Token-based influence user selection method under dynamic social network

Abstract

The invention provides a token-based influence user selection method under a dynamic social network, which comprises the following steps of: 1, acquiring information of users in the social network and relationship information among the users, generating the intimacy degree among the users into a weight of an edge in the network, and constructing a whole social network model; 2, each user in the social network can generate a certain number of tokens according to own original influence; 3, distributing tokens to neighbor users according to a network node relation and a self-influence rule after each round of continuous generation through an iteration mechanism; 4, the number of tokens absorbed by the social network users represents the influence of the users; 5, selecting a proper z seed set through final convergence judgment; and 6, verifying the effect of the result seed set through an IC diffusion model. According to the method, the token is used as a medium, the problem of influence maximization of the social network is solved from a distributed perspective, a user selection model is provided, and the method is suitable for large-scale and dynamically changing network scenes and is low in cost.

Description

Technical Field

[0001] The present invention belongs to the field of social network and influence maximization, and specifically relates to a token-based influential user selection method in a dynamic social network. Background Art

[0002] In the past two decades, with the development of the "word of mouth" effect, the influence maximization problem has become one of the hottest research topics in the field of social network research. This influence maximization problem is to select a group of influential users, namely seeds, who spread their influence to other users in the social network. With the support of influential users, businesses are launched to attract more customers and quickly gain high visibility and huge benefits in the social network.

[0003] In order to solve the problem of maximizing influence, many user selection models have been proposed. Current user selection models can be divided into two types of selection models: centralized and distributed, among which the centralized user selection model is the mainstream model. In the centralized user selection model, there is a central controller that takes a global view of the social network, collects the status and relationship information of users, and selects influential users based on the collected information. Some people use Monte Carlo to select nodes with the highest marginal benefits as influential nodes; there are also heuristic algorithms based on degree discount to select influential users. Once a user is selected as an influential user, its neighbor user relationship will be reduced, and influential users are selected iteratively based on this mechanism; others propose to use an integrated Pagerank algorithm to solve the problem of maximizing influence. In the past two decades, with the development of the Internet and social media, the user scale of social networks has become larger and larger, and the status and relationship of users in social networks are changing dynamically. In this case, the centralized user selection model has two main limitations. On the one hand, as the user scale grows, the network communication overhead of the central controller in collecting user information in social networks is increasing significantly; on the other hand, many centralized user selection models are based on computationally expensive mechanisms. The dynamic changes of users in social networks will lead to constant changes in their social influence, which requires the central controller to reselect influential users, thereby significantly increasing the computational overhead of the central controller.

[0004] In order to overcome the above limitations of the centralized user selection model, distributed user selection models have been proposed in recent years. For example, a distributed influence maximization model, SIMiner, has been proposed, which uses mobile agents to simulate ants based on the ant colony algorithm. In SIMiner, ants walk through users in a social network and select influential users by distributing pheromones. Although SIMiner overcomes the limitations of high communication and computational overhead of centralized user selection, it also has two limitations. First, SIMiner selects influential users in a social network by introducing external mobile agents, namely "ants". Therefore, the model overhead of SIMiner is high. Second, the "ants" in SIMiner are based on a random walk strategy, so many users in a social network cannot be obtained and updated in time. Therefore, the convergence speed of SIMiner is low. Summary of the invention

[0005] In response to the above-mentioned key issues that have not yet been resolved, the present invention proposes a token-based influential user selection method in a dynamic social network, using tokens as the medium of user influence in the social network. The purpose is to solve the complex and dynamic social network influence maximization problem from a distributed perspective without the help of external agents, while improving the convergence speed of the model and the coverage of user influence diffusion.

