Public safety multi-source risk factor association identification analysis method based on knowledge graph

Abstract

The invention provides a knowledge graph-based public security multi-source risk factor association identification analysis method, which relates to the technical field of risk identification, and comprises the steps of obtaining multi-source risk factor data, extracting information from unstructured data, constructing an initial association network, performing feature analysis and calculating a similarity matrix; and the close association subgroups are identified through community discovery, a multi-level association network is constructed, a conduction path is analyzed, a weight is calculated, and finally risk early warning information is generated. According to the invention, the complex association between public security risk factors can be effectively identified, and the risk prediction accuracy is improved.

Description

Technical Field

[0001] The present invention relates to risk identification technology, and in particular to a knowledge graph-based method for identifying and analyzing the association of multi-source risk factors in public safety. Background Art

[0002] Public safety is a crucial foundation for national security and social stability. With socioeconomic development, the risks facing public safety are becoming increasingly complex and dynamic. Public safety risk factors span multiple domains, including natural disasters, accidents and disasters, public health incidents, and social security. These risk factors are complexly interconnected and impactful. Traditional public safety risk analysis methods primarily rely on data analysis within a single domain, lacking comprehensive analysis of the correlations between multi-source risk factors. In recent years, the development of knowledge graph technology has provided new technical means for integrating and analyzing heterogeneous multi-source data, offering new research approaches for public safety risk analysis.

[0003] Existing public safety risk analysis methods find it difficult to effectively integrate and process structured and unstructured multi-source heterogeneous data, resulting in the inability to fully capture the characteristics and correlation information of various risk factors, affecting the comprehensiveness and accuracy of risk analysis.

[0004] Traditional risk analysis methods often adopt a static correlation analysis framework, which lacks the description of the dynamic transmission mechanism between risk factors, cannot accurately reflect the evolution and diffusion of public safety risks, and limits the ability to identify and warn of potential risks early.

[0005] In the process of constructing risk factor association networks, existing technologies lack effective community discovery and multi-level network representation methods, making it difficult to reveal the implicit association patterns and complex hierarchical structures between risk factors, and thus unable to provide accurate support and guidance for risk management and decision-making. Summary of the Invention

[0006] The embodiments of the present invention provide a method for identifying and analyzing the association of multi-source risk factors of public safety based on a knowledge graph, which can solve the problems in the prior art.

[0007] A first aspect of an embodiment of the present invention provides a method for identifying and analyzing associations of multi-source risk factors of public safety based on a knowledge graph, including: Acquiring public safety multi-source risk factor data, the public safety multi-source risk factor data including structured public safety risk factor data and unstructured public safety risk factor data; processing the unstructured public safety risk factor data to extract public safety risk factor information; integrating the structured public safety risk factor data and the processed unstructured public safety risk factor data to construct an initial public safety risk factor association network; Performing feature analysis on the public safety risk factor nodes in the initial public safety risk factor association network to obtain feature vectors of the public safety risk factor nodes; and calculating a similarity matrix based on the feature vectors. Performing community discovery on the similarity matrix to identify subgroups of public safety risk factors with closely associated relationships; and constructing a multi-level public safety risk factor association network based on the subgroups of public safety risk factors; Conducting path analysis on the multi-level public safety risk factor association network, calculating the conduction weights between public safety risk factor nodes, and generating public safety risk warning information based on the conduction weights and the analysis results of the multi-level public safety risk factor association network.

[0008] Integrating the structured public safety risk factor data and the processed unstructured public safety risk factor data to construct an initial public safety risk factor association network includes: Processing the unstructured public safety risk factor data, extracting security management unit entities, public safety attributes, and public safety relationship information from the unstructured public safety risk factor data, and converting the security management unit entities, the public safety attributes, and the public safety relationship information into a data structure that matches the structured public safety risk factor data according to a preset mapping rule; The structured public safety risk factor data and the converted unstructured public safety risk factor data are integrated according to a preset public safety system to construct an initial public safety risk factor association network.

[0009] Performing feature analysis on the public safety risk factor nodes in the initial public safety risk factor association network to obtain feature vectors of the public safety risk factor nodes; and calculating a similarity matrix between the public safety risk factor nodes based on the feature vectors includes: Performing feature analysis on public safety risk factor nodes in the initial public safety risk factor association network, extracting a security management unit static attribute feature vector, a public safety time series feature vector, and a security network topology feature vector of the public safety risk factor node, and combining the security management unit static attribute feature vector, the public safety time series feature vector, and the security network topology feature vector to construct a multi-dimensional feature vector of the public safety risk factor node; Based on the multidimensional feature vector, a sliding window adaptive clustering method is used to calculate the basic similarity value between the public safety risk factor nodes. The sliding window adaptive clustering method includes constructing a feature window of a preset size, dynamically grouping the features of the public safety risk factor nodes within the feature window, and calculating the similarity between the public safety risk factor nodes based on the dynamic grouping; based on the public safety time series feature vector, the sliding window adaptive clustering method is used to calculate the time series similarity value between the public safety risk factor nodes; based on the security network topology feature vector, the sliding window adaptive clustering method is used to calculate the topological similarity value between the public safety risk factor nodes; The basic similarity value, the temporal similarity value and the topological similarity value are used to optimize and reconstruct the initial public safety risk factor association network to obtain a similarity matrix between public safety risk factor nodes.

[0010] Based on the public safety time series feature vector, the sliding window adaptive clustering is used to calculate the time series similarity value between the public safety risk factor nodes; based on the safety network topology feature vector, the sliding window adaptive clustering is used to calculate the topology similarity value between the public safety risk factor nodes, including: Constructing a sliding window of a preset size, wherein the sliding window is used to extract the feature vectors of the public safety risk factor nodes in segments; Based on the sliding window, the temporal feature vectors of the public safety risk factor nodes are segmented and extracted to obtain public safety temporal local features, and the temporal similarity values ​​between the public safety risk factor nodes are calculated using adaptive clustering; based on the sliding window, the topological feature vectors of the public safety risk factor nodes are segmented and extracted to obtain security network topology local features, and the topological similarity values ​​between the public safety risk factor nodes are calculated using the adaptive clustering.

[0011] Performing community discovery on the similarity matrix to identify subgroups of public safety risk factors with close associations; and constructing a multi-level public safety risk factor association network based on the subgroups of public safety risk factors includes: Calculating, for each security management node in the similarity matrix, a community affiliation strength based on edge weights between adjacent security management nodes, iteratively calculating and continuously updating the community labels of the security management nodes based on the community affiliation strength until the distribution of the community labels tends to stabilize and converge, thereby obtaining a community division result including multiple initial security management communities; Calculating the cohesion of each security management community in the community division result. The cohesion is quantitatively calculated based on the edge weight relationship between security management nodes within the community. Security management communities with a cohesion greater than a preset cohesion threshold are marked as stable communities. Security management communities with a cohesion less than the preset cohesion threshold are recursively decomposed into multiple security management sub-communities with higher cohesion. The association strength of the security management sub-communities is calculated based on the distribution characteristics of the internal nodes of the security management sub-communities and the connection relationship between the sub-communities. Based on the stable community, the security management sub-community and the association strength, the inter-layer edge weights are dynamically updated and optimized to obtain a multi-level public security risk factor association network with a complete hierarchical structure.

[0012] Conducting a transmission path analysis on the multi-level public safety risk factor association network, calculating the transmission weights between public safety risk factor nodes, and generating public safety risk warning information based on the transmission weights and the analysis results of the multi-level public safety risk factor association network include: Obtaining a conduction path between pairs of public safety risk factor nodes in the multi-level public safety risk factor association network, obtaining historical conduction data of the public safety risk factor nodes at different times based on the conduction path to obtain a conduction feature vector, and calculating a temporal dependency relationship between the pairs of public safety risk factor nodes based on the conduction feature vector; Based on the temporal dependency, a temporal attenuation factor is calculated in combination with the time difference between the current moment and the historical moment, and the temporal dependency and the temporal attenuation factor are multiplied to obtain an initial transmission weight between the pairs of public safety risk factor nodes; the initial transmission weight is adjusted according to the connection relationship of the public safety risk factor nodes in the transmission path to obtain a dynamic transmission weight; The dynamic conduction weight is combined with the risk degree of the public safety risk element node along the conduction path to obtain the public safety risk conduction intensity, and the public safety risk conduction intensity is accumulated to obtain a public safety risk warning index. When the public safety risk warning index exceeds a preset warning threshold, public safety risk warning information is generated.

