Propagation suppression method and system based on key community recognition and terminal equipment

Abstract

The invention discloses a propagation suppression method and system based on key community recognition and terminal equipment, and the method comprises the steps: carrying out the community detection of multiple time slices in a sequential network, and constructing a community relation network; calculating a community time sequence scale gain index of each community node in the community relation network, and screening key community nodes; identifying a target propagation node set in the key community nodes; and releasing positive information through the network nodes in the target propagation node set so as to inhibit the propagation range of the inductive information. The method is applied to the field of social network analysis, comprehensively considers network density evolution and community density evolution, constructs cross-time random walk to obtain community importance scores, and preferentially considers large-scale communities with gains in time dimension, so that communities with higher influence can be obtained, influence nodes are identified according to key communities, and the social network analysis efficiency is improved. Therefore, propagation with higher influence is obtained, and effective suppression of multi-point propagation source induction information is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of social network analysis, in particular to a propagation suppression method and system based on key community identification and a terminal device. BACKGROUND

[0002] In the social network environment, the convenient forwarding characteristics of induced information make it difficult to completely eliminate the negative impact once the information spreads and diffuses, even if the original propagation source is removed. Moreover, in the case of induced information that has spread and diffused for multiple propagation sources, managing the propagation source cannot achieve the suppression purpose. Identifying the key community of the social network and identifying the core individual who publishes positive information in the community is a strategy to suppress the propagation of induced information, but the propagation suppression method based on key community identification has the following challenges:

[0003] 1) Compared with a single time slice, the community size in a multi-time slice network evolves, for example, community merging, expansion and extinction, which can cause changes in the influence gathering area in the network. However, the existing propagation suppression method based on key community identification pays less attention to the influence of community size changes on influential nodes, which can cause a single location node to lose the advantage of influence at a future time;

[0004] 2) The existing propagation suppression method based on key community identification focuses on optimizing the performance under a large-scale network, or improving the accuracy of seed node budget allocation according to community size, and only uses community detection as a preprocessing step. The research on the relationship between community characteristics and influential nodes is still insufficient. SUMMARY

[0005] In view of the deficiencies in the prior art, the present application provides a propagation suppression method and system based on key community identification to identify influential nodes based on key communities, thereby achieving effective suppression of induced information.

[0006] To achieve the above purpose, the present application provides a propagation suppression method based on key community identification, comprising the following steps:

[0007] Step 1: Community detection in a time series network for multiple time slices to construct a community relationship network;

[0008] Step 2: Calculate the community time series size gain index of each community node in the community relationship network, and filter key community nodes in the community relationship network based on the community time series size gain index;

[0009] Step 3: Identify a target propagation node set within the key community node;

[0010] Step 4, releasing positive information through the network nodes in the target propagation node set to suppress the propagation range of the induced information.

[0011] A propagation suppression method based on key community identification, characterized by comprising the following steps:

[0012] Step 1, community detection on multiple time slices in a time series network to construct a community relationship network;

[0013] Step 2, calculating a community time series scale gain index of each community node in the community relationship network, and screening key community nodes in the community relationship network based on the community time series scale gain index;

[0014] Step 3, identifying a target propagation node set in the key community nodes;

[0015] Step 4, releasing positive information through the network nodes in the target propagation node set to suppress the propagation range of the induced information.

[0016] In one of the embodiments, the community detection adopts Louvain method to form community division by iteratively merging network nodes to maximize modularity gain, and after deleting the communities with modularity of 0 or 1 and eliminating the communities with the number of network nodes less than a first threshold, the community relationship network is constructed based on the remaining communities as community nodes.

[0017] In one of the embodiments, in step 2, the community time series scale gain index is specifically:

[0018] WGC(u t )=DGR(G t )×DGC(u t )

[0019] Wherein, WGC(u t ) is the community time series scale gain index of community node u at t moment, DGR(G t ) is the network density growth rate of time series network G at t moment, and DGC(u t ) is the community centrality change rate of community node u at t moment.

[0020] In one of the embodiments, the calculation process of the network density growth rate is:

[0021]

[0022] Wherein, ρ(G t ) is the density of time series network G at t moment, and ρ(G t-ΔT ) is the density of time series network G at t-ΔT moment.

