基于复杂网络的文本抗毁性分析

Text invulnerability analysis based on complex network

针对中、英文有向加权文本同现网络的抗毁性能进行研究。选取著名的外文小说《飘》的中、英文版本作为基本语料,以章节作为基本单元,构造了12个中文字有向加权文本同现网络和12个英文词有向加权文本同现网络;分析了各文本同现网络基本参数的变化趋势,并且绘制了相应的分布曲线,并分析了各文本同现网络的抗毁性能。实验结果表明,中、英文的抗毁性能随着文章长度的增加都在逐渐降低,但中文网络的抗毁性下降的速率要比英文网络慢得多。从侧面反映了中文字之间的连接关系和协调性相对来说比较好。

This paper studied the anti-destroying ability of Chinese characters and English words text co-occurrence networks. Firstly, it selected the famous foreign novel ＂gone with the wind＂ in both Chinese and English versions as the basic corpus to structured 12 Chinese characters and 12 English words text co-occurrence networks, with the chapter as a basic unit. Second- ly, it analyzed the change tendency of basic parameters in the text co-occurrence network, and drawn the corresponding chan- ging curve. Finally, it researched the anti-damage performance of text co-occurrence networks. The experimental results show that the anti-damage performance gradually reduces with the increase of the length of the articles, but Chinese network anti-de- stroying ability falls speed much more slowly than English network, which reflects the connected relationship and coordination between the Chinese characters are relatively good from the side.