A method for constructing a group-based model for the spread of false information

Abstract

A method for constructing a group-based model for the spread of false information, comprising: First, constructing a network; Second, randomly selecting a certain proportion of intelligent nodes in the network; Third, assigning different abilities to filter information to ordinary nodes and the selected intelligent nodes; Fourth, setting the transition probabilities between node states; Fifth, performing propagation based on the method of cascading information propagation; Finally, calculating the information propagation range index. The present invention simulates the behavior of intelligent people in a group in the propagation model, sets intelligent nodes in the model to observe the influence of node behavior on the spread of false information, and explores the role of their remarks in the spread of group information.

Description

Technical Field

[0001] The present invention relates to the field of the spread of network false information, and in particular to a group-based social network false information spread model. Background Art

[0002] With the emergence of Web2.0, due to the popularity of various social media platforms (such as online social networks, Weibo, and WeChat), information can be spread quickly and widely. Millions of people discuss and share their topics on these social media, from public affairs to personal life, which greatly facilitates people's communication and information dissemination. Understanding the information dissemination process and revealing the dissemination characteristics of social media have important theoretical and application values. Therefore, it has also become one of the hottest topics. The spread of information can lead to large-scale contagion. True information is beneficial to people and can be widely spread; while the spread of some false information may cause serious social impacts and its spread should be weakened. Due to the high degree of freedom in social networks, it is more difficult to control the spread of information. Therefore, it is very important to study the characteristics and prevention and control of false information spread.

[0003] Chinese Patent Publication No. CN111797328 (a method for suppressing rumor spread in a social network) discloses a method for suppressing rumor spread in a social network. This method performs (k,η)-core decomposition on the network, calculates the number of immune individuals selected in each (k,η)-core based on the number of infected individuals and the core value of the (k,η)-core, and then calculates the rumor spreader influence of the infection set when the uninfected individuals are used as immune individuals according to the rumor spread model, and selects the individuals that minimize the rumor spread influence to join the immune individual set to obtain the final immune individual set. Immunize the individuals in the immune individual set to suppress rumor spread. This invention can quickly and effectively find the immune individuals in a large-scale social network to achieve the purpose of suppressing rumor spread. Publication No. CN119669958 (a method for establishing a rumor spread model) provides a rumor spread model that simultaneously considers individual activity and a rumor refutation mechanism. By improving the SIR model, a multi-factor model SIWISR-M model is proposed. The rumor refutation mechanism proposed by this invention can effectively reduce the peak value of the rumor spreader density and the rumor duration. Chinese Patent Publication No. CN111274496 (a method for constructing a rumor spread model considering groups on a heterogeneous network) discloses a rumor spread model based on the SEIR rumor spread model, combined with the group characteristics in today's mobile social network. This model randomly sets some groups in the network, and the spread mechanism includes group spread and node spread, and explores the role of groups in the spread in a heterogeneous network.

[0004] However, in the above-mentioned existing rumor propagation models or rumor suppression methods, there is no research on the role of the existence of intelligent nodes in information propagation, nor is the impact of their comments in the group on the information propagation process considered. Therefore, the model is not perfect.

[0005] The existence of groups in social networks enables information to be spread quickly and efficiently in a broadcast manner, such as in WeChat groups, QQ groups, and Weibo. At the same time, the existence of these groups also prompts individuals to conduct closed discussions on information, and such discussions play a role in fermenting or suppressing the spread of information. Summary of the Invention

[0006] In order to overcome the deficiencies in the prevention and control of false information in existing social networks, based on the independent cascade model and combined with the characteristics that people will widely participate in groups in social networks for discussion, the present invention proposes a method for implementing a false information propagation model based on groups, which is beneficial to the prevention and control of false information. The technical solutions adopted by the present invention to solve its technical problems are as follows:

[0007] A method for constructing a model for false information propagation based on groups, the method comprising the following steps:

[0008] S1. Construct a network: The network set is G = (V, E), and its node set and edge set are respectively V = {v 1 , v 2 , …, v n} and the total number of nodes N;

[0009] S2. Randomly select some intelligent nodes, and denote the set as V smart ;

