文中结合遗传算法和粒子群优化算法各自的优势,采用协同进化的思想,同时应用两种算法来遍历两个种群,并引入它们的信息交互机制。最后,实验和应用证明,在可接受的时间复杂度的前提下,协同进化算法不但能继承传统遗传算法的优越性,有效地减少扫描数据库的次数,和产生小规模的候选项目集;而且通过比较协同进化算法,传统的遗传算法和粒子群优化算法的属性,在关联规则挖掘中使用该算法,能避免早熟的现象。采取协同进化算法时可以发现高品质的关联规则,尤其是在高维数据库中。
This paper adopts a co-evolution algorithm, which utilizes improved genetic algorithm and particle swarm optimiza-tion algorithm to iterate two populations simultaneously. Meanwhile, the mechanism of information interaction between these two populations is introduced. Finally, experiments and application have been made to prove that on the premise of acceptable time complexity, not only does the co-evolution algorithm inherit the superiority of traditional genetic algorithm such as reduc-ing the number of scanning the database effectively and generating small-scale candidate item sets, but also avoid the phenome-non of premature through comparing the properties of co-evolution algorithm, traditional genetic algorithm and particle swarm optimization algorithm when used in association rules mining. High quality association rules can be found when adopted the co-evolution algorithm, especially in high-dimension database.