Social network information propagation governance system based on reverse reachable weakening score
By constructing a reverse graph and using reverse breadth-first search to calculate the weakness score of nodes, key nodes are screened out and their propagation probability is reduced. This solves the problem of accuracy and effectiveness in information propagation governance in social networks and is applicable to large-scale networks.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- GUANGZHOU UNIVERSITY
- Filing Date
- 2026-05-13
- Publication Date
- 2026-07-14
AI Technical Summary
Existing methods for governing the spread of information on social networks are insufficient to accurately quantify the impact of node weakening on overall network accessibility without deleting nodes or edges, and thus cannot effectively suppress the spread of rumors and extreme viewpoints.
A system based on reverse reachability weakening score is adopted. By constructing a reverse graph and reverse breadth-first search, the weak weakening score of the nodes is calculated, key nodes are screened out and their propagation probability is reduced, and the effect of the strategy is evaluated by combining Monte Carlo propagation simulation.
It achieves accurate quantification of the impact of node attenuation on information flow without changing the network structure, and provides an effective information propagation suppression scheme applicable to large-scale networks.
Smart Images
Figure CN122390732A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of social network propagation governance, and more particularly to a social network information propagation governance system based on reverse reachability weakening scoring. Background Technology
[0002] With the widespread use of online social media platforms, information spreads rapidly among users through relationships such as following, forwarding, and commenting, forming a high-density social network structure. On the one hand, businesses and organizations can utilize this structure for brand communication and event promotion; on the other hand, rumors and extreme viewpoints also spread rapidly through social networks, putting pressure on social governance and public safety.
[0003] Existing research largely focuses on the issue of "enhancing information dissemination," such as how to select a small number of seed nodes to maximize the spread. However, for the need to "govern information dissemination at a limited cost," common practices include directly deleting nodes or severing edges. This is often difficult to implement in real-world platforms due to legal and operational constraints. A more realistic solution is to identify key nodes through algorithms and "weaken" them by reducing their forwarding probability and limiting their visibility.
[0004] Traditional centrality metrics (such as degree centrality and betweenness centrality) can only roughly reflect the importance of a node and cannot directly quantify "how much impact weakening a node on the overall network reachability." Therefore, a new node weakening scoring method is needed, tightly coupled with structural reachability and probability propagation models, to guide the selection of weakened nodes. Summary of the Invention
[0005] The purpose of this invention is to disclose a social network information dissemination governance system based on reverse reachability weakening scoring, thereby solving the technical problems mentioned in the background art.
[0006] To achieve the above objectives, the present invention provides the following technical solution: This invention provides a social network information propagation governance system based on reverse reachability weakening scoring, comprising: Network Modeling Module: Used to construct directed graphs based on input social network data. and for each directed edge Assigning basic propagation probability ; Reverse reachability analysis module: used for analysis based on the directed graph Constructing a reverse graph And for each root node in the selected set of root nodes In the reverse diagram Perform a reverse breadth-first search to generate the corresponding set of nodes that can be reached in reverse. ; Weakness scoring module: used for each root node and its inverse reachable set. Simulate deletion of a set The impact of outgoing edges from candidate nodes on reachability is calculated and accumulated for each node. Weak rating ; Weakened Node Filtering Module: Used to budget for a given number of weakened nodes. Under constraints, nodes are sorted from highest to lowest based on their Weakness scores, and the node with the highest Weakness score is selected. Each node constitutes a weakened node set. ; Probability update module: used for updating the weakened node set For each node, all outgoing edges are updated with weakened propagation probabilities based on the base propagation probability according to a preset weakening function. ; The propagation simulation and evaluation module is used to perform multiple rounds of Monte Carlo propagation simulations on the graph structure before and after the propagation probability update, estimate and compare the information flow, and evaluate the suppression effect of the weakening strategy.
[0007] Preferably, the preset weakening function is: .
[0008] Preferably, the Weakness Score Calculation Module performs the following operations: For a single root node and its reverse reachable set , sequentially for the set Each candidate node in In the reverse diagram Remove all pointers to nodes from the temporary copy. The edges, re-established with the root node Perform a reverse breadth-first search to obtain the updated reachable set. ; If a node exists Accessible before deletion After deletion Unreachable Then determine the node This violates the reachability of the current root node and weakens the node's weak score. Increment the count value by one.