[0010] S3. Assign the ability of nodes to filter information: For the intelligent nodes in the set V smart , the probability of receiving false information is 0, and for ordinary nodes, the probability of receiving false information is P n ;

[0011] S4. Set the conversion probability between node states;

[0012] S5. Based on the method of cascading information propagation: Each node in the model has three different states: 0 (susceptible), 1 (adopted), 2 (immune). Starting from the adopted node for information propagation, this node forwards the information to the group centered on itself (the construction method of the group is: composed of any node and its first-order neighbors), and passes it layer by layer. When all the nodes in the network have been traversed, it stops;

[0013] S6. Calculate the information propagation range index: Calculate the average value of the false information propagation range under different proportions of intelligent nodes and different silence probabilities to obtain the final result of the propagation.

[0014] The scale of the network described in the stated step S1 is N = 5000, and the average degree k = 10. The initial state of all nodes is set to susceptible (state = 0).

[0015] The stated step S2 specifically includes: randomly selecting a part of the nodes in the network as intelligent nodes (the intelligent nodes in the group do not express their opinions with a silence probability λ, or comment with a probability of 1 - λ, expressing their doubts about false information to influence the decisions of individuals in the group), and adding them to the set V smart The remaining nodes are ordinary nodes.

[0016] The stated step S3 specifically includes:

[0017] S3.1. The initial probability P of an intelligent node receiving false information S = 0;

[0018] S3.2. The initial probability of an ordinary node receiving false information is a random number P within the range n ∈(0, 0.5).

[0019] The specific content of the stated step S4 includes:

[0020] S4.1. After a susceptible (state = 0) receives a message, it will convert to an adopter (state = 1) with a probability P n λ k where k represents the number of intelligent nodes in the group. When the number of intelligent nodes in the group is k, the greater the silence probability λ of the intelligent nodes, the greater the probability that a susceptible will convert to an adopter; at the same time, given the silence probability of the intelligent nodes, the more the number of intelligent nodes k in the group, the smaller the probability that a susceptible will convert to an adopter;

[0021] S4.2. The adopters in the group will also be affected by the comments of the intelligent nodes, thus doubting the authenticity of the information. Each adopter (state = 1) will convert to an immune (state = 2) with a probability of 1 - λ k and remain unchanged (state = 1) with a probability of λ k ;

[0022] The probability formulas are set as follows:

[0023] P 0→1 = P n λ k (1)

[0024] P 0→2 = 1 - P n λ k (2)

[0025] P 1→1 = λ k (3)

[0026] P 1→2 = 1 - λ k (4)

[0027] where P 0→1 is the probability that a susceptible node is converted to an adopted node, P 0→2 is the probability that a susceptible node is converted to an immune node, P 1→1 is the probability that an adopted node maintains its own state, P 1→2 is the probability that an adopted node is converted to an immune node.

[0028] The specific steps of step S5 are as follows:

[0029] S5.1. Randomly select a node as the initial adopted node, and start information dissemination from the adopted node. This node forwards the information to the group centered on itself (the group consists of the current node and its first-order neighbors);

[0030] S5.2. Denote the group centered on the adopter as V i ; Denote the set of adopted nodes in the network as V adopted ; Denote the set of immune nodes in the network as V immune ; Denote the set of nodes that newly adopt information at each step in the network as V new_adopted ; Denote the set of temporarily stored nodes as V temp ; After a certain node i forwards the information, every node in the group can see the information, and at the same time, node i will forward the information with probability P i The process is as follows:

[0031] S5.3. Randomly select a node i in the set of ordinary nodes, take it as the adopted node (state = 1), as the source of information dissemination, forward the false message, and add this node to V adopted and V new_adopted , add the nodes in V new_adopted to V temp ;

[0032] S5.4. Arbitrarily select a node i from V temp , denote the group centered on i as V i , calculate the number of intelligent nodes in the group, denoted as k. For any node belonging to V i , calculate the probability that the node receives the information according to the network information dissemination model. When there are no intelligent nodes in the group or no comments are made, the dissemination is carried out according to the general independent cascade model;

[0033] S5.5. Traverse the nodes in the group. For each susceptible node i, generate a random number random between 0 and 1. If P i ≥ random, then node i forwards, add node i to V new_adopted , and the node is transformed into an adopted node (state = 1). If P i < random, then node i does not forward, and node i enters V immune , and the node is transformed into an immune node (state = 2);

