Method and system for identifying key users in a social network

Abstract

The application provides a kind of key user identification method and system in social network, it is related to user identification technical field, comprising: according to social network data, construct social network graph model;Through graph theory algorithm, calculate basic centrality index;Social network data is divided into multiple communities, and the local role of user node in multiple communities is identified;Calculate the local centrality index of local role;Global score and local score are generated;Global score and local score are ranked respectively, and key user node is screened out.The application solves the problem of insufficient identification of cross-community bridge nodes and local core users in traditional methods by fusing two-dimensional analysis of global centrality index and local role features, significantly improves the comprehensiveness and accuracy of key user identification;Through the double check of time sequence feature analysis and rule engine verification, the abnormal state of disguised as high-influence user is effectively identified, and the result reliability is significantly improved.

Description

Technical Field

[0001] This invention relates to the field of user identification technology, and in particular to a key user identification method and system in social networks. Background Technology

[0002] Key user identification in social networks is an important branch of graph theory and complex network research, its background tracing back to the exploration of the complex structural characteristics and information dissemination patterns of social networks. With the rapid development of internet technology, social networks have evolved into ultra-large-scale heterogeneous graph structures containing billions of nodes, where users form dynamic systems of information dissemination, opinion exchange, and behavioral influence through complex relationship networks. In this context, identifying key users has become a core issue in understanding network topology, optimizing information dissemination efficiency, and predicting group behavior.

[0003] Existing technologies often use fixed thresholds or simple clustering algorithms to divide communities, failing to dynamically adapt to the evolution of social network topologies. When community structures split or merge due to changes in user behavior or malicious interference, traditional methods struggle to adjust community boundaries in a timely manner, leading to an underestimation or misjudgment of the local importance of key roles. Furthermore, existing technologies often detach the connection strength between communities and the role of bridging nodes when calculating node importance, analyzing only indicators within each community independently. This makes it difficult to identify key nodes for cross-community information transmission or coordinated action.

[0004] To address the shortcomings of the existing technology, this technical solution proposes a method and system for identifying key users in social networks. Summary of the Invention

[0005] This invention provides a method and system for identifying key users in social networks, in order to overcome the deficiencies in the prior art.

[0006] On one hand, the present invention provides a method for identifying key users in a social network, comprising:

[0007] Construct a social network graph model based on social network data; social network data includes user nodes.

[0008] By combining social network graph models, graph theory algorithms are used to quantify the importance of user nodes and calculate basic centrality indicators.

[0009] Divide social network data into multiple communities and identify the local roles of user nodes within these communities;

[0010] Calculate the local centrality index of local roles within each community;

[0011] The basic centrality index and the local centrality index are weighted and summed to generate global score and local score; the global score and local score are ranked separately, and key user nodes are selected based on the score ranking, and the set of key user nodes is output.

[0012] Score mutation detection is performed on key user nodes in the key user node set to identify malicious behavior, and the final key user node set is output based on the identification results.

[0013] According to the method for identifying key users in a social network provided by the present invention, the steps of constructing a social network graph model include:

[0014] Extract user nodes and relationship edges from social network data, and create a set of user nodes and a set of relationship edges;

[0015] Construct an adjacency matrix based on the set of user nodes and the set of relation edges;

[0016] Generate a graph structure based on the adjacency matrix, the set of user nodes, and the set of edges.

[0017] According to the method for identifying key users in a social network provided by the present invention, the steps for calculating the basic centrality index include:

[0018] Based on the graph structure, configure the basic indicator algorithm library and determine the indicator weight allocation rules;

[0019] By combining a basic index algorithm library, the index values ​​of each user node are calculated by traversing the graph structure.

[0020] The index values ​​are normalized to a unified dimension, and the basic centrality index and standardized basic score are output.

[0021] According to the present invention, a method for identifying key users in a social network includes the following steps for identifying local roles:

[0022] Hierarchical clustering algorithms are used to divide social network data into multiple communities;

[0023] By combining multiple communities and user nodes, output the local user nodes within each community and the connection strength between different communities;

[0024] Analyze the node location characteristics of local user nodes within their respective communities;

[0025] Based on the connection strength, determine the bridging nodes between the various communities;

[0026] Based on node location characteristics and bridging nodes, local roles within multiple communities are determined.

