CN115277115A - Method and system for solving robust information propagation problem on network - Google Patents
Method and system for solving robust information propagation problem on network Download PDFInfo
- Publication number
- CN115277115A CN115277115A CN202210799978.3A CN202210799978A CN115277115A CN 115277115 A CN115277115 A CN 115277115A CN 202210799978 A CN202210799978 A CN 202210799978A CN 115277115 A CN115277115 A CN 115277115A
- Authority
- CN
- China
- Prior art keywords
- population
- network
- influence
- robust
- solving
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L63/00—Network architectures or network communication protocols for network security
- H04L63/14—Network architectures or network communication protocols for network security for detecting or protecting against malicious traffic
- H04L63/1408—Network architectures or network communication protocols for network security for detecting or protecting against malicious traffic by monitoring network traffic
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L63/00—Network architectures or network communication protocols for network security
- H04L63/14—Network architectures or network communication protocols for network security for detecting or protecting against malicious traffic
- H04L63/1408—Network architectures or network communication protocols for network security for detecting or protecting against malicious traffic by monitoring network traffic
- H04L63/1416—Event detection, e.g. attack signature detection
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L63/00—Network architectures or network communication protocols for network security
- H04L63/20—Network architectures or network communication protocols for network security for managing network security; network security policies in general
Abstract
The invention discloses a method and a system for solving the problem of robust information propagation on a network. A method for solving a robust information dissemination problem over a network, comprising the steps of: s1, defining robust influence performance evaluation indexes of a seed node of a network structure under node attack; and S2, designing a method based on a diversity attention principle under the guidance of the influence performance evaluation index to solve the problem of robust influence maximization. The method has the advantages of low calculation cost, comparability of numerical values under different network scales and the like, and can provide reliable guidance for selecting the seed node set; the obtained seed node set can keep a good influence propagation effect in the network damage process.
Description
Technical Field
The invention belongs to the technical field of computer application, and particularly relates to a method and a system for solving the problem of robust information propagation on a network.
Background
With the popularization and rapid development of the internet in recent years, social networks such as WeChat and microblog enter the visual field of people. These platforms greatly enrich human socialization and have even become an integral part of human society. With the rapid increase of the user scale, the social platform becomes an important channel for people to release opinions and communicate; meanwhile, the characteristics of high efficiency of network spread information and huge number of users make online marketing become an important commercial marketing mode. Due to cost limitations such as capital and the like, a merchant cannot promote every user, often carries out product promotion aiming at a small part of 'seed' users with larger influence, and thus achieves the purpose of rapidly improving the influence of the product, and the means is called 'virus marketing' or 'public praise marketing'. Therefore, how to select the most influential "seed" user becomes the key to the problem, and this process can be modeled as an influential maximization problem.
Generally, there are the following common information propagation models for the influence propagation process on the network system: an independent cascade model (LC model), a weight cascade model (WC model), and a linear threshold model (LT model), etc. On the basis of the research, the influence maximization problem on the network can be regarded as an optimization problem, namely, under the guidance of a certain performance evaluation factor, an attempt is made to find the optimal seed node combination from all the members of the network. Heuristic algorithms such as hill climbing method and greedy algorithm can solve the problem, but all of them have the disadvantage of low efficiency; meanwhile, evaluation methods based on structural characteristics such as degree, pageRank, betweenness and the like can also provide reference for solving the problem of influence maximization, but the methods have the defect of influence overlapping. In addition, the searching methods based on the population, such as the discrete particle swarm algorithm, the Memetic algorithm and the like, also show a better effect on solving the problem, and the optimization methods provide a referable solution for solving the problem of influence maximization according to the characteristics of the network structure and the information of the network nodes.