[0034] S5.6. For the infected nodes in the group, generate a random number random between 0 and 1. If P i = λ k ≥ random, then the state of node i remains unchanged. If P i = λ k < random, then node i does not forward, node i enters V immune and node i is removed from the set V adopted , and the node state changes to an immune node (state = 2);

[0035] S5.7. Clear the nodes in V temp , add the nodes in V new_adopted to V adopted and V temp . Repeatedly repeat steps S5.4 - S5.7. When there are no more nodes in V new_adopted , the algorithm stops, indicating that the information dissemination ends at this time; V adopted represents the set of finally adopted nodes;

[0036] S5.8. Repeatedly repeat steps S5.3 - S5.7 until the number of iterations reaches 1000, and record the total number of transmissions;

[0037] S5.9. Change the proportion r of intelligent nodes and the silence probability λ of intelligent nodes, and repeat steps S3 - S6, and record the number of transmissions in each case.

[0038] The specific steps of step S6 are as follows:

[0039] S6.1. Denote the final number of adopters as N′, and the dissemination range of false information is as follows:

[0040]

[0041] S6.2. Obtain the dissemination situations of different networks under different proportions r of intelligent nodes and different silence probabilities λ of intelligent nodes.

[0042] The beneficial effects of the present invention are:

[0043] A method for constructing a group-based misinformation propagation model is proposed, which explains the role of intelligent nodes in a group. This model can help researchers better understand the speech of intelligent nodes in a group and the relationship between the existence of the group and the scope of misinformation propagation, and can provide ideas for suppressing misinformation propagation. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The drawings are used to provide an intuitive representation of the present invention and form a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation to the present invention.

[0045] Figure 1 It is a flowchart of the algorithm for a method for constructing a group-based misinformation propagation model of the present invention;

[0046] Figure 2 It is a graph showing the change of the misinformation propagation range with the silence probability of intelligent nodes at different p values in a random network;

[0047] Figure 3 It is a graph showing the change of the misinformation propagation range with time at different λ values in a random network;

[0048] Figure 4 It is a graph showing the change of the misinformation propagation range with the probability p of ordinary nodes receiving information at different λ and r values in a random network;

[0049] Figure 5 It is a graph showing the change of the information propagation range with the silence probability of intelligent nodes at different p values in a scale-free network;

[0050] Figure 6 It is a graph showing the change of the information propagation range with the silence probability of intelligent nodes at different p values in a real network. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0051] The following further describes in detail the specific embodiments of the present invention in conjunction with the drawings of the specification.

[0052] Embodiment 1

[0053] Refer to Figures 1 to 6 , a method for constructing a group-based misinformation propagation model. The present invention selects three different structural networks, namely a random network, a scale-free network, and a real network, as experimental networks. The above network structures are simple but representative, which can better help researchers understand the role of intelligent nodes and understand group-based misinformation propagation.

[0054] In this embodiment, a method for constructing a model for group-based false message propagation specifically includes the following steps:

[0055] A method for constructing a group-based model for the spread of false information, the method comprising the following steps:

[0056] S1. Construct a network: The network set is G=(V, E), and its node set and edge set are V={v 1 , v 2 , …, v n} and the total number of nodes N, the specific steps are as follows:

[0057] S1.1. Construct a random network with a scale of N = 5000 and an average degree of k = 10, and set the initial state of all nodes to susceptible (state = 0).

[0058] S2. Randomly select a certain proportion of intelligent nodes, and denote the set as V smart , the specific steps are as follows:

[0059] S2.1. Randomly select a part of the nodes in the network as intelligent nodes (the intelligent nodes in the group do not speak with a silence probability of λ, or make comments with a probability of 1 - λ, expressing their doubts about false information and influencing the decisions of individuals in the group), and add them to the set V smart , and the remaining nodes are ordinary nodes.

