A method to hide key figures in social networks based on degree-preserving attack

By applying guarantee attacks and genetic algorithms to change the node core value in social networks, combined with the K-core decomposition algorithm, the problem that traditional methods are easily discovered by key nodes is solved, and the hiddenness of key person nodes and network security is improved.

CN114301674BActive Publication Date: 2024-11-08ZHEJIANG UNIV OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111626397.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-12-28
Publication Date
2024-11-08
Estimated Expiration
2041-12-28

AI Technical Summary

Technical Problem

Traditional methods are easily easily discovered by hackers by using node sorting algorithms to discover key nodes in social networks, resulting in damage to network security and the existing technology cannot effectively conceal the nodes of key people.

Method used

Using a method based on guarantee attack, the links that meet specific conditions are screened by reconnecting the network links and using genetic algorithms to filter the nodes that meet specific conditions, and combining with the K-core decomposition algorithm, the hiddenness of key figure nodes in the social network is achieved. Keep the node degree unchanged.

Benefits of technology

Effectively hide key person nodes in social networks, improve network security, improve attack resistance, and avoid easy identification by hackers.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114301674B_ABST
    Figure CN114301674B_ABST
Patent Text Reader

Abstract

The method for hiding key figures in a social network based on a degree-preserving attack comprises: step S1, assuming an undirected and unweighted network G, using a K-core decomposition algorithm to assign a core value to each node in the network, and defining the node with the highest core value in the social network as a key figure; step S2, presetting a reconnection scheme based on the degree-preserving attack; step S3, selecting a link according to the reconnection scheme described in step S2, and performing a reconnection operation using a genetic algorithm; step S4, evaluating the hiding effect of the genetic algorithm, and outputting an optimal strategy for hiding the key figure; the present invention utilizes the degree-preserving attack method and integrates a classic evolutionary algorithm, so as to ensure that the nodes where the key figures in the social network are located are effectively hidden without being discovered by hacker attacks, thereby improving the security of the network.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to data mining and network node processing technology, and in particular to a method for hiding key figures in a social network based on a degree-preserving attack. Background Art

[0002] In recent years, with the rapid development of information technology and computer technology, network science plays an important role in various fields of life. Nowadays, the scale of these networks is getting larger and more complex, and the difficulty of analysis and the value of application are also increasing accordingly. Therefore, judging the importance and attention of certain key nodes in a network and other nodes has become a research focus in the field of network science, and plays an important role in various networks such as communication networks, power networks, social networks and transportation networks. For example, in social networks, friend attention plays an important role in recommendation systems, correlations between users and predicting user behavior patterns. Because the importance and attention of user nodes are different, some users are opinion leaders in a certain field, while some users are just ordinary netizens. Therefore, reasonable node importance ranking can help evaluate the importance of the current user node to the entire social network, and facilitate the subsequent processing of network division, such as community division, and also help improve the performance of the algorithm and the effect of the model. For example, some researchers claim that if a key individual species is accurately identified, then protecting key animals can help the population reproduce and maintain stability, achieve sustainable production of diversified food supply, and ultimately repair the ecosystem damaged by illegal hunting. In the field of network security and cloud computing, studies have shown that when 5% to 10% of key nodes are attacked, the entire network will be paralyzed. For example, in 2003, a tripped interconnection line between the Swiss and Italian power grids caused a large-scale power outage across Italy; there are often several important transportation hubs in a city's transportation network, and when these few important hubs are congested, it will indirectly cause traffic congestion throughout the city; in the current rapid development of short videos and live broadcasts, it is often a small number of self-media or anchors who control a large amount of platform resources and traffic. Protecting key nodes can not only improve the survivability and robustness of the entire network, but also improve the network's ability to resist conventional attacks.

[0003] In traditional networks, hackers can easily find key nodes in the network through node sorting algorithms. To this end, the present invention takes the classic K-core decomposition algorithm as an example, and can disrupt the core value of network nodes through the method of counterattack, so that the K-core decomposition algorithm cannot identify the key nodes in the network, thereby achieving an anonymity effect. The core value of a node is obtained by allocating the K-core decomposition algorithm to each node in the network during network initialization. The present invention defines the node with the highest core value in a social network as a key person. By using the method of degree-preserving attack, it can be ensured that when the attack disturbance meets the restrictions, the nodes where the key persons in the social network are located are effectively hidden. The so-called anonymity is achieved by the present invention to change the core value of the node.