[0006] To achieve the above object, the present invention adopts the following technical solution:

[0007] A token-based influential user selection method in a dynamic social network comprises the following steps:

[0008] Step 1: A token-based dynamic social network user selection model TBIUS is proposed. This method uses tokens as a medium and obtains a seed set through an iterative mechanism;

[0009] Step 2: Construct the existing real data set into a social network graph G, which describes in detail the user's personal information, unique identifier, and the strength and intimacy of the relationship between users, providing a basis for subsequent analysis;

[0010] Step 3: Determine the generation mechanism of user tokens and mark their sources to make each token identifiable. This process ensures that the tokens can accurately reflect their source and propagation path during propagation;

[0011] Step 4: Formulate an iterative token distribution strategy. Combining the user's own information, neighbor information, and the current token set, the user transfers his own token to his neighbor user according to the strategy in each iteration to simulate the information flow process;

[0012] Step 5: After each round of iteration, the tokens held by the user are filtered to remove redundant information, form an intermediate result view, and gradually approach the target seed set;

[0013] Step 6: Use J accar d and Dice similarity coefficients perform convergence analysis on the seed set to ensure that when the seed set reaches the preset convergence criterion, the final result seed set has the potential for propagation;

[0014] Step 7: Use the IC independent cascade diffusion model to test the influence diffusion of the final seed set. By activating other nodes in the network and comparing the diffusion coverage at the same time scale, the propagation efficiency can be evaluated and optimized.

[0015] Preferably, the steps of constructing the social network G in step 2 are as follows:

[0016] Step 2-1: A social network consists of many users who have different relationships. A social network is represented by a graph structure as described below.

[0017] G = {V, E}.

[0018] where V = {v1, v2, ...v n} is the user set, v i Corresponding to the i-th user in the social network, E = {e 12 ,…,e (n-1)n} is the edge set of users in the network, edge e ij Corresponding to v i and v j The relationship between them is initially the number of connections between them, which represents the strength and intimacy of the relationship between users, and e i，j =e j，i , that is, an undirected graph form.

[0019] Step 2-2: For the edge set E = {e 12 ,…,e (n-1)n} to perform normalization operation and convert e ij The value is mapped to (0, 1], where 0 and 1 represent the farthest and closest relationship between two users. The normalization formula is as follows:

[0020]

[0021] Step 2-3: The definition of the social network G is as follows:

[0022] v i ={id, inf i , L.E. i , Lv i , T i}

[0023] Where id is the user v i The logo of inf i Yes i The self-influence Yes i The adjacent edge set of Yes i A set of directly connected neighboring users, v j Yes i A user in the neighboring user set. i Yes i A token pool containing v i All tokens owned, T i The definition formula is: T i ={t1, t2, ... t n}, t i Indicates v i A token owned by t i ={s id , l}, s id It represents the node that generates the token, and l represents the path length of the token.

[0024] Preferably, the token generation mechanism in step 3 includes:

[0025] Step 3-1: In each iteration, all users in the social network G will generate a certain number of tokens. This user v i The number of tokens generated in an iteration depends mainly on v i Self-influence, i.e. inf i Since the token is the smallest unit in the model that represents the following effects and cannot be divided, in the subsequent token distribution process, the token generation coefficient tgc is introduced into the token generation calculation, where tgc is an integer greater than 1. Finally, the number of tokens generated in the iteration GT i The calculation is as follows:

[0026] ΔT i =inf i ×tgc

[0027] Where tgc is the token generation coefficient, which is in the range of 1 to 30, with a step size of 1. The appropriate value is obtained by fitting according to the convergence speed of the resulting seed set. The faster the convergence speed, the better the tgc. i Yes i After the token is generated, the newly generated token will be added to the user's token pool, i.e., T i =T i +ΔT i .

[0028] As a preference, the token iteration distribution strategy method of step 4 is as follows:

[0029] Step 4-1: v i calculate The influence ratio of each user node in the distribution token is calculated as follows:

[0030]

[0031] Among them, inf j Represents user v i Neighbor v j The influence of Represents user v i and the sum of the influence of its neighbors.

