CAJ | 학술논문

客观事物的信息内容主要体现为其对象属性特征及属性间的相互关系。本文尝试揭示客观事物对象成对属性间存在的关联性影响，提出一种构建属性空间的新方法。该方法将属性作为坐标空间的维度向量，元组作为坐标空间的中的点，通过计算任意成对属性间的关联程度，并将该关联程度非线性映射到属性向量间的夹角，可构建体现属性间相互影响程度的成对关联属性空间（PAAS）。将该属性空间运用于计算对象间的距离凝聚层次聚类（HAC）分析，结果表明比传统属性空间的结果更加有效。
객관사물적신식내용주요체현위기대상속성특정급속성간적상호관계。본문상시게시객관사물대상성대속성간존재적관련성영향，제출일충구건속성공간적신방법。해방법장속성작위좌표공간적유도향량，원조작위좌표공간적중적점，통과계산임의성대속성간적관련정도，병장해관련정도비선성영사도속성향량간적협각，가구건체현속성간상호영향정도적성대관련속성공간（PAAS）。장해속성공간운용우계산대상간적거리응취층차취류（HAC）분석，결과표명비전통속성공간적결과경가유효。
The information content of an objective entity mainly embodied in its attribute features and the mutual relationship between attributes. This paper try to reveal the association between the pairwise attributes of database tuple and propose a new method for the construction of attribute space. We try to set the attributes as basic vectors and objects as the points in the coordinate space. Depending on the conditional probability between the pairwise attributes, we could determine their association degree. Mapping the association degree between the pairwise attributes as the angle of pairwise attributes vector by using a nonlinear function, we could build the pairwise association attribute space (PAAS) which embodied the mutual influence between the pairwise attributes. In the end, we applied the attribute space to hierarchy clustering analysis (HAC). The result of analysis and comparison showed that it is more effective than the traditional attribute space.