[0004] Patent CN 108449311 A proposes a social relationship hiding method based on attack node similarity. The main core is to use node similarity to hide the links between nodes, and protect the user's social privacy from the perspective of link prediction, without performing key node hiding protection based on the topological properties of the network node itself and its neighboring nodes. Summary of the invention

[0005] The present invention aims to overcome the above-mentioned shortcomings of the prior art and provide a method for hiding key figures in social networks based on degree-preserving attacks.

[0006] The present invention is based on the classic network node importance ranking index and K-core decomposition algorithm. Through degree-preserving attack, that is, reconnecting some links of the original network without changing the degree value of the node, the genetic algorithm is integrated to screen the links that meet the restriction conditions. After countering the attack and based on the principle of K-core decomposition, the core value of the network node is changed, and an erroneous network node importance ranking is obtained, which effectively hides the nodes where the key figures in the social network are located.

[0007] The technical solutions adopted by the present invention to achieve the above-mentioned invention object are as follows:

[0008] The method for hiding key figures in social networks based on degree-preserving attack includes the following steps:

[0009] S1: Assume an undirected and unweighted network G, use the K-core decomposition algorithm to assign a core value to each node in the network, and define the node with the highest core value in the social network as the key person;

[0010] S2: pre-set a reconnection scheme based on degree-preserving attack;

[0011] S3: Select a link according to the reconnection scheme described in step S2, and perform a reconnection operation using a genetic algorithm;

[0012] S4: Evaluate the concealment effect of the genetic algorithm and output the optimal strategy for concealing key figures.

[0013] Preferably, in step S1: assuming an undirected unweighted social network G(V,E), V is a set of nodes in the network, E is a set of links in the network, and a K-core decomposition algorithm is used to assign a core value to each node in the network. The core value of a node is equivalent to the importance of the node. The higher the core value of a node, the higher the importance of the node. The node with the highest core value in the social network is defined as the key person V. k .

[0014] Preferably, in step S2: a reconnection scheme based on a degree-preserving attack is pre-set, the so-called degree-preserving attack means that during the attack, the degree value of the node is not changed, a pair of links to be reconnected are selected, the links to be reconnected include four different nodes, that is, there is no common node between the links to be reconnected, and the link after the four nodes are reconnected does not exist in the original network. The above are the restriction conditions of the reconnection scheme based on the degree-preserving attack, which can ensure that after the network attack, the degree value of each node in the network will not change.

[0015] Preferably, in step S3:

[0016] S3.1: For the selected link, determine whether it meets the restriction conditions described in step S2; the specific operation is: first randomly select two links from the link list to form a gene, and determine whether the following two restriction conditions are met: 1) whether the gene has a common vertex; 2) whether the link after the gene is reconnected exists in the original network; if one of the restriction conditions is not met, the gene is discarded; a number of genes constitute an individual, check whether there are identical genes in the individual, and if so, only retain one of them; set the gene length of the individual to include the number of genes that meet the restriction conditions 1) and 2), and iterate repeatedly until all genes meet the conditions; this step helps to avoid invalid reconnection operations from interfering with the overall network while reducing subsequent computing overhead; because genes with common nodes will generate self-loops during the reconnection process, destroying the connectivity of the network, and the existence of two identical genes in an individual will cause the effect of executing a round of reconnection to fail and increase the computing overhead;

[0017] S3.2: Based on the obtained population, calculate the fitness of each individual in the population; first, calculate the kernel value distribution of the original network node without executing the degree-preserving attack; then calculate the kernel value distribution of the network node obtained after the attack according to the reconnection scheme of the individuals in the population; record the node whose kernel value of the key person changes as the misclassified node V f ; The kernel value misclassification rate is calculated as shown in formula (1):

[0018]

[0019] Where |Vf |Total number of misclassified nodes,|V k | is the total number of key persons in the network before the attack using the reconnection scheme;

[0020] The kernel value is an important indicator for evaluating the importance of nodes and is widely used in the field of node classification. The fitness function is defined as formula (2):

[0021] f=1-d attack (2)

[0022] where d attack is the misclassification rate of the kernel value, and the fitness function represents the success rate of concealment. The smaller the value it returns, the higher the success rate of concealment.

