Dynamic method for minimizing influence of error information in social network

Abstract

The invention provides a dynamic method for minimizing the influence of error information in a social network, which comprises the following steps of: 1, constructing a user social network graph and an information set according to a real data set; 2, determining factors influencing the attention value of the user, wherein the factors mainly comprise the interestingness of the user on the information, the preference similarity between the users, the relationship strength between the users and the relative influence; and 3, formulating an exploration strategy of an information sending path. And 4, dynamically matching effective information for different paths. And according to the attention space condition of the target user, information to be spread is selected, and an anorexia factor is added to dynamically adjust information matched in each round of the path. And 5, changing the attention of the target user through multi-round information spreading, and finally removing error information from the attention space of the target user. According to the method, testing is carried out under an IC independent cascade diffusion model, and effectiveness is proved.

Description

Technical Field

[0001] The present invention belongs to the field of social networks and intelligent agent systems, and in particular relates to a dynamic method for minimizing the impact of false information in social networks. Background Art

[0002] In recent years, with the rapid development of network technology and social media, the scale and popularity of social networks such as Twitter, Facebook, Weibo, and WeChat have increased significantly. In these social networks, a large amount of information can be quickly and widely spread among users every day. These characteristics of social networks have attracted more researchers to pay attention to the influence diffusion model in social networks. The influence diffusion model can accelerate and expand the speed and coverage of information dissemination by identifying and selecting the most influential users in social networks. However, since users can post information freely in social networks, many users post a large amount of false information for different purposes. The widespread dissemination of false information may lead to serious public opinion bias and have a negative impact on society. Nowadays, false information in social networks has become a major social problem.

[0003] In order to minimize the impact of false information on society, early detection and prevention are crucial. Many researchers suggest identifying important nodes (i.e., users) and connections in social networks and blocking them when necessary to prevent the spread of false information. However, as the scale of social networks continues to expand, finding and monitoring important nodes becomes extremely time-consuming and resource-intensive. In addition, the dynamic changes in social networks may also lead to changes in important nodes and connections, and these detection methods cannot respond to dynamically changing social networks in a timely manner. Some researchers use influential users to prevent other users from spreading false information to minimize the impact of false information. However, the number of influential users in social networks is limited, and their distribution has a great impact on the detection and prevention of false information.

[0004] In recent years, diverting the attention of target users from paying attention to false information has been proven to be a practical and effective way to quickly stop the spread of false information. Researchers have proposed an influence minimization model that can divert the attention of target users by sending information related to other topics that users are interested in. However, their model only focuses on the impact of different information contents on the attention diversion of target users. In actual social networks, due to the large scale and complexity of the network structure, the information sending path also has a significant impact on the effect of diverting the attention of target users. Summary of the invention

[0005] In response to the above unresolved key issues, this paper proposes a decentralized method to minimize the impact of false information in social networks. First, the method uses intelligent agents to simulate users in social networks and uses an information sending path exploration mechanism to enable users who detect target users holding false information to explore efficient and effective information sending paths in a decentralized manner. In addition, a dynamic attention transfer mechanism is proposed to enable users to dynamically update and select appropriate information and its sending path based on the target user's boredom. Its effectiveness is explained under different cost constraints.

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

[0007] A dynamic method for minimizing the impact of misinformation in social networks includes the following steps:

[0008] Step 1: First, construct the user social network graph S and the information set msg based on the real data set. The information in the graph S contains the user’s unique identifier id, the user’s personal influence inf i , and the relationship strength r between users i，j .

[0009] Step 2: Define the user's interest in different information i , and the user's attention space mep i , providing a basis for subsequent analysis;

[0010] Step 3: Determine the factors that affect the user's attention value. Combined with sociological knowledge, it is determined that the user's attention value to different information is mainly affected by the user's own interest in the information. i，k , and the influence of information communicators on the recipients Effe i，j There are two main factors affecting this. i，j The preference similarity between user i and user j Pres i，j , relationship strength r i，j , and the relative influence Infd i，j The impact of

[0011] Step 4: Develop an exploration strategy for information transmission paths. Based on the relative influence between users, explore information transmission paths layer by layer, and calculate the influence value of the final path on the target user and the probability value of successful path transmission. This step ensures that the information transmission path we choose is the path with greater influence on users;

[0012] Step 5: Dynamically match effective information for different paths. Select the information to be disseminated based on the target user's attention space and add the boredom factor B. i，kTo dynamically adjust the path matching information in each round to achieve optimal efficiency.