[0032] Step 4-2: Through node v i Token Pool T i The composition of the neighbor nodes in one iteration is calculated from the user v i The number of tokens obtained. Since tokens are distinguished, v j From v i The obtained token set is calculated as:

[0033]

[0034] Among them, weight ji Represents node v j Occupied node v i The ratio of the sum of the influence of neighbor nodes, n is the number of nodes in the social network G, S k Indicates v i The token pool comes from v k The token set, e ij Indicates v j With v i The degree of intimacy between them, that is, the probability of transmission between them.

[0035] Step 4-3: Token normalization. Given the atomicity of tokens, the number of tokens must be an integer during the transfer process. Therefore, the calculation results need to be rounded down to maintain the integrity of the transfer process and the consistency of the model, that is, Indicates that Round down.

[0036] Preferably, after each iteration in step 5, the tokens held by the user are filtered to remove redundant information and form an intermediate result view, including:

[0037] Step 5-1: Calculate the evaluation function of the node score. The formula is as follows:

[0038]

[0039] Among them, d(v) represents the type of token in the node token pool, l k Indicates the v that generated the token k Node and v i Node distance, t k Indicates that the token pool comes from v k The number of tokens generated by the user, α and β are weight parameters, obtained by fitting.

[0040] Step 5-2: Filter out the tokens with less than 50 types in the token pool, calculate the user node evaluation function, and obtain the iterative intermediate result S.

[0041] Preferably, the seed set convergence calculation steps in step 6 are as follows:

[0042] Step 6-1: Perform convergence analysis on the seed set using the Jaccard and Dice similarity coefficients to ensure that when the seed set reaches the preset convergence standard, the final result seed set has propagation potential. The formula is as follows:

[0043]

[0044] Among them, S represents the intermediate result of the seed set obtained in this round of iteration, and S′ represents the intermediate result of the seed set obtained in the previous round of iteration.

[0045] When the Jaccard and Dice similarity coefficients are stably lower than 0.05, the model is considered to have reached the convergence condition.

[0046] As a preferred method, the seed set S obtained by iteration in step 7 is subjected to influence diffusion test on the final seed set through the IC independent cascade diffusion model. Assume that at time t, node v i There are x neighbor nodes a1, a2, ..., a that are activated at time t-1. n , the probabilities of these nodes activating node v are Then the node v i The probability of being activated is:

[0047]

[0048] If the node v i is activated, indicating that v i is covered, and the IC diffusion range increases.

[0049] Among the seed sets selected by different models, the coverage rate is the largest under the IC diffusion model, indicating that the effect of influence maximization selection is the best. The closer the seed set selected by this model is to the optimal seed set, the better the effect of this model.

[0050] Compared with the prior art, the present invention adopts the above technical solution and has the following technical effects:

[0051] Aiming at the problems of user selection models, such as centralized: when the user scale grows, the network communication overhead of the central controller to collect user information in the social network is large; the dynamic changes of users in the social network will cause their social influence to change constantly, and the central controller reselects influential users, which significantly increases the computational overhead. The distributed model based on the ant colony algorithm has the problems of agency cost and slow convergence. This paper proposes a decentralized user selection model TBIUS based on tokens. TBIUS uses tokens as a medium for user influence in social networks. In addition, the TBIUS model proposes an iterative mechanism in which the token generation and distribution process is iteratively performed by users in the social network. After a certain number of iterations, the user with the highest score evaluation is selected as an influential user. The contributions of the TBIUS model are summarized as follows: 1. The proposed model is a distributed user selection model that only allows users to select influential users in the social network using local views; 2. The iterative mechanism proposed by the proposed model enables users to generate and distribute tokens without involving external agents; 3. Experiments on real social networks: Facebook, Wikipedia voting and email show that the TBIUS model has excellent performance in terms of convergence speed and selected influence diffusion coverage. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 A flowchart of a token-based influential user selection method in a dynamic social network of the present invention;

[0053] Figure 2 A local schematic diagram for modeling a social network in the present invention;

[0054] Figure 3 Generate a local schematic diagram for the social network token of the present invention;

[0055] Figure 4 A partial schematic diagram of the distribution of social network tokens in the present invention;

