摘要
研究了中药色谱指纹图谱相似度的评价方法,提出了改良的程度相似度的计算方法。以模拟数据和实验数据研究了相关系数、夹角余弦和改良程度相似度的优劣,发现相关系数和夹角余弦对数据的差异不够敏感,经预处理之后仍然不灵敏;采用改良的程度相似度可以反映数据的差异,因此可以将其用于评价中药色谱指纹图谱共有峰的相似度。
The similarity evaluation of the fingerprints is one of the most important problems in the quality control of the traditional Chinese medicine (TCM). Similarity measures used to (evaluate) the similarity of the common peaks in the chromatogram of TCM have been discussed. Comparative studies were carried out among correlation coefficient, cosine of the angle and an improved extent similarity method using simulated data and experimental data. Correlation (coefficient) and cosine of the angle are not sensitive to the differences of the data set. They are still not sensitive to the differences of the data even after normalization. According to the similarity system theory, an improved extent similarity method was proposed. The improved extent similarity is more sensitive to the differences of the data sets than correlation coefficient and cosine of the angle. And the character of the data sets needs not to be changed compared with log-transformation. The improved extent similarity can be used to evaluate the similarity of the chromatographic fingerprints of TCM.
出处
《色谱》
CAS
CSCD
北大核心
2005年第2期158-163,共6页
Chinese Journal of Chromatography
关键词
指纹图谱
相似系统理论
程度相似度
相关系数
中药
fingerprint
similarity system theory
extent similarity
correlation coefficient
(traditional) Chinese medicine