[0023] In one of the embodiments, the process of calculating the community centrality change rate is as follows:

[0024]

[0025] wherein, π(u t ) is the centrality score of community node u at time t, π(u t-ΔT ) is the centrality score of community node u at time t-ΔT, V is the set of community nodes of the community relation network, α is the damping factor, Z(v,u t ) is the set of all walk paths from community node v to community node u at time t, Pr[z] is the unconditional probability of path z occurring.

[0026] In one of the embodiments, in step 3, the process of identifying the target propagation node set is as follows:

[0027] Step 301, establishing the target propagation node set S * and the candidate node set S;

[0028] Step 302, obtaining all the network nodes within the community corresponding to the key community node and adding them to the candidate node set S;

[0029] Step 303, calculating the discount degree of all the network nodes in the current candidate node set S, which is as follows:

[0030] D v =1+[d v -2t v -(d v -t v )t v p+o(t v )]p

[0031] wherein, D v is the discount degree of network node v, d v is the node degree of network node v, t v is the difference between the number of neighbor nodes and the active state neighbor nodes of network node v, p is the edge activation probability, and o(t v ) is a small correction term.

[0032] Step 304, selecting the network node with the maximum discount degree

[0033] Step 305, setting S * =S * ∪{v *}, S=S\{v *}, and then judging the current target propagation node set S * ​whether the number of network nodes in the current target propagation node set S reaches a second threshold value:

[0034] if yes, output the current target propagation node set S * ;

[0035] otherwise, mark the network node v * as an active state, and return to step 303.

[0036] To achieve the above object, the application further provides a propagation suppression system based on key community identification, which adopts the above method for propagation suppression, and the propagation suppression system comprises:

[0037] a community relationship network construction unit, configured to perform community detection on multiple time slices in a time sequence network and construct a community relationship network;

[0038] a key community screening unit, configured to calculate a community time sequence scale gain index of each community node in the community relationship network, and screen key community nodes in the community relationship network based on the community time sequence scale gain index;

[0039] a propagation node identification unit, configured to identify a target propagation node set in the key community nodes;

[0040] a propagation suppression unit, configured to publish positive information through network nodes in the target propagation node set to suppress the propagation range of induced information.

[0041] To achieve the above object, the application further provides a terminal device, which is provided with:

[0042] a memory, configured to store a program;

[0043] a processor, configured to execute the program stored in the memory, and when the program is executed, the processor is configured to execute the method as described above.

[0044] Compared with the prior art, the application has the following beneficial technical effects:

[0045] The application identifies key communities by designing a community time sequence scale gain index, comprehensively considers the network density evolution and the community density evolution, simultaneously constructs a cross-time random walk to obtain a community importance score, preferentially considers communities with scale gain in the time dimension, so as to obtain communities with higher influence, and identifies influence nodes according to the key communities to obtain higher influence propagation, thereby effectively suppressing induced information of multiple propagation sources. BRIEF DESCRIPTION OF DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description only show some of the embodiments of the present application, and other drawings can be obtained by those of ordinary skill in the art without any creative effort based on these drawings.

[0047] Figure 1 The flow chart of the propagation suppression method based on key community identification in the embodiment of the present application;

[0048] Figure 2 The structural block diagram of the propagation suppression system based on key community identification in the embodiment 2 of the present application;

[0049] Figure 3 The structural block diagram of the terminal device in the embodiment 3 of the present application.

[0050] The implementation, functional features and advantages of the present application will be further described with reference to the embodiments and the drawings. DETAILED DESCRIPTION

[0051] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without any creative effort fall within the protection scope of the present application.

[0052] In addition, the technical solutions in each embodiment of the present application can be combined with each other, but it must be based on the fact that a person of ordinary skill in the art can realize it. When the combination of technical solutions appears contradictory or unachievable, it should be considered that the combination of technical solutions does not exist and is not within the protection scope of the present application.

[0053] As Figure 1 The propagation suppression method based on key community identification disclosed in the present embodiment mainly includes the following steps:

[0054] Step 1, community detection is performed on multiple time slices in a time series network, and a community relationship network is constructed;

[0055] Step 2, the community time series scale gain index of each community node in the community relationship network is calculated, and a key community node is screened in the community relationship network based on the community time series scale gain index;

[0056] Step 3, a target propagation node set is identified in the key community node;

[0057] Step 4: Positive information is published by network nodes in the target set of propagation nodes to suppress the propagation range of induced information.