In addition, the network often faces a complex application environment, and sometimes the functions of the nodes and links are failed and fluctuated due to self or external factors. In fact, a properly functioning network should have the ability to tolerate faults and fluctuations, i.e., robustness. The robustness of the network has important significance for guaranteeing the safe operation of network systems such as a social network, a world wide web and a traffic network. Previous researches prove that the network structure is often closely related to the performance of the network, and for a social network, the change of the network structure also affects the interaction behavior of the members in the network, and the change of the interaction behavior is likely to further affect the influence propagation process. Therefore, how to evaluate and optimize the robustness of the network system is also one of the hot spots of research in the field of complex networks in recent years.
There are some deficiencies to the current work on the problem of maximizing influence. On the one hand, as mentioned above, in practical applications, the network structure is often subjected to unpredictable attacks, and both attacks on network links and network nodes may cause serious damage to the network structure. The current work has certain defects that the scene of the change of the network structure is not considered, or only the scene of the network under the link attack is considered, but the scene of the network under the node attack is not considered; on the other hand, the conventional genetic algorithm does not well apply population information obtained in the search process, and the population diversity is poor. In some extreme cases, problems of slow convergence efficiency, easy falling into local optimal solution and the like may occur. In the genetic algorithm, the diversity of the population has important influence on the calculation result and the calculation efficiency; to realize efficient optimization, the population should be dispersed as much as possible in the solution space. In addition, at present, some leading-edge works preliminarily discuss how to select seed nodes which simultaneously meet the maximization of robustness and influence, but the application scenarios of these works are only limited to the situations that system parameters such as node influence propagation probability and propagation models are uncertain, and certain limitations exist. Therefore, how to effectively evaluate the influence performance of the seed node under the condition that the network is attacked by the node and how to design an effective seed optimization selection scheme are still problems which are not solved by the current research.
Disclosure of Invention
The present invention is directed to overcoming the deficiencies of the prior art and providing a method and system for solving the problem of robust information propagation over a network.
In order to achieve the purpose, the invention is realized by the following technical scheme:
a method for solving a robust information dissemination problem over a network, comprising the steps of:
step S1,Robust influence performance evaluation index for defining network structure in seed node under node attack
The N represents the number of nodes of the current network, the
Representing the influence estimated value of the seed node set S after Q nodes are removed;
and S2, designing a method for solving the problem of the maximization of the robust influence based on a method of 'diversity attention' principle under the guidance of the evaluation index of the influence performance.
Preferably, the method based on the principle of "diversity attention" comprises the following steps: (2a) The execution relies on a priority strategy initialization operator to generate a series of seeds to initialize the population;
(2b) Executing a crossover operator combined with a crossover strategy, and then exchanging genetic information and expanding the population quantity so as to generate a temporary population with the same number as the original population;
(2c) Executing mutation operators on the original population and the temporary population, thereby increasing the population diversity and getting rid of the local optimal solution;
(2d) Executing a local search operator on the original population and the temporary population;
(2e) Executing a selection operator, and selecting an overall best individual (node set) according to a fitness function;
(2f) And repeatedly executing the steps until a set termination condition is met, and finally outputting an optimal solution which appears in the evolution process, namely the seed node set S with the maximum fitness and the fitness thereof.
Preferably, the priority strategy comprises a node degree priority strategy, a PageRank priority strategy and a clustering priority strategy.
The cross strategy comprises a uniform cross strategy, a two-point cross strategy and a single-point cross strategy.
The system comprises:
the defining module is used for defining the influence performance evaluation index of the seed node set S under node attack;
an initialization operator module for generating a series of seeds to initialize a population;
the temporary population generating module is used for generating a temporary population with the same number as the original population;
the mutation operator module is used for increasing the population diversity and getting rid of the local optimal solution;
the search module is used for executing local search operators on the original population and the temporary population;
the execution operator module is used for adopting an elite strategy to store or try to update each iteration to obtain an optimal solution;
and the output module is used for outputting the optimal solution appearing in the evolution process, wherein the optimal solution is the seed node set S with the maximum fitness and the fitness thereof.