[0023] S3.3: Roulette algorithm is set as the selection operator of the genetic algorithm. Through this selection method, the probability of individual mating is proportional to its fitness; single-point crossover is used as the crossover operator of the genetic algorithm; in order to reduce the time complexity of the experiment in this article and improve efficiency, the mutation rate of the genetic algorithm is set to 10%, and the population size is not more than 16; the next generation of population is obtained after the selection, crossover and mutation operations.

[0024] Preferably, in step S4: for the genetic algorithm described in step S3, the individual with the highest fitness in the best record population is used, and at the same time, a random reconnection attack is performed on the undirected unweighted network G, and the degree of the network node is kept unchanged, so that the genetic algorithm described in step S3 is compared with the random reconnection attack, and the hiding success rate of key figures in the social network and the disturbance rate of the social network are used as indicators to measure the hiding effect, wherein the disturbance rate of the social network is measured according to the number of reconnections of the links in the network, and its calculation formula is shown in formula (3):

[0025]

[0026] Where |E f | is the total number of links changed after the reconnection attack, |E| is the total number of links in the network;

[0027] When the hiding effect of the genetic algorithm is better than that of the random reconnection attack, it means that the hiding is successful, and the individual best with the highest fitness is output, which is the optimal strategy for successful hiding in the network graph after the attack; if the hiding effect of the genetic algorithm is worse than that of the random reconnection attack, the genetic algorithm is iterated again until the hiding effect of the genetic algorithm is better than that of the random reconnection attack.

[0028] For real social networks, the node importance ranking algorithm provided by the K-core decomposition algorithm is very sensitive to disturbances in the network topology, which makes it unreliable for incomplete or noisy systems. The present invention takes advantage of this characteristic of social networks and uses a degree-preserving attack. The attack process keeps the degree of the node unchanged. The motivation of this method also lies in the fact that the main contribution to the core value of the node is its degree. Therefore, the reconnection scheme based on the degree-preserving attack can not change the degree of the node. Such disturbance makes the attack not easy to be discovered by the defense detection algorithm. By using the links between neighboring nodes in the network topology structure, the link perturbation of the network is carried out. Under the node importance measurement based on the K-core decomposition algorithm, a better protection effect than random perturbation is achieved, and the security of the network is maintained more effectively.

[0029] The advantages of the present invention are: by utilizing the degree-preserving attack method and integrating the classic evolutionary algorithm, it is possible to effectively conceal the nodes where key figures in the social network are located without being discovered by hacker attacks, thereby improving the security of the network. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 is a flow chart of the method of the present invention;

[0031] Figure 2 is a flow chart of the genetic algorithm of step S3 of the present invention;

[0032] Figure 3 The present invention selects four different nodes in the network to form a pair of links. ij and e uv , after attacking according to the reconnection scheme, a new link e that does not exist in the original network is generated iv and e ju Schematic diagram of . DETAILED DESCRIPTION

[0033] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0034] In order to improve the anonymity of node importance, the present invention is based on the classic network node importance ranking index and K-core decomposition algorithm. Through degree-preserving attack, that is, reconnecting some links of the original network without changing the degree value of the node, the genetic algorithm is integrated to screen the links that meet the restriction conditions. After the counter-attack and based on the principle of K-core decomposition, the core value of the network node is changed, and the wrong network node importance ranking is obtained, which effectively hides the nodes where the key figures in the social network are located.

[0035] The technical solutions adopted by the present invention to achieve the above-mentioned invention object are as follows:

[0036] The method for hiding key figures in social networks based on degree-preserving attack includes the following steps:

[0037] S1: Assume an undirected and unweighted network G, use the K-core decomposition algorithm to assign a core value to each node in the network, and define the node with the highest core value in the social network as the key person;

[0038] S2: pre-set a reconnection scheme based on degree-preserving attack;

[0039] S3: Select a link according to the reconnection scheme described in step S2, and perform a reconnection operation using a genetic algorithm;

[0040] S4: Evaluate the concealment effect of the genetic algorithm and output the optimal strategy for concealing key figures.

