基于微粒群优化算法的文本模糊聚类方法

针对模糊C-均值算法(FCM)具有局部最优问题和初值敏感性的缺陷,将微粒群优化算法应用于文本模糊聚类, 提出了基于微粒群优化算法的模糊C-均值算法PFCM．该算法首先采用实数编码方式对聚类原型进行编码,利用微粒群优化算法的全局搜索性能对初始聚类原型的选取进行指导,然后利用模糊C-均值算法进行聚类．使用算法PFCM对文本集合进行聚类实验,并用目标函数值和划分系数来判断模糊划分的效果,实验结果表明,与FCM相比,该算法具有较好的全局收敛性和较好的聚类结果．

Document Fuzzy Clustering Algorithm based on Particle Swarm Optimization Algorithm

This paper analyzes local optimality and initialization dependence disadvantage of Fuzzy C-Means(FCM) and proposes an algorithm (PFCM) for document fuzzy clustering based on particle swarm optimization algorithm. Algorithm PFCM adopts real code for clustering prototype. Global searching of particle swarm optimization is used to instruct to choose clustering prototype and then clustering analysis is processed by FCM. This algorithm is used to conduct a clustering experiment on a document set. The quality of fuzzy partition is evaluated by objective function value and partition coefficient. The experimental results show that this algorithm can not only avoid local optima but also obtain better clustering result than FCM.