[0060] S3. Allocate the ability of nodes to filter information: For the intelligent nodes in the set V smart , the probability of accepting false information is 0, and for ordinary nodes, the probability of accepting false information is P n , the specific steps are as follows:

[0061] S3.1. The initial probability P S of an intelligent node accepting false information = 0;

[0062] S3.2. The initial probability of an ordinary node accepting false information is a random number P n ∈(0, 0.5).

[0063] S4. Set the transition probability between node states, the specific steps are as follows:

[0064] S4.1. After a susceptible (state = 0) receives a message, it will convert to an adopter (state = 1) with a probability of P n λ k , where k represents the number of intelligent nodes in the group. When the number of intelligent nodes in the group is k, the greater the silence probability λ of the intelligent nodes, the greater the probability that a susceptible will convert to an adopter; at the same time, given the silence probability of the intelligent nodes, the more the number of intelligent nodes k in the group, the smaller the probability that a susceptible will convert to an adopter;

[0065] S4.2. Adopters in the group are also influenced by the comments of intelligent nodes, thus doubting the authenticity of information. Each adopter (state = 1) becomes an immune node (state = 2) with probability 1 - λ k and remains unchanged (state = 1) with probability λ k ;

[0066] The probability formula is set as follows:

[0067] P o→1 = P n λ k (1)

[0068] P o→2 = 1 - P n λ k (2)

[0069] P 1→1 = λ k (3)

[0070] P 1→2 = 1 - λ k (4)

[0071] where P 0→1 is the probability that a susceptible node converts to an adopted node, P 0→2 is the probability that a susceptible node converts to an immune node, P 1→1 is the probability that an adopted node remains in its own state, and P 1→2 is the probability that an adopted node converts to an immune node.

[0072] S5. Method based on cascading information propagation: Each node in the model has three different states: 0 (susceptible), 1 (adopted), 2 (immune). Starting from an adopted node, information is propagated. This node forwards the information to the group centered on itself (the group is constructed as: consisting of any node and its first-order neighbors), and it is passed layer by layer until all nodes in the network have been traversed and then stops. The specific steps are as follows:

[0073] S5.1. Randomly select a node as the initial adopted node, and start information propagation from this adopted node. This node forwards the information to the group centered on itself (the group consists of the current node and its first-order neighbors);

[0074] S5.2. The group centered on the adopter is denoted as V i ; The set of adopted nodes in the network is denoted as V adopted ; The set of immune nodes in the network is denoted as V immune ; The set of nodes that newly adopt information at each step in the network is denoted as V new _ adopted; The set of temporary storage nodes is denoted as V temp ; When a certain node i forwards information, every node in the group can see the information, and at the same time, node i will forward the information with probability P i The forwarding process is as follows:

[0075] S5.3. Randomly select a node i from the set of ordinary nodes, use it as the adopted node (state = 1), which is the source of information dissemination, forward the false message, and add this node to V adopted and V new_adopted Add the nodes in V new_adopted to V temp ;

[0076] S5.4. Arbitrarily select a node i from V temp The group centered on i is denoted as V i , calculate the number of intelligent nodes in the group, denoted as k. For any node belonging to V i , calculate the probability of the node receiving information according to the network information dissemination model. When there are no intelligent nodes in the group or no comments are made, the dissemination is carried out according to the general independent cascade model;

[0077] S5.5. Traverse the nodes in the group. For each susceptible node i, generate a random number random between 0 and 1; if P i ≥ random, then node i forwards and adds node i to V new_adopted , and the node is transformed into an adopted node (state = 1); if P i < random, then node i does not forward, node i enters V immune , and the node is transformed into an immune node (state = 2);

[0078] S5.6. For the infected nodes in the group, generate a random number random between 0 and 1. If P i = λ k ≥ random, then the state of node i remains unchanged; if P i = λ k < random, then node i does not forward, node i enters V immune and node i is removed from the set V adopted , and the node state changes to an immune node (state = 2);

[0079] S5.7. Clear the nodes in V temp , add the nodes in V new_adopted to V adopted and V twmp , repeat steps S5.4 - S5.7 continuously. When V new_adoptedWhen there are no more nodes, the algorithm stops, indicating that the information dissemination ends at this time; V adopted Indicates the set of nodes finally adopted;

[0080] S5.8. Continuously repeat steps S5.3 - S5.7 until the number of iterations reaches 1000, and record the total number of transmissions;

[0081] S5.9. Change the proportion r of intelligent nodes and the silence probability of intelligent nodes, and repeat steps S3 - S6, recording the number of transmissions in each case.

