基于和声搜索机制的特征选择与文本聚类分析  

Feature selection and text clustering analysis based on harmony search

作　　者：王永刚[1] 李靖[2] 王文慧[3] 曹传剑[4] 王晓燕[1] WANG Yong-gang;LI Jing;WANG Wen-hui;CAO Chuan-jian;WANG Xiao-yan(College of General Education,Qingdao Huanghai University,Qingdao 266427,China;School of Data Science,Qingdao Huanghai University,Qingdao 266427,China;Teaching Department,Qingdao Huanghai University,Qingdao 266427,China;School of Intelligent Manufacturing,Qingdao Huanghai University,Qingdao 266427,China)

机构地区：[1]青岛黄海学院通识教育学院,山东青岛266427 [2]青岛黄海学院大数据学院,山东青岛266427 [3]青岛黄海学院教学工作部,山东青岛266427 [4]青岛黄海学院智能制造学院,山东青岛266427

出　　处：《计算机工程与设计》2022年第2期472-478,共7页Computer Engineering and Design

基　　金：山东省高等学校青创人才引育计划建设团队基金项目(201901)。

摘　　要：针对文本信息特征冗余多、噪声大问题,提出基于和声搜索机制的文本特征选择算法。以词频逆文本频率指数为目标函数评估特征词条;在初始文档集中通过和声搜索的记忆考虑、纵向倾角调整和随机选择3种特征选择新解更新规则,迭代搜索最优特征子集;以最优特征子集为基础,以K均值进行文本聚类。利用4种典型文档数据集进行仿真实验,实验结果表明,该算法可以有效降低文本特征维度,聚类准确率更高。Aiming at the problems of the redundancy and the big noise of features in text information,a text feature selection algorithm based on harmony search mechanism was proposed.The term frequency-inverse document frequency was used as an objective function to evaluate each text feature at the level of the document,and the original dataset was taken to obtain a new optimal feature subset by three update rules of new solutions,including the memory consideration,the longitudinal angle adjustment and the random selection in harmony search.Based on the optimal feature subset,K-mean was used to make text clustering.Simulation experiments were carried out using four typical text datasets on clustering test.The results show that,the proposed algorithm not only can effectively reduce the text feature dimension,but has higher accuracy of text clustering.

关 键 词：特征选择 文本聚类 和声搜索机制 K均值文本聚类 特征子集 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]