Aiming at the difficulty of obtaining global network information based on degree centrality in targeted immunization, and the high complexity of the algorithm based on network betweenness centrality, a multi-particle random walk algorithm is introduced to identify a group of susceptible nodes in network. Compared with the classical immune strategies, the SIR(suspected-infected-recovered/removed) and SI(suspected-infected) propagation experiments were carried out on eight real datasets. The results showed that the immune algorithm based on random walk has lower time complexity on networks of different scales and different structural characteristics, and it can achieve low propagation range without the need to obtain global information. The algorithm suppresses the network propagation range in low immune rate, which is equivalent to the targeted immunization, and has better effect in high immune rate.
基于以上描述,本文所选的无向无权网络可作为加权网络的一种特殊情况,即所有连边赋权为1。考虑到每一步的游走概率只与粒子所在节点度(加权网络中还需考虑边权大小)有关,并不会基于过去或当前的表现,且无法预知下一步的动态和方向,接近于布朗运动[9],根据其随机性命名为多粒子随机游走免疫算法(multi-particle random walk immunization,RWI)。
P 的计算与PageRank原理相似,是一种类马尔可夫过程,Google公司的PageRank算法[11]沿用随机游走思想,将游走的过程变为概率传播的过程,最后统计各节点的概率,但该算法最终会受网络出入度值的影响,形成某些无出度节点的“黑洞”,最终退化为节点度排序。本文所提的概率转移矩阵形式与PageRank算法的概率转移矩阵计算方式不同,PageRank表示在网络中的页面跳出关系,其统计的是最终的页面排序;而本文随机游走算法为统计粒子迁入概率,将每一个时间步的历史流量进行累加,考虑的是过程的流量。在游走过程中对每一步游走添加前提限制,考虑到现实传播受出度影响,因此算法步骤3对无出度或是无下一邻居节点做停止游走处理,流量记作0。
与本文的RWI作比对的免疫策略有:熟人免疫(AI)[3]、目标免疫中的度中心(degree centrality)免疫TI_DC以及介数中心(betweenness centrality)免疫 TI_BC[2]、PageRank筛选节点的免疫方式(PRI)以及文献[8]中的免疫粒子非独立随机游走免疫方法(dependent random walk particle immunization,DRWPI)。鉴于现实情况下免疫资源有限,且在实验准备过程中发现,在多数网络数据集中,当免疫率达到0.20时,网络传播范围已低于20%,甚至在有些情况下网络传播范围低于10%,控制传播效果明显,且已无明显的边际增益效果,因此本实验选取的免疫率均介于0.01至0.23之间,步长0.02。
COHENR, HAVLINS, BEN-AVRAHAMD. Efficient immunization strategies for computer networks and populations [J]. Physical Review Letters, 2003, 91(24): 247901. DOI:10.1103/PhysRevLett.91.247901 .
[5]
LIUB, YANS, XUEY M. Analysis of hybrid immunization strategy in complex networks [C]//Proceedings of the 2016 4th International Conference on Machinery, Materials and Computing Technology. Paris: Atlantis Press, 2016:6. DOI:10.2991/icmmct-16.2016.225 .
HUANGB, ZHAOX Y, QIK, et al. Coloring the complex networks and its application for immunization strategy[J]. Acta Physica Sinica, 2013, 62(21): 510-517. DOI:10.7498/aps.62.218902(Ch ).
[8]
SAXENAC, DOJAM N, AHMADT. Group based centrality for immunization of complex networks[J]. Physica A: Statistical Mechanics and its Applications, 2018, 508: 35-47. DOI: 10.1016/j.physa.2018.05.107 .
[9]
夏玲玲. 复杂网络上的信息传播动力学建模与免疫策略研究[D]. 南京: 南京邮电大学, 2017.
[10]
XIAL L. Modeling of Information Transmission Dynamics And Immunization Strategies on Complex Networks [D]. Nanjing: Nanjing University of Posts and Telecommunications, 2017(Ch).
ZHUY X, ZHANGF L, WANGR J, et al. Reseach on immunization strategy based on random walk mechanism in temporal networks [J]. Journal of University of Electronic Science and Technology of China, 2017, 46(1): 96-101. DOI:10.3969/j.issn.1001-0548.2017.01.015(Ch ).
FUW Y, LINGC D. Brownian motion based simulated annealing algorithm[J]. Chinese Journal of Computers, 2014, 37(6): 1301-1308. DOI:10.3724/SP.J.1016.2014.01301(Ch ).
WANGT, LID M. The gracefulness of some special graph classes [J]. Journal of Wuhan University (Natural Science Edition), 2012, 58(5): 437-440. DOI:10.14188/j.1671-8836.2012.05.012(Ch ).
[17]
LÜL S, ZHANGK, ZHANGT, et al. Nodes and layers PageRank centrality for multilayer networks [J]. Chinese Physics B, 2019, 28(2): 020501. DOI:10.1088/1674-1056/28/2/020501 .
[18]
NEWMANM. Network Data [N/OL].[2020-02-15].
[19]
KUMARS, SPEZZANOF, SUBRAHMANIANV S, et al. Edge weight prediction in weighted signed networks[C]// Proceedings of the 2016 IEEE 16th International Conference on Data Mining. New York: IEEE Press, 2016: 221-230. DOI:10.1109/ICDM.2016.0033 .
[20]
ANGRIMANE, GRINTEN AVAN DER, LOOZ MVON, et al. Guidelines for experimental algorithmics: A case study in network analysis [J]. Algorithms, 2019, 12(7): 127. DOI:10.3390/a12070127 .
LESKOVECJ, KLEINBERGJ, FALOUTSOSC. Graph evolution [J]. ACM Transactions on Knowledge Discovery from Data, 2007, 1(1): 2. DOI:10.1145/1217299.1217301 .
[23]
ROZEMBERCZKIB, SARKARR. Characteristic functions on graphs: Birds of a feather, from statistical descriptors to parametric models [C]// Proceedings of the 29th ACM International Conference on Information & Knowledge Management. New York: Association for Computing Machinery, 2020:1325-1334. DOI:10.1145/3340531.3411866 .
[24]
BATAGELJV, MRVARA, ZAVERSNIKM. Network analysis of texts [J]. Language Technologies, 2002, 40: 143-148.
[25]
BUD B, ZHAOY, CAIL, et al. Topological structure analysis of the protein-protein interaction network in budding yeast [J]. Nucleic Acids Research, 2003, 31(9): 2443-2450. DOI:10.1093/nar/gkg340 .
QUQ Q, HANH, LÜY N, et al. S2IR rumor dissemination model based on structural characteristics of social networks[J]. Complex Systems and Complexity Science, 2019, 16(3): 48-59. DOI:10.13306/j.1672-3813.2019.03.005(Ch ).