基于LSA与PCA的常见滋补中药NIR聚类

NIR Clustering of Common Tonic Medicine Based on LSA and PCA

潜在语义分析在信息检索领域应用较多,但在近红外光谱领域应用较少。利用近红外漫反射光谱技术,结合潜在语义分析(LSA)和主成分分析(PCA),比较了不同预处理方法、不同奇异值和主成分个数对所建模型的影响,最后确定的模型校正集误判数分别为4和3。用建立的校正模型对验证集进行验证,总的识别率分别达到了96.00%和96.50%。对于功效较近、难以聚类的滋补中药,潜在语义分析是一种新的有效的方法。

Latent semantic analysis （LSA）is widely applied in the field of information retrieval but rarely used in the field of near infrared spectroscopy. The effects of different pretreatment methods, singular values and number of principal components on the proposed model were compared using near-infrared diffuse reflectance spectroscopy technology,combined with latent semantic analysis and traditional principal component analysis （PCA） respectively. And the overall misclassification number of calibration sets is 4 and 3. Established calibration model was used to verify the validation set,and the overall recognition rate reached 96.00% and 96.50%. As for the effect close and difficult clustering Chinese herbal tonic,latent semantic analysis is a new and effective method.