[0056] Figure 5 It is a schematic diagram of the convergence of the TBIUS model in the iteration of the present invention;

[0057] Figure 6 This is a diagram showing the range of the TBIUS model in the present invention spreading in the Facebook dataset;

[0058] Figure 7 A schematic diagram of modeling the TBIUS model in a dynamic network in the present invention;

[0059] Figure 8This is a schematic diagram of the effect of the TBIUS model in the present invention in a dynamic network;

[0060] Fig. 9 This is a comparison diagram of the TBIUS model in the present invention with other models; DETAILED DESCRIPTION

[0061] The specific implementation of the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0062] like Figure 1 As shown, the technical route of the present invention mainly includes 6 steps, namely: building the user information and relationships between users of the social network into a social network model; the user nodes in the social network generate the corresponding initial number of tokens according to the generation formula; determining the token distribution mechanism; iterating multiple rounds of token generation and distribution until the convergence requirements are met; obtaining the result seed set; and verifying the IC diffusion model.

[0063] This example is based on the Email-Enron, Facebook, and Wiki Vote datasets. The Facebook dataset was collected from survey participants using this Facebook application. It contains 4039 nodes, 88234 edges, and 0 connected components. The Wiki Vote dataset is the historical data of all Wikipedia administrator elections and voting extracted from Wikipedia since January 3, 2008, and contains 7066 network nodes and 24 connected components. The nodes of the Email-Enron dataset network are email addresses, and the undirected edges of the network are emails sent between nodes. It contains 36692 nodes, 183831 edges, and 1065 connected components. The token-based influence user selection method under the dynamic social network of the present invention is used to solve the social network influence maximization problem. The following will introduce this example from four aspects: social network model establishment, iterative mechanism of token generation and distribution, IC independent cascade model diffusion test, and dynamic network adaptability.

[0064] 1) Social network model establishment

[0065] Step 1-1: A social network consists of many users who have different relationships. A social network is represented by a graph structure as described below.

[0066] G = {V, E}.

[0067] where V = {v1, v2, ...v n} is the user set, v i Corresponding to the i-th user in the social network, E = {e 12 ,…,e (n-1)n} is the edge set of users in the network, edge e ij Corresponding to v i and v j The relationship between them is initially the number of connections between them, which represents the strength and intimacy of the relationship between users, and e i，j =e j，i , that is, an undirected graph form.

[0068] Step 1-2: For the edge set E = {e 12 ,…,e (n-1)n} to perform normalization operation and convert e ij The value is mapped to (0, 1], where 0 and 1 represent the farthest and closest relationship between two users. The normalization formula is as follows:

[0069]

[0070] Step 1-3: The definition of the social network G is as follows:

[0071] v i ={id, inf i , L.E. i , LV i , T i}

[0072] Where id is the user v i The logo of inf i Yes i The self-influence Yes i The adjacent edge set of Yes i A set of directly connected neighboring users, v j Yes i A user in the neighboring user set. i Yes i A token pool containing v i All tokens owned, T i The definition formula is: T i ={t1, t2, ... t n}, t i Indicates v i A token owned by t i ={s id , l}, s id It represents the node that generates the token, and l represents the path length of the token.

[0073] In the embodiment, the initial modeling local schematic diagram is as follows Figure 2As shown, the edges between nodes have weights, which represent the closeness between nodes. In the model, they are related to the probability of successful information transmission between nodes and are mapped to (0, 1) through normalization operations. The nodes themselves have the attributes of tokens and influence, and the number of tokens is initialized to 0. The relationship between nodes is represented by an undirected graph. The examples Email-Enron, Facebook, and Wiki Vote are modeled as three undirected social network graphs respectively.

