基于数据挖掘的及物性和单宾语句典型性关系研究

Transitivity and Category of Transitive Sentences:Based on Data Mining of Multi-variables Analysis

本文以[±意愿]、[±生命]、[±源头]、[±具体]、[±瞬时]、[±受力]、[±变化]这七个及物性特征参数为变量解构单宾语句的语义表现,通过分析100例样本句在这些变量上的强弱表现以获得样本赋值数据。(1)SPSS软件的量化分析发现,及物性特征可降维成三个维度,分别对应于主语、谓语和宾语,三个维度具有良好的信度和效度;(2)通过分析样本句在及物性特征上的表现差异,100例样本句分为三个类别;(3)以数据挖掘软件Clementine构建及物性特征和单宾语句典型性类别之间的CART决策树,发现[±生命]、[±变化]、[±受力]和[±瞬时]是参与典型性类别决策的有效变量,这四个特征对单宾语句的类别判定至关重要。

The paper applies seven semantic parameters analysis of the transitive sentence and collects data from 100 Chinese transitive sentences according to their performance on each parameter. By using SPSS, the paper finds out:(1) the seven parameters fall into three dimensions, which are of good reliability and validity;(2) k-cluster method helps to divide the 100 sentences into three categories;(3) CART decision tree filters four parameters that have significant influence upon the decision of sentences’ categories.