Harmful node identification and community division method and system for online social network

Abstract

The invention discloses an online social network harmful node identification and community division method and system. The method comprises the following steps: acquiring a social platform user speaking and interactive information data set; on the basis of the data set, a social network directed graph structure is constructed with users as nodes and interaction relations as edges; based on the social network directed graph structure, introducing evaluation indexes of edges for calculation, and determining weights of the edges; based on the directed graph structure of the social network, introducing degree centrality, betweenness centrality and a Kshell value to carry out total graph node calculation, and determining key nodes in social network propagation; and based on the weight of the edge and the key node, using a Louvain algorithm to realize community division of the optimal modularity. According to the method, the potential manipulation behaviors are identified by analyzing the graph structure of the dynamic social network, so that the identification capability of the manipulation behaviors of the social network can be improved.

Description

Technical Field

[0001] The present invention relates to the field of social network analysis and data mining, and in particular to a method and system for identifying harmful nodes and dividing communities in online social networks based on the Louvain algorithm. Background Art

[0002] At present, the detection theory of manipulation behavior in social networks mainly relies on methods such as user behavior analysis, content feature analysis and social relationship detection. The main detection methods include user profiling, content review and user association calculation (Reference 1). Specifically, (1) User profiling identifies abnormal activities by analyzing user interaction behaviors such as likes, comments and forwarding. This method usually uses machine learning algorithms to model behavior patterns and detect frequent or unnatural interaction behaviors. (2) Content review performs text analysis on published content and uses natural language processing technology to identify malicious information or fake news. This method evaluates the authenticity and tendency of content through keyword monitoring and sentiment analysis. (3) User association calculation uses the graph structure of social networks to analyze the relationship between accounts and identify key nodes and closely connected groups.

[0003] However, the above methods have obvious limitations (Reference 2). First, user behavior monitoring often only focuses on the abnormal behavior of a single account and cannot fully capture the complexity of coordinated inauthentic behavior (CIB), because these behaviors usually involve collaboration between multiple accounts. Second, content review methods may be affected by content diversity and information complexity, resulting in inaccurate detection of fake news and malicious information (Reference 3). In addition, when facing dynamic social networks, user association calculations may ignore the real-time interactive relationship between accounts, which greatly reduces the monitoring effect.

[0004] [1] Luo Xin. Computational propaganda: a new form of public opinion in the era of artificial intelligence[J]. Academic Frontiers, 2020(15):13.DOI:10.16619 / j.cnki.rmltxsqy.2020.15.003.

[0005] [2] Magelinski, T., Ng, LH, & Carley, K. (2021). A SynchronizedAction Framework for Responsible Detection of Coordination on Social Media. ArXiv, abs / 2105.07454.

[0006] [3] Luis Vargas, Patrick Emami, and Patrick Traynor. 2020. On theDetection of Disinformation Campaign Activity with Network Analysis. InProceedings of the 2020 ACM SIGSAC Conference on Cloud Computing SecurityWorkshop (CCSW'20). Association for Computing Machinery, New York, NY, USA, 133–146. https: / / doi.org / 10.1145 / 3411495.3421363. Summary of the invention

[0007] In order to overcome the deficiencies of the above-mentioned prior art, the present invention provides a method and system for identifying harmful nodes and dividing communities in online social networks, which can improve the ability to identify manipulative behaviors in social networks by analyzing the graph structure of dynamic social networks.

[0008] According to one aspect of the present invention, a method for identifying harmful nodes and dividing communities in an online social network is provided, comprising: Obtain a dataset of user speech and interaction information on social platforms; Based on the data set, a directed graph structure of a social network is constructed with users as nodes and interaction relationships as edges; Based on the directed graph structure of the social network, edge evaluation indicators are introduced for calculation to determine edge weights; Based on the directed graph structure of the social network, degree centrality, betweenness centrality and Kshell value are introduced to calculate the nodes of the whole graph and determine the key nodes in the propagation of the social network; Based on the weights of the edges and key nodes, the Louvain algorithm is used to achieve community division with optimal modularity.

[0009] As a further technical solution, the method further includes: Based on the trained SVM support vector machine model, the user groups with strong internal interaction are judged to determine whether their collaboration is driven by malicious intent.