[0027] According to the key user identification method in a social network provided by the present invention, the steps of generating global scores and local scores include:

[0028] Calculate the local centrality index of local roles within the independent scope of each community;

[0029] Adjust the weight of the local centrality index based on the node location characteristics;

[0030] The role type of the local role is determined based on the bridging node, and a weight coefficient is assigned according to the role type;

[0031] By combining the weighting coefficients, a weighted summation is performed to generate a local score;

[0032] Based on the standardized base score and local score, a weighted fusion function is constructed, and the weight parameters are output.

[0033] The global score is calculated based on the weight parameters, standardized base score, and data score.

[0034] According to the present invention, a method for identifying key users in a social network includes the following steps for filtering a set of key user nodes:

[0035] Based on the global score and the local score, perform a two-dimensional descending sort and output the sorting result;

[0036] Calculate the overall ranking of the local roles based on the sorting results;

[0037] The top 10% of the overall ranking are used as the pre-selected set of key user nodes. Low-scoring user nodes are filtered out, and the set of key user nodes is output.

[0038] According to the method for identifying key users in a social network provided by the present invention, the step of detecting score mutations in a set of key user nodes includes:

[0039] Based on social network data, extract the behavioral temporal features of all key user subsets in the key user node set;

[0040] By combining behavioral temporal features, a change point detection algorithm is used to identify scores for abnormal mutations;

[0041] Based on the abnormal mutation scores, the relevant key user nodes of mutation are marked and the marking information is output;

[0042] Based on the tagging information and log data from social network data, the rule engine performs verification and outputs abnormal key user nodes;

[0043] Exclude abnormal critical user nodes from the critical user node set and output the final critical user node set.

[0044] This invention also provides a key user identification system in a social network, comprising:

[0045] The graph data modeling module is used to construct social network graph models based on social network data.

[0046] Centrality calculation engine, which combines social network graph models and uses graph theory algorithms to quantify the importance of user nodes and calculate basic centrality metrics;

[0047] The community segmentation module is used to divide social network data into multiple communities;

[0048] The role recognition module is used to identify the local roles of user nodes across multiple communities;

[0049] The local indicator calculation module is used to calculate the local centrality indicator of local roles within each community;

[0050] The score calculation and fusion module is used to calculate the global score and local score based on the basic centrality index and the local centrality index; and to rank the scores, filter key user nodes, and output the set of key user nodes.

[0051] The anomaly detection module performs score mutation detection on key user nodes in the key user node set and outputs the final key user node set based on the detection results.

[0052] According to the present invention, a key user identification system in a social network includes a centrality calculation engine comprising an algorithm selection unit and an indicator calculation unit. The algorithm selection unit is used to select different quantization algorithms based on different centrality indicators by combining the social network graph model. The indicator calculation unit is used to calculate multiple centrality indicators of user nodes based on different quantization algorithms.

[0053] According to the present invention, a key user identification system in a social network includes an anomaly detection module comprising a feature extraction unit, an anomaly identification unit, an anomaly labeling unit, and an anomaly feedback unit. The feature extraction unit is used to extract the behavioral temporal features of all key user subsets in the key user node set. The anomaly identification unit is used to combine the behavioral temporal features and use a change point detection algorithm to identify anomalous mutation scores. The anomaly labeling unit is used to label relevant mutated key user nodes according to the anomalous mutation scores and output labeling information. The anomaly feedback unit is used to exclude mutated key user nodes in the key user node set according to the labeling information and output the final key user node set.

[0054] This invention provides a method and system for identifying key users in social networks. By integrating global centrality indicators and local role characteristics into a two-dimensional analysis, it retains the advantages of traditional global influence assessment while accurately capturing the structural characteristics of nodes within the community. This effectively solves the problem of insufficient identification of cross-community bridging nodes and local core users in traditional methods, significantly improving the comprehensiveness and accuracy of key user identification. By dynamically adjusting the weights of local indicators based on node location characteristics and combining them with role type allocation coefficients, the indicator system achieves scenario-adaptive optimization. This ensures the system maintains stable identification performance even under scenarios such as changes in social network structure and malicious interference. Through dual verification of temporal feature analysis and rule engine validation, it effectively identifies abnormal states disguised as high-influence users. Compared to traditional single-dimensional detection methods, the false positive rate is significantly reduced, significantly improving the reliability of the results. Attached Figure Description