[0041] In step S1: Assume an undirected and unweighted social network G(V,E), where V is the set of nodes in the network and E is the set of links in the network. Use the K-core decomposition algorithm to assign a core value to each node in the network. The core value of a node is equivalent to the importance of the node. The higher the core value of a node, the higher the importance of the node. The node with the highest core value in the social network is defined as the key person V. k .

[0042] In step S2: a reconnection scheme based on degree-preserving attack is preset. The so-called degree-preserving attack means that the degree value of the node is not changed during the attack process. A pair of links to be reconnected are selected, and the links to be reconnected include four different nodes, that is, there is no common node between the links to be reconnected, and the link after the four nodes are reconnected does not exist in the original network. The above are the restriction conditions of the reconnection scheme based on degree-preserving attack, which can ensure that the degree value of each node in the network will not change after the network attack.

[0043] Step S3 specifically includes:

[0044] S3.1: For the selected link, determine whether it meets the restriction conditions described in step S2; the specific operation is: first randomly select two links from the link list to form a gene, and determine whether the following two restriction conditions are met: 1) whether the gene has a common vertex; 2) whether the link after the gene is reconnected exists in the original network; if one of the restriction conditions is not met, the gene is discarded; a number of genes constitute an individual, check whether there are identical genes in the individual, and if so, only retain one of them; set the gene length of the individual to include the number of genes that meet the restriction conditions 1) and 2), and iterate repeatedly until all genes meet the conditions; this step helps to avoid invalid reconnection operations from interfering with the overall network while reducing subsequent computing overhead; because genes with common nodes will generate self-loops during the reconnection process, destroying the connectivity of the network, and the existence of two identical genes in an individual will cause the effect of executing a round of reconnection to fail and increase the computing overhead;

[0045] S3.2: Based on the obtained population, calculate the fitness of each individual in the population; first, calculate the kernel value distribution of the original network node without executing the degree-preserving attack; then calculate the kernel value distribution of the network node obtained after the attack according to the reconnection scheme of the individuals in the population; record the node whose kernel value of the key person changes as the misclassified node V f ; The kernel value misclassification rate is calculated as shown in formula (1):

[0046]

[0047] Where |V f |Total number of misclassified nodes,|V k | is the total number of key persons in the network before the attack using the reconnection scheme;

[0048] The kernel value is an important indicator for evaluating the importance of nodes and is widely used in the field of node classification. The fitness function is defined as formula (2):

[0049] f=1-d attack (2)

[0050] where d attack is the misclassification rate of the kernel value, and the fitness function represents the success rate of concealment. The smaller the value it returns, the higher the success rate of concealment.

[0051] S3.3: Roulette algorithm is set as the selection operator of the genetic algorithm. Through this selection method, the probability of individual mating is proportional to its fitness; single-point crossover is used as the crossover operator of the genetic algorithm; in order to reduce the time complexity of the experiment in this article and improve efficiency, the mutation rate of the genetic algorithm is set to 10%, and the population size is not more than 16; the next generation of population is obtained after the selection, crossover and mutation operations.

[0052] In step S4: for the genetic algorithm described in step S3, the individual with the highest fitness in the best record population is used. At the same time, a random reconnection attack is performed on the undirected unweighted network G, and the degree of the network node is kept unchanged. The genetic algorithm described in step S3 is compared with the random reconnection attack, and the hiding success rate of key figures in the social network and the disturbance rate of the social network are used as indicators to measure the hiding effect. The disturbance rate of the social network is measured according to the number of reconnections of the links in the network, and its calculation formula is shown in formula (3):

[0053]

[0054] Where |E f | is the total number of links changed after the reconnection attack, |E| is the total number of links in the network;

[0055] When the hiding effect of the genetic algorithm is better than that of the random reconnection attack, it means that the hiding is successful, and the individual best with the highest fitness is output, which is the optimal strategy for successful hiding in the network graph after the attack; if the hiding effect of the genetic algorithm is worse than that of the random reconnection attack, the genetic algorithm is iterated again until the hiding effect of the genetic algorithm is better than that of the random reconnection attack.

[0056] As described above, the present invention is introduced in an embodiment. The present invention is a method for hiding key figures in a social network based on a degree-preserving attack. The final results show that the method for hiding key figures in a social network based on a degree-preserving attack is effective, can successfully hide key figure nodes in a social network, effectively maintain the security of the network, and meets the requirements of practical use.

