Data Mining's Research And Application In TCM Patent Data Set

Traditional Chinese medice (TCM) is the heritage culture. It has long hisroty of thousands of years and accumulates a great deal of data. These data are applied in providing simple search and statistics as always. The information hidden in these data could not be made full use of. With the tendency towards nature, natural medicine's development will be taken more and more account of, which provides great opportunity to the TCM's development. In order to increase the technological content of TCM, it is a key to discovery valuable information from TCM database.This paper presents the realizing process of KDD system on the TCM patent information platform. This process includes three modules: prescription information's data preprocessing (PIDP), medical compatibility regulation's finding and assisted deciding.On the basis of the characteristic of the TCM patent, we adopt the process of PIDP to standardizing the prescription data: 1 long text is dealt with by the approach of orthotropic dispartment, so the single herb can be stored separately; 2 using the TCM's dictionary, the problem of herb's alias is tackled rightly; 3 fuzzy set is applied to depicting the medical dose.This paper makes progress on the FTDA algorithm proposed by T.P.Hong, and details the FTDA2 algorithm. FTDA2 only pays attention to the items, which can contribute to the support's value, thus degrades the time complexity of exacting association rules from transaction database, and draws the medical compatibility regularity from standardized prescription data. On this basis, we build the pattern and rule's library.Using the library, the system brings the assisted deciding module into effect. The medical researchers can obtain reference in the process of exploring new compound medical.Innovations of this paper:1 We design the algorithm FTDA2 to discover fuzzy association rules from TCM data containing quantitative attribute. Compared to the FTDA, the ratio ofrunning time is when computing the set of k-itemset, therebyeffectively degraded the time complexity.2 Aimed to the need of disposal on prescription's long text, we design a preprocessing process -PIDP, appropriate to TCM data's feature, and thus standardize the original TCM data. We are separating the prescription data first, and adopt different tactics to different attributes: different names of the same kind of traditional Chinese medicine is replaced with the same name, fuzzy set is in place of TCM's dose.3 This system applies DM to the process of exploring new compound medicine deeply, and implements the defects of the current system. This paper introduces two major functions of the KDD system mainly: First, offer the reference about crude drugs and dose to research and development of the new medicine; Second, according to the crude drugs that allowed in the prescription, this system can analyse the other treatment function of the prescription.