咨询对话系统。通过实体的拼音、模糊拼音、音调和正确汉字数纠正用户输入中易错实体,并给出一种构建噪声数据的方法,以提高对话系统识别未知意图的能力。实验表明,所构建的系统在实体纠错、内存占用和响应速度上都较对比算法更优,且整体测评结果证明构建噪声数据方法的有效性。
In order to improve the entity error correction ability and the correct rate of intention recognition of the dialogue system,a college psychological counseling dialogue system based on entity fuzzy recognition is proposed.Correcting the error-prone entity in user input through the entity’s Pinyin,fuzzy Pinyin,tone and correct number of Chinese characters.And a method of constructing noise data is given to improve the ability of the dialogue system to identify unknown intent.Experiments show that the constructed system is better than the comparison algorithm in terms of entity error correction,memory occupation and speed of response,and the overall evaluation results prove the effectiveness of the method of constructing noise data.
作者
章亮
徐戈
陈芳
ZHANG Liang;XU Ge;CHEN Fang(Department of Laboratory and Equipment Management, Minjiang University, Fuzhou, Fujian 350108, China;College of Computer and Control Engineering, Minjiang University, Fuzhou, Fujian 350108, China;Student Affairs Department, Minjiang University, Fuzhou, Fujian 350108, China)
出处
《闽江学院学报》
2022年第2期33-42,共10页
Journal of Minjiang University
基金
福建省中青年教师教育科研项目(JAT200446)
中央引导地方项目“福建省心理健康人机交互技术研究中心”(2020L3024)
大数据分析系统国家工程实验室开放课题(CASNDST202006)
福建省科技厅引导性科技项目(2019H0026)。
关键词
对话系统
实体提取
模糊识别
心理咨询
dialogue system
entity extraction
fuzzy recognition
psychological counseling