[0009] Preferably, the reverse reachability analysis module selects the root node using any of the following methods: Traversing the directed graph All nodes are taken as the root node; or, From the node set A predetermined number of nodes are randomly sampled as the root node.
[0010] Preferably, the reverse reachability analysis module and the weak score calculation module support a parallel computing architecture, which can allocate the reverse breadth-first search and the corresponding weak score calculation tasks for different root nodes to different processing threads or computing nodes for simultaneous execution.
[0011] Preferably, the Monte Carlo propagation simulation includes: Based on the propagation probability of each edge in the current graph structure, independent sampling is used to determine whether the edge is activated, generating a propagation subgraph consisting only of activated edges; In the propagation subgraph, a breadth-first search or depth-first search is performed on each node to count the number of other nodes that it can reach, and the information flow value of this round of simulation is obtained by summing the reachable numbers of all nodes.
[0012] Preferably, it also includes a data interface module for connecting to the backend database of the social network platform to periodically acquire or update user relationship data and interaction behavior data, so as to enable the network modeling module to construct or update the basic propagation probability of the directed graph G and its edges.
[0013] Preferably, it also includes a strategy output and auditing module, used to output the final determined set of weakened nodes. The corresponding weakened propagation probability is recorded, along with the generation time, execution parameters, and operation logs of the weakening strategy, for traceability and auditing purposes.
[0014] Preferred, basic propagation probability Let be the probability that node u successfully influences node v in a single propagation attempt.
[0015] Preferably, after establishing a directed graph G, the nodes in the directed graph G are propagated through an independent cascade propagation model. The propagation is carried out in discrete time steps, and some nodes are activated at the initial time. In the same time step, the nodes in the activated state will initiate an independent propagation attempt along their outgoing edges to the unactivated neighbor nodes. After success, the neighbor nodes will become activated in the next time step. When no new nodes are activated in a certain time step, the propagation process terminates. Beneficial effects
[0016] (1) The indicators are strongly correlated with the weakening operation: The Weak weakening score is directly based on the definition of "weakening the change in reachability after the node leaves the edge", which can accurately reflect the criticality of the node in the propagation governance scenario.
[0017] (2) No need to delete nodes or edges: through probability transformation Implementing a softer approach better aligns with the requirements of real-world platforms for retaining users and connections.
[0018] (3) Adaptable to large-scale networks: Weak score calculation is based on reverse breadth-first search, which can be combined with root node sampling and parallelization to achieve linear time complexity.
[0019] (4) The effect can be quantified and evaluated: combined with Monte Carlo simulation, the numerical difference in information flow before and after weakening can be directly given, providing a basis for strategy selection and parameter adjustment. Attached Figure Description
[0020] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0021] Figure 1 This is a schematic diagram of the social network information propagation governance system based on reverse reachability weakening score according to the present invention.
[0022] Figure 2 This is a schematic diagram of the topological statistics of the five real social network datasets used in this invention.
[0023] Figure 3 This is a schematic diagram comparing the propagation inhibition effects of the present invention. Detailed Implementation
[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0025] refer to Figure 1 This invention provides a social network information propagation governance system based on reverse reachability weakening scoring, including... Network Modeling Module: Used to construct directed graphs based on input social network data. and for each directed edge Assigning basic propagation probability ; Reverse reachability analysis module: used for analysis based on the directed graph Constructing a reverse graph And for each root node in the selected set of root nodes In the reverse diagram Perform a reverse breadth-first search to generate the corresponding set of nodes that can be reached in reverse. ; Weakness scoring module: used for each root node and its inverse reachable set. Simulate deletion of a set The impact of outgoing edges from candidate nodes on reachability is calculated and accumulated for each node. Weak rating ; Weakened Node Filtering Module: Used to budget for a given number of weakened nodes. Under constraints, nodes are sorted from highest to lowest based on their Weakness scores, and the node with the highest Weakness score is selected. Each node constitutes a weakened node set. ; Probability update module: used for updating the weakened node set For each node, all outgoing edges are updated with weakened propagation probabilities based on the base propagation probability according to a preset weakening function. ; The propagation simulation and evaluation module is used to perform multiple rounds of Monte Carlo propagation simulations on the graph structure before and after the propagation probability update, estimate and compare the information flow, and evaluate the suppression effect of the weakening strategy.