[0013] Step 6: Change the target user's attention through multiple rounds of information propagation, and finally remove the wrong information from his attention space; test under the IC independent cascade diffusion model to obtain the information propagation rounds required to remove the wrong information from the target user's attention space to evaluate the effectiveness of the method;

[0014] Preferably, the steps of constructing the social network S in step 1 are as follows:

[0015] Step 1-1: In a social network, there are a large number of users, and there are different relationships between these users. Usually, a social network can be represented by an undirected graph, where nodes and edges are used to represent users and their relationships, respectively. The social network S can be defined as follows:

[0016] S = {U, R}

[0017] Where U={u1, u2,...u i , ...} represents the user set, u i is the i-th user in S; R = {r 1，2 , r 1，3 , ..., r i，j , ...} represents a set of relations, r i，j Is user u i and u j The degree of relationship between i，j The value of is between 0 and 1, where 0 and 1 represent the farthest and closest relationship between users, respectively. Since S is represented by an undirected graph, r i，j =r j，i· .

[0018] Step 1-2: The information set msg consists of multiple messages. k It can be defined according to its theme and keywords as follows:

[0019] msg k ={topic k}

[0020] The topic k ={kw1, kw2, ..., kw n} is the message msg k Theme kw t It is the tth keyword in the topic.

[0021] Step 1-3: The definition of social network S is as follows:

[0022] u i={id,pre i , inf i ,nu i , mep i}

[0023] Among them, id is user u i The unique identifier of i Is user u i Preference, inf i Is user u i Personal influence, nu i Is user u i The neighbor user set of mep i Is user u i The message pool includes user u i All messages received.

[0024] Preferably, the definition information in step 2 includes:

[0025] Step 2-1: User interest level i It can be described as follows:

[0026] pre i ={kw1, kw2, ..., kw t , ...}

[0027] Among them, kw t is the tth keyword describing the preference of user i.

[0028] Step 2-2: User’s attention space mep i as follows:

[0029] mep i =(msg1, msg2, ..., msg k , ...}

[0030] mep i is the message pool of user i, which includes all the messages he has received.

[0031] Preferably, the influencing factors of the user attention value in step 3 are calculated as follows:

[0032] Step 3-1: User u i For information msg k The attention of users is mainly affected by their own interest in the information. i，k , and information disseminator u j To u i The influence of Effe i，j Two main factors influence this.

[0033] Step 3-2: User i’s interest in message k i，k By selecting the preference from i i and the topic of k k The Jaccard similarity between them is calculated as follows:

[0034]

[0035] Among them, pre i ={kw1, kw2, ..., kw t , ...} is the set of user i’s preferred keywords, topic k ={kw1, kw2, ..., kw n} is the set of topic keywords of message k, and the ratio of the number of elements in the intersection and the number of elements in the union of the two sets is taken.

[0036] Step 3-3: User j’s relative personal influence on i Effe i , j is calculated as follows:

[0037] Effe i，j =(r i，j +Pres i，j +Infd i，j ) / 3,

[0038] Among them, r i，j is the closeness of the connection between users i and j, obtained by processing the information in the data set, pres i，j is the preference similarity between users i and j, infd i，j is the relative personal influence of user j on i, and the calculation methods of the two are as follows:

[0039]

[0040]

[0041] Among them, pre i and pre j are the personal preferences of users i and j, respectively, inf j and inf i are the personal influence values ​​of users j and i respectively, taking the ratio of the number of elements in the intersection and the number of elements in the union of the two sets, and λ is the relative coefficient, which takes a value between 0 and 1.

