The associational rule refers to the sets of attribute-values,which frequently appeared in data set recognition,also named as frequent item-sets.using these frequent sets Description association relation rules process is set up.Based on the analysis of the classical algorithm for mining association rules,we summarize classical Apriori algorithm of association rules,put forward improved Apriori algorithm of association rules,whose performance was tested by testing learning data sets.The results demonstrate that the improved Apriori algorithm of association rules works better in both running speed and mining capability.