[0074] 2) Iterative mechanism of token generation and distribution

[0075] Step 2-1: In each iteration, all users in the social network G generate a certain number of tokens. This user v i The number of tokens generated in an iteration depends mainly on v i Self-influence, i.e. inf i Since the token is the smallest unit in the model that represents the following effects and cannot be divided, in the subsequent token distribution process, the token generation coefficient tgc is introduced into the token generation calculation, where tgc is an integer greater than 1. Finally, the number of tokens generated in the iteration GT i The calculation is as follows:

[0076] ΔT i =inf i ×tgc

[0077] Where tgc is the token generation coefficient, which is in the range of 1 to 30, with a step size of 1. The appropriate value is obtained by fitting according to the convergence speed of the resulting seed set. The faster the convergence speed, the better the tgc. i Yes i After the token is generated, the newly generated token will be added to the user’s token pool, i.e., T i =T i +ΔT i .

[0078] Table 1. Experimental parameter settings

[0079]

[0080]

[0081] Figure 3 A local implementation scheme for generating tokens is shown. Here, the tgc parameter is set to 10, and a large number of comparative experiments are performed to obtain the optimal value of the tgc parameter.

[0082] Step 2-2: v i calculate The influence ratio of each user node in the distribution token is calculated as follows:

[0083]

[0084] Among them, inf j Represents user v i Neighbor v j The influence of Represents user v i and the sum of the influence of its neighbors.

[0085] Step 2-3: Through node v i Token Pool T i The composition of the neighbor nodes in one iteration is calculated from the user v i The number of tokens obtained. Since tokens are distinguished, v j From v i The obtained token set is calculated as:

[0086]

[0087] Among them, weight ji Represents node v j Occupied node v i The ratio of the sum of the influence of neighbor nodes, n is the number of nodes in the social network G, S k Indicates v i The token pool comes from v k The token set, e ij Indicates v j With v i The degree of intimacy between them, that is, the probability of transmission between them.

[0088] Step 2-4: Token normalization. Given the atomicity of tokens, the number of tokens must be an integer during the transfer process. Therefore, the calculation results need to be rounded down to maintain the integrity of the transfer process and the consistency of the model, that is, Indicates that Round down.

[0089] Figure 4 The token distribution of a local embodiment after the first distribution is illustrated, where v1 includes tokens from v1 and v3.

[0090] Step 2-5 calculates the evaluation function of the node score, and the formula is as follows:

[0091]

[0092] Among them, d(v) represents the type of token in the node token pool, l k Indicates the v that generated the token k Node and v i Node distance, tk Indicates that the token pool comes from v k The number of tokens generated by the user, α and β are weight parameters, obtained by fitting.

[0093] Step 2-6: Filter out the tokens with less than 50 types in the token pool, calculate the user node evaluation function, and obtain the iterative intermediate result S.

[0094] Step 2-7: Perform convergence analysis on the seed set using the Jaccard and Dice similarity coefficients to ensure that when the seed set reaches the preset convergence criteria, the final result seed set has propagation potential. The formula is as follows:

[0095]

[0096] Among them, S represents the intermediate result of the seed set obtained in this round of iteration, and S′ represents the intermediate result of the seed set obtained in the previous round of iteration.

[0097] When the Jaccard and Dice similarity coefficients are stably lower than 0.05, the model is considered to have reached the convergence condition.

[0098] like Figure 5 As shown in the figure, when tgc is 10, the entire model converges in 15 rounds, and the subsequent fluctuations of individual nodes with similar influence can be ignored.

[0099] 3)IC independent cascade model diffusion test

[0100] Step 3-1 Perform influence diffusion test on the final seed set through IC independent cascade diffusion model. Assume that at time t, node v i There are x neighbor nodes a1, a2, ..., a that are activated at time t-1. n , these nodes activate node v i The probabilities are Then the node v i The probability of being activated is:

[0101]

[0102] If the node v i is activated, indicating that v i is covered and the IC diffusion range increases.

[0103] Among the seed sets selected by different models, the coverage rate is the largest under the IC diffusion model, indicating that the effect of influence maximization selection is the best. The closer the seed set selected by this model is to the optimal seed set, the better the effect of this model.

