科技项目评审专家分组匹配算法

为了解决科技项目评审中申请书与专家的分组匹配问题，提出了一个基于二部图谱划分异构对象分组匹配算法。 该算法用二部图描述申请书与专家之间的对应关系，提出了关联强度计算公式，计算图中每条边的权重值，求出图的关联矩阵，对关联矩阵进行奇异值分解得到奇异特征向量，用k-means分组算法对奇异特征向量进行分组。 利用申请书与专家分组匹配算法能够实现项目评审过程中申请书与专家的自动分组与匹配，并且分组匹配结果有较高的准确性和合理性。 

Expert Grouping and Matching Algorithm in Science and Technology Project Review

In order to group and match the applications and experts in project reviewing, the heterogeneous objects grouping and matching algorithm has been proposed, which was based on the co-clustering using bipartite spectral graph partitioning. In this algorithm, the relationship between experts and applications were depicted as a bipartite graph model. The edge-weight was calculated by the relationship weight calculating formula which was proposed in this paper. After decomposing the matrix which was formed by the edge-weight, eigenvectors were gotten. At last, experts and applications were grouped and matched through mapping to the groups of the eigenvectors which were created by k-means grouping algorithm. Applications and experts grouping and matching automatically in the process of project evaluation can be achieved by the applications and experts grouping and matching algorithm, and the grouping and matching results have a higher accuracy and rationality.