| 一种适合检索词推荐的K-means算法最佳聚类数确定方法 |
| |
| 引用本文： | 边鹏,赵妍,苏玉召.一种适合检索词推荐的K-means算法最佳聚类数确定方法[J].图书情报工作,2012,56(4):107-111. |
| |
| 作者姓名： | 边鹏 赵妍 苏玉召 |
| |
| 作者单位： | 1. 中国科学院文献情报中心;2. 郑州航空工业管理学院计算机科学与应用系; |
| |
| 摘 要： | 从嵌入式NSTL文本推荐系统的检索词推荐需求入手，分析原有方法的不足，引入共词分析方法和分化理论，提出一种新的最佳聚类数确定方法，改进最小类间距离和平均类内距离的计算方法，强化聚类结果的推荐效果，同时使推荐效果可以随着样本数据的变化而动态调整。最后，运用实验验证该方法的有效性。
|
| 关 键 词： | K-means聚类 聚类数 文本聚类 个性化推荐 |
| 收稿时间： | 2011-08-26 |
| 修稿时间： | 2011-10-26 |
An Improved Method of Determining Optimal Number of Clusters in K-means Clustering Algorithm for Dynamic Text Recommending |
| |
| Bian Peng,Zhao Yan,Su Yuzhao.An Improved Method of Determining Optimal Number of Clusters in K-means Clustering Algorithm for Dynamic Text Recommending[J].Library and Information Service,2012,56(4):107-111. |
| |
| Authors: | Bian Peng Zhao Yan Su Yuzhao |
| |
| Institution: | 1. Center for Documentation and Information, Chinese Academy of Sciences,;2. Computer Science and Application Department, Zhengzhou Institute of Aeronautical Industry Management,;3. Center for Documentation and Information, Chinese Academy of Sciences,; |
| |
| Abstract: | The authors analyze the shortage of the original method,based on the text recommending requirement from the embedded NSTL recommending system.Then the authors raise a novel algorithm to determine optimal number of clusters,and optimize the calculation of the minimal distance between clusters and the average distance within one cluster,introducing the co-word analysis method and the differentiation theory.The improved algorithm enforces the effect of the text recommending in the cluster result.Moreover it can adjust the recommending effect according to the change of the amount of sampling data.At last,the lab result shows it is effective. |
| |
| Keywords: | K-means clustering cluster number text clustering personalized recommending |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
| 点击此处可从《图书情报工作》浏览原始摘要信息 |
| 点击此处可从《图书情报工作》下载免费的PDF全文 |
