| The development of Chemistry, Instrument Science and Information Science spurs accumulation of chemical data. Such data contains lots of chemical knowledge and information, if them can be extracted properly, chemical science will be propelled to more bright future. Being analysts we must confront this opportunity. Therefore, the aim of the thesis is to develop new methods in chemical information extracting and apply the appropriate matching methods to analyze the complex herbal medicines systems. There are three main departs of this paper: Datamining in Mass databases, comparison for matching algorithms in complex herbal medicine system and research in herbal medicine fingerprint technology.1. Datamining in Mass databases(chapter2): Aim of this study is to predict the presence or absence of the certain substructure by learning from Mass spectra library. In section 1, a new try was applied to deal with information in substructure prediction by adjusting block variables and CCA (Canonical Correlation Analysis) thought from QSAR/QSPR research to Mass classification, such special variable selection algorithm showed a good performance in saturated alcohol and ether classification, served a reference for development of Mass classification and prediction. In section 2, a new classifier is devised bythought of drawing information from Mass fragments. It is different from traditional classifiers. Such new classifier gives substructure classification a good result. It is a supplement to traditional classifiers. Section 3 is statistical results of all alcohol and ether in NIST62 Mass Database, those results reflect some relationship between Mass peak and substructure, furthermore in this section some rules of Mass spectral are proved by Bayes theory.2. Comparison for matching algorithms in complex herbal medicine system: for complex analytical system, such as Chinese herbal medicine, chemical components are always contaminated by others in hyphenated Mass detection. Traditional matching algorithms for high purity component are not fit for such complex system, so results of matching are always not reliable. Many methods from both Mass and Chemometrics fields have been developed to solve this problem. In chapter 3, several classical methods are compared for herbal complex system. Matching results show that combination of those methods is more convincible than relying on single one.3. Fingerprint technology research in Chinese herbal medicine (chapter 4): modernization of Chinese medicine mostly relies on quality control, and fingerprint technology is a powerful tool to nature product's quality control. In mis chapter Volatile oil of 11 Gaoben (Ligusticum sinense Oliv) samples was tested by GC and GC-MS experiments to setup fingerprint analysis model. Conditions of GC fingerprint experiments are confirmed and some main ingredients are identified by GCMS. During similarity computation the thought of total chromatography computation and weighting computation were all implemented for more thoroughly analysis. Similarity and PCA analysis all distinguished inferior and confusable samples from normal ones.
|
PDF Full Text Request