Objective: To analyze and excavate the medication rule in herbal prescriptions for diabetes by using Traditional Chinese Medicine in Heritance Support System (TCMISS). Method: Herbal prescriptions for diabetes in recent 10 years were collected from papers

and Apriori algorithm was used to analyze correlation between herbs

which was based on improved mutual information method. And according to the relevant constraints and punishment degree

core combination of herbs was analyzed by complex system entropy clustering method based on the correlation analysis between drugs. New prescriptions were acquired by using unsupervised clustering algorithm. The network diagrams of medicine association rules and core combination herbs in the new prescriptions were drawn based on the analysis results. The above operations were completed besed on TCMISS V1.1. Result: The frequency of each herb and the association rules among the herbs were found out

16 core combinations and 8 new prescriptions were excavated from the database about the analysis of 110 medicine prescriptions of diabetes. Conclusion: The data mining method based on Apriori algorithm and cluster analysis is suitable for mining

analysis and summary of the medication rules in herbal prescriptions. Herbs of nourishing qi and yin

combined with herbs for heat-clearing and activating blood circulation in accordance with different types is important to improve the curative effect on diabetes.