基于传递闭包的知识发现方法和Swanson知识发现方法的比较研究——以癌药物靶点为例  

Comparative Study on Knowledge Discovery Methods of Transitive Closure-based and Swanson: Taking Cancer Drug Target for an Example

作　　者：杨渊[1] 李扬[1] 孙晓北[1] 高柳滨[2] 池慧[1] 

机构地区：[1]中国医学科学院医学信息研究所 [2]中国科学院上海药物研究所

出　　处：《图书情报工作》2013年第1期136-141,共6页Library and Information Service

摘　　要：以探寻癌药物靶点间的潜在关联为例,对基于传递闭包的知识发现方法和Swanson的一阶知识发现方法进行了比较,结果得到,运用传递闭包的方法获得知识假设,可能发现更多有意义的新关联,且得到较高查全率的同时,并没有牺牲过多的查准率。The mechanism of method to find the transitive closure in discrete mathematics and Swanson＇ s method to discovery disjoint literature knowledge is similar, but use the former to discovery disjoint literatures can develop the one step to multi steps at one time to get more potential relevance. Taking the cancer drug target discovery for an example, this paper compares the two methods. The results show that using the transitive closure method to acquire knowledge assumptions may find more meaningful association; it can get a higher recall rate, while not sacrifice too much precision.

关 键 词：传递闭包 知识发现 WARSHALL算法 向量空间模型 药物靶点 

分 类 号：R-5[医药卫生] G353[文化科学—情报学]