The invention has the following beneficial effects: under the condition that a network structure is attacked by nodes, the method designs an evaluation index for measuring the robust influence performance of the seed nodes under the attack of the network nodes, the evaluation index can effectively evaluate the influence performance of the seed nodes under the condition that the network is attacked by the nodes, the calculation cost is low, the result has comparability under different network scales, and a reliable evaluation standard can be provided for the selection of a seed node set;
under the guidance of performance indexes for measuring the robust influence of the seed nodes under the attack of the network nodes, a method based on a 'diversity attention' principle is designed to solve the problem of maximization of the robust influence, and a seed node set obtained by the method can keep a good influence propagation effect in the process of network damage.
Drawings
FIG. 1 is a flow chart of the present invention.
Detailed Description
The technical scheme of the invention is further explained by combining the attached drawings of the specification:
as shown in the figure 1 of the drawings,
(1) And the robust influence performance evaluation indexes of the seed nodes of the network structure under the node attack are as follows:
(1a) In the process that the influence is spread outwards by the seed nodes, after the seed nodes are attacked each time, the influence of the seed node set S is reevaluated;
(1b) Following the formula
The comprehensive influence performance of the seed node set S under the network node attack is obtained through the addition and normalization operations in the method, wherein S (Q) is the maximum network connected cluster without Q nodes, N is the total number of the nodes in the network, and 1/N is a normalization factor;
(1c) According to the mechanism, the robust influence performance evaluation index of the seed node of the network structure under the node attack is defined as follows:
here, N denotes the number of nodes of the current network,
and representing the influence estimation value of the seed node set S after Q nodes are removed.
(2) The method for solving the problem of maximizing the robust influence based on the principle of 'diversity attention' comprises the following steps:
(2a) And (3) following an individual diversity principle, and combining a node degree priority strategy, a PageRank priority strategy and a clustering priority strategy to execute an initialization operator to generate an individual (node set) population with the largest possible influence. Specifically, a node degree priority strategy, a PageRank priority strategy and a clustering priority strategy are respectively executed on the initialization of the individuals of the first 1/3, the middle 1/3 and the rest 1/3 of the population, namely, a given number of nodes are selected as genes of the individuals by using a roulette method according to the node degree, the PageRank and the clustering as index information for judging the importance of the nodes;
(2b) According to the principle of 'gene diversity', three strategies of uniform crossing, two-point crossing and single-point crossing are combined to execute a crossing operator, then gene information is exchanged, the population quantity is expanded, so that a temporary population with the same number as the original population is generated, and individuals with the size of PopSize are generated. It should be noted that if a duplicate node appears in an individual after the crossover, a correction operation needs to be performed on the individual. The correction operation is as follows: randomly selecting a node in the initial network to replace a repeated node in the individual until no repeated node exists in the individual;
(2c) And (3) dynamically adjusting the mutation probability according to the fitness of the individual according to the principle of probability diversity, and executing mutation operators on the original population and the temporary population. Specifically, fitness values (i.e., an influential performance index R) for all individuals in the population are calculated as followsS) And calculating the sum of fitness of all individuals. Then, for each individual in the population, the mutation operator is represented by pm*β(pmIs the probability of variation, β = 1-fitness value of the individual/fitness value of all individuals in the population), replacing a random one-gene node in an individual with another node in the network. Note that the principle of no repeated node in each individual is also followed in the mutation process, and if the repeated node occurs, the above-mentioned correction operation is executed;
(2d) And (3) expanding the search range from local to global according to a search space diversity principle, namely searching mutation operators for the original population and the temporary population, so that the population diversity is increased, and the local optimal solution is got rid of. Specifically, a local search operator is designed into three parts, and different search strategies are dynamically adopted according to the iteration times: the search space of the first part is a 2-hop neighborhood of the current seed node, namely the node is considered to be randomly replaced by a neighbor or a neighbor of the neighbor; the second part of the search space is a set of several nodes (accounting for 10% of the network scale) with the maximum fitness value in the network, and the nodes are considered to be replaced by one of the nodes; the third part of the search space is a set of randomly selecting a certain number of nodes (10% of the network size) from the network, and considering replacing the nodes with one. According to the three search strategies, if the seed set after replacement has better fitness, the replacement operation is accepted;
(2e) And executing a selection operator, and selecting the globally optimal individual (node set) according to the fitness function. Specifically, the superior individuals are selected into offspring populations using a roulette method or algorithm. Some individuals with poor performance also have the opportunity of being selected, and the diversity of the population can be improved to a certain extent. In addition, the first individual of the offspring population stores the global optimal solution found in the current search process, so as to prevent population fitness from degrading;
(2f) And repeatedly executing the steps until the set termination condition is met. And finally, outputting the optimal solution (namely the seed node set S with the maximum fitness and the fitness thereof) appearing in the evolution process.
