基于小生境混合遗传算法的文本特征词聚类研究  

Research of Text Feature Words Clustering Based on Niche Hybrid Genetic Algorithm

作　　者：戴文华[1] 何婷婷[1] 焦翠珍[2] 

机构地区：[1]华中师范大学计算机科学系,武汉430079 [2]咸宁学院计算机系,咸宁437000

出　　处：《计算机科学》2008年第1期202-203,223,共3页Computer Science

基　　金：咸宁学院科研重点项目(No.KZ0637);国家自然科学基金(No.60442005,No.60673040);国家社会科学基金(No.06BYY029);教育部科学技术研究重点项目(No.105117)基金资助

摘　　要：提出一种基于小生境混合遗传算法的文本特征词聚类方法。该方法首先采用贝叶斯语义模型对语料库进行统计分析,并以K-L距离度量特征词间的距离,然后将小生境遗传算法与K-Means算法相结合,对文本特征词进行聚类,为文本特征词聚类提供了较高的效率和精确度。实验表明该方法是一种高效可行的文本特征词聚类方法。Combined with the global optimization ability of Niche Genetic Algorithm and the efficiency of K-Means Algorithm, a new Text Feature Words Clustering method based on Niche Hybrid Genetic Algorithm is proposed. This method first uses the Bayesian Semantics Model to carry out statistical analysis on the corpus. K-L distance is used for measuring distances between feature words. Using this method, we can provide a higher efficiency and precision for Text Feature Words Clustering. Experiments indicate that Niche Hybrid Genetic Algorithm is an effective and feasible method for Text Feature Words Clustering.

关 键 词：小生境 遗传算法 K—L距离 K-MEANS聚类 特征词聚类 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]