Objective To explore the feasibility of application of data mining in the study on selection and specificity of acupoints.Methods The ancient and modern prescription data of acupuncture treating migraine and functional dyspepsia were regarded as sample data. Database techniques and natural language processing techniques were introduced to structurize and standardize acupuncture prescription data. Based on association rule data mining

the self-developed Acupuncture Data Mining System was applied to discover the rules of selection of acupoints.Results There were collected 634 and 386 prescriptions of acupuncture treating migraine and functional dyspepsia respectively. In the prescriptions treating migraine

acupoints distributed on Gallbladder meridian and Triple Energizer meridian accounted for 40.5% and 50.6% in ancient time and modern time respectively

and specific points accounted for 84.0% and 82.8% in ancient time and modern time respectively. In the prescriptions treating functional dyspepsia

acupoints distributed on Conception vessel

Bladder meridian and Stomach meridian accounted for 62.1% and 75.7% in ancient time and modern time respectively

and specific points accounted for 70.0% and 83.3% in ancient time and modern time respectively.Conclusion The result of data mining proves that the rule of selection of acupoints is "selecting acupoints along the affected meridian" and "mostly selecting specific points"

and the specificity of acupoints is "where the meridian goes

where the qi accumulates".