[0026] Preferably, social networks are represented as directed graphs. ,in A set of nodes (representing social network users). The set of edges (representing social connections between users); Each directed edge Corresponding basic propagation probability This indicates that node u influences node u through this edge. The probability of.
[0027] Preferably, the k-weakening problem is defined as follows: Given an upper limit k for the number of weakened nodes, find a set W of weakened nodes such that after adjusting the propagation probability of outgoing edges from nodes in set W from p(u,v) to p′(u,v), the information flow of the social network is reduced by the greatest extent. The information flow can be measured by the expected value of the number of reachable node pairs or the number of reachable relations of all ordered node pairs after propagation terminates.
[0028] Preferably, a reverse graph is constructed based on the directed graph G. The process includes: For each edge Create an edge (v, u) in the reverse graph.
[0029] Preferably, the preset weakening function is: .
[0030] In the present invention, node weakening does not delete nodes or edges, but applies a probability transformation to all outgoing edges (v, x) of the selected node v: . Since 0 < p(v, x) ≤ 1, it can be proved that 0 < p'(v, x) ≤ p(v, x), thereby reducing the ability to propagate information through node v without changing the graph structure.
[0031] Information propagation or propagation simulation can be performed on the updated probability graph to achieve governance of the information flow volume in the social network, and the weakened node set W and corresponding weakening parameters are output for system applications.
[0032] Preferably, the Weak weakening score calculation module specifically performs the following operations: For a single root node and its reverse reachable set , for each candidate node in the set in turn, in the temporary copy of the reverse graph , delete all edges pointing to the node , and perform a reverse breadth-first search again with the root node to obtain the updated reachable set ; If there exists a node that is reachable before deletion but unreachable in after deletion , it is determined that the reachability of the node to the current root node is damaged, and the Weak weakening score of the node is increased by one count value.
[0033] Restore the deleted edges and continue to process other nodes in the set S. Repeat the above process for multiple root nodes root to obtain the Weak weakening score Weak(v) of each node.
[0034] The core idea of the Weak weakening score is to examine the impact of "deleting the outgoing edges of a node" on reachability in the reverse graph. For a given root node root, perform a breadth-first search with root as the source in the reverse graph to obtain the reachable set S; Once any node v in the reachable set S is weakened, its outgoing edges have a probability of being considered deleted, which is equivalent to deleting all edges pointing to v in the reverse graph.
[0035] After deleting an edge, perform a breadth-first search again from the root. If there is a node u in set S that is reachable from the root before deletion but not reachable from the root after deletion, then node v is considered to have a non-zero destructive effect on network reachability from the current root perspective, and its Weak score is increased by 1.
[0036] The pseudocode for the Weak rating process is as follows: Input: a probabilistic directed graph G=(V,E,p), a root node set R, and a budget k; Output: Weak(v) score for each node; 1. For each node v∈V, initialize Weak(v)=0; 2. For each root ∈ R: 2.1 Perform BFS on the reverse graph GR starting from the root to obtain the reverse reachable set S; 2.2 For each v∈S: a) Delete all edges pointed to by v in the reverse graph with probability 1-p(u,v); b) Perform reverse BFS again starting from the root to obtain a new reachable set S'; c) If there exists u∈S such that u is reachable from the root in the original BFS but not from the root in S', then let Weak(v) = Weak(v) + 1; d) Restore the deleted edges; 3. Return Weak(v) of all nodes, sort them in descending order of score, and select the top k nodes to form the weakening set W.
[0037] Preferably, the reverse reachability analysis module selects the root node using any of the following methods: Traversing the directed graph All nodes are taken as the root node; or, From the node set A predetermined number of nodes are randomly sampled as the root node.
[0038] In the node set In the graph, a root node is randomly selected or sequentially traversed. For each root, a reverse breadth-first search (BFS) is performed starting from the root in the reverse graph. The set S of all nodes that can be reached from the root is recorded. The set S is equivalent to the set of nodes in the original graph that can be reached from the root by a directed path.
[0039] Preferably, the reverse reachability analysis module and the weak score calculation module support a parallel computing architecture, which can allocate the reverse breadth-first search and the corresponding weak score calculation tasks for different root nodes to different processing threads or computing nodes for simultaneous execution.