The method designs an index for measuring the robust influence performance of the seed node under the attack of the network node under the condition that the network structure is attacked by the node, the evaluation index can effectively evaluate the influence performance of the seed node under the condition that the network is attacked by the node, the calculation cost is low, the result has comparability under different network scales, and a reliable evaluation standard can be provided for the selection of the seed node set.
A method for solving the problem of robust influence maximization based on a 'diversity attention' principle is designed under the guidance of performance indexes for measuring the robust influence of seed nodes under the attack of network nodes, and a seed node set obtained by the method can obtain a good influence propagation effect in the network damage process.
It should be noted that the above list is only one specific embodiment of the present invention. It is clear that the invention is not limited to the embodiments described above, but that many variations are possible, all of which can be derived or suggested directly from the disclosure of the invention by a person skilled in the art, and are considered to be within the scope of the invention.
Claims (5)
1. A method for solving a problem of robust information dissemination over a network, comprising the steps of:
step S1, defining network structure at nodeAttacking the robust influence performance evaluation indexes of the following child nodes:
the N represents the number of nodes of the current network, the
Representing the influence estimated value of the seed node set S after Q nodes are removed;
and S2, designing a method based on a diversity attention principle under the guidance of the influence performance evaluation index to solve the problem of the maximization of the robust influence.
2. The method for solving the problem of robust information dissemination over a network as claimed in claim 1, wherein said "diversity attention" principle based method comprises the steps of:
(2a) Executing an initialization operator depending on a priority strategy, and generating a series of seeds to initialize a population;
(2b) Executing a crossover operator combined with a crossover strategy, and then exchanging gene information and expanding population quantity so as to generate a temporary population with the same number as the original population;
(2c) Executing mutation operators on the original population and the temporary population, thereby increasing the population diversity and getting rid of the local optimal solution;
(2d) Executing a local search operator on the original population and the temporary population;
(2e) Executing a selection operator, and selecting an overall best individual (node set) according to a fitness function;
(2f) And repeatedly executing the steps until a set termination condition is met, and finally outputting an optimal solution appearing in the evolution process.
3. The method for solving the problem of robust information dissemination over a network as claimed in claim 2, wherein said prioritization policies comprise a node-degree prioritization policy, a PageRank prioritization policy, a clustering prioritization policy.
4. The method for solving the problem of robust information dissemination over a network as recited in claim 2, wherein said crossover strategy comprises a uniform crossover strategy, a two-point crossover strategy, and a one-point crossover strategy.
