k-部排序本体相似度计算.docVIP

k-部排序本体相似度计算 摘要:将本体图中每个顶点的相关信息用一个向量表示。根据本体图自身的结构将顶点分成k个部分。在每个部分中选取样本点组成s，并选择相应的排序亏损函数。运用k.部排序学习算法得到最优排序函数，从而将本体结构图中每个顶点映射成一个实数，通过比较实数间的差值判断两概念的相似程度。实验表明该方法对于计算本体概念间的相对相似度是有效的。 关键词:本体；相似度计算；k.部排序；排序函数；排序亏损函数 ontology similarity computation using k.partite ranking method lan mei.hui1* , ren you.jun1, xu jian1, gao wei2,3 ( 1．college of computer science and engineering, qujing normal university, qujing yunnan 655011, china; 2．college of information, yunnan normal university , kunming yunan 650092, china; 3．college of mathematical sciences, soochow university, suzhou jiangsu 215006, china abstract: this paper represents the information of each vertex in ontology graph as a vector. according to its structure of ontology graph, the vertices are divided intok parts. it chooses vertices from each part, and chooses the function as ranking loss. it uses k-partite ranking learning algorithm to get the optimization ranking function, maps each vertex of ontology structure graph into a real number, then calculates the relative similarities of concepts by comparing the difference between real numbers. experimental result shows that the method for calculating the relative similarity between the concepts of ontology is effective. this paper represented the information of each vertex in ontology graph as a vector. according to its structure of ontology graph, the vertices were divided into k parts. it chose vertices from each part, and chose the ranking loss function. it used k.partite ranking learning algorithm to get the optimization ranking function, mapped each vertex of ontology structure graph into a real number, and then calculated the relative similarities of concepts by comparing the difference between real numbers. the experimental results show that the method for calculating the relative similarity between the concepts of ontology is effective. key words: ontology; similarity computation; k.partite ranking; ranking function; ranking loss functio