[0042] Preferably, the information sending path selection strategy described in step 4 includes:

[0043] Step 4-1: Set the current user d as the detector of the error information and traverse its neighboring users. If the target user t is encountered, calculate the impact effect Effe and the probability of successful path propagation P l , and add this information to the current path.

[0044] The probability of successful information dissemination of the path P l The calculation formula is as follows:

[0045]

[0046] Step 4-2: If the target user is not encountered and the length of the current path does not exceed the maximum number of exploration layers Max, continue to explore the neighboring users of the neighboring users and gradually build the path Path. If the path length reaches the limit, that is, it is greater than the maximum number of exploration layers Max, the current path will be abandoned.

[0047] Step 4-3: After exploring all possible paths, check whether there are paths with the same impact in the path set. If so, compare the propagation success probabilities of these paths, retain the path with the highest probability, and remove the path with the lowest probability.

[0048] Finally, a set of filtered paths is returned, which represent the most effective influence paths from misinformation detectors to target users. This process ensures the efficiency and effectiveness of information dissemination while taking into account the depth limit of information dissemination.

[0049] Preferably, the dynamic information matching strategy described in step 5 is as follows:

[0050] Step 5-1: Select the information set in the target user’s attention space as the information to be propagated to enhance their attention.

[0051] Step 5-2: Traverse the path set, and for the selected path, calculate the potential change in attention of the target user caused by the information k propagated through this path l. The calculation formula is as follows:

[0052] Δatt i，k，l =pinte i，k B i，k +qEffe i，j +mP l

[0053] Among them, p, q, and m are the balance coefficients of each factor, ranging from 0 to 1. i，k is the user i’s current boredom with information k. Its value decreases as the number of times the user receives the information increases. The smaller the value, the higher the boredom, and the smaller the impact on attention change. It is calculated as follows:

[0054] B i，k =1-n×numk

[0055] where num k is the number of times the user has received the message, n is the scaling factor, based on the integer num k The value range of is scaled to between 0 and 1.

[0056] Through calculation, each path is matched with the information with the best transmission effect in the current sending round.

[0057] Step 5-3: Update the target user's attention to the message.

[0058] This process can continue until the target user's attention value for the erroneous information drops below that of all the information in the attention space, and it is considered that the erroneous information fades out of the user's attention space and the impact is eliminated.

[0059] Preferably, in step 6, the target node and the erroneous information in its attention space are selected under the IC independent cascade diffusion model, and the source node is determined according to the target node and its surrounding neighborhood users. The source node detects that the target node is contaminated, starts to explore possible paths for propagating correct information, and determines the correct information to be propagated. Until the target user's attention value to the erroneous information is lower than all the information in the attention space, the target's attention space is exited, and the impact is eliminated. The number of information propagation rounds required for the elimination of the impact is obtained, and the fewer the rounds, the better the effect of the proof method.

[0060] Compared with the prior art, the present invention adopts the above technical solution and has the following technical effects:

[0061] Compared with previous methods, the biggest advantage of the present invention is that it solves the problems caused by the changeable network structure and the excessive network scale to a certain extent. Whether it is the traditional centralized method of shielding user nodes or cutting off the connection between users or the previous method of using the information environment to reduce the impact of false information, there are problems such as the limitation of network structure and scale, resulting in the computational complexity of the method and unstable results. The method we proposed is decentralized, and the detection of the target and the initiation of the behavior are all in the local area. Based on the above factors, the contributions of the proposed method can be summarized as follows: 1. The information sending path exploration mechanism of the proposed method enables users to explore the efficient and effective information sending path of the target user in a decentralized manner; 2. The dynamic attention transfer mechanism of the proposed method enables users to dynamically update and select appropriate information and its sending path according to the boredom of the target user; 3. Comparative experiments on real social network datasets show that the proposed method has good performance in minimizing the impact of false information with high efficiency, high performance and low cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1It is a flow chart of the method for minimizing the impact of dynamic error information in the present invention;

[0063] Figure 2 A local schematic diagram for modeling a social network in the present invention;