[0013] The public safety risk transmission intensity is obtained by combining the dynamic transmission weight with the risk level of the public safety risk factor node along the transmission path, and the public safety risk early warning index is obtained by accumulating the public safety risk transmission intensity, including: Based on the dynamic transmission weight and risk level of the public safety risk factor node, the PageRank algorithm is used to calculate the centrality score of the public safety risk factor node, and the centrality score is combined with the dynamic transmission weight to obtain the carrying capacity of the public safety risk factor node; the structural importance of the public safety risk factor node is calculated based on the PageRank algorithm, and the structural importance is combined with the carrying capacity to obtain the failure threshold of the public safety risk factor node; Combining the dynamic transmission weight with the risk level of the public safety risk factor node along the transmission path to obtain a public safety risk transmission intensity; determining whether the public safety risk factor node has reached the failure threshold based on the public safety risk transmission intensity, and marking the public safety risk factor node that has reached the failure threshold as a trigger node; The PageRank algorithm is used to calculate the degree of influence of the trigger node on the adjacent public safety risk factor nodes on the transmission path, the propagation impact is determined based on the degree of influence, and the propagation impact is accumulated to obtain the public safety risk cumulative effect. The public safety risk warning index is calculated based on the public safety risk cumulative effect and the public safety risk transmission intensity.

[0014] According to a second aspect of an embodiment of the present invention, an electronic device is provided, including: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0015] According to a third aspect of an embodiment of the present invention, a computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the method described above is implemented.

[0016] The beneficial effects of this application are as follows: The present invention integrates structured and unstructured public safety risk factor data to construct a multi-level public safety risk factor association network, which can comprehensively capture the complex associations between risk factors from different sources and improve the comprehensiveness and accuracy of risk identification.

[0017] The present invention adopts feature analysis and community discovery technology, which can effectively identify subgroups of risk factors with close correlations, reveal the implicit relationships and clustering characteristics between risk factors, and enhance the ability to identify potential security risks and the accuracy of early warning.

[0018] The present invention calculates the conduction weights between risk factor nodes through conduction path analysis, realizes dynamic propagation simulation and prediction of risks, makes public safety risk warning more forward-looking and operational, and provides a scientific basis for government departments to formulate risk prevention and control measures. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a flow chart of a method for identifying and analyzing the association of multiple risk factors of public safety based on a knowledge graph according to an embodiment of the present invention; Figure 2 Schematic diagram showing a comparison of feature vector similarity calculation performance in an embodiment of the present invention; Figure 3 This is a bar chart comparing network performance associated with multi-level public security risk factors according to an embodiment of the present invention; Figure 4 A flowchart of generating risk warning information based on a multi-level public safety risk factor association network according to an embodiment of the present invention; Figure 5 This is a bar chart comparing and analyzing the performance of the public safety risk transmission model according to an embodiment of the present invention. DETAILED DESCRIPTION