[0058] In the implementation of step 1, the time-series network is constructed by time-series slicing of the social network. The time-series network is a dynamic network model that considers the dynamics of nodes and edges in the time dimension. Unlike static networks, the connection relationship in the time-series network changes over time, and the connection between nodes has timestamp information, reflecting the time dependence of the network topology. A typical time-series model based on network snapshots can be regarded as an extension of static graphs, reflecting the evolution process of the network from the time dimension. This type of model divides the entire time period of study into multiple network snapshots. Each snapshot can cover the network nodes, edges, and attribute states at a specific time, or it can be a static aggregation of all point and edge changes within a specific time window.

[0059] Due to the difficulty of data collection in reality, the present embodiment adopts a multi-network snapshot modeling method to establish a time-series network. Specifically, the time-series network can be represented as G=(G1, G2, …, G t ), G1, G2, …, G t represent the time-slice network at time nodes 1, 2, …, t. Among them, G t =(V t ,E t ,t), V t is the node set of the time-slice network G t , E t is the edge set of the time-slice network G t , and t is the time-slice number of the time-slice network G t . For example, in a social media network, the nodes of the time-slice network are social media users, and the edges of the time-slice network are information interaction behaviors of social media users. For example, if the number of information interaction behaviors between two information interaction behaviors exceeds a threshold within a specified period, the network edge corresponding to the two nodes exists in the time-slice network, and the two nodes are adjacent nodes, otherwise the network edge corresponding to the two nodes does not exist.

[0060] The community detection in the embodiment adopts the Louvain method to form community division by iteratively merging network nodes to maximize the modularity gain. That is, each network node in the time-series network is regarded as a separate community first, and the initial modularity is calculated, and then adjacent network nodes are iteratively merged into larger communities until the modularity cannot be increased any more. After completing the community division, each community has a certain number of network nodes, and all communities are preliminarily screened and removed on this basis to avoid redundant traversal of the entire network in the subsequent calculation stage. The preliminary screening and removal includes deleting communities with a modularity of 0 or 1, and eliminating communities with a number of network nodes less than a first threshold. Among them, from the physical meaning, the degree value of 0 indicates that it has no interaction with the core community, and the degree value of 1 indicates that it is only a node attached to the core community, not a core community. In addition, the first threshold can be set to 10, that is, the community with a small size and weak propagation potential. After completing the preliminary screening and removal, the remaining communities can be used as community nodes to build a community relationship network, and the connection edges of the community nodes in the community relationship network are synchronous with the connection edges of the network nodes in the network.

[0061] In the specific implementation process of step 2, the community time-series scale gain index is used to eliminate some communities with low density and small number of nodes. The community time-series scale gain index in the embodiment is composed of the DGR index and the RDC index. The DGR index evaluates the increase and decrease of the density of the time-series network, and the RDC index evaluates the importance of the community, and is used to select community nodes with gradually enhanced time-series influence, that is, the community time-series scale gain index is:

[0062] WGC(u t )=DGR(G t )×DGC(u t )

[0063] Wherein, WGC(u t ) is the community time-series scale gain index of the community node u at t time, DGR(G t ) is the network density growth rate of the time-series network G at t time, and DGC(v t ) is the community centrality change rate of the community node u at t time.

[0064] The DGR index mainly measures the growth speed of the time-series network in a fixed time interval ΔT, which is expressed by the density change between two consecutive time steps, that is:

[0065]

[0066] Wherein, ρ(G t ) is the density of the time-series network G at t time, and ρ(G t-ΔT) is the density of the time-ordered network G at time t-ΔT, the density of the network at a certain time ρ = actual existing network edges / maximum possible existing network edges;

[0067] Therefore, DGR(G t ) can capture the change of density between time windows of time-ordered network. If DGR(G t ) is positive, it indicates that the network density increases in the specified time interval; if DGR(G t ) is negative, it indicates that the density decreases.

[0068] The RDC index mainly measures the change of community centrality of community node u in a fixed time interval ΔT, which is represented by the change of community centrality between two consecutive time steps, and is:

[0069]

[0070] Wherein, π(u t ) is the centrality score of community node u at time t, and π(u t-ΔT ) is the centrality score of community node u at time t-ΔT.

[0071] DGC(u t ) is positive, indicating that the importance of community node u is increasing over time, and if DGC(u t ) is close to zero, it indicates that the importance of community node u has not changed in the set time interval. It can capture the magnitude and direction of change and observe the evolution of the community over time.

