Research On Technology Of Data Mining For Traditional Chinese Medicine Analysis

Traditional Chinese medicine(TCM)is a broad range of medicine practices sharing common concepts which have been developed in China and are based on a tradition of more than 2,000 years.As its main form,traditional Chinese prescription is formulated under the guidance of compatible regularity of the TCM.Therefore,it is shows a strong regularity,especially on same disease.Mining them using the data mining technology is very significant to the development of TCM.This paper gives a research on technology of data mining for TCM analysis.The major work and conclusions of present paper are as follows:1)Proposed a parallel co-evolution algorithm which helps for discover the relationship between herbs and symptoms.Co-evolution algorithm is a search heuristic that mimics the process of natural coevolution,where each party in a co-evolutionary relationship exerts selective pressures on the other,thereby affecting each other's evolution.we have studied and discussed the parallelization of the co-evolution algorithm using the technology of Map-Reduce.Experimental results show that the method can enhance the working speed.2)Proposed a missing value estimation method for traditional Chinese prescription records based on k-nearest neighbor algorithm,caused by the dissimilarity of measurement used in TCM records.3)Discovering the inconsistency knowledge between herbs using the technology of outlier detection on the basis of our preceding work.We adopted a outlier detection for high dimensional data at the characteristics of traditional Chinese prescription,which dividing the data space into grids and searching the most sparse one.Based on this,this paper proposed an improved outlier detection algorithm for high dimensional data aimed to avoiding double counting problem.Experimental results shows that the method can obvious enhance the working speed.