[0104] like Figure 6As shown in the figure, the diffusion range of the TBIUS model in the Facebook data set can reach 4,000 node users in about 11 rounds of diffusion, which is a very objective diffusion effect, indicating that the influential seed set selected by the TBIUS model is more influential.

[0105] 4) Dynamic network adaptability

[0106] In a real social network, the relationships between users are highly dynamic. In the embodiment, first, the social structure network is as follows: Figure 2 After 50 iterations, users v1 and v2 and their edges are deleted, as shown in Figure 7 At this point, the TBIUS model processes and analyzes the network, as shown in Figure 8 As shown in the figure, after 10 iterations, the relationship between the number of node tokens tends to be stable, that is, the influence changes between nodes are completed and can be quickly captured by tokens, which shows that the TBIUS model can quickly respond to dynamic social networks.

[0107] The TBIUS model is compared with other related models to comprehensively evaluate the effect of the model of the present invention. The number of selected seeds is set to 10, 20, 30, 40, and 50. Based on the IC model, the number of other users in the social network influenced by the selected user is calculated as an indicator of influence diffusion coverage. The influence diffusion coverage experiments are all based on multiple Monte Carlo simulations. Fig. 9 As shown, the X-axis represents the number of selected users, and the Y-axis represents the influence of other users in the social network on the selected users based on the IC model. The TBIUS model of the present invention is always better than other centralized and distributed model results.

[0108] Based on the above analysis, after a certain number of iterations, the experiments on the real social network dataset show that the proposed model has good performance speed and model overhead in terms of impact diffusion coverage and convergence, and is suitable for large-scale and dynamic social networks.

[0109] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the art should understand that certain modifications and changes can be made to the present invention without exceeding the spirit and essence of the present invention, but they should be included in the protection scope of the present invention.

Claims

1. A token-based influential user selection method in a dynamic social network, characterized in that: The following steps are involved: Step 1: Propose a token-based dynamic social network user selection model TBIUS, which uses tokens as the medium and obtains the seed set through an iterative mechanism; Step 2: Construct the existing real data set into a social network graph G, which describes in detail the user's personal information, unique identifier, and the strength and intimacy of the relationship between users, providing a basis for subsequent analysis; Step 3: Determine the mechanism for generating user tokens and mark their sources to make each token identifiable; this process ensures that the tokens can accurately reflect their source and propagation path during propagation; Step 4: Develop an iterative token distribution strategy; Combining the user's own information, neighbor information and the current token set, the user transfers his own token to his neighbor users according to the strategy in each iteration to simulate the information flow process; Step 5: After each round of iteration, the tokens held by the user are filtered to remove redundant information, form an intermediate result view, and gradually approach the target seed set; Step 6: Perform convergence analysis on the seed set using Jaccard and Dice similarity coefficients to ensure that when the seed set reaches the preset convergence criteria, the final result seed set has propagation potential.

2. The method according to claim 1, characterized in that The steps to construct a social network G are as follows: Step 2-1: A social network consists of many users who have different relationships. The social network is represented by a graph structure as described below. G = {V, E} where V = {v1, v2, ...v n } is the user set, v i Corresponding to the i-th user in the social network, E = {e 12 ,…,e (n-1)n } is the edge set of users in the network, edge e ij Corresponding to v i and v j The relationship between them is initially the number of connections between them, which represents the strength and intimacy of the relationship between users, and e i，j =e j，i , that is, undirected graph form; Step 2-2: For the edge set E = {e 12 ,…,e (n-1)n } to perform normalization operation and convert e ij The value is mapped to (0, 1], where 0 and 1 represent the farthest and closest relationship between two users. The normalization formula is as follows: Step 2-3: The definition of the social network G is as follows: v i ={id,inf i ,LE i ,LV i ,T i } Where id is the user v i The logo of inf i Yes i The self-influence Yes i The adjacent edge set of Yes i A set of directly connected neighboring users, v j Yes i A user in the neighboring user set; T i Yes i A token pool containing v i All tokens owned, T i The definition formula is: T i ={t1, t2, ... t n }, t i Indicates v i A token owned by t i ={s id , l}, s id It represents the node that generates the token, and l represents the path length of the token.