[0064] Figure 3 This is an example of the results of the path exploration mechanism;

[0065] Figure 4 This is an example of the results of the path exploration mechanism;

[0066] Figure 5 An example of the change in the user's attention space before information matching propagation;

[0067] Figure 6 This is an example of the change in the user's attention space after information matching and propagation;

[0068] Figure 7 A comparison chart of the effects of our method and other path selection methods on real datasets;

[0069] Figure 8 This is a comparison chart of the effects of our method and the method without considering information matching in a real dataset;

[0070] Fig. 9 The effect diagram of different costs under different data sets; DETAILED DESCRIPTION

[0071] 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.

[0072] like Figure 1 As shown, the technical route of the present invention mainly includes 6 steps, namely: building a social network model based on user information, relationships between users, and information sets of the social network; the intelligent agent explores and selects the information propagation path according to the mechanism; dynamically matches messages; information propagation; observes the attention space of the target node after each round of propagation, and if the wrong information is still concerned, continues the next round of information matching and propagation until the wrong information fades out of the user's attention space; obtains the number of information propagation rounds required to remove the wrong information.

[0073] This example uses Twitter, Wiki Vote, and Epinions datasets as examples. The Twitter dataset is collected from survey participants using this Twitter application. It contains 272,861 nodes and 433,439 edges. The Wiki Vote dataset is the historical data of all Wikipedia administrator elections and votes extracted from Wikipedia since January 3, 2008, and contains 7,115 network nodes and 103,689 edges. The Epinions dataset network is a trust network obtained from a consumer review website Epinions.com. Website members can decide whether to "trust" each other, and it contains 75,879 nodes and 508,837 edges. The method of the present invention is used to solve the problem of minimizing the impact of false information on social networks. The following will introduce this example from four aspects: social network model establishment, propagation path exploration and dynamic information matching, and experimental testing.

[0074] 1) Social network model establishment

[0075] Step 1-1: In a social network, there are a large number of users, and there are different relationships between these users. Usually, a social network can be represented by an undirected graph, where nodes and edges are used to represent users and their relationships, respectively. The social network S can be defined as follows:

[0076] S = {U, R}

[0077] Where U={u1, u2,...u i , ...} represents the user set, u i is the i-th user in S; R = {r 1，2 , r 1，3 , ..., r i，j , ...} represents a set of relations, r i，j Is user u i and u j The degree of relationship between i，j The value of is between 0 and 1, where 0 and 1 represent the farthest and closest relationship between users, respectively. This value can be obtained indirectly through the network connection status. In the experimental processing, its value is taken as user u i and u j The ratio of the intersection size to the union size of the neighbor nodes between them. Since S is represented by an undirected graph, r i，j =r j，i· .

[0078] Step 1-2: The information set msg consists of multiple messages. k It can be defined according to its theme and keywords as follows:

[0079] msg k={topic k}

[0080] The topic k ={kw1, kw2, ..., kw n} is the message msg k Theme kw t It is the tth keyword in the topic.

[0081] Step 1-3: User interest level i It can be described as follows:

[0082] pre i ={kw1, kw2, ..., kw t , ...}

[0083] Among them, kw t is the tth keyword describing the preference of user i.

[0084] Step 1-4: User’s attention space mep i as follows:

[0085] mep i ={msg1, msg2, ..., msg k , ...}

[0086] mep i is the message pool of user i, which includes all the messages he has received.

[0087] Step 1-5: The definition of social network S is as follows:

[0088] u i ={id,pre i , inf i ,nu i , mep i}

[0089] Among them, id is user u i The unique identifier of i Is user u i Preference, inf i Is user u i The impact of self i Is user u i The neighbor user set of mep i Is user u i The message pool includes user u i All messages received.

[0090] In our embodiment, the local schematic diagram of the initial modeling is as follows Figure 2As shown in the figure, the edges between nodes have weights, which represent the degree of connection between nodes. In the model, it is related to the probability of successful information transmission between nodes. In addition to its own attributes, each node also has its attention space information. The relationship between nodes is represented by an undirected graph.

