基于LDA主题模型的中医药方剂相似度计算

评价中医药方剂的相似度的常用方法是基于方剂的功效和主治功能进行相似度分析，但存在相似度无法定量化的问题，并且没有考虑中医方剂组成成分的影响。提出了利用LDA主题模型发掘“方剂-证型-组成成分”的隐含关系的方法，将“方剂-组成成分”转换成“方剂-证型”和“证型-组成成分”两个概率分布，并利用KL距离来计算相似度。实验结果表明基于LDA主题模型的方法能够更好地计算方剂间的相似度，并且能较好地反映中医辩证论治。...

Full description

Saved in:
Bibliographic Details
Published in计算机应用研究 Vol. 34; no. 6; pp. 1668 - 1670
Main Author 朱志鹏 杜建强 刘英锋 喻芳 罗计根
Format Journal Article
LanguageChinese
Published 江西中医药大学计算机学院,南昌,330004%江西中医药大学岐黄医学院,南昌,330004 2017
Subjects
LDA主题模型
中医药信息
古布斯采样
方剂相似度
LDA主题模型
prescription similarity
LDA topic model
TCM information
古布斯采样
方剂相似度
Gibbs sampling
中医药信息
Online AccessGet full text
ISSN1001-3695
DOI10.3969/j.issn.1001-3695.2017.06.015

Cover

More Information
Summary:评价中医药方剂的相似度的常用方法是基于方剂的功效和主治功能进行相似度分析，但存在相似度无法定量化的问题，并且没有考虑中医方剂组成成分的影响。提出了利用LDA主题模型发掘“方剂-证型-组成成分”的隐含关系的方法，将“方剂-组成成分”转换成“方剂-证型”和“证型-组成成分”两个概率分布，并利用KL距离来计算相似度。实验结果表明基于LDA主题模型的方法能够更好地计算方剂间的相似度，并且能较好地反映中医辩证论治。
Bibliography:Zhu Zhipeng1, Du Jianqiang1, Liu Yingfeng2, Yu Fang1, Luo Jigen1 （ 1. School of Computer Science, 2. Qihnang Medical College, Jiangxi University of Traditional Chinese Medicine, Nanchang 330004, China）
51-1196/TP
Evaluation of TCM （traditional Chinese Medicine）prescription similarity mainly used the efficacy of prescriptions and indications function for similarity analysis, but the similarity was not quantitative and did not consider TCM prescription composition and implication relation of syndrome. This paper proposed LDA topic model to explore ＂prescription-syndrome typecomposition＂ implicit relations ,transformed the ＂prescription-composition＂ into two probability distribution of ＂prescriptionyndrome type＂ and ＂syndrome type-composition＂, and used the KL divergence distance to calculate the similarity. In the TCM prescription data analysis and processing, the experimental results show that TCM prescription based on LDA topic model can be better to calculate the similarity, and can be a good reflection of the
ISSN:1001-3695
DOI:10.3969/j.issn.1001-3695.2017.06.015