本文将粗糙集理论与模糊集理论结合起来 ， 给出一种图书馆最优选书算法。该算法首先从已知数据的初始信息系统出发 ， 计算各选书方案之间的相似度，从而构造相似矩阵 ， 然后根据相似矩阵的传递闭包对各方案进行聚类，并根据粗糙集理论求各属性重要性 ， 最后利用加权综合的思想及最小距离方法选择最优买书方案。 Combining rough set and fuzzy set theory, an optimal decision algorithm of library to select books is put forward. During this algorithm, firstly, we construct the similarity matrix from the original information system; secondly, we classify all the programs according to fuzzy clustering; thirdly, an algorithm to choose the optimal program is put forward according to the minimum distance of weighted relative deviation.