[0010] As a further technical solution, based on the directed graph structure of the social network, edge evaluation indicators are introduced for calculation to determine edge weights, including: Calculate text similarity based on TF-IDF method; Calculate active time similarity based on time normalization mechanism; Calculate the forwarding similarity based on the Jaccard similarity coefficient; Based on the text similarity, active time similarity and forwarding similarity, the weight of the edge is obtained by weighting.

[0011] As a further technical solution, based on the directed graph structure of the social network, degree centrality, betweenness centrality and Kshell value are introduced to calculate the nodes of the entire graph and determine the key nodes in the propagation of the social network, including: Construct the adjacency matrix of the directed graph structure of the social network; Based on the adjacency matrix, respectively calculating the degree centrality, betweenness centrality and Kshell value of the nodes in the directed graph structure of the social network; Based on the calculated degree centrality, betweenness centrality and Kshell value, the importance of nodes is propagated in combination with the PageRank algorithm to determine the key nodes that promote the development of topics in social network communication.

[0012] As a further technical solution, based on the weights of the edges and key nodes, when using the Louvain algorithm to achieve the optimal modularity community division, the following modularity Q maximization solution is performed: , Where m represents the total number of edges in the graph; represents the actual edge weight between node i and node j; Represents the indicator function, when the node and nodes The value is 1 when they belong to the same community, otherwise it is 0; Representation Node The degree centrality of Representation Node The degree centrality of .

[0013] As a further technical solution, the method further includes: Put degree centrality, betweenness centrality or Kshell value in front θ% The key points are filtered into the key node set, and the remaining nodes are filtered into the non-key node set; Construct key point feature vector , non-critical node feature vector , the edge similarity feature vector , and the time-development eigenvector , the feature vector and Combine to get the final distinguishing vector for each community : , The final distinguishing vector is used to train an SVM support vector machine model to determine whether there is malicious collaboration behavior in the community.

[0014] According to one aspect of the present invention, a system for identifying harmful nodes and dividing communities in an online social network is provided, comprising: The data acquisition module is used to obtain the data set of user speech and interaction information on the social platform; A graph construction module, for constructing a directed graph structure of a social network based on the data set, with users as nodes and interaction relationships as edges; A weight determination module, used to introduce edge evaluation indicators for calculation based on the directed graph structure of the social network to determine the edge weight; A key node determination module is used to introduce degree centrality, betweenness centrality and Kshell value to calculate the nodes in the whole graph based on the directed graph structure of the social network, so as to determine the key nodes in the propagation of the social network; The community partitioning module is used to implement community partitioning with optimal modularity using the Louvain algorithm based on the weights of the edges and key nodes.

[0015] As a further technical solution, it also includes: The collaborative behavior determination module is used to determine the strong internal interactive user groups based on the trained SVM support vector machine model to determine whether their collaboration is driven by malicious intent.

[0016] According to one aspect of the present invention specification, there is provided a device for identifying harmful nodes and dividing communities in an online social network, comprising: one or more processors; a memory for storing one or more programs, wherein when one or more of the programs are executed by one or more of the processors, the one or more processors implement the steps of the method for identifying harmful nodes and dividing communities in an online social network.

[0017] According to one aspect of the present specification, a non-transitory computer-readable storage medium is provided, wherein the non-transitory computer-readable storage medium stores computer instructions, wherein the computer instructions enable the computer to execute the steps of the method for identifying harmful nodes and dividing communities in an online social network.

[0018] Compared with the prior art, the present invention has the following beneficial effects: The present invention identifies potential manipulation behaviors by analyzing the graph structure of dynamic social networks. The provided method focuses on the relationship and interaction patterns between accounts in social networks, and reveals the potential network of manipulation behaviors by identifying key nodes and community structures. The graph-based method provided by the present invention can integrate user behaviors, content features, and network relationships, providing more powerful support for identifying and responding to complex manipulation behaviors, thereby improving the ability to identify manipulation behaviors on social networks. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, a brief introduction is given below to the drawings used in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0020] Figure 1 A flowchart of a method for identifying harmful nodes and dividing communities in an online social network provided by an embodiment of the present invention.