[0072] In the specific implementation process, the centrality score of the community node is calculated by combining the random walk in time sequence. The community relationship network constructed by the time-ordered network belongs to the extension of static graph, in which the edge has a time stamp recording the interaction time between the design nodes. This embodiment extends the PageRank idea to the time sequence field, integrates random walk, time information and network dynamics characteristics, and constructs the centrality score of the community node in each time slice, and the calculation process is:

[0073]

[0074] Wherein, V is the community node set of the community relationship network, Z(v,u t ) is the set of all walk paths from community node v to community node u at time t, Pr[z] is the unconditional probability of path z occurring, and α is the damping factor, which is the probability of visiting edges in the walk.

[0075] The centrality score of the community node in this embodiment combines the importance of the community node itself in the time-series network, and considers the strength of the connected other community nodes, reflecting the importance of the community node in the entire network. In addition, for the scenario of local influence propagation, the centrality score in this embodiment is based on global information of the entire community network, avoiding the case that some nodes have excessive influence in the local range, and better reflecting the importance of the nodes in the entire network.

[0076] In the specific implementation of step 3, the identification of the target propagation node set is mainly that the network with local density and certain aggregation relationship can promote the information propagation in a larger range. If these nodes have a network position that facilitates the propagation of information, then in the network with aggregation structure, the propagation distance and speed of influence are better than those of the corresponding random network. Therefore, this embodiment adopts a degree discount method to identify the target propagation node set, the main idea of which is to assign a "discount degree" to each node, considering the degree of its neighbor. The higher the degree value of the neighbor node of a node, the lower the discount degree of the node. Then the nodes are selected in descending order of the discount degree until the required number of nodes is reached. The purpose of this method is to balance between the nodes with high influence and the nodes with low centrality, and to obtain a more diversified target propagation node set. In the specific implementation process, the identification process of the target propagation node set is as follows:

[0077] Step 301, establishing a target propagation node set S * and the candidate node set S;

[0078] Step 302, let the target propagation node set S get all network nodes in the community corresponding to the key community node and add them to the candidate node set S;

[0079] Step 303, calculate the discount degree of all network nodes in the current candidate node set S, which is:

[0080] D v =1+[d v -2t v -(d v -t v )t v p+o(t v )]p

[0081] where D v is the discount degree of the network node v, d v is the node degree of the network node v, t v is the difference between the number of neighbor nodes and the active state neighbor nodes of the network node v, p is the edge activation probability, and o(t v ) is a small correction term.

[0082] Step 304, select the network node with the maximum discount degree

[0083] Step 305: Let S * =S * ∪{v *}、S=S\{v *}, determine the current target propagation node set S * Whether the number of network nodes in reaches the second threshold:

[0084] If so, output the current target propagation node set S * ;

[0085] Otherwise, the network node v * After marking as active, return to step 303.

[0086] After identifying the target propagation node set, positive information can be released through the network nodes in the target propagation node set to suppress the spread of inductive information.

[0087] It is worth noting that although this embodiment Figure 1 The steps in the diagram are shown in the order indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. In addition, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0088] Example 2

[0089] Based on the propagation suppression method based on key community identification in Example 1, this embodiment discloses a propagation suppression system based on key community identification, referring to Figure 2 The propagation suppression system based on key community identification includes a community relationship network construction unit, a key community screening unit, a propagation node identification unit, and a propagation suppression unit. Specifically:

[0090] The community relationship network construction unit is used to perform community detection on multiple time slices in the time series network and construct a community relationship network;

[0091] The key community screening unit is used to calculate the community time series scale gain index of each community node in the community relationship network, and screen key community nodes in the community relationship network based on the community time series scale gain index;

[0092] The propagation node identifying unit is configured to identify a target propagation node set in the key community node;

[0093] The propagation inhibiting unit is configured to publish positive information through the network nodes in the target propagation node set to inhibit the propagation range of the induced information.

[0094] In this embodiment, the specific working process and working principle of the community relationship network constructing unit, the key community screening unit, the propagation node identifying unit and the propagation inhibiting unit are the same as those in the method of embodiment 1, and thus the above will not be described herein. The various unit modules can be realized by software, hardware and combinations thereof in whole or in part, and the various unit modules can be embedded in or independent of the processor in the computer device in hardware form, or can be stored in the memory in the computer device in software form, so as to be called and executed by the processor to perform the operations corresponding to the various unit modules.

[0095] Embodiment 3

[0096] As shown in Figure 3 A terminal device disclosed in this embodiment includes a transmitter, a receiver, a memory and a processor. The transmitter is configured to transmit instructions and data, the receiver is configured to receive instructions and data, the memory is configured to store computer execution instructions, and the processor is configured to execute the computer execution instructions stored in the memory to implement the method in embodiment 1.

[0097] It should be noted that the above memory can be independent or integrated with the processor. When the memory is independently arranged, the terminal device further includes a bus for connecting the memory and the processor.

[0098] The above only describes the preferred embodiments of the present application, and does not limit the protection scope of the present application. Any equivalent structure transformation made under the inventive concept of the present application, or direct / indirect application in other related technical fields is included in the protection scope of the present application.

Claims

1. A method for suppressing propagation based on key community identification, characterized in that: The steps include: Step 1: Perform community detection on multiple time slices in the time series network and construct a community relationship network; Step 2: Calculate the community time series scale gain index of each community node in the community relationship network, and screen key community nodes in the community relationship network based on the community time series scale gain index; Step 3, identifying a target propagation node set within the key community node; Step 4: Release positive information through the network nodes in the target propagation node set to suppress the propagation range of the inductive information.

2. The method for suppressing propagation based on key community identification according to claim 1, characterized in that: The community detection adopts the Louvain method to form community divisions by iteratively merging network nodes to maximize modularity gain. After deleting communities with a modularity of 0 or 1 and eliminating communities with a number of network nodes less than a first threshold, the community relationship network is constructed based on the remaining communities as community nodes.

3. The method for suppressing propagation based on key community identification according to claim 1, characterized in that: In step 2, the community time series scale gain index is specifically: WGC(u t )=DGR(G t )×DGC(u t ) Among them, WGC(u t ) is the community time series scale gain index of community node u at time t, DGR(G t ) is the network density growth rate of the sequential network G at time t, DGC(v t ) is the rate of change of the community centrality of community node u at time t.

4. The method for suppressing propagation based on key community identification according to claim 3, characterized in that: The calculation process of the network density growth rate is: Among them, ρ(G t ) is the density of the sequential network G at time t, ρ(G t-ΔT ) is the density of the sequential network G at time t-ΔT.

5. The method for suppressing propagation based on key community identification according to claim 4, characterized in that: The calculation process of the community centrality change rate is: Among them, π(u t ) is the centrality score of community node u at time t, π(u t-ΔT ) is the centrality score of community node u at time t-ΔT, V is the set of community nodes in the community relationship network, α is the damping factor, Z(v,u t ) is the set of all travel paths from community node v to community node u at time t, and Pr[z] is the unconditional probability of path z occurring.

6. The method for suppressing propagation based on key community identification according to any one of claims 1 to 5, characterized in that: In step 3, the identification process of the target propagation node set is: Step 301: Establish target propagation node set S * and the candidate node set S; Step 302: Set the target propagation node set Obtain all network nodes in the community corresponding to all key community nodes and add them to the candidate node set S; Step 303: Calculate the discount degree of all network nodes in the current candidate node set S, which is: D v =1+[d v -2t v -(d v -t v )t v p+o(t v )]p Among them, D v is the discount degree of network node v, d v is the node degree of network node v, t v is the difference between the number of neighbor nodes of network node v and the number of neighbor nodes in the activated state, p is the edge activation probability, o(t v ) is an infinitesimal correction term; Step 304: Select the network node with the maximum discount Step 305: Let S * =S * ∪{v * }、S=S\{v * }, determine the current target propagation node set S * Whether the number of network nodes in reaches the second threshold: If so, output the current target propagation node set S * ; Otherwise, the network node v * After marking as active, return to step 303.

7. A propagation suppression system based on key community identification, characterized in that: Propagation suppression is performed using the method according to any one of claims 1 to 6, wherein the propagation suppression system comprises: Community relationship network construction unit, used to perform community detection on multiple time slices in the time series network and build a community relationship network; a key community screening unit, configured to calculate a community time-series scale gain index of each community node in the community relationship network, and screen key community nodes in the community relationship network based on the community time-series scale gain index; a propagation node identification unit, configured to identify a target propagation node set within the key community node; The propagation suppression unit is used to release positive information through the network nodes in the target propagation node set to suppress the propagation range of the inductive information.

8. A terminal device, characterized in that: The terminal device is provided with: Memory, used to store programs; A processor is configured to execute the program stored in the memory, wherein when the program is executed, the processor is configured to execute the method according to any one of claims 1 to 6.

Images