3. The method according to claim 1, characterized in that The token generation mechanism includes: Step 3-1: In each iteration, all users in the social network G will generate a certain number of tokens; this user v i The number of tokens generated in an iteration depends mainly on v i Self-influence, i.e. inf i ; Since the token is the smallest unit in the model that represents the following effects and cannot be divided, in the subsequent token distribution process, the token generation coefficient tgc is introduced into the token generation calculation, where tgc is an integer greater than 1; Finally, the number of tokens generated in the iteration GT i The calculation is as follows: ΔT i =inf i ×tgc Where tgc is the token generation coefficient, which is in the range of 1 to 30, with a step size of 1. The appropriate value is obtained by fitting according to the convergence speed of the resulting seed set. The faster the convergence speed, the better the tgc; inf i Yes i After the token is generated, the newly generated token will be added to the user's token pool, i.e., T i =T i +ΔT i .

4. The token iterative distribution strategy method as claimed in claim 1 is as follows: Step 4-1: v i calculate The influence ratio of each user node in the distribution token is calculated as follows: in, inf j Represents user v i Neighbor v j The influence of Represents user v i and the sum of the influence of the neighbors; Step 4-2: Through node v i Token Pool T i The composition of the neighbor nodes in one iteration is calculated from the user v i The number of tokens obtained. Since tokens are distinguished, v j From v i The obtained token set is calculated as: Among them, weight ji Represents node v j Occupied node v i The ratio of the sum of the influence of neighbor nodes, n is the number of nodes in the social network G, S k Indicates v i The token pool comes from v k The token set, e ij Indicates v j With v i The degree of intimacy, that is, the probability of transmission between them; Step 4-3: Token regularization. Given the atomicity of tokens, the number of tokens must be an integer during the transfer process. Therefore, the calculation results need to be rounded down to maintain the integrity of the transfer process and the consistency of the model, that is, Indicates that Round down.

5. The method according to claim 1, characterized in that The intermediate result view generation method is as follows: after each round of iteration, the tokens held by the user are filtered to remove redundant information and form an intermediate result view, including: Step 5-1: Calculate the evaluation function of the node score. The formula is as follows: Among them, d(v) represents the type of token in the node token pool, l k Indicates the v that generated the token k Node and v i Node distance, t k Indicates that the token pool comes from v k The number of tokens generated by users, α and β are weight parameters, which are fitted; Step 5-2: Filter out the tokens with less than 50 types in the token pool, calculate the user node evaluation function, and obtain the iterative intermediate result S.

6. The method according to claim 1, characterized in that The seed set convergence calculation steps are as follows: Step 6-1: Perform convergence analysis on the seed set using Jaccard and Dice similarity coefficients to ensure that when the seed set reaches the preset convergence standard, the final result seed set has propagation potential. The formula is as follows: Among them, S represents the intermediate result of the seed set obtained in this round of iteration, and S′ represents the intermediate result of the seed set obtained in the previous round of iteration; When the Jaccard and Dice similarity coefficients are stably lower than 0.05, the model is considered to have reached the convergence condition.

7. The method according to claim 1, characterized in that You can also add an IC model diffusion step. The IC model diffusion test method is as follows: The seed set S obtained by iteration is used to perform influence diffusion test on the final seed set through the IC independent cascade diffusion model. Assume that at time t, node v i There are x neighbor nodes a1, a2, ..., a that are activated at time t-1. n , the probabilities of these nodes activating node v are Then the node v i The probability of being activated is: If the node v i is activated, indicating that v i Being covered, the IC diffusion range increases; Among the seed sets selected by different models, the IC diffusion model has the largest coverage, indicating that the effect of influence maximization selection is the best. The closer the seed set selected by this model is to the optimal seed set, the better the effect of this model.

Images