提出了基于空间映射的顶点带属性网络的链接预测算法.顶点带属性网络包含拓扑及顶点属性两种信息,为了综合考虑这两种信息,将二者同时映射到另一空间.完成空间映射后,在新的空间计算顶点的相似度,并以此来预测链接存在的可能性.提出分步交叉迭代的方法来取得最优的映射矩阵,以在新的空间中有效融合拓扑信息与顶点属性信息.实验结果证明了空间映射方法的正确性,所提出的基于空间映射的方法能够取得较高质量的预测结果.%A link prediction algorithm in networks with node attributes based on space mapping was proposed in this paper.In general,networks with node attributes have two types of information,i.e.topology and node attributes.To integrate these two types of information,we mapped them into a new space.After information being mapped,similarities between nodes were calculated in the new space,which are used to predict the possibility of links between nodes.Alternative iteration method was proposed to get the optimal mapping matrix,so as to integrate topology information and node attribute information effectively.The experimental results verify the correctness of space mapping approach,and show that the proposed space mapping based algorithm can obtain high quality prediction results.
展开▼