[0021] Figure 2 A structural diagram of a system for identifying harmful nodes and dividing communities in an online social network provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0022] The invention provides a method for identifying harmful user nodes and dividing communities in an online social network: firstly, user speech information and user interaction relationships of a specific social platform are collected, and a directed graph structure of a social network is constructed with users as nodes and interaction relationships between users as edges; secondly, information such as text similarity, active time similarity and forwarding similarity is quantitatively scored as the weight value of the adjacent edges between nodes; thirdly, indicators such as node degree, Kshell and betweenness in the social network graph structure are statistically calculated, and key nodes that affect community development are calculated by PageRank algorithm; finally, the Louvain algorithm is applied to divide the graph structure into communities, and the modularity is calculated to evaluate the effect of community division, and the community division results are visualized and interactive, so as to facilitate viewing of the community composition of the social network.

[0023] The present invention constructs a social network graph with multi-dimensional similarity indicators and applies the Louvain algorithm to accurately identify key nodes and efficiently divide communities from the perspective of social network propagation structure, thereby realizing intelligent and refined detection and visual analysis of harmful nodes, and has the advantages of high efficiency, accuracy, and automation.

[0024] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in 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. In addition, the technical features in the various embodiments or single embodiments provided by the present invention are arbitrarily combined with each other to form a new technical solution. This combination is not restricted by the sequence of steps and / or the structural composition mode, but must be based on the ability of ordinary technicians in this field to achieve. When the combination of technical solutions is contradictory or cannot be achieved, it should be considered that this combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0025] See also Figure 1 The embodiment of the present invention provides a method for identifying harmful nodes and dividing communities in an online social network, including: Step 1: Obtain a dataset of user speeches and interactive information on social platforms. This step first collects user speeches and interactive information on a specific topic on a social user platform to obtain a preliminary dataset; then preprocesses the preliminary dataset using methods such as missing value filling and non-text character filtering to obtain the final dataset.

[0026] In step 1, the data structure constructed contains the following fields:

[0027] Among them, i represents the index, u represents the user ID, and t represents the name of the replying or interacting user. represents the unique ID of the replying or interacting user, s represents the name of the replied or interacted user, represents the unique ID of the user name who is replied or interacted with, r represents the name of the user who created the topic tag, The unique ID of the user who created the topic tag, p represents the time when the reply or interaction event occurred, and o represents the original text of the reply or interaction. The text representing the reply or interaction, represents the interval between the occurrence of the reply or interaction event and the topic tag creation event, and d represents the cascade depth of the reply or interaction event relative to the topic tag creation event. A unique number representing the event. Represents the forwarded link of the social user platform, Represents the cascade width of a reply or interaction event.

[0028] Step 2: Based on the data set, a directed graph structure of a social network is constructed with users as nodes and interaction relationships as edges. This step maps user interactions into a weighted directed graph structure based on the preliminary data set. The graph structure uses users as nodes and interaction relationship information as edges to construct directed propagation relationship edges.

[0029] In step 2, the structure of the constructed social network directed graph is:

[0030] Among them, V represents a set of nodes, and nodes represent users in the social network, including user identifiers , the corresponding user name E represents the edge set, and the edge represents the interactive relationship between users. The edge direction of the directed graph is from the replying or interacting user to the replied or interacted user.

[0031] Step 3, based on the directed graph structure of the social network, introduce the evaluation index of the edge for calculation and determine the weight of the edge. This step first introduces the TF-IDF method to analyze the text similarity, introduces the time normalization mechanism to analyze the active time similarity, and introduces the Jaccard similarity coefficient to analyze the forwarding similarity; then these indicators are used as the evaluation indicators of the edge, and the interactive behavior between nodes is evaluated based on specific weights, and the evaluation results are used as the total weight of the edge.

[0032] Step 3 further includes the following steps: Step 3.1, calculate TF-IDF vector: for the original text content and forwarded texts (i.e. texts that are replies or interactions) Perform word segmentation and convert it into vector representation using TF-IDF:

[0033] Calculate text similarity using cosine similarity :

[0034] Step 3.2, active time similarity analysis: Assume that edge E(i, j) represents user Original text published on and users Forwarding time , then the active time similarity The calculation method is as follows:

[0035] Among them, max(p)) and min(p) are the maximum and minimum release time in the dataset, respectively, to ensure the normalization of the time difference.