[0091] 2) Exploration of propagation paths

[0092] Step 2-1: Set the current user d as the detector of the error information and traverse its neighboring users. If the target user t is encountered, calculate the impact effect Effe and the probability of successful path propagation P1, and add this information to the current path.

[0093] In the path, the relative personal influence Effe of information disseminator user j on information receiver i i , j is calculated as follows:

[0094] Effe i，j =(r i，j +Pres i，j +Infd i，j ) / 3,

[0095] Among them, r i，j is the closeness of the connection between users i and j obtained by processing the data set, pres i，j is the preference similarity between users i and j, infd i，j is the relative personal influence of user j on i, and the calculation methods of the two are as follows:

[0096]

[0097]

[0098] Among them, pre i and pre j are the personal preferences of users i and j, respectively, inf j and inf i are the personal influence values ​​of users j and i, respectively, λ is the relative coefficient, which is taken as 1 in the experiment to ensure that if (inf j -inf i ) is a positive number, the relative influence of j on i is also greater.

[0099] The probability of successful information dissemination of the path P l The calculation formula is as follows:

[0100]

[0101] Step 2-2: If the target user is not encountered and the length of the current path does not exceed the maximum number of exploration layers Max, continue to explore the neighboring users of the neighboring users and gradually build the path Path. If the path length reaches the limit, that is, it is greater than the maximum number of exploration layers Max, the current path will be abandoned.

[0102] Figure 3 An example of the path exploration results between users u2 to u5 is shown.

[0103] Figure 4 It is the detailed information of the path generated during the exploration process, including the two adjacent users and their edge values ​​of each path, as well as the calculated value of the path's influence effect Effe and the probability of successful information propagation of the path P l .

[0104] Step 2-3: After exploring all possible paths, check whether there are paths with the same impact in the path set. If so, compare the propagation success probabilities of these paths, retain the path with the highest probability, and remove the path with the lowest probability.

[0105] Figure 4 Path 1 and path 4 both spread information to target user u5 through user u4 and have the same impact, but the probability of successful information dissemination of path 4 is lower than that of path 1, so path 4 is discarded.

[0106] Finally, a set of filtered paths is returned, which represent the most effective influence paths from misinformation detectors to target users. This process ensures the efficiency and effectiveness of information dissemination while taking into account the depth limit of information dissemination.

[0107] 3) Dynamic information matching

[0108] Step 6-1: Select the information set in the target user’s attention space as the information to be propagated to enhance their attention.

[0109] like Figure 5 As shown, the error message in the attention space of user u5 is msg2. You can choose messages msg1 and msg4 with lower attention values ​​than the error message to fade the error message out of the user's attention space by increasing the user's attention to them, because the user's attention is limited. For example, as shown in the figure, only the first three messages are of interest to the user.

[0110] Step 6-2: Traverse the path set, and for the selected path, calculate the potential change in attention of the target user caused by the information k propagated through this path 1. The calculation formula is as follows:

[0111] Δatt i，k，l =pinte i，k Bi，k +qEffe i，j +mP l

[0112] Among them, p, q, and m are the balance coefficients of various factors. After multiple experiments, their values ​​are determined to be 0.5, 0.2, and 0.3 respectively. i，k is the user i’s current boredom with information k. Its value decreases as the number of times the user receives the information increases. The smaller the value, the higher the boredom, and the smaller the impact on attention change. It is calculated as follows:

[0113] B i，k =1-n×num k

[0114] where num k is the number of times the user has received the message, n is the scaling factor, based on the integer num k The value range of is scaled to between 0 and 1.

[0115] Through calculation, each path is matched with the information with the best transmission effect in the current sending round.

[0116] Step 6-3: Update the target user's attention to the message.

[0117] like Figure 6 As shown, according to the mechanism calculation, the propagation information with the best propagation effect is matched to different paths. After one round of propagation, the user's attention space has changed, and the user's attention value to msg1 is greater than the error information msg2.

