一种基于时延约束的社会网络信用分布优化模型

发布时间：2018-04-06 22:34

本文选题：社会网络　切入点：影响力最大化　出处：《计算机研究与发展》2017年02期

【摘要】：基于时延约束的影响力最大化问题(influence maximization with time-delay constraint,IMTC)定义为在时延约束条件下,选取网络中一部分初始用户,使得影响力传播过程结束后网络中被成功影响的用户数量最多.现有研究工作主要依据网络结构优化影响力传播模型,或改进启发式算法提高初始节点的选取质量,影响力传播过程中的时间延迟特性及时延约束条件往往被忽略.针对这点不足,基于时延约束的信用分布模型(credit distribution with time-delay constraint model,CDTC)综合考虑见面概率和条件激活概率对信用分配进行优化定义,同时将相邻节点之间不断见面并激活对信用分配的阻碍作用映射到传播增量路径中,最后根据信用分布函数,使用基于时延约束的贪心算法GA-TC,递归选取边际收益最大的节点组成初始节点集合.实验结果表明:在CDTC模型上使用GA-TC算法不仅能够保证初始节点的选取质量,而且具有更高的执行效率及更好的行为执行预测能力.
[Abstract]:The influence maximization with time-delay constraint (IMTC-based) is defined as the selection of some initial users in the network under the condition of delay constraints, which makes the number of successful users in the network after the end of the process of influence propagation is the largest.The existing research mainly optimizes the influence propagation model based on the network structure or improves the heuristic algorithm to improve the quality of the initial node selection. The time delay characteristics and delay constraints in the process of influence propagation are often ignored.In order to solve this problem, credit distribution with time-delay constraint model based on delay constraint is used to optimize credit allocation by considering the probability of meeting and conditional activation.At the same time, we map the function of blocking the credit distribution between the adjacent nodes to the path of propagation increment, finally, according to the credit distribution function,The greedy algorithm GA-TC based on delay constraint is used to recursively select the nodes with the largest marginal income to form the initial set of nodes.The experimental results show that using GA-TC algorithm in CDTC model can not only guarantee the quality of initial node selection, but also have higher execution efficiency and better performance prediction ability.
【作者单位】：中南大学信息科学与工程学院;中南大学软件学院;
【基金】：国家自然科学基金项目(61379058,61272149,61379057,61350011)~~
【分类号】：TP393.09

【相似文献】