[0040] Preferably, the Monte Carlo propagation simulation includes: Based on the propagation probability of each edge in the current graph structure, independent sampling is used to determine whether the edge is activated, generating a propagation subgraph consisting only of activated edges; In the propagation subgraph, a breadth-first search or depth-first search is performed on each node to count the number of other nodes that it can reach, and the information flow value of this round of simulation is obtained by summing the reachable numbers of all nodes.
[0041] Preferably, multiple rounds of Monte Carlo propagation simulations are performed on the graph structure before and after the propagation probability update to estimate and compare the information flow, including: Multiple independent simulations were performed on the probabilistic graph before and after weakening. In each simulation, each edge (u,v) was independently sampled for activation based on the current propagation probability, resulting in a subgraph containing only activated edges. In this subgraph, a breadth-first search or depth-first search was performed on each node to count the number of other nodes it could reach. The reachability of all nodes was then summed to obtain the information flow value for that simulation. After multiple simulations, the average information flow was taken as the information flow estimate for the corresponding graph structure. The suppression effect of the weakened node set W was evaluated by comparing the average information flow before and after weakening.
[0042] The preferred pseudocode for the Monte Carlo propagation simulation process is as follows: enter: Basic probability directed graph (Not weakened); weakened probabilistic directed graph (Selected weakening nodes have been processed) (Update edge probability); Seed node set Sc; The number of simulation wheels is T.
[0043] Output: Average propagation attenuation (The average difference in the number of propagable nodes before and after the reduction).
[0044] Algorithm steps: 1. Initialization ; 2. Repeat for t from 1 to T: 2.1 On the basic probabilistic graph G, with the seed node set Sc as the initial activation set, a random propagation simulation is performed according to the independent cascade model. Let the final set of nodes activated in this round be denoted as . ,make ; 2.2 On the weakened probabilistic graph G′, using the same seed node set Sc as the initial activation set, a random propagation simulation is performed according to the same propagation model. Let the final set of nodes activated in this round be denoted as . ,make ; 2.3 Calculate the propagation difference caused by this round of weakening: , And update the cumulative differences: = + , Calculate the average weakening effect: , Will As an evaluation value of the average suppression effect of the weakening node set on the scale of information propagation in a multi-round Monte Carlo simulation, given a seed node set Sc.
[0045] Preferably, it also includes a data interface module for connecting to the backend database of the social network platform to periodically acquire or update user relationship data and interaction behavior data, so as to enable the network modeling module to construct or update the basic propagation probability of the directed graph G and its edges.
[0046] Preferably, it also includes a strategy output and auditing module, used to output the final determined set of weakened nodes. The corresponding weakened propagation probability is recorded, along with the generation time, execution parameters, and operation logs of the weakening strategy, for traceability and auditing purposes.
[0047] Preferred, basic propagation probability Let be the probability that node u successfully influences node v in a single propagation attempt.
[0048] Preferably, after establishing a directed graph G, the nodes in the directed graph G are propagated through an independent cascade propagation model. The propagation is carried out in discrete time steps, and some nodes are activated at the initial time. In the same time step, the nodes in the activated state will initiate an independent propagation attempt along their outgoing edges to the unactivated neighbor nodes. After success, the neighbor nodes will become activated in the next time step. When no new nodes are activated in a certain time step, the propagation process terminates.
[0049] Independent Cascade (IC) Propagation Model: All nodes initially possess potential propagation capabilities (no preset propagation starting point required). Nodes are categorized into ordinary nodes (unweakened) and weakened nodes based on whether they are selected as weakening targets. The outbound propagation probability of weakened nodes is determined by the base propagation probability in subsequent steps. Adjust to weaken probability Ordinary nodes maintain the same basic propagation probability; Furthermore, propagation proceeds in discrete time steps. Initially, all nodes are in an "inactive but potentially propagating" state; at time step t, for all nodes at time t... Newly activated node u: If u is a normal node, then for each outgoing edge (u,v), attempt to activate the unactivated neighbor node v with the basic propagation probability p(u,v); if u is a weakened node, then for each outgoing edge (u,v), use the weakened propagation probability p(u,v). Attempt to activate an unactivated neighbor node v; each edge is attempted at most once during the entire propagation process. When no new node is activated at a certain time step, the propagation terminates, and the total number of activated nodes or reachable node pairs is counted as the information flow of this propagation or simulation.