5. System for solving the problem of robust information dissemination over a network, according to any of the claims 1-4, characterized in that said system comprises:
the defining module is used for defining the influence performance evaluation index of the seed node set S under node attack;
an initialization operator module for generating a series of seeds to initialize a population;
the temporary population generating module is used for generating a temporary population with the same number as the original population;
the mutation operator module is used for increasing the population diversity and getting rid of the local optimal solution;
the search module is used for executing a local search operator on the original population and the temporary population;
the execution operator module is used for storing or trying to update each iteration by adopting an elite strategy to obtain an optimal solution;
and the output module is used for outputting the optimal solution appearing in the evolution process.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210799978.3A CN115277115A (en) | 2022-07-06 | 2022-07-06 | Method and system for solving robust information propagation problem on network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210799978.3A CN115277115A (en) | 2022-07-06 | 2022-07-06 | Method and system for solving robust information propagation problem on network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115277115A true CN115277115A (en) | 2022-11-01 |
Family
ID=83765778
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210799978.3A Pending CN115277115A (en) | 2022-07-06 | 2022-07-06 | Method and system for solving robust information propagation problem on network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115277115A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116720293A (en) * | 2023-05-30 | 2023-09-08 | 中山大学 | Robust influence optimization method and device, electronic equipment and storage medium |
| CN117155786A (en) * | 2023-08-09 | 2023-12-01 | 中山大学 | Directed network optimization method and system for screening robust influence nodes |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105337773A (en) * | 2015-11-19 | 2016-02-17 | 南京邮电大学 | ReciprocityRank algorithm based microblogging network influence node discovering method |
-
2022
- 2022-07-06 CN CN202210799978.3A patent/CN115277115A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105337773A (en) * | 2015-11-19 | 2016-02-17 | 南京邮电大学 | ReciprocityRank algorithm based microblogging network influence node discovering method |
Non-Patent Citations (2)
| Title |
|---|
| 王帅: "复杂网络鲁棒性的分析、进化优化与应用研究" * |
| 王帅等: "一种针对网络结构破坏下鲁棒影响力最大化问题的Memetic算法" * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116720293A (en) * | 2023-05-30 | 2023-09-08 | 中山大学 | Robust influence optimization method and device, electronic equipment and storage medium |
| CN117155786A (en) * | 2023-08-09 | 2023-12-01 | 中山大学 | Directed network optimization method and system for screening robust influence nodes |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| He et al. | Modeling propagation dynamics and developing optimized countermeasures for rumor spreading in online social networks | |
| Buchegger et al. | The effect of rumor spreading in reputation systems for mobile ad-hoc networks | |
| CN115277115A (en) | Method and system for solving robust information propagation problem on network | |
| Gao et al. | Network immunization with distributed autonomy-oriented entities | |
| Hota et al. | Game-theoretic vaccination against networked SIS epidemics and impacts of human decision-making | |
| Fletcher et al. | Unstructured peer-to-peer networks: Topological properties and search performance | |
| Despotovic et al. | Maximum likelihood estimation of peers' performance in P2P networks | |
| CN109413710B (en) | Clustering method and device of wireless sensor network based on genetic algorithm optimization | |
| Chaudhry et al. | FZ enabled Multi-objective PSO for multicasting in IoT based Wireless Sensor Networks | |
| CN111723399B (en) | Large-scale social network directed graph privacy protection method based on k-kernel | |
| Zhu et al. | Optimization of cascading failure on complex network based on NNIA | |
| CN112149967A (en) | Power communication network vulnerability assessment method and system based on complex system theory | |
| CN113963195A (en) | Method for identifying key nodes in complex network | |
| CN108965287B (en) | Virus propagation control method based on limited temporary edge deletion | |
| Yan et al. | Structure optimization based on memetic algorithm for adjusting epidemic threshold on complex networks | |
| Przewoźniczek et al. | The transformation of the k-Shortest Steiner trees search problem into binary dynamic problem for effective evolutionary methods application | |
| Przewoźniczek | Subpopulation initialization driven by linkage learning for dealing with the Long-Way-To-Stuck effect | |
| Przewozniczek et al. | Empirical problem decomposition—the key to the evolutionary effectiveness in solving a large-scale non-binary discrete real-world problem | |
| CN111865793B (en) | IPv6 network service customized reliable routing system and method based on function learning | |
| CN107623586B (en) | Network link prediction method and device | |
| Xie et al. | A monte carlo graph search algorithm with ant colony optimization for optimal attack path analysis | |
| Zhao et al. | An approach to generate damage strategies for inter-domain routing systems based on multi-objective optimization | |
| Rezazad | Computer network optimization | |
| CN110232641B (en) | Privacy protection method based on network regulation and control mechanism of power information system | |
| Dai et al. | Group-based competitive influence maximization |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |