A message control method, product, device and medium

Abstract

The present invention discloses a message control method, product, device and medium, and relates to the technical field of information dissemination control. This scheme defines the concept of gravity value for each node of a social network; the gravity value represents the degree of information dissemination correlation between two nodes; the corresponding clustering label is updated according to the gravity value of each node, and each node in the social network can be divided into multiple sub-networks according to the clustering label, and finally the message control is performed on each sub-network respectively, which greatly reduces the computational complexity. At the same time, since the split sub-networks are independent of each other, different sub-networks can concurrently and synchronously execute related operations such as negative message control, thereby improving the concurrency of the system, shortening the algorithm execution cycle and enhancing the scalability of the algorithm, and is suitable for large-scale social networks.

Description

Technical Field

[0001] The present invention relates to the technical field of information dissemination control, and in particular to a message control method, product, device and medium. Background Art

[0002] Compared with traditional media, online social networks spread information very quickly, especially in reporting major news or public emergencies. Users can participate in discussions and express their opinions in real time, thus influencing the course of events. However, there is also the phenomenon of spreading negative information such as rumors on the Internet. These false news may cause economic losses and disrupt social order. To address this problem, scholars have conducted research on negative information control (NIC), and the results have been applied to rumor prevention, social media management, and network security supervision.

[0003] However, due to the complex structure of social networks and the large number of nodes and edges, the computational process of centralized algorithms to screen key edges in the global network is extremely complex. Especially when the network scale expands, the computational complexity grows exponentially, making the algorithm difficult to apply to large-scale networks and limiting its scalability.

[0004] In view of the above, how to solve the current complex structure of social networks, the extremely complex calculation process of screening key edges in the global network, poor scalability, and unfavorable for negative message control is an urgent problem to be solved by technicians in this field. Summary of the invention

[0005] The purpose of the present invention is to provide a message control method, product, device and medium to solve the problem that the current social network structure is complex, the calculation process of screening key edges in the global network is extremely complex, the scalability is poor, and it is not conducive to negative message control.

[0006] In order to solve the above technical problems, the present invention provides a message control method, comprising:

[0007] Obtaining a social network and determining a gravity value of each node in the social network; wherein the gravity value represents the degree of information propagation correlation between two nodes;

[0008] Get the initial cluster label of each node, and update the cluster label of the corresponding node according to each gravity value;

[0009] Divide each node in the social network into multiple sub-networks according to the updated cluster labels; wherein the cluster labels of each node in the same sub-network are the same;

[0010] Message control is performed on each sub-network separately.

[0011] On the one hand, determine the gravity value of each node in the social network, including:

[0012] Obtain the similarity between nodes and obtain the information entropy of each node;

[0013] The gravity value of each node is determined according to the preset non-negative parameters, the similarity between nodes and the information entropy of each node.

[0014] On the other hand, the information entropy of each node is obtained, including:

[0015] Get the total number of nodes, total number of edges and out-degree of each node in the social network;

[0016] Determine the first information sending probability of each node according to the out-degree and total number of edges of each node;

[0017] The information entropy of each node is determined according to the total number of nodes and the probability of each first information being sent.

[0018] On the other hand, the cluster labels of the corresponding nodes are updated according to the gravity values, including:

[0019] Calculate the sum of gravity values ​​between each node and the remaining nodes to obtain a set of total gravity values;

[0020] Select a starting node from all nodes;

[0021] Determine the maximum total gravity value among the total gravity values, and determine the node corresponding to the maximum total gravity value as the end node;

[0022] Update the cluster label of the starting node according to the cluster label of the ending node;

[0023] Delete the maximum gravity value sum from the gravity value sum set, and determine whether the current gravity value sum set is an empty set;

[0024] If not, a new starting node is selected from the remaining nodes, and the process returns to the step of determining the maximum sum of gravity values ​​among the sums of gravity values;

[0025] If yes, the cluster label update process ends.

[0026] On the other hand, message control is performed on each sub-network separately, including:

[0027] Obtain a seed set; wherein each node in the seed set is a negative message node; and the seed set is a subset of the social network;

[0028] Determine whether the intersection of the sub-network and the seed set is an empty set;

[0029] If not, determine the influence propagation value of each edge in the subnetwork;

[0030] Determine the candidate edges to be removed from the subnetwork based on the influence propagation value of each edge;

[0031] Remove candidate edges from the subnetwork.

[0032] On the other hand, the influence propagation value of each edge in the subnetwork is determined, including:

[0033] Determine the influence propagation value of each node in the subnetwork based on the two-hop neighbor information;

[0034] The influence propagation value of each edge is determined according to the influence propagation value of each node.

[0035] On the other hand, the influence propagation value of each node in the subnetwork based on the two-hop neighbor information is determined, including:

[0036] Determine the one-hop neighbor node and the two-hop neighbor node corresponding to each node in the subnetwork;

[0037] Determine the influence propagation value of each one-hop neighbor node and the influence propagation value of each two-hop neighbor node; wherein the initial influence propagation value of each node in the subnetwork is 1;

[0038] The influence propagation value of the corresponding node is determined according to the influence propagation value of each one-hop neighbor node and the influence propagation value of each two-hop neighbor node.

[0039] On the other hand, according to the influence propagation value of each one-hop neighbor node and the influence propagation value of each two-hop neighbor node, the influence propagation value of the corresponding node is determined, including:

[0040] Determine a probability of each node sending a second message to a corresponding one-hop neighbor node;

[0041] Determine a probability of sending third information from each one-hop neighbor node to a corresponding two-hop neighbor node;

[0042] Determine the probability value of the seed set activating each node; wherein the initial probability value of the seed set activating each node is 0;

[0043] The influence propagation value of the corresponding node is determined according to the influence propagation value of each one-hop neighbor node, the influence propagation value of each two-hop neighbor node, each second information sending probability, each third information sending probability and each probability value.

[0044] On the other hand, before determining the candidate removal edges of the sub-network according to the influence propagation value of each edge, it also includes:

[0045] Get the total number of edges in the social network, and get the number of edges and removal parameters of the subnetwork;

[0046] Determine the removal threshold of the candidate removal edges based on the total number of edges, the number of edges, and the removal parameters.

[0047] On the other hand, after removing the candidate removal edges from the sub-network, it also includes:

[0048] Determine whether the number of candidate removal edges that have been removed is greater than the removal threshold;

[0049] If not, return to the step of determining the influence propagation value of each edge in the subnetwork;

[0050] If yes, the edge removal process of the subnetwork ends.

[0051] On the other hand, if it is confirmed that the number of candidate removal edges that have been removed is not greater than the removal threshold, it also includes:

[0052] Update the influence propagation value of each node in the sub-network;

[0053] Update the probability value of activating each node in the seed set.

[0054] On the other hand, after determining the candidate removal edges of the sub-network according to the influence propagation value of each edge, it also includes:

[0055] Update the removed edge set of the subnetwork according to the candidate removed edges; wherein the removed edge set includes the candidate removed edges;

[0056] Correspondingly, the candidate edges are removed from the sub-network, including:

[0057] The removed edge sets corresponding to each sub-network are removed as a whole.

[0058] In order to solve the above technical problem, the present invention also provides a computer program product, including a computer program or instructions, which implement the steps of the above message control method when executed by a processor.

[0059] In order to solve the above technical problems, the present invention further provides a message control device, comprising:

[0060] Memory for storing computer programs;

[0061] The processor is used to implement the steps of the above-mentioned message control method when executing a computer program.

[0062] In order to solve the above technical problem, the present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above message control method are implemented.

[0063] The message control method provided by the present invention obtains a social network and determines the gravity value of each node in the social network; wherein the gravity value represents the degree of information transmission correlation between two nodes; obtains the initial clustering label of each node, and updates the clustering label of the corresponding node according to each gravity value; divides each node in the social network into multiple sub-networks according to each updated clustering label; wherein the clustering labels of each node in the same sub-network are the same; and performs message control on each sub-network respectively.

[0064] The beneficial effect of the present invention is that the concept of gravity value is defined for each node of the social network; the gravity value represents the degree of information transmission correlation between two nodes; the corresponding clustering label is updated according to the gravity value of each node, and each node in the social network can be divided into multiple sub-networks according to the clustering label, and finally the message control is performed on each sub-network respectively, which greatly reduces the computational complexity. At the same time, since the sub-networks after the split are independent of each other, the different sub-networks can concurrently and synchronously perform related operations such as negative message control, thereby improving the concurrency of the system, shortening the algorithm execution cycle and enhancing the scalability of the algorithm, and being suitable for large-scale social networks.

[0065] On the other hand, the present invention specifically achieves the determination of the node gravity value by obtaining the information entropy and similarity of each node in the social network. The gravity value can measure the attraction between nodes. By updating the node clustering label through the gravity value, closer nodes can be classified into the same sub-network, thereby improving the accuracy of sub-network division. The clustering of nodes is divided by the gravity value between nodes. The process of iteratively updating the clustering label makes the node attracted to the cluster where the node similar to it is located, thereby improving the accuracy of sub-network division. By obtaining the seed set, it is determined whether the intersection of the sub-network and the seed set is an empty set; if not, the influence propagation value of each edge in the sub-network is determined, and the candidate removal edge of the sub-network is determined according to the influence propagation value of each edge, and the candidate removal edge is removed from the sub-network, thereby minimizing the message propagation. By determining the influence propagation value of each node in the sub-network based on the two-hop neighbor information, the influence propagation value of each edge is determined according to the influence propagation value of each node, and the two-hop neighbor node information is considered, so that the node influence determination is more accurate.

[0066] In addition, the present invention also provides a computer program product, a message control device and a medium, with the same effects as above. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to more clearly illustrate the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0068] Figure 1 A flow chart of a message control method provided by an embodiment of the present invention;

[0069] Figure 2 A schematic diagram of calculating the node influence propagation value provided by an embodiment of the present invention;

[0070] Figure 3 A schematic diagram of a message control device provided by an embodiment of the present invention;

[0071] Figure 4 A schematic diagram of a message control device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0072] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0073] The core of the present invention is to provide a message control method, product, device and medium to solve the problems that the current social network structure is complex, the calculation process of screening key edges in the global network is extremely complex, the scalability is poor, and it is not conducive to negative message control.

[0074] In order to enable those skilled in the art to better understand the solution of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0075] Compared with traditional media, online social networks spread information very quickly, especially in reporting major news or public emergencies. Users can participate in discussions and express their opinions in real time, thus influencing the course of events. However, there is also the phenomenon of spreading negative information such as rumors on the Internet. These false news may cause economic losses and disrupt social order. To address this problem, scholars have conducted research on the NIC problem, and the results have been applied to the fields of rumor prevention, social media management, and network security supervision.

[0076] The so-called NIC problem is that given an initial seed set of rumor carriers in a social network and an information propagation model (which describes how information propagates in the network), under the condition of a limited budget, at most k key edges are allowed to be selected (these key edges are considered to be the key paths for information propagation), and these edges are removed from the network to cut off the key paths for information propagation, so that the number of network users who receive the rumor information is minimized.

[0077] However, due to the complex structure of social networks and the large number of nodes and edges, the calculation process of centralized algorithms to screen key edges in the global network is extremely complex. Especially when the network scale expands, the calculation complexity increases exponentially, making the algorithm difficult to apply to large-scale networks and limiting its scalability. Therefore, in order to solve the above problems, the present invention provides a message control method.

[0078] Figure 1 Flow chart of a message control method provided by an embodiment of the present invention. Figure 1 As shown, the method includes:

[0079] S10: Obtain a social network and determine the gravity value of each node in the social network.

[0080] Among them, the gravity value represents the degree of information propagation correlation between two nodes.

[0081] Specifically, first obtain the social network. A social network is a complex network consisting of users (nodes) and their relationships (edges), in which information is propagated through interactions between users. At the same time, determine the gravity value of each node in the social network.

[0082] It should be noted that the gravity value represents the degree of information transmission correlation between two nodes. In this embodiment, there is no limitation on the specific method of determining the node gravity value, for example, it can be determined based on information entropy, similarity, etc. of the nodes.

[0083] S11: Obtain the initial clustering label of each node, and update the clustering label of the corresponding node according to each gravity value.

[0084] Furthermore, the initial clustering label of each node is obtained. It should be noted that the clustering label represents the category of the node in the social network; initially, it is assumed that each node belongs to a category, and when the sub-network is subsequently divided, "similar" nodes will be clustered into one category, and the labels will remain consistent after clustering.

[0085] Therefore, in order to realize the sub-network division, nodes of different categories need to be clustered. In this embodiment, the clustering label of the corresponding node is updated according to the gravity value of each node. In this embodiment, there is no restriction on the updating process of the clustering label.

[0086] S12: Divide each node in the social network into multiple sub-networks according to the updated cluster labels.

[0087] Among them, the clustering labels of each node in the same sub-network are the same.

[0088] S13: Execute message control on each sub-network respectively.

[0089] Finally, the nodes in the social network are divided into multiple sub-networks according to the updated cluster labels, that is, the nodes with the same cluster label are divided into the same sub-network, and the cluster labels of the nodes in the obtained multiple sub-networks are the same. In this way, the sub-network splitting of the social network is realized, so that message control can be performed on each sub-network separately.

[0090] It should be noted that the specific process of executing message control on each sub-network in this embodiment is not limited. The sub-networks can be sorted and message control can be executed in sequence, or message control can be executed on multiple sub-networks in parallel, depending on the specific implementation situation.

[0091] In this embodiment, the concept of gravity value is defined for each node of the social network; the gravity value represents the degree of information transmission correlation between two nodes; further, the corresponding clustering label is updated according to the gravity value of each node, and each node in the social network can be divided into multiple sub-networks according to the clustering label, and finally the message control is performed on each sub-network respectively, which greatly reduces the computational complexity. At the same time, since the sub-networks after the split are independent of each other, the different sub-networks can concurrently and synchronously perform related operations such as negative message control, thereby improving the concurrency of the system, shortening the algorithm execution cycle and enhancing the scalability of the algorithm, which is suitable for large-scale social networks.

[0092] Based on the above embodiments, in some embodiments, determining the gravity value of each node in the social network includes:

[0093] S101: Obtain the similarity between each node, and obtain the information entropy of each node;

[0094] S102: Determine the gravity value of each node according to a preset non-negative parameter, the similarity between each node and the information entropy of each node.

[0095] Specifically, each node in the social network is represented by a d-dimensional feature vector. Assume that the feature vector of node i is , the feature vector of node j is In order to determine the node gravity value, in this embodiment, it is first necessary to obtain the similarity between the nodes and obtain the information entropy of each node.

[0096] Specifically, the cosine similarity between nodes is obtained, and the formula is as follows:

[0097] ;

[0098] in, is the cosine similarity between node i and node j, is the modulus of the feature vector of node j, i and j are 1 to Integers between 1 and ). Therefore, for a The social network of nodes can be obtained by using the above formula. The network node cosine similarity matrix is ​​as follows:

[0099] ;

[0100] in, is the cosine similarity matrix of network nodes.

[0101] Furthermore, the information entropy of each node is obtained. The calculation formula of information entropy is as follows:

[0102] ;

[0103] in, is the information entropy of the node, is the total number of network nodes, is the probability of sending the first information from node i to node j.

[0104] In some embodiments, in order to determine the information entropy of each node, it is specifically necessary to obtain the total number of nodes in the social network. , the total number of edges and the out-degree of each node ; The probability of sending the first information of each node is determined according to the out-degree and total number of edges of each node. The formula is as follows:

[0105] ;

[0106] in, is the probability of sending the first information from node i to node j, is the number of edges from node i to node j, that is, the out-degree of the node. Thus, the information entropy of each node can be determined according to the total number of nodes and the probability of sending each first message in combination with the information entropy calculation formula.

[0107] Finally, the gravity value of each node is determined according to the preset non-negative parameters, the similarity between nodes and the information entropy of each node. The formula is as follows:

[0108] ;

[0109] in, and Represent the information entropy of node i and node j respectively, Represents the cosine similarity between node i and node j; preset non-negative parameter , In this embodiment, there is no restriction on the specific values ​​of the preset non-negative parameters, which depend on the specific implementation situation.

[0110] In summary, the gravitational value matrix between any two nodes in a social network can be calculated as follows:

[0111] ;

[0112] In this embodiment, the node gravity value is determined by obtaining the information entropy and similarity of each node in the social network. The gravity value can measure the attraction between nodes. By updating the node clustering label through the gravity value, closer nodes can be classified into the same sub-network, thereby improving the accuracy of sub-network division.

[0113] Based on the above embodiments, in some embodiments, updating the clustering label of the corresponding node according to each gravity value includes:

[0114] S111: Calculate the sum of gravity values ​​between each node and the remaining nodes to obtain a set of gravity value sums.

[0115] S112: Select a starting node from all nodes.

[0116] S113: Determine the maximum total gravity value among the total gravity values, and determine the node corresponding to the maximum total gravity value as the end node.

[0117] S114: Update the cluster label of the start node according to the cluster label of the end node.

[0118] S115: Delete the maximum gravity value sum from the gravity value sum set, and determine whether the current gravity value sum set is an empty set; if not, proceed to step S116. If yes, the cluster label update process ends.

[0119] S116: Select a new starting node from the remaining nodes and return to step S113.

[0120] In order to update the clustering label of the node, a starting node is selected from all the nodes in the social network. It should be noted that the starting node is any node among all the nodes, and the type is the same as the rest of the nodes in the social network. At the same time, the sum of the gravitational values ​​between each node and the rest of the nodes is calculated. The formula is as follows:

[0121] ;

[0122] in, is the sum of the gravitational values ​​corresponding to node i.

[0123] Furthermore, the maximum gravity value sum is determined among the gravity value sums, and the node corresponding to the maximum gravity value sum is determined as the end node. The clustering label of the starting node is updated according to the clustering label of the end node:

[0124] ;

[0125] in, is the cluster label of node i.

[0126] Then, the largest sum of gravity values ​​is deleted from the sum of gravity values ​​set, and it is determined whether the current sum of gravity values ​​set is an empty set. If it is confirmed that the current sum of gravity values ​​set is not an empty set, it is considered that the cluster labels of the existing nodes have not been updated, and a new starting node needs to be selected from the remaining nodes, and the step of determining the largest sum of gravity values ​​in the sum of gravity values ​​is returned to iteratively update the cluster labels; if it is confirmed that the current sum of gravity values ​​set is an empty set, it is considered that the cluster labels of all current nodes have been updated, and the cluster label update process ends.

[0127] In this embodiment, the nodes are divided into clusters according to the gravitational values ​​between the nodes, and the process of iteratively updating the cluster labels makes the nodes attracted to the clusters where the nodes similar to them are located, thereby improving the accuracy of the division of the sub-network.

[0128] At present, when dealing with key edge screening in online social networks, traditional centralized algorithms mainly start from a global perspective and ignore the local characteristics of the network, such as community structure. In online social networks, users are often clustered into communities due to common attributes or interests. The communities are closely connected and there are significant differences between communities. Therefore, relying solely on global methods may not be able to fully capture these local characteristics, resulting in low algorithm efficiency and difficulty in expanding to large-scale networks. Therefore, in order to solve the above problems, after using the sub-network splitting process in the above embodiments to divide the social network into multiple sub-networks, in some embodiments, message control is performed on each sub-network separately, including:

[0129] S131: Obtain a seed set; wherein each node in the seed set is a negative message node; and the seed set is a subset of the social network.

[0130] S132: Determine whether the intersection of the sub-network and the seed set is an empty set; if not, proceed to step S133. If yes, end the message control process.

[0131] S133: Determine the influence propagation value of each edge in the sub-network.

[0132] S134: Determine candidate edges to be removed from the sub-network according to the influence propagation value of each edge.

[0133] S135: Remove the candidate removal edge from the sub-network.

[0134] Specifically, in order to achieve message control on the sub-network, it is first necessary to obtain the seed set of the social network It should be noted that the seed set Each node in is a negative message node, and the seed set is a subset of the social network. At the same time, initialize the directed social network ,in Form a set for all users, i.e., a node set; It is the complex set of relationships between all users, i.e., the edge set; is the information propagation probability set. The weight of the social network is standardized, that is, the sum of all edge weights is calculated, and then each edge weight is divided by the total weight, so that each weight value is between 0 and 1.

[0135] Further determine the sub-network and seed set Is the intersection of is an empty set, that is, to determine whether the seed set Is the node in the sub-network not in the sub-network? If the sub-network is confirmed to be consistent with the seed set If the intersection of is an empty set, then the seed set is considered The nodes in are not in the sub-network, so they will not affect other users in the sub-network, and the message control process of the sub-network ends. If the intersection of is not an empty set, then the seed set is considered The nodes in the sub-network need to perform message control on the sub-network.

[0136] Specifically, determine the influence propagation value of each edge in the subnetwork. It should be noted that the influence propagation value of an edge represents the influence of the edge in the process of information propagation in the subnetwork. In this embodiment, there is no restriction on the process of determining the influence propagation value of each edge. Further, determine the candidate removal edges of the subnetwork based on the influence propagation value of each edge, and the candidate removal edges are related to the nodes of the seed set. Finally, remove the candidate removal edges from the subnetwork, thereby cutting off the propagation path of negative messages in the network. It should be noted that in this embodiment, there is no restriction on the method of determining the candidate removal edges.

[0137] In this embodiment, by obtaining the seed set, it is determined whether the intersection of the sub-network and the seed set is an empty set; if not, the influence propagation value of each edge in the sub-network is determined, and the candidate removal edges of the sub-network are determined according to the influence propagation value of each edge, and the candidate removal edges are removed from the sub-network, thereby minimizing the message propagation.

[0138] Based on the above embodiments, in some embodiments, determining the influence propagation value of each edge in the subnetwork includes:

[0139] S141: Determine the influence propagation value of each node in the subnetwork based on the two-hop neighbor information;

[0140] S142: Determine the influence propagation value of each edge according to the influence propagation value of each node.

[0141] In order to determine the influence propagation value of each edge in the subnetwork, in this embodiment, the influence propagation value of each node in the subnetwork based on the two-hop neighbor information is first determined. It should be noted that the influence propagation value based on the two-hop neighbor information is specifically generated according to the two-hop neighbor node information of the node, and represents the ability of the node to influence other nodes during the information propagation process. In this embodiment, there is no restriction on the determination process of the influence propagation value of each node based on the two-hop neighbor information.

[0142] Furthermore, the influence propagation value of each edge is determined according to the influence propagation value of each node. The specific formula is as follows:

[0143] ;

[0144] in, is the influence propagation value of edge (u,v), For Node The influence spread value of For Node The influence spread value.

[0145] In this embodiment, by determining the influence propagation value of each node in the subnetwork based on the two-hop neighbor information, the influence propagation value of each edge is determined according to the influence propagation value of each node, and the two-hop neighbor node information is taken into consideration, so that the node influence determination is more accurate.

[0146] Based on the above embodiments, in some embodiments, determining the influence propagation value of each node in the subnetwork based on two-hop neighbor information includes:

[0147] S151: Determine the one-hop neighbor node and the two-hop neighbor node corresponding to each node in the sub-network;

[0148] S152: Determine the influence propagation value of each one-hop neighbor node and the influence propagation value of each two-hop neighbor node; wherein the initial influence propagation value of each node in the subnetwork is 1;

[0149] S153: Determine the influence propagation value of the corresponding node according to the influence propagation value of each one-hop neighbor node and the influence propagation value of each two-hop neighbor node.

[0150] In order to determine the influence propagation value of each node based on the two-hop neighbor information, in this embodiment, the one-hop neighbor node and the two-hop neighbor node corresponding to each node in the sub-network are specifically determined, and the influence propagation value of each one-hop neighbor node and the influence propagation value of each two-hop neighbor node are determined. It should be noted that the initial influence propagation value of each node in the sub-network is 1.

[0151] Further, according to the influence propagation value of each one-hop neighbor node and the influence propagation value of each two-hop neighbor node, the influence propagation value of the corresponding node is determined. The specific process of determining the influence propagation value of the node is described below:

[0152] First, determine the probability of each node sending the second information to the corresponding one-hop neighbor node. and Respectively represent nodes One-hop neighbor node and two-hop neighbor node, then For Node To the corresponding one-hop neighbor node The probability of sending the second information.

[0153] Further determine the probability of sending the third information from each one-hop neighbor node to the corresponding two-hop neighbor node. and Respectively represent nodes One-hop neighbor node and two-hop neighbor node, then For Node One-hop neighbor node To the corresponding two-hop neighbor node The probability of sending the third information.

[0154] Then the probability value of the seed set activating each node is determined. It should be noted that the initial probability value of the seed set activating each node is 0. The probability value of the seed set activating each node is calculated as follows:

[0155] ;

[0156] in, For seed collection Activate Node The probability value of , s is the seed set Nodes in .

[0157] Finally, according to the influence propagation value of each one-hop neighbor node, the influence propagation value of each two-hop neighbor node, each second information sending probability, each third information sending probability and each probability value, the influence propagation value of the corresponding node is determined. The calculation formula is as follows:

[0158] ;

[0159] in, For Node The influence spread value of and Node One-hop neighbor nodes and two-hop neighbor nodes, represents an adjustable coefficient, Representation Node The neighbor nodes of Representation Node The neighbor nodes of For Node To the corresponding one-hop neighbor node The probability of sending the second information is, For Node One-hop neighbor node To the corresponding two-hop neighbor node The probability of sending the third information is For Node The influence spread value of For Node The influence spread value.

[0160] Figure 2 Schematic diagram of node influence propagation value calculation provided by an embodiment of the present invention. Figure 2 As shown in the figure, a simple social network contains 4 nodes, and the value on each edge represents the propagation probability value between nodes. After using the standard independent cascade model, the influence propagation value of node 1 is 2.54, the influence propagation value of node 2 is 1.2, the influence propagation value of node 3 is 1, and the influence propagation value of node 4 is 1.4. When only the influence ranking and influence estimation (IRIE) algorithm of one-hop neighbor nodes is used, the influence propagation value of node 1 is only 1.3, the influence propagation value of node 2 is 1.1, the influence propagation value of node 3 is 1, and the influence propagation value of node 4 is 1.2.

[0161] In comparison, the present invention uses two-hop neighbor information, and the influence propagation value of node 1 is 1.6, the influence propagation value of node 2 is 1.1, the influence propagation value of node 3 is 1, and the influence propagation value of node 4 is 1.2. In summary, compared with the classical method, the node influence propagation value calculated by the present invention is closer to the ideal independent cascade model, and thus shows a certain degree of accuracy.

[0162] In order to determine the number of edges that need to be removed from each sub-network, based on the above embodiments, in some embodiments, before determining the candidate edges to be removed from the sub-network according to the influence propagation value of each edge, the method further includes:

[0163] S161: Obtain the total number of edges of the social network, and obtain the number of edges and removal parameters of the subnetwork;

[0164] S162: Determine a removal threshold of a candidate edge to be removed according to the total number of edges, the number of edges, and the removal parameter.

[0165] Specifically, first, the total number of edges of the social network is obtained, and the number of edges and removal parameters of the subnetwork are obtained. In this embodiment, there is no limit on the size of the removal parameter, which depends on the specific implementation situation. Then, according to the total number of edges, the number of edges and the removal parameter, the removal threshold of the candidate removal edge is determined, and the specific formula is as follows:

[0166] ;

[0167] in, For subnetwork The removal threshold is For subnetwork The number of edges, To remove a parameter, For social networks In this embodiment, the total number of edges of the social network Divide into sub-network, , Is a positive integer.

[0168] Therefore, after determining the removal threshold of the candidate removal edge of the sub-network and removing the candidate removal edge from the sub-network, the method further includes:

[0169] S163: Determine whether the number of candidate edges that have been removed is greater than the removal threshold; if not, return to step S133; if so, end the edge removal process of the sub-network.

[0170] Specifically, after the candidate edges for removal of the sub-network are removed, it is determined whether the number of candidate edges for removal that have been removed is greater than the removal threshold. If it is confirmed that the number of candidate edges for removal that have been removed is not greater than the removal threshold, it is considered that other candidate edges for removal need to be removed from the sub-network, and the process returns to the step of determining the influence propagation value of each edge in the sub-network. If it is confirmed that the number of candidate edges for removal that have been removed is greater than the removal threshold, it is considered that a sufficient number of edges have been removed from the sub-network, and no further removal is required, and the edge removal process of the sub-network can be terminated.

[0171] In this embodiment, by setting a removal threshold of candidate removal edges of a sub-network, a corresponding number of edges can be accurately removed when executing message control on the sub-network, thereby ensuring normal communication between nodes on the basis of preventing the propagation of negative messages.

[0172] Furthermore, if it is confirmed that the number of candidate removal edges that have been removed is not greater than the removal threshold, the method further includes:

[0173] S164: Update the influence propagation value of each node in the sub-network;

[0174] S165: Update the probability value of activating each node in the seed set.

[0175] In order to ensure the accuracy of selecting the next candidate edge to be removed, in this embodiment, if it is confirmed that the number of candidate edges to be removed is not greater than the removal threshold, it is also necessary to update the influence propagation value of each node in the subnetwork and update the probability value of each node activated by the seed set. It can be understood that the process of updating the influence propagation value of each node is the same as the process of determining the influence propagation value of the node in the above embodiment, and the process of updating the probability value of each node activated by the seed set is the same as the process of determining the probability value of each node activated by the seed set in the above embodiment. In this embodiment, by updating the influence propagation value of each node in the subnetwork and the probability value of each node activated by the seed set, the accuracy of selecting the next candidate edge to be removed in the subnetwork can be guaranteed.

[0176] In order to better manage the candidate removal edges and more conveniently remove the candidate removal edges, based on the above embodiments, in some embodiments, after determining the candidate removal edges of the sub-network according to the influence propagation value of each edge, the method further includes:

[0177] S171: Update the removed edge set of the subnetwork according to the candidate removed edges; wherein the removed edge set includes the candidate removed edges.

[0178] Specifically, when executing message control on a sub-network, the removal edge set of the sub-network may be initialized first. It is understandable that the initial removal edge set of the sub-network is an empty set. Further, after determining the candidate removal edges of the sub-network according to the influence propagation value of each edge, the removal edge set of the sub-network is updated according to the candidate removal edges, that is, the candidate removal edges are put into the removal edge set.

[0179] Correspondingly, in order to remove the candidate removal edges from the sub-network, the removal edge set corresponding to each sub-network is removed as a whole, without removing the candidate removal edges of each sub-network multiple times, which greatly improves the convenience of edge removal.

[0180] In summary, this paper proposes for the first time a node influence calculation method based on two-hop neighbor information, overcoming the inefficiency of the approximate independent cascade model of the traditional heuristic method. On the basis of the node influence evaluation method, a method for calculating the influence propagation value of the network edge is defined, thereby effectively screening the key edges. Since the algorithm can be executed simultaneously in multiple subnetworks, the solution space and solution time of the traditional centralized algorithm are greatly reduced.

[0181] In addition, in the process of negative information control, source node control and positive information injection are also two key strategies. Specifically, source node control first relies on monitoring and identification technology, which quickly locates the initial disseminator of negative information by tracking the information flow in the network in real time and using advanced algorithms such as natural language processing and machine learning. Once the source node is identified, a variety of control measures can be taken, including but not limited to warnings, account bans and content deletions, to directly prevent the further spread of negative information. Positive information injection hedges the impact of negative information by increasing the spread of positive and accurate information. Its core is to guide the public to correctly understand events or issues by providing objective, fair and fact-based content, thereby reducing the trust in negative information. The specific implementation includes content creation and publishing, that is, organizing professional teams or using automated tools to create positive content, and ensuring that these contents can provide comprehensive background information. Through the platform's recommendation algorithm and advertising system, positive information is accurately pushed to users who may be affected by negative information, increasing their exposure and dissemination range, thereby more effectively offsetting the negative effects of negative information.

[0182] In the above embodiments, the message control method is described in detail. The present invention also provides a corresponding embodiment of the message control device.

[0183] Figure 3 Schematic diagram of a message control device provided by an embodiment of the present invention. Figure 3 As shown, the device comprises:

[0184] The acquisition module 10 is used to acquire the social network and determine the gravity value of each node in the social network; wherein the gravity value represents the degree of information propagation correlation between two nodes.

[0185] The updating module 11 is used to obtain the initial clustering label of each node and update the clustering label of the corresponding node according to each gravity value.

[0186] The division module 12 is used to divide each node in the social network into multiple sub-networks according to the updated clustering labels; wherein the clustering labels of each node in the same sub-network are the same.

[0187] The control module 13 is used to perform message control on each sub-network respectively.

[0188] In some embodiments, the acquisition module 10 includes:

[0189] The first acquisition submodule is used to obtain the similarity between each node and obtain the information entropy of each node;

[0190] The first determination submodule is used to determine the gravity value of each node according to a preset non-negative parameter, the similarity between each node and the information entropy of each node.

[0191] In some embodiments, the first acquisition submodule is specifically used to obtain the total number of nodes, the total number of edges and the out-degree of each node in the social network; determine the first information sending probability of each node based on the out-degree and the total number of edges of each node; determine the information entropy of each node based on the total number of nodes and the first information sending probability.

[0192] In some embodiments, the update module 11 includes:

[0193] The first calculation submodule is used to calculate the sum of gravity values ​​between each node and other nodes to obtain a set of gravity value sums;

[0194] The first selection submodule is used to select a starting node from all nodes;

[0195] The second determination submodule is used to determine the maximum sum of gravity values ​​among the sums of gravity values, and determine the node corresponding to the maximum sum of gravity values ​​as the end node;

[0196] A first updating submodule, used to update the clustering label of the starting node according to the clustering label of the ending node;

[0197] The first judgment submodule is used to delete the maximum gravity value sum from the gravity value sum set and determine whether the current gravity value sum set is an empty set; if not, a new starting node is selected from the remaining nodes to trigger the second determination submodule; if yes, the cluster label update process ends.

[0198] In some embodiments, the control module 13 includes:

[0199] The second acquisition submodule is used to acquire a seed set; wherein each node in the seed set is a negative message node; and the seed set is a subset of the social network;

[0200] The second judgment submodule is used to judge whether the intersection of the subnetwork and the seed set is an empty set; if not, the third determination submodule is triggered;

[0201] The third determination submodule is used to determine the influence propagation value of each edge in the subnetwork;

[0202] The fourth determination submodule is used to determine the candidate removal edges of the subnetwork according to the influence propagation value of each edge;

[0203] The removal submodule is used to remove candidate removal edges from the subnetwork.

[0204] In some embodiments, the third determining submodule includes:

[0205] A fifth determination submodule, used to determine the influence propagation value of each node in the subnetwork based on two-hop neighbor information;

[0206] The sixth determination submodule is used to determine the influence propagation value of each edge according to the influence propagation value of each node.

[0207] In some embodiments, the fifth determining submodule includes:

[0208] A seventh determination submodule is used to determine a one-hop neighbor node and a two-hop neighbor node corresponding to each node in the subnetwork;

[0209] An eighth determination submodule is used to determine the influence propagation value of each one-hop neighbor node and the influence propagation value of each two-hop neighbor node; wherein the initial influence propagation value of each node in the subnetwork is 1;

[0210] The ninth determination submodule is used to determine the influence propagation value of the corresponding node according to the influence propagation value of each one-hop neighbor node and the influence propagation value of each two-hop neighbor node.

[0211] In some embodiments, the ninth determination submodule is specifically used to determine the probability of each node sending the second information to the corresponding one-hop neighbor node; determine the probability of each one-hop neighbor node sending the third information to the corresponding two-hop neighbor node; determine the probability value of each node activated by the seed set; wherein the initial probability value of each node activated by the seed set is 0; determine the influence propagation value of the corresponding node according to the influence propagation value of each one-hop neighbor node, the influence propagation value of each two-hop neighbor node, each second information sending probability, each third information sending probability and each probability value.

[0212] In some embodiments, it also includes:

[0213] The third acquisition submodule is used to obtain the total number of edges of the social network and the number of edges and removal parameters of the subnetwork;

[0214] The tenth determination submodule is used to determine a removal threshold of a candidate removal edge according to the total number of edges, the number of edges and the removal parameter.

[0215] In some embodiments, it also includes:

[0216] The third judgment submodule is used to judge whether the number of candidate removal edges that have been removed is greater than the removal threshold; if not, the third determination submodule is triggered; if so, the edge removal process of the subnetwork is terminated.

[0217] In some embodiments, it also includes:

[0218] The second updating submodule is used to update the influence propagation value of each node in the subnetwork;

[0219] The third updating submodule is used to update the probability value of each node activated by the seed set.

[0220] In some embodiments, it also includes:

[0221] A fourth updating submodule is used to update a removal edge set of the subnetwork according to the candidate removal edge; wherein the removal edge set includes the candidate removal edge;

[0222] Correspondingly, the removal submodule is specifically used to remove the removal edge sets corresponding to each sub-network as a whole.

[0223] Since the embodiments of the apparatus part correspond to the embodiments of the method part, please refer to the description of the embodiments of the method part for the embodiments of the apparatus part, which will not be repeated here.

[0224] In addition, the present invention also provides a computer program product, including a computer program or instructions, which implement the steps of the above message control method when executed by a processor.

[0225] Figure 4 A schematic diagram of a message control device provided by an embodiment of the present invention. Figure 4 As shown, the message control device includes:

[0226] A memory 20, for storing computer programs;

[0227] The processor 21 is used to implement the steps of the message control method mentioned in the above embodiment when executing the computer program.

[0228] The message control device provided in this embodiment may include but is not limited to a smart phone, a tablet computer, a laptop computer, or a desktop computer.

[0229] Among them, the processor 21 may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor 21 can be implemented in at least one hardware form of a digital signal processor (DSP), a field programmable gate array (FPGA), and a programmable logic array. The processor 21 may also include a main processor and a coprocessor. The main processor is a processor for processing data in an awake state, also known as a central processing unit (CPU); the coprocessor is a low-power processor for processing data in a standby state. In some embodiments, the processor 21 may be integrated with a graphics processing unit (GPU), and the GPU is responsible for rendering and drawing the content to be displayed on the display screen. In some embodiments, the processor 21 may also include an artificial intelligence (AI) processor, which is used to process computing operations related to machine learning.

[0230] The memory 20 may include one or more computer-readable storage media, which may be non-transitory. The memory 20 may also include a high-speed random access memory, and a non-volatile memory, such as one or more disk storage devices, flash memory storage devices. In this embodiment, the memory 20 is at least used to store the following computer program 201, wherein, after the computer program is loaded and executed by the processor 21, it can implement the relevant steps of the message control method disclosed in any of the aforementioned embodiments. In addition, the resources stored in the memory 20 may also include an operating system 202 and data 203, etc., and the storage method may be temporary storage or permanent storage. Among them, the operating system 202 may include Windows, Unix, Linux, etc. Data 203 may include but is not limited to data related to the message control method.

[0231] In some embodiments, the message control device may further include a display screen 22 , an input / output interface 23 , a communication interface 24 , a power supply 25 , and a communication bus 26 .

[0232] Those skilled in the art will understand that Figure 4 The structure shown in the figure does not constitute a limitation on the message control device, and may include more or less components than those shown in the figure.

[0233] Finally, the present invention also provides an embodiment corresponding to a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps recorded in the above method embodiment are implemented.

[0234] It is understandable that if the method in the above embodiment is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc. Various media that can store program codes.

[0235] The above is a detailed introduction to a message control method, product, device and medium provided by the present invention. The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same and similar parts between the embodiments can be referenced to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description. It should be pointed out that for ordinary technicians in this technical field, without departing from the principle of the present invention, several improvements and modifications can be made to the present invention, and these improvements and modifications also fall within the scope of protection of the present invention.

[0236] It should also be noted that, in this specification, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the statement "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device including the element.

Claims

1. A message control method, characterized in that: include: Obtaining a social network and determining a gravity value of each node in the social network; wherein the gravity value represents the degree of information propagation correlation between two nodes; Obtaining an initial cluster label of each of the nodes, and updating the cluster label corresponding to the node according to each of the gravity values; Dividing each of the nodes in the social network into a plurality of sub-networks according to the updated cluster labels; wherein the cluster labels of the nodes in the same sub-network are the same; Performing message control on each of the sub-networks respectively; Executing message control on each of the sub-networks respectively, including: Obtain a seed set; wherein each node in the seed set is a negative message node; and the seed set is a subset of the social network; Determine whether the intersection of the sub-network and the seed set is an empty set; If not, determining the influence propagation value of each edge in the subnetwork; Determine the candidate removal edges of the sub-network according to the influence propagation value of each edge; removing the candidate removal edge from the sub-network; Wherein, determining the influence propagation value of each edge in the subnetwork includes: Determine a one-hop neighbor node and a two-hop neighbor node corresponding to each of the nodes in the subnetwork; Determine the influence propagation value of each of the one-hop neighbor nodes and the influence propagation value of each of the two-hop neighbor nodes; wherein the initial influence propagation value of each of the nodes in the subnetwork is 1; Determine a second information sending probability of each of the nodes to the corresponding one-hop neighbor node; Determine a probability of sending third information from each of the one-hop neighbor nodes to the corresponding two-hop neighbor node; Determining a probability value of the seed set activating each of the nodes; wherein the initial probability values ​​of the seed set activating each of the nodes are all 0; According to the influence propagation value of each of the one-hop neighbor nodes, the influence propagation value of each of the two-hop neighbor nodes, each of the second information sending probabilities, each of the third information sending probabilities and each of the probability values, the influence propagation value of the corresponding node is determined, and the formula is as follows: ； in, For Node The influence spread value of and Node One-hop neighbor nodes and two-hop neighbor nodes, represents an adjustable coefficient, Representation Node The neighbor nodes of Representation Node The neighbor nodes of For Node To the corresponding one-hop neighbor node The probability of sending the second information is, For Node One-hop neighbor node To the corresponding two-hop neighbor node The probability of sending the third information is For Node The influence spread value of For Node The influence spread value of For seed collection Activate Node The probability value of s For seed collection Nodes in The influence propagation value of each edge is determined according to the influence propagation value of each node.

2. The message control method according to claim 1, characterized in that: Determining the gravity value of each node in the social network includes: Obtaining the similarity between the nodes and obtaining the information entropy of the nodes; The gravitational value of each node is determined according to a preset non-negative parameter, the similarity between each node and the information entropy of each node.

3. The message control method according to claim 2, characterized in that: Obtaining the information entropy of each of the nodes, including: Obtaining the total number of nodes, the total number of edges and the out-degree of each node in the social network; Determine the first information sending probability of each of the nodes according to the out-degree of each of the nodes and the total number of edges; The information entropy of each of the nodes is determined according to the total number of the nodes and the probability of each of the first information being sent.

4. The message control method according to claim 1, characterized in that: Updating the cluster label corresponding to the node according to each gravity value includes: Calculating the sum of gravity values ​​between each of the nodes and the remaining nodes to obtain a set of gravity value sums; Select a starting node from all the nodes; Determine the largest sum of gravity values ​​among the sums of gravity values, and determine the node corresponding to the largest sum of gravity values ​​as the end node; Updating the cluster label of the starting node according to the cluster label of the ending node; Deleting the largest gravity value sum from the gravity value sum set, and determining whether the current gravity value sum set is an empty set; If not, a new starting node is selected from the remaining nodes, and the process returns to the step of determining the maximum sum of gravity values ​​among the sums of gravity values; If so, the cluster label update process ends.

5. The message control method according to claim 1, characterized in that: Before determining the candidate removal edges of the sub-network according to the influence propagation value of each edge, the method further includes: Obtaining the total number of edges of the social network, and obtaining the number of edges and removal parameters of the subnetwork; A removal threshold of the candidate removal edge is determined according to the total number of edges, the number of edges and the removal parameter.

6. The message control method according to claim 5, characterized in that: After removing the candidate removal edge from the sub-network, the method further includes: Determine whether the number of the candidate removal edges that have been removed is greater than the removal threshold; If not, return to the step of determining the influence propagation value of each edge in the subnetwork; If yes, the edge removal process of the sub-network ends.

7. The message control method according to claim 6, characterized in that: If it is confirmed that the number of the candidate removal edges that have been removed is not greater than the removal threshold, the method further includes: Updating the influence propagation value of each of the nodes in the sub-network; The probability value of each of the nodes activated by the seed set is updated.

8. The message control method according to any one of claims 1 to 7, characterized in that: After determining the candidate removal edges of the sub-network according to the influence propagation value of each edge, the method further includes: Update the removal edge set of the subnetwork according to the candidate removal edge; wherein the removal edge set includes the candidate removal edge; Correspondingly, removing the candidate removal edge from the sub-network includes: The removed edge sets corresponding to each of the sub-networks are removed as a whole.

9. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the steps of the message control method according to any one of claims 1 to 8 are implemented.

10. A message control device, characterized in that: include: Memory for storing computer programs; A processor, configured to implement the steps of the message control method according to any one of claims 1 to 8 when executing the computer program.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the message control method according to any one of claims 1 to 8 are implemented.

Images