[0055] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0056] Figure 1 This is a flowchart of a key user identification method in a social network provided by an embodiment of the present invention;

[0057] Figure 2 This is a flowchart of score mutation detection for a set of key user nodes provided in an embodiment of the present invention;

[0058] Figure 3 This is a schematic diagram of the structure of a key user identification system in a social network provided by an embodiment of the present invention. Detailed Implementation

[0059] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0060] Example 1:

[0061] The following is combined Figures 1-3 This invention describes a method and system for identifying key users in a social network.

[0062] like Figures 1-2As shown in the figure, an embodiment of the present invention provides a method for identifying key users in a social network, comprising:

[0063] Construct a social network graph model based on social network data. Social network data includes user nodes. The steps for constructing a social network graph model include:

[0064] Extract user nodes and relationship edges from social network data, creating a set of user nodes and a set of relationship edges. User nodes include unique identifiers and attribute fields (such as registration time and geographic location), while relationship edges include interaction type (following / being followed, liking / commenting), timestamp, and weight value. Separate individual users and their associated behaviors from the data to form user sets and relationship sets.

[0065] An adjacency matrix is ​​constructed based on the set of user nodes and the set of relation edges. A sparse matrix storage structure is used to process large-scale network data, and the weighted relation edges are normalized. The connection relationships between users are recorded in tabular form, and the matrix elements represent the interaction strength between users (such as the number of mutual friends or the number of messages sent).

[0066] A graph structure is generated based on the adjacency matrix, the set of user nodes, and the set of edges. The graph structure is then visualized, stored, and dynamically updated using a graph database (such as Neo4j) or an in-memory computing framework (such as NetworkX).

[0067] By combining social network graph models and using graph theory algorithms, the importance of user nodes is quantified, and the fundamental centrality index is calculated. The steps for calculating the fundamental centrality index include:

[0068] Based on the graph structure, a basic indicator algorithm library is configured, and the indicator weight allocation rules are determined. The basic indicator algorithm library includes core indicators such as degree centrality (number of direct connections), proximity centrality (average distance to other nodes), and betweenness centrality (path control capability). The weight allocation uses the Analytic Hierarchy Process (AHP) combined with the entropy weight method for dynamic calibration.

[0069] Degree centrality is used to measure local influence based on the number of neighbors, and it is calculated as follows:

[0070]

[0071] In the formula, C D (v) represents the degree centrality score of node v, and deg(v) represents the number of direct neighbors of node v, i.e., the number of relation edges. n is the total number of user nodes in the social network, and (n-1) is the normalization factor, indicating that a node may have at most n-1 neighbors.

[0072] Proximity centrality reflects the average proximity of node v to other user nodes, and is expressed by the formula:

[0073]

[0074] In the formula, C C (v) represents the proximity centrality score of node v, V represents the set of all user nodes in the social network, and d(v,u) represents the shortest path length from node v to node u.

[0075] Betweenness centrality measures the mediating role of node v in the shortest path among all user node pairs. A higher value indicates that node v is closer to the key hubs of the social network. The formula is expressed as:

[0076]

[0077] In the formula, C B (v) is the betweenness centrality score of node v, representing the sum of v's mediation effects on all nodes in the shortest path (the larger the value, the stronger v's mediation effect). ∑ s,v,t This represents the mediating effect of accumulating v over all possible node pairs (s, t). Iterate through all possible node triples (s, v, t), where s and t are any two distinct nodes in the network other than v (i.e., s, t ∈ V\{v}), requiring s = t (to avoid self-loop paths). σ st σ represents the total number of shortest paths from node s to t (regardless of whether the path passes through v). st (v) represents the number of paths that pass through v in the shortest path from node s to t.

[0078] This typically also includes eigenvector centrality, used to measure the global influence of user nodes, which takes into account the network topology, and is expressed by the formula:

[0079]

[0080] The above formula is solved using the power iteration method, where A represents the adjacency matrix. If A ij =1 indicates that user nodes i and j are connected. x represents the eigenvector centrality score vector, where each element corresponds to a user node's score. λ is the maximum value of the centrality score for the corresponding eigenvector. The score xv of node v is proportional to the sum of the scores of its neighboring nodes; that is, nodes connected to high-influence neighbors have higher scores.

