基于免疫克隆算法的蒙古文形近元素聚类

Mongolian Similar Elements Clustering via Immune Clone Algorithm

中文关键词:  免疫克隆算法  蒙古文形近元素聚类  $K$-均值

英文关键词:immune clone algorithm  Mongolian similar elements clustering  K-means

中文摘要:

      文本聚类是聚类技术的重要研究领域.该技术根据文本的相似特征或相似表达式对文本进行聚类,使得属于同类的文本具有最大的相似性,而属不同类文本具有最大的差异性.与其它文字相比,蒙古文的结构和书写方式具有许多特征.本文结合K-means与克隆免疫算法提出了一种称为ICKM的新型聚类技术.四种元素集上的仿真实验说明了我们提出的方法在蒙古文聚类的有效性.

英文摘要:

      Text clustering is an important research issue of clustering technique. It aims to use the similar characteristics or similar expression to group the text so that the texts in the same clusters have the greatest similarity, and those in different clusters have the greatest dissimilarity. There are many characteristics in Mongolian structure and writing-mode compared with other kinds of characters. By combining K-means and clone immune algorithm, we propose a novel clustering technique called ICKM. Numerical experiments on four elements sets illustrate the validity of our method in the clustering task for Mongolian.