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0021] The technical solution of the present invention is described in detail below with reference to specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0022] Figure 1 This is a flow chart of a method for identifying and analyzing the association of multiple risk factors of public safety based on a knowledge graph according to an embodiment of the present invention. Figure 1 As shown, the method includes: Acquiring public safety multi-source risk factor data, the public safety multi-source risk factor data including structured public safety risk factor data and unstructured public safety risk factor data; processing the unstructured public safety risk factor data to extract public safety risk factor information; integrating the structured public safety risk factor data and the processed unstructured public safety risk factor data to construct an initial public safety risk factor association network; Performing feature analysis on the public safety risk factor nodes in the initial public safety risk factor association network to obtain feature vectors of the public safety risk factor nodes; and calculating a similarity matrix based on the feature vectors. Performing community discovery on the similarity matrix to identify subgroups of public safety risk factors with closely associated relationships; and constructing a multi-level public safety risk factor association network based on the subgroups of public safety risk factors; Conducting path analysis on the multi-level public safety risk factor association network, calculating the conduction weights between public safety risk factor nodes, and generating public safety risk warning information based on the conduction weights and the analysis results of the multi-level public safety risk factor association network.

[0023] In an optional embodiment, integrating the structured public safety risk factor data and the processed unstructured public safety risk factor data to construct an initial public safety risk factor association network includes: Processing the unstructured public safety risk factor data, extracting security management unit entities, public safety attributes, and public safety relationship information from the unstructured public safety risk factor data, and converting the security management unit entities, the public safety attributes, and the public safety relationship information into a data structure that matches the structured public safety risk factor data according to a preset mapping rule; The structured public safety risk factor data and the converted unstructured public safety risk factor data are integrated according to a preset public safety system to construct an initial public safety risk factor association network.

[0024] Acquire both structured and unstructured data on public safety risk factors. Structured data comes from government statistical databases and structured records in enterprise safety management systems. This data is typically stored in tables and databases, with well-defined fields and formats. For example, a chemical company's safety inspection record database contains fields such as inspection time, inspector, inspection item, risk level, and corrective measures. Unstructured data, on the other hand, comes from text materials such as safety incident reports, news reports, expert opinions, and meeting minutes. These data do not have predefined data models or organizational structures.

[0025] Named entity recognition technology is used to identify safety management unit entities from unstructured text. For example, in the text "A leakage accident occurred in production workshop A due to aging equipment," the system can identify "A production workshop" as the safety management unit entity. The system maintains a dictionary of entity types, including enterprise, department, workshop, and team. By matching words in the text with entity types in the dictionary and combining contextual analysis, the system can identify the safety management unit entity in the text.

[0026] Using keyword extraction and semantic analysis techniques, public safety attribute information is extracted from unstructured text. For example, from the text "The operating temperature of this equipment is too high, the pressure value has exceeded the safety threshold, and there is an explosion risk," the system extracts safety attributes such as "overtemperature," "excessive pressure," and "explosion risk." The system has a pre-established safety attribute classification system, including categories such as equipment status, environmental factors, human operation, and management measures. Each category has a specific attribute vocabulary. By matching words in the text with the attribute vocabulary and combining it with semantic understanding, the system can identify the public safety attributes described in the text.

[0027] Using dependency parsing and relation extraction techniques, common safety relationship information is extracted from unstructured text. For example, from the text "Safety valve failure prevents pressure release, which in turn causes container rupture," the system extracts the causal chain of "safety valve failure - pressure failure - container rupture." The system predefines a series of relationship types, such as causal relationships, inclusion relationships, and temporal relationships. By analyzing relational terms (such as "lead to," "trigger," and "include") in the text and their context, it identifies the relationship type between entities.

[0028] After extracting information about security management unit entities, public security attributes, and public security relationships, the system converts this information into a data structure that matches the structured public security risk factor data based on pre-set mapping rules. Mapping rules define how entities, attributes, and relationships in unstructured data are mapped to the fields and formats of structured data.

[0029] Specifically, the system maps the security management unit entity to fields such as "Unit ID," "Unit Name," and "Unit Type" in structured data. For example, "Production Workshop A" extracted from the text is mapped to a structured record with the unit ID "WS001," the unit name "Production Workshop A," and the unit type "Production Unit."

[0030] For public safety attributes, the system maps them to fields such as "attribute category," "attribute value," and "risk level" in structured data. For example, the phrase "temperature is too high" extracted from text is mapped to a structured record with the attribute category "device status," the attribute value "temperature is too high," and the risk level "high."