[0057] The contents described in the embodiments of this specification are merely an enumeration of the implementation forms of the inventive concept. The protection scope of the present invention should not be regarded as limited to the specific forms described in the embodiments. The protection scope of the present invention also extends to equivalent technical means that can be conceived by those skilled in the art based on the inventive concept.

Claims

1. A method for hiding key figures in social networks based on degree-preserving attacks, characterized by: The following steps are involved: S1: In the undirected and unweighted network G, a core value is assigned to each node in the network using the K-core decomposition algorithm, and the node with the highest core value in the social network is defined as the key person; S2: Preset a reconnection scheme based on the degree-preserving attack; specifically including: A reconnection scheme based on a degree-preserving attack is preset. The so-called degree-preserving attack means that the degree value of the node is not changed during the attack process; a pair of links to be reconnected are selected, the links to be reconnected include four different nodes, there is no common node between the links to be reconnected, and the link after the four nodes are reconnected does not exist in the original network. The above are the restriction conditions of the reconnection scheme based on the degree-preserving attack, which ensures that the degree value of each node in the network will not change after the network attack; S3: Select a link according to the reconnection scheme described in step S2, and perform a reconnection operation using a genetic algorithm; specifically including: S3.1: For the selected link, determine whether it meets the restriction conditions described in step S2; the specific operation is: first randomly select two links from the link list to form a gene, and determine whether the following two restriction conditions are met: 1) whether the gene has a common vertex; 2) whether the link after the gene is reconnected exists in the original network; if one of the restriction conditions is not met, the gene is discarded; several genes form an individual, check whether there are the same genes in the individual, if so, only keep one of them; set the gene length of the individual, the gene length includes the number of genes that meet the restriction conditions 1) and 2), and iterate repeatedly until all genes meet the conditions; S3.2: Based on the obtained population, calculate the fitness of each individual in the population; first, calculate the kernel value distribution of the original network nodes that have not performed the degree-preserving attack; then calculate the kernel value distribution of the network nodes obtained after the attack based on the reconnection scheme of the individuals in the population; record the node whose kernel value of the key person changes as the misclassified node V f ; The kernel value misclassification rate is calculated as shown in formula (1): Where |V f |Total number of misclassified nodes,|V k | is the total number of key persons in the network before the attack using the reconnection scheme; The fitness function is defined as formula (2): f=1-d attack (2) where d attack is the misclassification rate of the kernel value, and the fitness function represents the success rate of concealment. The smaller the value it returns, the higher the success rate of concealment. S3.3: Set the roulette algorithm as the selection operator of the genetic algorithm; use single-point crossover as the crossover operator of the genetic algorithm; set the mutation rate of the genetic algorithm to 10%, and the population size is not more than 16; after the population is subjected to selection, crossover, and mutation operations, the next generation population is obtained; S4: Evaluate the concealment effect of the genetic algorithm and output the optimal strategy for concealing key figures, including: For the genetic algorithm described in step S3, the best record is used to record the individual with the highest fitness in the population. At the same time, a random reconnection attack is performed on the undirected unweighted network G, and the degree of the network node is kept unchanged. The genetic algorithm described in step S3 is compared with the random reconnection attack. The hiding success rate of key figures in the social network and the disturbance rate of the social network are used as indicators to measure the hiding effect. The disturbance rate of the social network is measured according to the number of reconnections of the links in the network. The calculation formula is shown in formula (3): Where |E f | is the total number of links changed after the reconnection attack, |E| is the total number of links in the network; When the hiding effect of the genetic algorithm is better than that of the random reconnection attack, it means that the hiding is successful, and the individual best with the highest fitness is output, which is the optimal strategy for successful hiding in the network graph after the attack; if the hiding effect of the genetic algorithm is worse than that of the random reconnection attack, the genetic algorithm is iterated again until the hiding effect of the genetic algorithm is better than that of the random reconnection attack.

2. The method for hiding key figures in a social network based on a degree-preserving attack as claimed in claim 1, characterized in that: The step S1 comprises: Undirected unweighted network G(V,E), V is the set of nodes in the network, E is the set of links in the network, and a core value is assigned to each node in the network using the K-core decomposition algorithm. The core value of a node is equivalent to the importance of the node. The higher the core value of a node, the higher the importance of the node. The node with the highest core value in the social network is defined as the key person V k .