Research On DataMing Algorithm And Its Application In TCM Formula System

Traditional Chinese Medicine (TCM) is a medicine of Chinese, which has several thousands of year's history, and people have accumulated plenty of experience, as well as literature. The prosperity of Chinese proved that TCM has great live and value in its existence. Recently, the nation has high regard for the technological content of TCM, a mass of databases have been established. How to take good use of the data and how to quicken discovering new knowledge with the help of computer are the keystones of TCM research.Data mining and Knowledge discovery is the technology that can extraction of implicit, previously unknown, and potential useful information from data. Continue the information platform of Nanjing TCM research center; DMKD is applied to discovery pairs of medicine. Design and implementation of the TCM formula system, supporting the new medicine development researcher, and accelerating the development of TCM towards internationalization and modernization.The main contributions of the paper are following:1. The original TCM data is too atactic to data mining. Integrate with data reduction technology, clustering method and fuzzy set theory, a series of data preprocessing methods are promoted to standardize the patent data of TCM. Then provide appropriate data for data mining algorithm.2. According to the peculiarity of TCM data and our goal, improve the FP-growth algorithm, import multidimensional nodes in the same tree and promote construction algorithm of TCMFP hybrid-dimension tree. Take the fuzzy grade of membership as the support's value of medicine node, and add quantitative attribute to the rules.3. Promote TCMA algorithm to mine the maximal frequent itemsets in TCMFP tree. Set double-support for the medicine dimension, not only reduces the mining dataset but also make the rules more meaningful. Moreover, promote a whole new search strategy for the Maximal frequent itemset, omit the search of unnecessary nodes, as well as the construction of their conditional pattern base and conditional pattern tree. The algorithm accord with the actual meaning of TCM rules. Furthermore, it is faster than FP-growth algorithm. We keep the discovered TCM rules into database as a knowledge foundation.4. Design and implementation of the TCM formula system. First, design and...