[0082] S6. Calculate the information dissemination range index: Calculate the average value of the false information dissemination range under different proportions of intelligent nodes and different silence probabilities to obtain the final result of the dissemination. The specific steps are as follows:

[0083] S6.1. Denote the final number of adopters as N′, and the dissemination range of false information is as follows:

[0084]

[0085] S6.2. Obtain the dissemination situations of different networks under different proportions r of intelligent nodes and different silence probabilities λ of intelligent nodes.

[0086] Figure 2 is the graph of the change of the false information dissemination range with the silence probability of intelligent nodes at different p values in a random network. Figure 3 is the graph of the change of the false information dissemination range with time at different λ values in a random network. Figure 4 is the graph of the change of the false information dissemination range with the probability p of ordinary nodes receiving information at different λ and r values in a random network. Figure 5 , 6 are respectively the graphs of the information dissemination range with the silence probability of intelligent nodes at different p values in a scale - free network and a real network. From Figure 2 it can be obtained that when the silence probability of intelligent nodes is greater than a certain threshold, false information can spread. Comparing Figure 2 , 5, 6 shows that false information is more difficult to spread in a scale - free network and there is a larger threshold. Figure 3 Indicates that with the increase of time steps, the dissemination range of false information expands rapidly, and then the network remains stable. As λ increases, the dissemination range of false information gradually decreases. Figure 4 Indicates that the more intelligent nodes there are in the network, the more difficult it is for false information to spread, and only when it is greater than a certain threshold can false information spread.

[0087] As described above, the embodiments of the present invention on random networks, scale-free networks and real networks are introduced. We propose a method for constructing a group-based false information propagation model. When there are intelligent nodes in the social network, the speech of the intelligent nodes may have an impact on the nodes in the group. Based on this, we propose the above method. By combining it with simulation experiments, the model can help researchers better understand the role of intelligent nodes in group propagation and play the role of intelligent nodes in suppressing the spread of false information.

[0088] Example 2

[0089] This embodiment provides a method for suppressing the spread of false information using a method for constructing a group-based false information propagation model of the present invention, comprising the following steps:

[0090] Steps S1 to S6 of this embodiment are the same as those of Embodiment 1, and further include the following steps:

[0091] S7. Specific application method. The specific steps are as follows:

[0092] S7.1. Select influential nodes in social networks as "smart nodes", such as professors and experts from well-known universities. These people have relative authority and knowledge, their opinions are more credible, and they also have the attributes of "smart nodes".

[0093] S7.2. Influential people make comments. When they see false information on the Internet, they can post their opinions on the false information on platforms such as WeChat Moments, WeChat groups or Weibo to influence the decisions of the people around them.

[0094] S7.3. The demise of false information. When the truth relative to false information begins to spread on the Internet, the masses who would have believed in the false information will no longer believe in the false information and will not spread it. As time goes by, the false information will not be able to reach a cascade phenomenon on the Internet, that is, it will only spread on a small scale. This will achieve the purpose of suppressing the spread of information.

[0095] The above description is only one embodiment of the present invention and is not intended to limit the present invention. Any modification, equivalent substitution, improvement, etc. made within the concept and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for constructing a group-based model for false information dissemination, comprising the following steps: S1. Build a network: The network set is , and its node set and edge set are and respectively. The total number of nodes is . S2. Randomly select some intelligent nodes and denote the set as ; The intelligent nodes are nodes with influence in the social network; S3. Ability to distribute filtering information of nodes: For the set of intelligent nodes, the probability of receiving false information is 0. For ordinary nodes, the probability of receiving false information is ； Specifically, it includes: S3.

1. Initial probability of the intelligent node accepting false information ; S3.

2. The initial probability of an ordinary node accepting false information is a random number within the interval range ; S4. Set the transition probabilities between node states; S5. Based on the method of cascading information dissemination: Each node in the model has three different states: 0, 1, 2, where 0 represents susceptible, 1 represents adopted, and 2 represents immune. Information dissemination starts from the adopted node, and this node forwards the information to the group centered on itself, passing it layer by layer, and stops when all nodes in the network have been traversed; specifically, it includes: S5.