[0036] Step 3.3, forwarding similarity analysis: using Jaccard similarity coefficient calculation The collection of all forwarding users and The collection of all forwarding users The similarity between :

[0037] in, represents the number of intersections of the forwarding user sets, Indicates the union number of forwarding user sets.

[0038] Step 3.4, comprehensive weight calculation: Edge weight By weighting the three similarity values ​​[ , , ]Weighted to get:

[0039] in , and They are the preset weight coefficients for text similarity, active time similarity and forwarding similarity respectively.

[0040] Step 4: Based on the directed graph structure of the social network, degree centrality, betweenness centrality and Kshell value are introduced to calculate the nodes of the entire graph to determine the key nodes in the social network communication. This step first extracts the degree, betweenness and Kshell indicators of the nodes of the entire graph based on the above graph structure to determine the key nodes in the social network communication; then the graph structure is analyzed through the PageRank algorithm to calculate the key nodes that promote the development of the topic.

[0041] The specific implementation of step 4 includes the following sub-steps: Step 4.1, construct the adjacency matrix in the social network directed graph.

[0042]

[0043] Among them, A is an N×N matrix, which represents the connection relationship between nodes in the network, and N is the total number of nodes.

[0044] Step 4.2, calculate the degree centrality of the node based on the weighted directed graph structure: for the node , degree centrality The calculation formula is:

[0045] in, Representation Node The degree of a node is the number of nodes connected to it.

[0046] Step 4.3, calculate the betweenness centrality of the node based on the weighted directed graph structure: for the node , betweenness centrality The calculation formula is:

[0047] in, Represents a slave node To Node The total number of shortest paths, Represents a slave node To Node And through the node The total number of shortest paths.

[0048] Step 4.4, construct KShell decomposition (k-shell decomposition) and solve the KShell value of each node. Given a graph G = (V, E) we can solve the node The Kshell value is .

[0049]

[0050] By iterating the G directed graph, the following solution can be obtained: Initialize k=1; Iteratively remove nodes , for the removed node , let its Kshell value be k; Increase the k value and repeat the above steps until all nodes are removed to meet the convergence condition.

[0051] Step 4.5, propagate the importance of nodes based on the PageRank propagation formula.

[0052]

[0053] in: Representation Node The degree centrality PageRank value at the t+1th iteration, Representation Node The degree centrality PageRank value at iteration t, Indicates The number of edges starting from .

[0054] Representation Node The betweenness centrality PageRank value at iteration t+1. Representation Node The betweenness centrality PageRank value at iteration t.

[0055] Representation Node The Kshell PageRank value at the t+1th iteration, Representation Node Kshell PageRank value at iteration t.

[0056] Representation Node The initial degree centrality PageRank value. Representation Node The betweenness centrality PageRank value at the initial time. Representation Node Kshell PageRank value at the initial time.

[0057] For PageRank value, iterate continuously until convergence condition: .

[0058] Step 5: Based on the weights of the edges and key nodes, the Louvain algorithm is used to achieve the optimal modularity community division. This step is based on the above graph structure and key node analysis results, and the Louvain algorithm is used to divide the graph structure into communities. It is continuously iterated until the modularity index meets the convergence, so as to achieve the discovery of user groups with strong internal interactions.

[0059] In step 5, the community division for the optimal modularity is realized through the Louvain algorithm to achieve the maximum solution of the modularity Q:

[0060] Where m represents the total number of edges in the graph. represents the actual edge weight between node i and node j, is the indicator function, when the node and nodes The value is 1 if they belong to the same community, otherwise it is 0.

[0061] After performing community division and finding a user group with strong internal interaction, the embodiment of the present invention further includes: Step 6: Based on the trained SVM support vector machine model, the strong internal interaction user group is judged to determine whether there is any malicious promotion in their collaboration. In this step, the SVM support vector machine model is trained based on the key node indicators and some marked labels to judge the strong internal interaction user group to determine whether there is any malicious promotion in their collaboration. The labels here can be manually marked.

