语义、句法网络作为语体分类知识源的对比研究

Comparison study of using semantic and syntactic network characteristics to do text clustering

基于6种语体的句法和语义树库分别构建了依存句法和语义网络,对这些网络的边数、节点数、节点平均度、聚类系数、平均最短路径长度、网络中心势、直径、节点度幂律分布的幂指数、度分布与幂律拟合的决定系数等整体特征进行了对比分析。以这些整体特征为变量,采用不同的聚类方法,对这6种语体的句法和语义网络进行了聚类分析。研究结果显示,同样是基于语言学原则构建起来的网络结构,依存句法网络和依存语义网络之间有明显差异。其参数的含义不尽相同,依据其各项参数所做的聚类实验的结果也不相同。采用语义网络的一些主要参数组合,可以获得相对合理的聚类结果,但不能很好地区分书面语体和口语体;通过句法网络的一些主要参数组合,可以很好地区分不同语体的文本,获得较为合理的文本聚类结果。

The study builds six dependence syntactic networks and semantic networks based on syntactic and semantic treebanks of different genres and does a comparative analysis of overall features of the networks, including the number of edges, the number of the nodes, the average degree, the clustering coefficient, the average path length, the centraliza- tion, the diameter, the index of power-law, and the coefficient of determination. The article tries multi-methods, with fea- tures as variables, to do clustering analysis of these networks. The results show that, although the syntactic and semantic networks all follow the linguistic principles, there are obvious differences between syntax and semantic networks. The meanings of the network parameters vary and the clustering results according to the parameters are different. Using the combinations of main semantic network parameters can obtain relatively reasonable clustering results, but it cannot distin- guish well written style from colloquialism while using the combinations of main syntactic network parameters can well distinguish different styles of texts and obtain reasonable text clustering results.