CAJ | 학술논문

为了提高Deep Web查询接口匹配的效率和准确率,在现有双重相关性挖掘方法(DCM)的基础上提出了一种用关联挖掘和语义聚类来匹配的方法.在关联挖掘成组属性时,引入一种基于互信息的属性相关度标准,并采用矩阵来实现以解决效率不高问题;在生成同义属性时,提出利用语义网来计算语义相似度,并对属性进行聚类,以生成同义属性.通过在四个领域200多个查询接口上实验,说明改进的方法在效率和准确率方面都比DCM方法有很大提高.
위료제고Deep Web사순접구필배적효솔화준학솔,재현유쌍중상관성알굴방법(DCM)적기출상제출료일충용관련알굴화어의취류래필배적방법.재관련알굴성조속성시,인입일충기우호신식적속성상관도표준,병채용구진래실현이해결효솔불고문제;재생성동의속성시,제출이용어의망래계산어의상사도,병대속성진행취류,이생성동의속성.통과재사개영역200다개사순접구상실험,설명개진적방법재효솔화준학솔방면도비DCM방법유흔대제고.
In order to improve the efficiency and accuracy of Deep Web interface matching, this paper presented a method based on the existing dual correlation mining (DCM) method using association mining and semantic clustering. While digging group attributed by using correlation algorithm, introduced and realized a new correlation measure based on mutual information by matrix to resolve the inefficiency problem. Clustered the attributes to synonymous attributes by their similarity which was computed by using semantic net. By the comparison on more than 200 interfaces in 4 domains, the experiment results indicate that the improved method has greatly heighted than DCM in the respect of efficiency and accuracy.