[0062] In step 6, when training the SVM model, the Keshell value, degree centrality or betweenness centrality can be first screened to obtain the top θ% The key node set is K, the non-key node set is N, and then perform the following sub-steps: Step 6.1: Construct key point feature vector , non-critical node feature vector , the edge similarity feature vector , the time-development feature vector .

[0063]

[0064] in, represents the average value of feature m in the key node set, represents the variance of feature m in the key node set, represents the maximum value of feature m in the key node set, Represents the minimum value of feature m in the key node set.

[0065] Step 6.2, non-critical node feature vector It also includes the statistical characteristics of Kshell value, degree centrality and betweenness centrality, which are defined as follows:

[0066] in, They represent the mean, variance, maximum and minimum values ​​of the features in the set of non-critical nodes respectively.

[0067] Step 6.3, edge similarity feature vector Contains content similarity , time similarity Similarity to forwarding The statistical characteristics of are defined as follows:

[0068] in, They represent the mean, variance, maximum and minimum values ​​of the similarity feature m in the edge set respectively.

[0069] Step 6.4: Count the number of responses and interaction events in the community every hour to get a time series ,in represents the number of interactions in the i-th hour. The coefficients are obtained by fitting a 10th-order polynomial The temporal development characteristic vector It is expressed as:

[0070] Step 6.5: transform the above feature vector and The final distinguishing vector X for each community is obtained by combining:

[0071] The above final distinguishing vector X is used to train the SVM support vector machine model to determine whether there is malicious collaboration behavior in the community.

[0072] The implementation basis of each embodiment of the present invention is to implement programmed processing through a device with a processor function. Therefore, in engineering practice, the technical solutions and functions of each embodiment of the present invention are encapsulated into various modules. Based on this reality, on the basis of the above embodiments, an embodiment of the present invention provides an online social network harmful node identification and community division system, which is used to execute an online social network harmful node identification and community division method in the above method embodiment.

[0073] See also Figure 2 The system includes: a data acquisition module, which is used to obtain a data set of user speeches and interactive information on a social platform; a graph construction module, which is used to construct a directed graph structure of a social network based on the data set, with users as nodes and interactive relationships as edges; a weight determination module, which is used to introduce edge evaluation indicators for calculation based on the directed graph structure of the social network to determine the weight of the edge; a key node determination module, which is used to introduce degree centrality, betweenness centrality and Kshell value for full-graph node calculation based on the directed graph structure of the social network to determine the key nodes in social network communication; a community division module, which is used to use the Louvain algorithm to achieve community division with optimal modularity based on the weights of the edges and key nodes.

[0074] The embodiment of the present invention provides a system for identifying harmful nodes and dividing communities in online social networks, which aims to solve the limitations of existing methods for detecting manipulation behaviors in social networks. Figure 2 Several modules in the system can identify potential manipulation behaviors by analyzing the graph structure of dynamic social networks, which can improve the ability to identify manipulation behaviors on social networks.

[0075] It should be noted that the system embodiment provided by the present invention is used to implement the methods in the above method embodiment as well as the methods in other method embodiments provided by the present invention. The only difference is that the corresponding functional modules are set. The principle is basically the same as the principle of the above system embodiment provided by the present invention. As long as the technical personnel in this field refer to the specific technical solutions in other method embodiments on the basis of the above system embodiment, obtain the corresponding technical means and the technical solutions composed of these technical means by combining the technical features, and on the premise of ensuring the practicality of the technical solutions, improve the modules in the above system embodiment to obtain the corresponding system class embodiments, which are used to implement the methods in other method class embodiments. For example: Based on the content of the above system embodiment, as a preferred embodiment, an online social network harmful node identification and community division system provided in the embodiment of the present invention further includes: The collaborative behavior determination module is used to determine the strong internal interactive user groups based on the trained SVM support vector machine model to determine whether their collaboration is driven by malicious intent.

[0076] Furthermore, the collaborative behavior determination module is used to execute the following instructions: Put degree centrality, betweenness centrality or Kshell value in front θ% The key points are filtered into the key node set, and the remaining nodes are filtered into the non-key node set; Construct key point feature vector , non-critical node feature vector , the edge similarity feature vector , and the time-development eigenvector , the feature vector and Combine to get the final distinguishing vector for each community : , The final distinguishing vector is used to train an SVM support vector machine model to determine whether there is malicious collaboration behavior in the community.

