CASIA OpenIR  > 毕业生  > 博士学位论文
面向医疗场景的对话理解与生成方法研究
李梅
2022-06-19
Pages124
Subtype博士
Abstract

人机对话系统是指人类用户通过自然语言与计算机进行对话的交互系统。整 个系统通常由多个模块组成,其中对话理解和生成是人机对话系统中关键的组 成部分。近年来,随着人工智能技术的蓬勃发展,对话理解与生成技术的研究取 得了较为丰硕的成果。然而,已有的工作主要关注常规领域的对话理解和生成方 法,对医疗领域关注较少。医疗领域的对话理解任务面临着对话表达口语化、信 息冗余等诸多问题。此外,在医疗领域中,已有的情感陪护系统大多只适用于处 理简单的情感类别,缺乏对细粒度情感的建模。针对上述不足,本文围绕医疗领 域对话文本的建模、上下文选择和信息融合方法展开,旨在改善已有方法的上下 文利用不足的缺陷,使模型能够更好地对上下文进行建模和利用,实现对话系统 性能的提升。

Other Abstract

Human-machine dialogue systems use natural language to communicate with humans. Typically, it consists of multiple modules, of which the  understanding and generation modules are key components. In recent years, there has been significant progress in the dialogue understanding and generation tasks. However, most existing work mainly focuses on conventional fields and pays less attention to the medical field. Dialogue understanding and generation in the medical field still face several challenges, including conversational characteristics like colloquialism, redundancy, and so on. Fur- thermore, most of the existing dialogue generation models of emotional chat systems are only suitable for dealing with simple emotion categories and lack fine-grained emotion modeling. To address the above problems, this dissertation focuses on the approaches of modeling, selecting, and incorporating information into dialogue systems, so as to alleviate the deficiencies of context utilization in existing methods and make better use of contextual information to improve the quality of the dialogue system.

Keyword自然语言处理
Subject Area人工智能 ; 自然语言处理
MOST Discipline Catalogue工学 ; 工学::计算机科学与技术(可授工学、理学学位)
DOIno
URL查看原文
Language中文
Citation statistics
Document Type学位论文
Identifierhttp://ir.ia.ac.cn/handle/173211/48945
Collection毕业生_博士学位论文
Recommended Citation
GB/T 7714
李梅. 面向医疗场景的对话理解与生成方法研究[D]. 中国科学院自动化研究所. 中国科学院自动化研究所,2022.
Files in This Item:
File Name/Size DocType Version Access License
李梅-面向医疗场景的对话理解与生成方法研(4843KB)学位论文 限制开放CC BY-NC-SA
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[李梅]'s Articles
Baidu academic
Similar articles in Baidu academic
[李梅]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[李梅]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.