[0118] This process can continue until the target user's attention value for the erroneous information drops below that of all the information in the attention space, and it is considered that the erroneous information fades out of the user's attention space and the impact is eliminated.

[0119] 4) Experimental testing

[0120] We tested our method on a real dataset.

[0121] like Figure 7 As shown, the X-axis represents the number of rounds of information dissemination, and the Y-axis represents the position of the erroneous information in the user's attention space. The higher the position, the more attention the user pays, and 0 indicates that it has moved out of the user's attention space. From the results, it can be seen that the path exploration selection mechanism of the present invention is superior to the path selection results based on other indicators.

[0122] Figure 8It demonstrates the superiority of our dynamic information matching mechanism. Similarly, the X-axis represents the number of rounds of information dissemination, and the Y-axis represents the position of the wrong information in the user's attention space. The higher the position, the more attention it receives from the user, and 0 indicates that it has been moved out of the user's attention space. We compared the two situations of randomly matching messages to paths and matching messages without considering the user's boredom. The results show that the dynamic information matching mechanism of the present invention has a certain effect improvement.

[0123] Fig. 9 The results under different cost limits in multiple data sets are shown. The X-axis represents different costs. We use the maximum number of layers of path exploration as the cost, because one more user in the path means that we need to persuade one user to spread information, which requires a certain actual cost. The Y-axis represents the number of information dissemination rounds required to remove the wrong information. The results show that when the cost is 7, the effect tends to be stable.

[0124] Based on the above analysis, our method has certain advantages in minimizing the impact of false information.

[0125] 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 dynamic method for minimizing the impact of misinformation in social networks, characterized by The following steps are involved: Step 1: First, construct the user social network graph S and the information set msg based on the real data set; the information in the graph S contains the user's unique identifier id, the user's personal influence inf i , and the relationship strength r between users i,j ; Step 2: Define the user's interest in different information i , and the user's attention space mep i , providing a basis for subsequent analysis; Step 3: Determine the factors that affect the user's attention value; Combined with sociological knowledge, determine that the user's attention value to different information is mainly affected by the user's own interest in the information. i,k , and the influence of information communicators on the recipients Effe i,j The influence of two main factors; Effe i,j The preference similarity between user i and user j Pres i,j , relationship strength r i,j , and the relative influence Infd i,j The impact of Step 4: Develop an exploration strategy for information transmission paths. Based on the relative influence between users, explore information transmission paths layer by layer, and calculate the influence value of the final path on the target user and the probability value of successful path transmission. This step ensures that the information transmission path we choose is the path with greater influence on users. Step 5: Dynamically match effective information for different paths; select the information to be disseminated based on the target user's attention space, and add the boredom factor B i,k To dynamically adjust the path matching information for each round; Step 6: Through multiple rounds of information propagation, the target user’s attention is changed and the wrong information is eventually removed from his attention space.

2. The method according to claim 1, characterized in that The steps to construct a social network S are as follows: Step 2-1: In a social network, there are a large number of users, and there are different relationships between these users. Generally, a social network can be represented by an undirected graph, where nodes and edges are used to represent users and their relationships, respectively. The social network S can be defined as follows: S = {U, R} Where U={u1, u2,...u i , ...} represents the user set, u i is the i-th user in S; R = {r 1，2 , r 1，3 , ..., r i，j , ...} represents a set of relations, r i,j Is user u i and u j The degree of relationship between i,j The value of is between 0 and 1, where 0 and 1 represent the farthest and closest relationship between users, respectively. Since S is represented by an undirected graph, r i，j =r j，i ; Step 2-2: The information set msg consists of multiple messages. k It can be defined according to its theme and keywords as follows: msg k ={topic k } The topic k ={kω1, kω2, ..., kω n } is the message msg k Theme kw t It is the tth keyword in the topic.

3. The method according to claim 1, characterized in that User interest level i , and the attention space mep i The construction is as follows: Step 3-1: User interest level i It can be described as follows: leek ＝ {kw1,kw2,...,kw t ...} Among them, kw t is the tth keyword describing the preference of user i; Step 3-2: User’s attention space mep i as follows: mep i ={msg1,msg2,...,msg k ,...} mep i is the message pool of user i, which includes all the messages he receives.