1. Randomly select a node as the initial adopted node, and start information dissemination from the adopted node. This node forwards the information to the group centered on itself, and the group consists of the current node and its first-order neighbors; S5.

2. The adopter-centered group is denoted as ; The set of adopted nodes in the network is denoted as ; The set of immune nodes in the network is denoted as ; The set of nodes that newly adopt information at each step in the network is denoted as ; the set of nodes for temporary storage is denoted as ; when a certain node forwards information, every node in the group can see the information, and at the same time, node will forward the information with probability : S5.

3. Randomly select a node from the set of ordinary nodes , use it as the adopted node, which serves as the source of information dissemination, forward the false message, and add this node to and . Add the nodes in to ; S5.

4. Select any one node from and denote the group centered on as . Calculate the number of intelligent nodes in the group, denoted as . For any node belonging to , calculate the probability of the node receiving information according to the network information propagation model. When there are no intelligent nodes in the group or no comments are made, the propagation is carried out according to the general independent cascade model;​ S5.

5. Traverse the nodes in the group. For each susceptible node , generate a random number between 0 and 1 ; if ≥ , then node performs forwarding, adds node to , and the node is transformed into an adopted node < , then node does not perform forwarding, and node enters , and the node is transformed into an immune node S5.

6. For the infected nodes in the group, generate a random number between 0 and 1 , if ≥ , then node remains in the same state; if < , then node does not perform forwarding, node enters and node is removed from the set , and the node state changes to an immune node S5.

7. Clear the nodes in , add the nodes in to and , and repeat steps S5.4 - S5.7 continuously. When there are no more nodes in , the algorithm stops, indicating that the information dissemination ends at this time; represents the final set of adopted nodes; S5.

8. Continuously repeat steps S5.3 - S5.7 until the number of iterations reaches 1000, and record the total number of disseminations; S5.9, Change the proportion of intelligent nodes , and the silent probability of intelligent nodes, repeat steps S2 - S5.8, and record the number of transmissions in each case; S6. Calculate the information dissemination range index: Calculate the average value of the false information dissemination range under different proportions of intelligent nodes and different silence probabilities to obtain the final result of dissemination.

2. A method for constructing a group-based model for false information dissemination according to claim 1, wherein, The total number of nodes in the network described in step S1 , the average degree , set the initial state of all nodes to susceptible, that is .

3. A method for constructing a group-based model for false information dissemination according to claim 1, wherein, In the step S2, some intelligent nodes are randomly selected, and the intelligent nodes in the group do not make statements with a silence probability or make comments with to question the false information and influence the decision-making of individuals in the group, that is, randomly select a part of nodes in the network as intelligent nodes and add them to the set while the remaining nodes are ordinary nodes.

4. A method for constructing a group-based model for false information dissemination according to claim 1, wherein, In step S4, when setting the transition probabilities between node states, the specific steps are as follows: S4.

1. After receiving the message, the susceptibles will convert into adopters with a probability That is, , where represents the number of intelligent nodes in the group. When the number of intelligent nodes in the group is , the larger the silence probability of the intelligent nodes, the greater the probability that the susceptibles will convert into adopters. At the same time, given the silence probability of the intelligent nodes, the more the number of intelligent nodes in the group, the smaller the probability that the susceptibles will convert into adopters. S4.

2. Adopters in the group are also influenced by the comments of the intelligent nodes, thus doubting the authenticity of the information. Each adopter converts to an immune individual with probability , that is , and remains unchanged with probability . The probability formula is set as follows: , where is the probability that a susceptible node is converted into an adopted node, is the probability that a susceptible node is converted into an immune node, is the probability that an adopted node maintains its own state, is the probability that an adopted node is converted into an immune node.

5. A method for constructing a group-based model for false information dissemination according to claim 1, wherein, In step S6, the specific steps are as follows: S6.

1. Denote the final number of adopters as , and the scope of spread of false information is as follows: ; S6.

2. Obtain different intelligent node ratios , different intelligent node silence probabilities and the propagation conditions of different networks under

Images