To discover the frequent item sets from the huge data sets, one of the most popular techniques of data mining, called association rule mining technique used. For generating association rules from huge database using association rule mining technique, Computer system takes too much. This can be enhanced, if the number of association rules generated using association rule mining technique from a huge dataset can be optimized. So here in this work, firstly association rules are generated using standard Apriori algorithm and then optimized these association rules using modified artificial bee colony (ABC) algorithm. In this modified ABC algorithm, one additional operator, called crossover operator, is used after the third phase, called scout bee phase, of ABC algorithm. Due to the better exploration property of crossover operator, it is used in this work. Experimental results show that the proposed schemes performance better than previously proposed schemes like K-Nearest Neighbor algorithm (KNN) and ABC algorithm. Keywordsbee colony algorithm, ABC, Crossover operator, Association rules, Support, Confidence, Frequent item sets