[0077] Based on the content of the above system embodiment, as a preferred embodiment, in an online social network harmful node identification and community division system provided in an embodiment of the present invention, the weight determination module is further used to execute the following instructions: Calculate text similarity based on TF-IDF method; Calculate active time similarity based on time normalization mechanism; Calculate the forwarding similarity based on the Jaccard similarity coefficient; Based on the text similarity, active time similarity and forwarding similarity, the weight of the edge is obtained by weighting.

[0078] Based on the content of the above system embodiment, as a preferred embodiment, in an online social network harmful node identification and community division system provided in an embodiment of the present invention, the key node determination module is further used to execute the following instructions: Construct the adjacency matrix of the directed graph structure of the social network; Based on the adjacency matrix, respectively calculating the degree centrality, betweenness centrality and Kshell value of the nodes in the directed graph structure of the social network; Based on the calculated degree centrality, betweenness centrality and Kshell value, the importance of nodes is propagated in combination with the PageRank algorithm to determine the key nodes that promote the development of topics in social network communication.

[0079] Based on the content of the above system embodiment, as a preferred embodiment, an online social network harmful node identification and community division system provided in the embodiment of the present invention, the community division module is also used to execute the following instructions: Based on the weights of the edges and key nodes, when the Louvain algorithm is used to achieve the optimal modularity community division, the following modularity Q maximization solution is performed: , Where m represents the total number of edges in the graph; represents the actual edge weight between node i and node j; Represents the indicator function, when the node and nodes The value is 1 when they belong to the same community, otherwise it is 0; Representation Node The degree centrality of Representation Node The degree centrality of .

[0080] Based on the same inventive concept as the above-mentioned embodiment, an embodiment of the present invention also provides a device for identifying harmful nodes and dividing communities in an online social network, including: one or more processors; a memory for storing one or more programs, when one or more of the programs are executed by one or more of the processors, the one or more processors implement the steps of the method for identifying harmful nodes and dividing communities in an online social network.

[0081] In an embodiment of the present invention, the memory may be a non-volatile memory, such as a hard disk drive (HDD) or a solid-state drive (SSD), etc., or a volatile memory (volatile memory), such as a random-access memory (RAM). The memory is any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory in an embodiment of the present invention may also be a circuit or any other device that can implement a storage function, for storing program instructions and / or data.

[0082] In the embodiments of the present invention, the processor may be a general-purpose processor, a digital signal processor, an application-specific integrated circuit, a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component, and may implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. A general-purpose processor may be a microprocessor or any conventional processor, etc. The steps of the method disclosed in the embodiments of the present invention may be directly embodied as being executed by a hardware processor, or may be executed by a combination of hardware and software modules in the processor.

[0083] Based on the same inventive concept as the above-mentioned embodiment, an embodiment of the present invention also provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions enable the computer to execute the steps of the method for identifying harmful nodes and dividing communities in an online social network.

[0084] When the above-mentioned computer instructions are implemented in the form of software functional units and sold or used as independent products, they are stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or partly in other words, the part that contributes to the prior art or part of the technical solution is embodied in the form of a software product, and the computer software product is stored in a storage medium, including several instructions for enabling a computer device (a personal computer, a server, or a network device) to perform all or part of the steps of the method described in each method embodiment of the present invention. The aforementioned storage medium includes: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media for storing program codes.

[0085] In summary, the present invention constructs a social network graph with multi-dimensional similarity indicators and applies the Louvain algorithm to accurately identify key nodes and efficiently divide communities from the perspective of social network communication structure; it realizes intelligent, refined detection and visual analysis of harmful nodes, and has the advantages of high efficiency, accuracy, and automation.

