CAJ | 학술논문

案例聚类是按照案例库中案例的相似度进行归类，目的是减少案例推理系统搜索相似案例的时间、提高案例推理系统的性能和降低案例库维护的复杂度。该问题的难度在于案例库的案例规模比较大和不同的聚类算法的选择对于聚类结果的影响。文中在粒子群算法与细菌觅食算法基础上，将两者结合起来，综合两个算法的优点，并将其应用在k-pro-totypes方法上对案例库中案例进行聚类。与流行的聚类算法进行比较，实验结果显示文中的算法具有更高的效率并且性能相对而言更加优秀。
안례취류시안조안례고중안례적상사도진행귀류，목적시감소안례추리계통수색상사안례적시간、제고안례추리계통적성능화강저안례고유호적복잡도。해문제적난도재우안례고적안례규모비교대화불동적취류산법적선택대우취류결과적영향。문중재입자군산법여세균멱식산법기출상，장량자결합기래，종합량개산법적우점，병장기응용재k-pro-totypes방법상대안례고중안례진행취류。여류행적취류산법진행비교，실험결과현시문중적산법구유경고적효솔병차성능상대이언경가우수。
Case clustering is classified by the similarity to cases in case-base,the object is to reduce the time for searching similar case, improve the performance of case-base system and reduce the complexity of maintaining the case-base. The difficulty problem lies in that the size of case base is very large,and the clustering results is influenced by the choice of the clustering algorithm. In this paper,combined the advantages of particle swarm algorithm and bacterial foraging algorithm,use in case clustering with k-prototypes. Compared with pop-ular clustering algorithm show that this algorithm is efficient,has better performance.