基于反向标签传播的移动终端用户群体发现

针对现有方法在移动终端用户群体发现中不能兼顾社会关系和位置属性的问题，提出基于反向标签传播算法的重叠群体发现方法.根据移动终端用户的位置信息推断社会关系拓扑图，提取时空共现区.将时空共现区作为位置属性标签，标注社会关系拓扑图.在标签拓扑图上进行反向标签传播，消除节点伴随标签.经过反复迭代，在标签稳定状态下的每一个节点保留所属群体的主标签.根据用户社会关系和稳定状态下的节点主标签完成群体划分与识别.在4个真实数据集上比较反向标签传播算法与3种同类方法，实验结果表明，反向标签传播算法较好地兼顾了用户社会关系和位置属性，群体发现结果的标准互信息（NMI）与综合评价函数（F）分别比次优者平均高8.97%和3.87%.
Abstract: A new method was proposed for overlapping group discovery based on reverse-label propagation algorithm to solve the problem that the social relationship and location attribute cannot be taken into account simultaneously when using the existing methods to discover the groups of mobile terminal users. According to the location information of mobile terminal users, the topological graph of social relationship was inferred and the spatio-temporal co-occurrence areas were extracted. The spatio-temporal co-occurrence areas were used as position attribute labels to mark the topological graph. The label graph was processed with the reverse-label propagation algorithm to remove companion-labels for nodes. With repeated iterations, each node preserved the main-labels of the groups when the state of the labels was stable. According to the user social relationship and node's main label under stable state, the groups of mobile users were divided and recognized.The experiments were carried out to compare reverse-label propagation algorithm with three similar methods on four real datasets. Results showed that the reverse-label propagation algorithm took better account of social relationship and location attribute simultaneously, and the normalized mutual information (NMI) and comprehensive evaluation function (F), the evaluating indicators of group discovery, were increased averagely by 8.97% and 3.87% respectively than the suboptimal algorithm.