The Research About Data Mining Patterns And Algorithms On TCM Prescription Databases

TCM (Traditional Chinese Medicine) is the excellent culture of China and is the results of accumulating science. TCM play an important role in people's life and health. Science of prescriptions is one of the important basic courses in TCM. Prescription is an important part of theory, method, prescription and drug in TCM. The pricinples of forming a prescription have rich contents of science.During thousands of years, people have accumulated a large number of prescriptions by numerous clinical practises and TCM theory researches. These prescriptions were included in TCM ancient books, magazines and current clinical research journals. The college of computer science of Zhejiang University's CCNT laboratory cooperated with China Academy of Chinese Medical Science has created a lot of prescription databases since these years. One of the big problems in TCM is how to make full use of these database resources. Data mining is defined as the process of discovering significant and potentially useful patterns in large volume of data. Therefore, dataming is the proper technique that is used to analyse these prescriptions. Thus, in this paper, we use the dataming techniques to extract useful patterns from prescription databases in order that we can provide useful patterns for prescription researches and clinical practise researches.One of our main works in this paper is to propose many new kinds of mining patterns for precrisption data mining and develop algorithms for discovering these patterns. The other main works is to intergrate all algorithms proposed in this paper and develop prescription data mining systems.Frequent associated patterns can reflect association relationships between items. However, like frequent pattern mining, when the minimum association threshold is low, frequent associated pattern mining still generates a large number of patterns. It is difficult for human to analyse such a large number of patterns. Therefore, we mine maximal frequent associated patterns. Experimental results show that maximal frequent associated pattern mining can dramatically decrease...