4. The method according to claim 1, characterized in that The factors affecting the user's attention value are calculated as follows: Step 4-1: User u i For information msg k The attention of users is mainly affected by their own interest in the information. i,k , and information disseminator u j To u i The influence of Effe i,j The influence of two main factors; Step 4-2: User i’s interest in message k i，k By selecting the preference from i i and the topic of k k The Jaccard similarity between them is calculated as follows: Where, prei={ke1,kw2,...,kw t , ...} is the set of user i’s preferred keywords, topic k ={kw1, kw2, ..., kw n } is the set of topic keywords of message k, and the ratio of the number of elements in the intersection and the number of elements in the union of the two sets is taken; Step 4-3: User j’s relative personal influence on i Effe i，j The calculation method is as follows: Effe i，j =(r i，j +Pres i，j +Infd i，j ) / 3, Among them, r i，j is the closeness of the connection between users i and j obtained by processing the data set, pres i,j is the preference similarity between users i and j, infd i，j is the relative personal influence of user j on i, and the calculation methods of the two are as follows: Among them, pre i and pre j are the personal preferences of user i and j respectively, and the ratio of the number of elements in the intersection and the number of elements in the union of the two sets is taken, inf j and inf i are the personal influence values ​​of users j and i respectively, and λ is the relative coefficient, which ranges from 0 to 1.

5. The method according to claim 1, characterized in that The information sending path selection strategy is as follows: Step 5-1: Set the current user d as the detector of error information and traverse its neighboring users; if the target user t is encountered, calculate the path impact effect Effe and the probability of successful path propagation P l , and add this information to the current path; The probability of successful information dissemination of the path P l The calculation formula is as follows: Step 5-2: If the target user is not encountered and the length of the current path does not exceed the maximum number of exploration layers Max, continue to explore the neighboring users of the neighboring users and gradually build the path Path; if the path length reaches the limit, that is, it is greater than the maximum number of exploration layers Max, the current path will be abandoned; Step 5-3: After exploring all possible paths, check whether there are paths with the same impact in the path set; if so, compare the propagation success probabilities of these paths, retain the path with the highest probability, and remove the path with the lowest probability; Finally, the filtered path collection is returned.

6. The method according to claim 1, characterized in that The dynamic information matching strategy is as follows: Step 6-1: Select the information set in the target user's attention space as the information to be disseminated to enhance their attention; Step 6-2: Traverse the path set, and for the selected path, calculate the potential change in attention of the target user caused by the information k propagated through this path l. The calculation formula is as follows: Δatt i，k，l =point i，k B i，k +qEffect i，j +mP l Among them, p, q, and m are the balance coefficients of each factor, ranging from 0 to 1. i，k is the user i’s current boredom with information k. Its value decreases as the number of times the user receives the information increases. The smaller the value, the higher the boredom, and the smaller the impact on attention change. It is calculated as follows: B i，k =1-n×num k where num k is the number of times the user has received the message, n is the scaling factor, based on the integer num k The value range of is scaled to between 0 and 1; Through calculation, match each path with the information with the best transmission effect in the current sending round; Step 6-3: Update the target user's attention to the message; This process can continue until the target user's attention value for the erroneous information drops below that of all the information in the attention space, and it is considered that the erroneous information fades out of the user's attention space and the impact is eliminated.

7. The method according to claim 1, characterized in that The evaluation method under the IC model is as follows: Step 7-1: Select the target node and its error information in the attention space; Step 7-2: According to the target node and its surrounding neighborhood users, the source node is determined, which detects that the target node is contaminated, starts to explore possible paths for propagating correct information, and determines the correct information to be propagated; Step 7-3: Start to spread the correct information until the target user's attention value to the wrong information is lower than all the information in the attention space, exit the target's attention space, and the influence is eliminated; The fewer rounds of information dissemination required to eliminate the impact, the better the effect of the proof method.

Images