基于模式聚合和广义粒子群的网页文本属性约简  

Attribute reduction in webpage categorization using general particle swarm optimization and pattern aggregation

作　　者：童亚拉[1] 刘磊[1] 

机构地区：[1]湖北工业大学理学院,湖北武汉430068

出　　处：《计算机工程与设计》2009年第15期3670-3672,3675,共4页Computer Engineering and Design

基　　金：国家自然科学基金项目(60773009);湖北工业大学校基金项目(200601)

摘　　要：将模式聚合和自适应广义粒子群算法相结合,提出了一种文本属性约简新方法。设计交叉操作模拟粒子飞行速度的变化,变异操作保持种群的多样性,同时引入自适应策略动态调整变异概率,以求最优特征子集。在用自适应广义粒子群算法约简前,先用模式聚合理论对原始特征空间约简,得到中间特征子集,然后再用自适应广义粒子群算法继续约简,充分发挥两者的优势。实验结果表明,此算法能有效降低文本维数,提高分类精度。A new method of text dimension reduction is brought forward, based on pattern aggregation and adaptive general particle swarm optimization （AGPSO）. A crossover operator is designed to simulate flying velocity alteration and a mutation operator is used to keep population diversity. Besides these, an adaptive strategy is introduced to adjust probability of crossover and mutation just in order to obtain optimal feature set. Before applying APSO to text feature space, a middle attribute subset will be produced by using pattern aggregation theory to original feature space and then continuously reduce attributes by general PSO. Therefore, benefit of general PSO and pattern aggregation theory is adequately employed. The experimental results indicate that the algorithm can not only reduce dimension, but also imorove catezorization orecision.

关 键 词：网页分类 属性约简 广义粒子群 模式聚合 自适应策略 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]