一种机会网络动态社区检测及演化方法研究

发布时间：2018-06-27 02:39

本文选题：机会网络 + 社区划分　；参考：《新疆大学》2017年硕士论文

【摘要】：复杂网络中存在着一个重要的特性,即社区特性。机会网络作为复杂网络的一种特殊形式,也具有相似特征节点聚集的现象,即也呈现出了社区结构的特性。由于机会网络是根据节点相遇的机会进行通信的,因此机会网络的拓扑不断在改变,所以社区结构也随着网络拓扑的改变处于不断地变化之中。研究这些动态社区有利于更好地理解网络结构以及更好的利用网络,针对这个问题,本文主要做了以下研究工作:1、从节点间社会联系,关系强度以及亲密度的综合考虑,提出了一种基于亲密度的社区检测方法(CDMI)。该方法首先根据单个周期内节点间的相遇历史信息计算节点间的社会压力指标以及关系强度的值,从而确定相应周期内网络中哪些节点间有边相连。然后计算节点与节点之间、节点与社区间的亲密度。最后根据聚集系数得出种子节点,从种子节点进行局部扩展从而完成社区结构检测。将仿真结果与节点动态归属性算法(NBDE)比较,验证了该方法的准确性与可行性,此外,该方法还能够得到重叠社区结构。2、节点归属性是机会网络中社区研究的一个重要方面,提出一种基于神经网络的节点归属性判断方法。通过将节点间的相遇频率、相遇持续时间、相遇次数作为神经网络的输入向量,不断地调整模型的权值和阀值来进行模型的训练,训练完成后,把新节点组成的向量输入该模型经过网络计算即可得出获胜的神经元,获胜的神经元就代表输入数据的分类,以此即可判断新节点的归属性。在人工数据集LFK基准网络上测试,结果表明,该方法可以有效地判断新节点的归属性。
[Abstract]:There is an important characteristic in complex networks, that is, community characteristics. As a special form of complex network, opportunistic network also has the phenomenon of similar characteristic node aggregation, that is, it also presents the characteristics of community structure. Because the opportunistic network communicates according to the chance that the nodes meet, the topology of the opportunistic network is constantly changing, so the community structure is constantly changing with the change of the network topology. Studying these dynamic communities is conducive to a better understanding of the network structure and better use of the network. In view of this problem, this paper mainly does the following research work: 1, from the social connection between nodes, the relationship strength and the comprehensive consideration of the affinity. A community detection method based on affinity (CDMI) is proposed. The method first calculates the social pressure index and the relationship strength between nodes according to the historical information of the encounter between nodes in a single cycle, and then determines which nodes in the corresponding cycle are connected with each other. Then the affinity between nodes and communities is calculated. Finally, the seed node is obtained according to the aggregation coefficient, and the community structure is detected by the local expansion from the seed node. The simulation results are compared with the node dynamic attribute algorithm (NBDE), and the accuracy and feasibility of the method are verified. In addition, the method can obtain overlapping community structure. The node attribute is an important aspect of community research in the opportunistic network. In this paper, a neural network based method for determining the attribute of nodes is proposed. The frequency, duration and number of encounters are used as input vectors of the neural network, and the weights and thresholds of the model are constantly adjusted to train the model. Input the vector of the new node into the model, the winning neuron can be obtained by network calculation, and the winning neuron can represent the classification of the input data, and then the attribute of the new node can be judged. The test results on the LFK benchmark network show that the proposed method can effectively judge the attribute of the new node.
【学位授予单位】：新疆大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：O157.5

【参考文献】