融合节点状态信息的跨社交网络用户对齐

提出一种融合节点状态信息的跨社交网络用户对齐方法, 通过网络表示捕获节点的局部特征和节点状态信息得到每个账户的嵌入向量, 计算不同账户对应表示之间的相似性发现对齐用户。在2个真实数据集上的试验结果表明, 提出的方法相对于其他方法可以对齐更多的用户。在预测不同尺度的top-k时, 提出的方法在网络结构较稠密的Twitter-Foursquare数据集上能够在top-9时对齐准确率达到50%且在稀疏且大网络数据集DM-ML上相比其他方法对齐准确率提高12.06%~36.62%;在分析F1-score时, 提出的方法能够有效提高用户对齐的性能。

Cross social network user alignment via fusing node state information

A cross social network user alignment method by fusing node state information was proposed. The local characteristics of nodes and node state information were captured through network representation to obtain the embedded vector of each account, and the aligned users were found by calculating the similarity between corresponding representations of different accounts. Experimental results on two real data sets showed that the proposed method could align more users than other methods. When predicting top-k of different scales, the proposed method could achieve an alignment precision of 50% at top-9 on the data set Twitter-Foursquare with dense network structure. Compared with other methods on the sparse and large network data set DM-ML, the improvement on alignment precision was 12.06%-36.62%. The analysis of F1-score also showed that the proposed method could effectively improve the performance of user alignment.