[0086] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying harmful nodes and dividing communities in online social networks, characterized in that: include: Obtain a dataset of user speech and interaction information on social platforms; Based on the data set, a directed graph structure of a social network is constructed with users as nodes and interaction relationships as edges; Based on the directed graph structure of the social network, edge evaluation indicators are introduced for calculation to determine edge weights; Based on the directed graph structure of the social network, degree centrality, betweenness centrality and Kshell value are introduced to calculate the nodes of the whole graph and determine the key nodes in the propagation of the social network; Based on the weights of the edges and key nodes, the Louvain algorithm is used to achieve community division with optimal modularity.

2. According to claim 1, a method for identifying harmful nodes and dividing communities in an online social network is characterized in that: The method further comprises: Based on the trained SVM support vector machine model, the user groups with strong internal interaction are judged to determine whether their collaboration is driven by malicious intent.

3. According to claim 1, a method for identifying harmful nodes and dividing communities in an online social network is characterized in that: Based on the directed graph structure of the social network, edge evaluation indicators are introduced for calculation to determine edge weights, including: Calculate text similarity based on TF-IDF method; Calculate active time similarity based on time normalization mechanism; Calculate the forwarding similarity based on the Jaccard similarity coefficient; Based on the text similarity, active time similarity and forwarding similarity, the weight of the edge is obtained by weighting.

4. According to claim 1, a method for identifying harmful nodes and dividing communities in an online social network is characterized in that: Based on the directed graph structure of the social network, degree centrality, betweenness centrality and Kshell value are introduced to calculate the nodes of the whole graph and determine the key nodes in the propagation of the social network, including: Construct the adjacency matrix of the directed graph structure of the social network; Based on the adjacency matrix, respectively calculating the degree centrality, betweenness centrality and Kshell value of the nodes in the directed graph structure of the social network; Based on the calculated degree centrality, betweenness centrality and Kshell value, the importance of nodes is propagated in combination with the PageRank algorithm to determine the key nodes that promote the development of topics in social network communication.

5. According to claim 1, a method for identifying harmful nodes and dividing communities in an online social network is characterized in that: Based on the weights of the edges and key nodes, when the Louvain algorithm is used to achieve the optimal modularity community division, the following modularity Q maximization solution is performed: ， Where m represents the total number of edges in the graph; represents the actual edge weight between node i and node j; Represents the indicator function, when the node and nodes The value is 1 when they belong to the same community, otherwise it is 0; Representation Node The degree centrality of Representation Node The degree centrality of .

6. According to claim 2, a method for identifying harmful nodes and dividing communities in an online social network is characterized in that: The method further comprises: Put degree centrality, betweenness centrality or Kshell value in front θ% The key points are filtered into the key node set, and the remaining nodes are filtered into the non-key node set; Construct key point feature vector , non-critical node feature vector , the edge similarity feature vector , and the time-development eigenvector , the feature vector and Combine to get the final distinguishing vector for each community : ， The final distinguishing vector is used to train an SVM support vector machine model to determine whether there is malicious collaboration behavior in the community.

7. A system for identifying harmful nodes and dividing communities in online social networks, characterized in that: include: The data acquisition module is used to obtain the data set of user speech and interaction information on the social platform; A graph construction module, for constructing a directed graph structure of a social network based on the data set, with users as nodes and interaction relationships as edges; A weight determination module, used to introduce edge evaluation indicators for calculation based on the directed graph structure of the social network to determine the edge weight; A key node determination module is used to introduce degree centrality, betweenness centrality and Kshell value to calculate the nodes in the whole graph based on the directed graph structure of the social network, so as to determine the key nodes in the propagation of the social network; The community partitioning module is used to implement community partitioning with optimal modularity using the Louvain algorithm based on the weights of the edges and key nodes.

8. The system for identifying harmful nodes and dividing communities in online social networks according to claim 7, characterized in that: Also includes: The collaborative behavior determination module is used to determine the strong internal interactive user group based on the trained SVM support vector machine model to determine whether there is any malicious promotion in their collaboration.

9. A device for identifying harmful nodes and dividing communities in an online social network, characterized in that: include: one or more processors; A memory for storing one or more programs, which, when executed by one or more processors, enables one or more processors to implement the steps of a method for identifying harmful nodes and dividing communities in an online social network as described in any one of claims 1 to 6.

10. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions enable the computer to execute the steps of the method for identifying harmful nodes and dividing communities in an online social network as described in any one of claims 1 to 6.

Images