论文部分内容阅读
考察了在语言精细分类中使用复杂网络以及在基于复杂网络的语言分类中使用平行词同现网络替代句法依存网络的可行性.采用12种斯拉夫语言和2种非斯拉夫语言的平行文本,构建了14个词同现网络.通过这些网络的主要参数的恰当组合,聚类分析能够将斯拉夫诸语言与非斯拉夫语言区分开来,并能将12种斯拉夫语言正确地划分到各自的语支中去.另外,聚类也能反映某些斯拉夫语言在其语支内部的亲缘关系.结果表明,平行词同现网络能够被用于语言的精细分类,而且在基于复杂网络的语言分类中可被用作句法依存网络的一种更为便捷的替代品.
The feasibility of using complex network in linguistic fine classification and using parallel word co-occurrence network to replace syntactic dependency network in linguistic classification based on complex networks was investigated.Parallel texts of 12 Slavic languages and 2 kinds of non-Slavic languages were constructed to construct With the proper combination of the main parameters of these networks, cluster analysis can distinguish the Slavic languages from the Fisic languages and correctly classify the 12 Slavic languages into their own branches of speech In addition, the clustering can also reflect the kinship of some Slavic languages within their own dialects.The results show that the parallel words co-occurrence network can be used for the fine classification of languages and can be used in language classification based on complex networks A more convenient alternative to the syntactic dependent network.