[0081] By combining a basic index algorithm library, the index values ​​of each user node are calculated by traversing the graph structure. A breadth-first search (BFS) algorithm is used to optimize traversal efficiency, and a distributed computing framework is enabled for ultra-large-scale networks (number of nodes > 1M).

[0082] The index values ​​are normalized to a unified dimension, and the basic centrality index and standardized basic score are output. The Min-Max normalization method is used, and a dynamic correction coefficient is set to deal with outlier interference.

[0083] Dividing social network data into multiple communities and identifying the local roles of user nodes within these communities involves the following steps:

[0084] Hierarchical clustering algorithms are used to divide social network data into multiple communities. Hierarchical clustering is an unsupervised learning method based on the progressive merging or splitting of clusters according to inter-cluster similarity, achieving data grouping by constructing a tree-like hierarchical structure. The quality of the partitioning is evaluated using a modularity optimization function, while a resolution parameter controls the granularity of the communities. Modularity is a metric for measuring the quality of community partitioning, reflecting the difference between the internal connection density of a community and the expected value of a random network. The resolution parameter adjusts the coarseness of the community partitioning; higher values ​​tend to generate more small communities.

[0085] By combining multiple communities and user nodes, the system outputs the local user nodes within each community and the connection strength between communities. The Jaccard similarity of inter-community bridging edges is calculated as a connection strength indicator. Jaccard similarity measures similarity by calculating the ratio of the intersection to the union of two sets; here, it is used to quantify the degree of association between bridging edges in the communities.

[0086] This study analyzes the location characteristics of local user nodes within their respective communities. These characteristics include topological properties such as degree centrality and betweenness centrality, reflecting their influence and mediating role within the community.

[0087] Based on connectivity strength, bridging nodes are determined between communities. Bridging nodes are nodes that connect different communities; their removal will lead to community isolation.

[0088] Based on node location characteristics and bridging nodes, local roles within multiple communities are determined. A role classification model (such as decision tree + cluster fusion) is established to divide role types into core hub nodes, boundary propagation nodes, and isolated response nodes.

[0089] Calculate the local centrality metric for each local role within each community. Design differentiated metrics for different role types: calculate flow betweenness centrality for hub nodes and semi-local centrality for boundary nodes. Flow betweenness centrality measures a node's control over network traffic; semi-local centrality for boundary nodes focuses on the node's connectivity strength within neighboring communities.

[0090] The basic centrality metric and the local centrality metric are weighted and summed to generate a global score and a local score. The global score and the local score are then ranked, and key user nodes are selected based on the ranking, outputting a set of key user nodes. The steps for generating the global score and the local score include:

[0091] Calculate the local centrality index of local roles within the independent scope of each community. Establish a local centrality index system, including quantifying the dispersion of the tight centrality of nodes within the community, using the percentile of betweenness centrality within the community to measure the node's control ability, and evaluating the information diffusion efficiency based on the average path length of the community network.

[0092] The weights of local centrality indicators are adjusted based on node location characteristics. The contribution of these indicators is dynamically adjusted by analyzing the depth and breadth of nodes within the hierarchical structure. Hierarchical weight generation rules are defined, including: constructing a depth-breadth adjustment matrix where depth weight decreases exponentially with increasing hierarchy level, and breadth weight increases linearly with expanding coverage. Example adjustment rule: for every increase in depth, the weight coefficient decreases by 15%; for every doubling of breadth, the weight coefficient increases by 10%.

[0093] The role type of local roles is determined based on bridging nodes, and weight coefficients are assigned according to role type. Role types include core hub nodes, edge coordinators, and structural mediator nodes. Core hub nodes are those with the highest depth weight and centrality index in the community. Edge coordinators are those with bridging index exceeding the threshold and located on the edge of the community. Structural mediator nodes are those with connectivity across more than 3 communities and abnormal HDA values.