[0050] Preferably, the root node selection and reverse BFS execution method are as follows: In the random sampling method, a preset number of roots are randomly selected from the node set V, and reverse BFS is performed independently for each root. In the reverse BFS process, when a node that has been marked as reachable is visited, its incoming edges are not traversed again to avoid redundant visits, thereby keeping the time complexity of a single reverse BFS within a range that is linearly related to the number of edges |E|.
[0051] Furthermore, the present invention also includes a propagation gain estimation module, which is used for: Define RRS weakening function : For a single RRS (Reverse Reachable Set) and potential weakening candidate sets If the node is weakened And after weakening With the root node By activating edge connectivity, then To weaken the root node "Reverse reachability reduces the number of nodes", otherwise ; sampling Group of independent RRS, calculating propagation attenuation estimates based on the RRS set: where Guarantee through the Chernov boundary Compared with true propagation gain The error does not exceed the preset threshold, and the confidence level meets the preset requirements; Furthermore, the present invention also includes a screening module for: Executing the RR-Boost algorithm: based on the RRS set A greedy strategy is used to iteratively select weakening nodes—initializing the set of weakening nodes. Each time from Select to enable: The node with the largest (marginal gain) join in until ; Furthermore, the present invention also includes a weakening processing module, used for: Social networks based on weakened node sets Implement weakening operations to reduce the scope of information dissemination.
[0052] The experimental procedure of this invention is as follows: System deployment requirements: Hardware environment: The server needs to be equipped with at least a multi-core CPU (no less than 8 cores) and 16GB or more of memory to support multiple rounds of reverse BFS and Monte Carlo simulation at a scale of millions of nodes; for larger-scale networks, multi-machine clusters or cloud computing platforms can be used for deployment.
[0053] Software environment: The operating system can be a Linux distribution, and the algorithm implementation can be based on languages such as C++, Java or Python, combined with distributed computing frameworks (such as Spark, Flink, etc.) to improve parallel processing capabilities.
[0054] Data Interface: The system needs to establish a stable data synchronization channel with the social platform's user relationship database and behavior log database, and support periodic incremental updates of graph structure and probability parameters.
[0055] Security and permissions: The weakening operation involves adjustments to user visibility and content dissemination permissions. The system should provide audit logs and permission control mechanisms to ensure that the policy execution process is traceable and rollbackable.
[0056] In the experimental setup, the Influence Maximization via Martingales (IMM) algorithm was first used to estimate the set of most influential nodes on a given social network, with 1% of the network size serving as the upper limit for the number of seed nodes, thus obtaining the baseline seed node set. Subsequently, the top 1% of nodes ranked by importance in the proposed Weak weakening algorithm were selected as the targets for weakening. Based on this, multiple rounds of Monte Carlo propagation simulations were performed on the network before and after weakening. By comparing the differences in information flow or final activation scale in the two cases, the inhibitory effect of the weakening strategy was quantitatively evaluated.
[0057] Experimental results are as follows Figure 2 and 3 express, Figure 2This table presents the topological statistics of five real-world social network datasets used in this experiment to verify the effectiveness of the system in networks of different sizes. The meanings of the fields in the table are as follows: Name (Dataset Name): Indicates the source of the data, namely the European Email Network (euall), the Stanford Network (stanford), the Twitter Social Graph (twitter), the Academic Collaboration Network (dblp), and the Google Network Graph (google); n (number of nodes): represents the total number of nodes in each network, i.e. the user scale in the social network, covering different scales from about 80,000 nodes to 870,000 nodes. m (number of edges): represents the total number of directed edges in each network, that is, the number of attention or interaction relationships between users, reflecting the connection density of the network.
[0058] Figure 3 This paper presents a comparison of the propagation suppression effects of the proposed Weak Score-based weakening strategy on the five different datasets mentioned above. The horizontal axis represents the names of the five social network datasets involved in the experiment; the vertical axis represents the average propagation weakening amount. (That is, the Monte Carlo simulation index defined in the aforementioned embodiments), its value is equal to "average information flow before weakening" minus "average information flow after weakening". Graphical analysis: The height of the bar chart intuitively reflects the effectiveness of the governance system. The higher the bar, the more nodes the strategy successfully prevents from being activated under the same weakening budget k, i.e., the more significant the suppression effect on information propagation. As can be seen from the figure, this system achieved significant weakening gains in large-scale dense networks such as Google and DBLP.
[0059] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.
