Thesaurus retrieval is fundamental in Chinese information processing.After a brief review of the current technique,this pa-per made a deep analysis to the design of Chinese thesaurus Hash function based on chain addre...Thesaurus retrieval is fundamental in Chinese information processing.After a brief review of the current technique,this pa-per made a deep analysis to the design of Chinese thesaurus Hash function based on chain address conflict dissolving method,and several criteria,as well as the theoretic expectation of these criteria,were proposed to evaluate different Hash functions.According these values,some experimental Hash functions were proposed which had high efficiency in our test.展开更多
海量电子病历(Electronic Medical Record,EMR)数据是支撑医疗智能化研究的重要原料,然而电子病历文本数据的半结构化甚至无结构化特点,造成后续对其分析利用的极大困难.虽然近年来基于深度学习的命名实体识别(Named Entity Recognition...海量电子病历(Electronic Medical Record,EMR)数据是支撑医疗智能化研究的重要原料,然而电子病历文本数据的半结构化甚至无结构化特点,造成后续对其分析利用的极大困难.虽然近年来基于深度学习的命名实体识别(Named Entity Recognition,NER)成为对电子病历进行自动化信息抽取的核心技术,但鉴于中文电子病历(Chinese Electronic Medical Record,CEMR)具有包括病历文本的非规范性与专业性、医疗实体的独特性和标注语料的稀缺性在内的独特文本数据特征,该研究目前仍存在诸多挑战.本文对中文电子病历命名实体识别的研究与进展进行了综述,系统梳理了命名实体识别的概念、相关理论模型以及制约中文电子病历命名实体识别准确率和识别效率的主要原因;从技术发展角度详细分析了中文电子病历命名实体识别方法的变革历程;并对中文电子病历命名实体识别效果做了实验验证与深入分析,指出了现有模型的不足与改进方向.鉴于国内近年来与中文信息学处理相关的测评会议CCKS持续关注中文电子病历命名实体识别,本文特别对CCKS在该领域五年来的全部代表性测评论文做了纵横对比分析,并通过在主流模型上的深入实验与研究,为后续该领域的继续推进寻求了思路.展开更多
文摘Thesaurus retrieval is fundamental in Chinese information processing.After a brief review of the current technique,this pa-per made a deep analysis to the design of Chinese thesaurus Hash function based on chain address conflict dissolving method,and several criteria,as well as the theoretic expectation of these criteria,were proposed to evaluate different Hash functions.According these values,some experimental Hash functions were proposed which had high efficiency in our test.
文摘海量电子病历(Electronic Medical Record,EMR)数据是支撑医疗智能化研究的重要原料,然而电子病历文本数据的半结构化甚至无结构化特点,造成后续对其分析利用的极大困难.虽然近年来基于深度学习的命名实体识别(Named Entity Recognition,NER)成为对电子病历进行自动化信息抽取的核心技术,但鉴于中文电子病历(Chinese Electronic Medical Record,CEMR)具有包括病历文本的非规范性与专业性、医疗实体的独特性和标注语料的稀缺性在内的独特文本数据特征,该研究目前仍存在诸多挑战.本文对中文电子病历命名实体识别的研究与进展进行了综述,系统梳理了命名实体识别的概念、相关理论模型以及制约中文电子病历命名实体识别准确率和识别效率的主要原因;从技术发展角度详细分析了中文电子病历命名实体识别方法的变革历程;并对中文电子病历命名实体识别效果做了实验验证与深入分析,指出了现有模型的不足与改进方向.鉴于国内近年来与中文信息学处理相关的测评会议CCKS持续关注中文电子病历命名实体识别,本文特别对CCKS在该领域五年来的全部代表性测评论文做了纵横对比分析,并通过在主流模型上的深入实验与研究,为后续该领域的继续推进寻求了思路.