[0094] By combining weighting coefficients, a weighted summation is performed to generate local scores. Min-Max normalization is used to eliminate dimensional differences in the local scores, and Z-Score standardization is applied to outlier-sensitive indicators. Based on the standardized base score and local scores, a weighted fusion function is constructed to output the weight parameters. The weighted fusion function generates a comprehensive evaluation value by linearly combining and normalizing multi-dimensional scores. Global score = α × standardized base score + β × local score. The optimal α / β ratio is solved using the NSGA-II multi-objective optimization algorithm, and then the weight fluctuation range is constrained by a regularization term (0.4 ≤ α ≤ 0.6, 0.4 ≤ β ≤ 0.6).

[0095] Finally, the global score is calculated based on the weight parameters, the standardized base score, and the data score.

[0096] The steps for filtering the set of key user nodes include:

[0097] Based on the global score and local score, perform a two-dimensional descending sort and output the ranking results. The global score reflects a node's strategic position within the overall network topology, while the local score reflects the node's connectivity within a specific community. The ranking results are output in a structured data table format, including node ID, global score, local score, and community identifier.

[0098] Based on the ranking results, the overall ranking of local roles is calculated. An Analytic Hierarchy Process (AHP) decision model is constructed, with a global weight coefficient of 0.6 and a local weight coefficient of 0.4, to perform multi-attribute decision calculations on the ranking results. The entropy method is used to eliminate the influence of dimensions, generating a comprehensive ranking table containing confidence intervals.

[0099] The top 10% of the overall ranking is used as a pre-selection set of key user nodes. Low-scoring user nodes are filtered out, resulting in the output set of key user nodes. Based on the Pareto optimality principle, the top 10% of nodes in the overall ranking are selected as the pre-selection set, with a dynamic threshold filtering mechanism: when the network size is <100,000, the top 10% is retained; when the network size is 100,000 ≤ network size ≤ 500,000, Top 10% + standard deviation filtering is used; and when the network size is >500,000, centrality cross-validation is added. The exclusion criteria include: local scores below the community average, activity decay exceeding 35% in the past 30 days, and zombie nodes with isolated edge connections.

[0100] Score mutation detection is performed on key user nodes in the key user node set to identify malicious behavior, and the final key user node set is output based on the identification results. The steps for score mutation detection on the key user node set include:

[0101] Based on social network data, behavioral temporal features of all key user subsets within the key user node set are extracted. Four-dimensional temporal features are extracted from the social network data: behavioral pattern features (time-series autocorrelation of posting / commenting / liking); social relationship features (Hurst exponent of friend growth rate); content features (text sentiment polarity volatility); and device features (entropy value of IP address change frequency). The TSFEL toolkit is used for feature extraction, generating a rolling dataset with a 72-hour time window.

[0102] Combining behavioral temporal features, a change point detection algorithm is used to identify anomalous mutation scores. A hybrid model of Facebook Prophet and LSTM is deployed for change point detection, and Bayesian change point analysis with saliency levels is performed. For the detected anomalous mutations, the Isolation Forest algorithm is used for secondary validation, and the anomalous score S = 1 / (1 + e^(-x)) is calculated, where x is the model output value.

[0103] Based on the abnormal mutation scores, relevant key user nodes are marked, and the marking information is output. Based on the marking information and log data from the social network data, a rule engine is executed for verification, and abnormal key user nodes are output. Abnormal key user nodes in the key user node set are excluded, and the final key user node set is output.

[0104] like Figure 3 As shown, the present invention also provides a key user identification system in a social network, comprising:

[0105] The graph data modeling module is used to construct social network graph models based on social network data. It employs a heterogeneous graph construction framework, integrating user attribute data, social relationship data, and time series data. A multi-dimensional relationship encoding mechanism is implemented, supporting unified modeling of directed weighted graphs and multi-relationship networks through a hybrid storage structure of adjacency and association matrices.

[0106] The centrality calculation engine combines social network graph models with graph theory algorithms to quantify the importance of user nodes and calculate basic centrality metrics. Basic centrality metrics typically include traditional metrics, higher-order metrics, and dynamic metrics. Traditional metrics include degree centrality, proximity centrality, and betweenness centrality. Higher-order metrics include eigenvector centrality, PageRank, and Katz centrality. Dynamic metrics include time decay centrality and propagation influence centrality.

