The Application of Hybrid Ant Colony Algorithm in Association Rule Mining

Abstract

For the problem that recommendation result is less accurate, this paper designs the process of the hybrid ant colony particle swarm algorithm(ACO-MPSO) in detail. Using the results of PSO algorithm to determine the initial phenomenon of colony algorithm to avoid the blindness, and importing Metropolis mechanism to avoid the premature phenomenon. Choosing quality of strong association rules as evaluation indicators, and then compares ACO-MPSO algorithm with classical Apriori algorithm and hybrid simulated annealing particle swarm algorithm in association rule mining issues. Experimental results show that ACO-MPSO algorithm can dig out strong association rules having better quality. Using association rule mining method based on ACO-MPSO to solve the problem of MovieLens's personalized recommendation, and regards the hit rate as evaluation index of accuracy rate, and then compares ACO-PSO algorithm with Apriori algorithm, hybrid simulated annealing particle swarm algorithm. Experimental results shows that association rule mining method based on hybrid ant colony particle swarm algorithm has higher accuracy.