中文摘要：

为了快速发现可重用产品结构，提出了基于非负矩阵分解的产品结构相似性判断方法。通过将产品结构邻接矩阵转化为邻接向量，构建包含全部结构信息的库矩阵；利用Multiplicative Updates（MU）算法对库矩阵进行非负矩阵分解，实现以低维空间向量描述的产品结构；在此基础上，通过计算低维向量的欧氏距离，可以判断产品结构之间的相似性；最后通过实例对所提出原理和方法进行了验证，结果表明，该方法比目前的相似性判断方法更高效。

英文摘要：

In order to find reusable product structure promptly, an approach of measuring the similarity among product structures was proposed based on non-negative matrix factorization. A com- prehensive matrix which containsed all structures was constructed on the basic of adjacent vectors that were transformed from the adjacent matrices of product structures. The non-negative matrix factoriza- tion for the comprehensive matrix was implemented based on Multiplicative Updates（MU） algorithm. Then all product structures might be described in low dimensional space. On the basis of these, the similarity between two product structures could be measured by calculating the Euclidean distance a- mong these low dimensional vectors. Finally, an example was presented to verify the principles and methods mentioned above. The results show that the proposed methodologies are more effective than those of the existing methods.