[0107] The community segmentation module is used to divide social network data into multiple communities. Specifically, it first performs fast coarse-grained segmentation based on the Louvain algorithm, then uses a label propagation algorithm for fine-grained optimization, and finally introduces a module density optimization criterion to correct the segmentation results.

[0108] The role recognition module is used to identify the local roles of user nodes within multiple communities. Role types include core hub nodes, edge coordinators, and structural mediator nodes.

[0109] The local indicator calculation module is used to calculate the local centrality indicator of local roles within each community.

[0110] The score calculation and fusion module is used to calculate the global score and local score based on the basic centrality index and the local centrality index. It also ranks the scores, filters key user nodes, and outputs a set of key user nodes.

[0111] The anomaly detection module performs score mutation detection on key user nodes in the key user node set and outputs the final key user node set based on the detection results.

[0112] According to the present invention, a key user identification system in a social network includes a centrality calculation engine comprising an algorithm selection unit and an indicator calculation unit. The algorithm selection unit is used to select different quantization algorithms based on different centrality indicators by combining the social network graph model. The indicator calculation unit is used to calculate multiple centrality indicators of user nodes based on different quantization algorithms.

[0113] In summary, this invention provides a key user identification method and system in social networks. By integrating global centrality indicators and local role characteristics into a two-dimensional analysis, it retains the advantages of traditional global influence assessment while accurately capturing the structural characteristics of nodes within the community. This effectively solves the problem of insufficient identification of cross-community bridging nodes and local core users in traditional methods, significantly improving the comprehensiveness and accuracy of key user identification. By dynamically adjusting the weights of local indicators based on node location features and combining role type allocation coefficients, the indicator system achieves scenario-adaptive optimization. This ensures stable identification performance even under scenarios such as changes in social network structure and malicious interference. Through dual verification of temporal feature analysis and rule engine validation, it effectively identifies abnormal states disguised as high-influence users. Compared to traditional single-dimensional detection methods, the false positive rate is significantly reduced, significantly improving the reliability of the results. Adopting a three-level evaluation framework of role-community-global, through role-aware community division, dynamic weight fusion mechanism, and adversarial anomaly detection, it addresses the technical shortcomings of traditional methods in cross-community influence assessment, dynamic role evolution capture, and fake node identification.

[0114] Example 1:

[0115] The graph structure is a directed weighted graph A→B(0.5), B→C(0.8), A→C(0.3), C→D(0.7).

[0116] The shortest paths and traffic distribution between nodes are shown in Table 1.

[0117] Table 1:

[0118]

[0119] The calculation steps are as follows:

[0120] 1. Enumerate the shortest path for all node pairs.

[0121] From A to D: Two paths (A→B→C→D and A→C→D);

[0122] From B to D: One path (B→C→D);

[0123] A to C: A path (A→C);

[0124] From C to D: a path (C→D).

[0125] 2. Traffic is allocated according to the proportion of the total path weight.

[0126] The total weight from A to D is 2.0 + 1.0 = 3.0, of which the traffic passing through C accounts for (2.0 / 3.0) + (1.0 / 3.0) = 1.0 (all traffic passes through C).

[0127] 3. Accumulate the traffic contribution of node C.

[0128] The A→D path contributes 1.0; the B→D path contributes 1.0 (all traffic passes through C); the total flow betweenness score is 1.0 + 1.0 = 2.0 (normalized to 0.5).

[0129] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0130] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying key users in a social network, characterized in that, include: Construct a social network graph model based on social network data; The social network data includes user nodes; Combining the aforementioned social network graph model, the importance of user nodes is quantified using graph theory algorithms, and the basic centrality index is calculated. The social network data is divided into multiple communities, and the local roles of user nodes within these communities are identified. This includes using a hierarchical clustering algorithm to divide the social network data into multiple communities; combining the multiple communities and the user nodes to output the connection strength between local user nodes within each community and between each community; and analyzing the node location characteristics of the local user nodes within their respective communities. Based on the connection strength, the bridging nodes between the various communities are determined; Based on the node location characteristics and the bridging node, multiple local roles within the community are determined; Calculate the local centrality index of the local role within each community; The process involves weighted summation of the basic centrality index and the local centrality index to generate a global score and a local score. This includes calculating the local centrality index of the local roles within each community's independent scope; adjusting the weights of the local centrality indexes based on the node's location characteristics; determining the role type of the local role based on the bridging node and assigning weight coefficients accordingly; combining the weight coefficients to generate the local score through weighted summation; constructing a weighted fusion function based on the standardized basic score and the local score to output weight parameters; calculating the global score based on the weight parameters, the standardized basic score, and the local score; ranking the global score and the local score respectively; and filtering key user nodes based on the score ranking to output a set of key user nodes. Score mutation detection is performed on the key user nodes in the set of key user nodes to identify malicious behavior, and the final set of key user nodes is output based on the identification results.

