The Knowledge Discovery Of Chinese Herbology’s Near-infared Spectroscopy Based On Attribute Partial Order Structure Diagram

Currently,the management system of Chinese herbology market is not that good and the quality of Chinese herbology is mixed. Thus, as a pivotal part in control of Chinese herbology quality, the analysis of Chinese herbology is particularly important. Near-infrared spectroscopy is a technique which overcomes traditional means of subjective experience and semi-empirical in this field. Although Chinese herbology’s near-infrared spectroscopy contains abundant information, the data is too large and hard to anayze. In this paper, on the research basis of knowledge discovery in many fields of attribute partial order structure diagram, an academic idea is put forward that attribute partial order structure diagram is applied to the knowledge discovery of Chinese herbology’s near-infrared spectroscopy. This idea is a new way of Chinese herbology’s knowledge discovery.In this paper, first, the theory of near-infrared spectroscopy and some related concepts of Formal Concept Analysis are introduced. Then, the paper gives some related definitions of attribute partial order structure and construct attribute partial order structure diagram of “Bodies of water” according to the algorithm principle from Formal Analysis home page and the superiority of this mathematical structure is analyzed. The discretization algorithms and principles of formal context are given as well.Next, the expression of near-infrared spectroscopy’s multi-dimensional radar chart and the feature selection of radar charts are introduced in detail. The feature of Iris dataset is extracted and analyzed.Finally, 28 samples of Ephedrae Herba spectral data is expressed in radar charts and the features of the radar charts are extracted. Then, the attribute is divided into several regions to obtain the formal context. And the corresponding attribute partial order structure diagram is generated for knowledge discovery and verified the feasibility of this method.The last part is experiment, which used the MCS600 serials of near infrared spectrometer made in Carl Zeiss, Germany. The experiment collected the spectrum data of 33 kinds of Chinese herbology. These medicine herbs contain cold, hot and warm ones from 2015 of Chinese Pharmacopoeia. Radar charts are used to express these data and then these charts’ feature are extracted to generate formal context. The corresponding attribute partial order structure diagrams are computed for the 33 samples’ knowledge discovery and prove that the method is scientific.