重叠社区与强弱边关系研究

RESEARCH ON THE RELATIONSHIP BETWEEN OVERLAPPING AREA AND STRONG-WEAK EDGES

近年来重叠区域结构的发现使得相对独立的社区关系变得日趋复杂化。对于社区内部以及社区之间的用户关系也变得更加多样性,评估用户关系的紧密程度以及用户关系对信息交流和传播的促进作用已成为目前研究的热点。重叠节点由于自身的特殊性,重叠节点之间以及它们与非重叠节点之间的关系很难用传统的强弱关系进行刻画。针对此问题提出重叠系数指标,以数值形式量化节点之间的强弱关系。在此基础上,对节点度进行重新定义并分析重叠节点与强弱边之间的关系。同时对重叠与非重叠社区结构中强弱边的数量进行了比较分析,发现在重叠社区结构中节点之间隐藏着丰富的隐性关系。最后在对重叠社区结构中弱关系的深入研究发现,移除弱边的过程也是重叠社区结构向非重叠社区结构转化的过程。结果表明重叠结构与非重叠结构并不是相互独立,而是存在着密切联系的,即非重叠社区结构是重叠社区结构网络中的一个特例。

In recent years, the discovery of overlapping regional structures has made the relatively independent community relations become more and more complicated and the user relationship between the community and the users also have been more diversity. It is a hot research to evaluate the tightness of user relations and the promotion of user relations to information exchange and communication. Overlapping nodes the overlap between the nodes and their relations with non-overlapping between nodes is hard to be described by using the traditional strength of the relationship, because of its particularity. In this thesis, we proposed overlap coefficient index between nodes in the strength of the relationship between the quantitative numerical forms to solve this problem. On this basis, we redefined the nodes and analysis of the relationship between the overlapping nodes and the strength of the edge, the number of overlapping and non-overlapping community structure of strong and weak edge. We found a hidden relationship between the hidden nodes in the rich overlapping community structure; finally found there are plentiful weak relationships in deep research of the overlapping community structure, in the process of remove the weak edge and overlapping community structure to process non overlapping community structure transformation. The results showed that the overlapping structure and non- overlapping structure were not mutually independent, but closely related, namely non overlapping community structure is a special case of overlapping community structure in the network.