基于跳步的增量式影响力最大化算法  

Hop-based incremental influence maximization algorithm

作　　者：黄颖 梁春泉[1] 杨泽宽 曹晓旭 武文君 HUANG Ying;LIANG Chun-quan;YANG Ze-kuan;CAO Xiao-xu;WU Wen-jun(College of Information Engineering,Northwest A&F University,Yangling 712100,China)

机构地区：[1]西北农林科技大学信息工程学院,陕西杨陵712100

出　　处：《计算机工程与设计》2021年第1期89-95,共7页Computer Engineering and Design

基　　金：国家自然科学基金项目(61402375);陕西省重点研发计划基金项目(2019ZDLNY07-02-01);西北农林科技大学中央高校基本科研业务费专项基金项目(2452019065)。

摘　　要：对动态在线社交网络中的影响力最大化问题进行研究,提出一种基于跳步的增量式算法,快速跟踪动态网络最具有影响力的用户集。为应对网络结构变化,基于跳步,一方面评估变化用户影响力上限值,快速识别和保留无需变动的影响力用户;另一方面增量式地计算有潜力用户的实际影响力,替换不再属于最具影响力的用户。在真实数据集上进行实验和分析,其结果表明,相比其它最新同类算法,所提算法能以更快速度在动态网络中维护最具影响力用户集。To solve the problem of influence maximization in dynamic online social networks,a hop-based incremental algorithm was proposed to track the most influential users of the networks.To respond to the continuously changing of topologies,based on hop propagation,the upper bounds of the influence spread of users related to the changing were evaluated,and influence users who maintained the same were identified and retained.The influence gains of potential users were calculated incrementally,and the users who were no longer the most influential were updated.Experiments were conducted on real data sets.The results show that the proposed algorithm sustains the most influential users in dynamic networks significantly faster than the state-of-the-arts methods.

关 键 词：影响力最大化 社交网络 基于跳步的增量式算法 动态网络 用户集合 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]