Claims
1. A social network information dissemination governance system based on reverse reachability weakening scoring, characterized in that, include: Network Modeling Module: Used to construct directed graphs based on input social network data. and for each directed edge Assigning basic propagation probability ; Reverse reachability analysis module: used for analysis based on the directed graph Constructing a reverse graph And for each root node in the selected set of root nodes In the reverse diagram Perform a reverse breadth-first search to generate the corresponding set of nodes that can be reached in reverse. ; Weakness scoring module: used for each root node and its inverse reachable set. Simulate deletion of a set The impact of outgoing edges from candidate nodes on reachability is calculated and accumulated for each node. Weak rating ; Weakened Node Filtering Module: Used to budget for a given number of weakened nodes. Under constraints, nodes are sorted from highest to lowest based on their Weakness scores, and the node with the highest Weakness score is selected. Each node constitutes a weakened node set. ; Probability update module: used for updating the weakened node set For each node, all outgoing edges are updated with weakened propagation probabilities based on the base propagation probability according to a preset weakening function. ; The propagation simulation and evaluation module is used to perform multiple rounds of Monte Carlo propagation simulations on the graph structure before and after the propagation probability update, estimate and compare the information flow, and evaluate the suppression effect of the weakening strategy.
2. The social network information dissemination governance system based on reverse reachability weakening scoring according to claim 1, characterized in that, The preset weakening function is: .
3. The social network information dissemination governance system based on reverse reachability weakening scoring according to claim 1, characterized in that, The Weakness Score Calculation Module performs the following operations: For a single root node and its reverse reachable set , sequentially for the set Each candidate node in In the reverse diagram Remove all pointers to nodes from the temporary copy. The edges, re-established with the root node Perform a reverse breadth-first search to obtain the updated reachable set. ; If a node exists Accessible before deletion After deletion Unreachable Then determine the node This violates the reachability of the current root node and will affect the node. Weak rating Increment the count value by one.
4. The social network information dissemination governance system based on reverse reachability weakening scoring according to claim 1, characterized in that, The reverse reachability analysis module selects the root node using any of the following methods: Traversing the directed graph All nodes are taken as the root node; or, From the node set A predetermined number of nodes are randomly sampled as the root node.
5. The social network information dissemination governance system based on reverse reachability weakening scoring according to claim 1, characterized in that, The reverse reachability analysis module and the weak score calculation module support a parallel computing architecture, which can allocate the reverse breadth-first search and corresponding weak score calculation tasks for different root nodes to different processing threads or computing nodes for simultaneous execution.
6. The social network information dissemination governance system based on reverse reachability weakening scoring according to claim 1, characterized in that, The Monte Carlo propagation simulation includes: Based on the propagation probability of each edge in the current graph structure, independent sampling is used to determine whether the edge is activated, generating a propagation subgraph consisting only of activated edges; In the propagation subgraph, a breadth-first search or depth-first search is performed on each node to count the number of other nodes that it can reach, and the information flow value of this round of simulation is obtained by summing the reachable numbers of all nodes.
7. The social network information dissemination governance system based on reverse reachability weakening scoring according to claim 1, characterized in that, It also includes a data interface module for connecting to the backend database of the social network platform to periodically acquire or update user relationship data and interaction behavior data, which are then used by the network modeling module to construct or update the basic propagation probabilities of the directed graph G and its edges.
8. The social network information dissemination governance system based on reverse reachability weakening scoring according to claim 1, characterized in that, It also includes a strategy output and auditing module, used to output the final set of weakened nodes. The corresponding weakened propagation probability is recorded, along with the generation time, execution parameters, and operation logs of the weakening strategy, for traceability and auditing purposes.
9. The social network information dissemination governance system based on reverse reachability weakening scoring according to claim 1, characterized in that, Basic propagation probability Let be the probability that node u successfully influences node v in a single propagation attempt.
10. The social network information dissemination governance system based on reverse reachability weakening scoring according to claim 1, characterized in that, After establishing a directed graph G, the nodes in the directed graph G are propagated through an independent cascaded propagation model. The propagation is carried out in discrete time steps. At the initial time step, some nodes are activated. Nodes that are activated in the same time step will initiate an independent propagation attempt along their outgoing edges to their unactivated neighbor nodes. If successful, the neighbor nodes will become activated in the next time step. When no new nodes are activated in a certain time step, the propagation process terminates.