Objective: To explore the rules among external therapy and food therapy of traditional Chinese medicine (TCM) in treating diabetes by applying text mining technology. Method: The relative literatures were collected from Sinomed database (CBM)

and an Access database was established. The data was processed by SQL using frequency statistical method based on sensitive keywords hierarchical algorithm

and noise was manually removed according to the medical professional knowledge. The rules among external therapy and food therapy of TCM were deeply analysed and comprehensively displayed by the frequency of one-dimensional and two-dimensional network diagram. Result: The results showed that the most common external therapies of diabetes were acupuncture (frequency 2 079)

phototherapy (frequency 1 237)

external application with TCM (frequency 348)

acupiont injection (frequency 327) and massage (frequency 207) in turn. The most common food therapies were followed by panax notoginseng (frequency 446)

Chinese yam (frequency 262)

lemon (frequency 234) and wolfberry (frequency 143). Conclusion: Text mining

together with artificial reading for anti-noising

is an useful approach in exploring the clinical rules of TCM in treating diabetes. However

its exact mechanism and certain indication in treating diabetes is worthy of further study.