基于汉字与部件聚类的计算机模拟研究  

Computer simulation research based on Chinese characters and their components clustering

作　　者：方新[1] 穆志纯[1] 陈静[1] 杜大鹏[1] 

机构地区：[1]北京科技大学信息工程学院,北京100083

出　　处：《计算机应用》2005年第12期2951-2953,共3页journal of Computer Applications

摘　　要：以自组织特征映射算法为理论基础,提出了一种新的SOMNET算法,利用人工神经网络实现了具有相似特征的汉字的聚类以及汉字部件的聚类。并对汉字及其部件的显示效果进行了分析,从不同角度刻画了SOM模型的应用。研究结果对于汉字认知研究以及留学生第二语言习得教学具有一定的参考价值。In this paper, a new algorithm named SOMNET was presented based on the theoretical foundation of Selforganizing Feature Map algorithm. Using Artificial Neural Network, SOMNET algorithm realizes similar-featured Chinese characters clustering and Chinese character components clustering. The display effect of Chinese characters and their components was analyzed, and the application of SOM model was described from multiple points of view. The research outcome is expected to be helpful to Chinese character cognition and teaching of foreign students＇ second language acquisition.

关 键 词：SOMNET算法 汉字聚类 自组织特征映射 第二语言习得 

分 类 号：TP391.12[自动化与计算机技术—计算机应用技术] TP183[自动化与计算机技术—计算机科学与技术]