2. The method for identifying key users in a social network according to claim 1, characterized in that, The steps for constructing the social network graph model include: Extract user nodes and relationship edges from the social network data, and create a set of user nodes and a set of relationship edges; Construct an adjacency matrix based on the user node set and the relation edge set; A graph structure is generated based on the adjacency matrix, the set of user nodes, and the set of edges.

3. The method for identifying key users in a social network according to claim 2, characterized in that, The steps for calculating the fundamental centrality index include: Based on the graph structure, configure the basic indicator algorithm library and determine the indicator weight allocation rules; By combining the aforementioned basic indicator algorithm library and traversing the graph structure, the indicator values ​​of each user node are calculated. The index values ​​are normalized to a unified dimension, and the basic centrality index and standardized basic score are output.

4. The method for identifying key users in a social network according to claim 1, characterized in that, The steps for filtering the set of key user nodes include: Based on the global score and local score, perform a two-dimensional descending sort and output the sorting result; Based on the sorting results, calculate the overall ranking of the local roles; The top 10% of the overall ranking are used as the pre-selected set of key user nodes. Low-scoring user nodes are filtered out, and the set of key user nodes is output.

5. The method for identifying key users in a social network according to claim 1, characterized in that, The steps for detecting score mutations in the key user node set include: Based on the social network data, extract the behavioral temporal features of all key user subsets in the key user node set; Combining the aforementioned behavioral temporal features, a change point detection algorithm is used to identify abnormal mutation scores; Based on the abnormal mutation scores, the relevant key user nodes of mutations are marked and the marking information is output. Based on the tagging information and the log data in the social network data, the rule engine is executed to verify and output the abnormal key user nodes; Exclude the abnormal key user nodes from the set of key user nodes, and output the final set of key user nodes.

6. A key user identification system in a social network, comprising employing a key user identification method in a social network as described in any one of claims 1 to 5, characterized in that, include: The graph data modeling module is used to construct social network graph models based on social network data. A centrality calculation engine is used to combine the aforementioned social network graph model, quantify the importance of user nodes through graph theory algorithms, and calculate basic centrality metrics. The community segmentation module is used to divide the social network data into multiple communities; The role recognition module is used to identify the local roles of the user node within multiple communities; The local indicator calculation module is used to calculate the local centrality indicator of each local role within the community. The score calculation and fusion module is used to calculate the global score and the local score based on the basic centrality index and the local centrality index. The system then ranks users by scores, filters key user nodes, and outputs a set of key user nodes. The anomaly detection module performs score mutation detection on the key user nodes in the key user node set, and outputs the final key user node set based on the detection results.

7. A key user identification system in a social network according to claim 6, characterized in that, The centrality calculation engine includes an algorithm selection unit and an indicator calculation unit; the algorithm selection unit is used to select different quantization algorithms based on different centrality indicators in conjunction with the social network graph model; the indicator calculation unit is used to calculate multiple centrality indicators of the user node based on different quantization algorithms.

8. A key user identification system in a social network according to claim 6, characterized in that, The anomaly detection module includes a feature extraction unit, an anomaly identification unit, an anomaly labeling unit, and an anomaly feedback unit. The feature extraction unit is used to extract the behavioral temporal features of all key user subsets in the key user node set. The anomaly identification unit is used to combine the behavioral temporal features and use a change point detection algorithm to identify anomalous mutation scores. The anomaly labeling unit is used to label relevant mutated key user nodes according to the anomalous mutation scores and output labeling information. The anomaly feedback unit is used to exclude mutated key user nodes in the key user node set according to the labeling information and output the final key user node set.

Images