[0031] The system maps public safety relationship information to structured data fields such as "Relationship Type," "Source Entity," "Target Entity," and "Relationship Description." For example, the text extracted from "Safety Valve Failure Leads to Unable to Release Pressure" is mapped to a structured record with the relationship type "Cause and Effect," the source entity "Safety Valve Failure," the target entity "Unable to Release Pressure," and the relationship description "Causes."

[0032] After completing the structured conversion of unstructured data, the system will integrate the structured public safety risk factor data and the converted unstructured public safety risk factor data according to the preset public safety system to build an initial public safety risk factor association network.

[0033] The pre-designed public safety system is a multi-layered classification framework, consisting of a safety management unit layer, a safety risk factor layer, and a safety relationship layer. The safety management unit layer includes safety management responsible units at all levels; the safety risk factor layer includes categories such as equipment and facility risks, environmental risks, personnel risks, and management risks; and the safety relationship layer defines the relationships between units and risk factors.

[0034] Based on a pre-defined public safety framework, the system constructs a network structure from the integrated data. In this network, nodes represent safety management units and safety risk factors, and edges represent the relationships between them. For example, "Production Workshop A," as a node, has a "Risk Exists" edge with the "Overtemperature" risk factor node; and a "Causes" edge exists between the "Overtemperature" node and the "Explosion Risk" node.

[0035] In this way, the system successfully integrates structured and unstructured public safety risk factor data into an initial public safety risk factor association network, which intuitively displays the complex relationships between security management units, security risk factors and their relationship, providing a basis for subsequent risk assessment and early warning.

[0036] In an optional embodiment, performing feature analysis on the public safety risk factor nodes in the initial public safety risk factor association network to obtain feature vectors of the public safety risk factor nodes; and calculating the similarity matrix between the public safety risk factor nodes based on the feature vectors includes: Performing feature analysis on public safety risk factor nodes in the initial public safety risk factor association network, extracting a security management unit static attribute feature vector, a public safety time series feature vector, and a security network topology feature vector of the public safety risk factor node, and combining the security management unit static attribute feature vector, the public safety time series feature vector, and the security network topology feature vector to construct a multi-dimensional feature vector of the public safety risk factor node; Based on the multidimensional feature vector, a sliding window adaptive clustering method is used to calculate the basic similarity value between the public safety risk factor nodes. The sliding window adaptive clustering method includes constructing a feature window of a preset size, dynamically grouping the features of the public safety risk factor nodes within the feature window, and calculating the similarity between the public safety risk factor nodes based on the dynamic grouping; based on the public safety time series feature vector, the sliding window adaptive clustering method is used to calculate the time series similarity value between the public safety risk factor nodes; based on the security network topology feature vector, the sliding window adaptive clustering method is used to calculate the topological similarity value between the public safety risk factor nodes; The basic similarity value, the temporal similarity value and the topological similarity value are used to optimize and reconstruct the initial public safety risk factor association network to obtain a similarity matrix between public safety risk factor nodes.

[0037] In the feature analysis stage, for each risk factor node in the initial public safety risk factor association network, three types of feature vectors are extracted: security management unit static attribute feature vector, public safety time series feature vector and security network topology feature vector.

[0038] The extraction of static attribute feature vectors of security management units is achieved by processing the basic attribute data of public security management units. The feature vectors contain characteristic dimensions such as unit type, security level, jurisdiction, and staff size. For example, for a city fire management department, its static attribute feature vector can be expressed as [3, 4, 2, 5], which respectively represents that the unit is of fire type (3), security level is Class A (4), jurisdiction is at the municipal level (2), and staff size is 500 people (5).

[0039] Public safety time series feature vectors are extracted based on historical security incident data, quantifying risk trends over time. These feature vectors include dimensions such as incident frequency, severity fluctuations, and seasonality. Specifically, security incident data from the past 24 months is divided by month, and the number and severity of each month are counted to form a 48-character time series feature vector. For example, the first 12 values ​​of a traffic safety time series feature vector for a particular area are [5, 7, 3, 4, 8, 10, 12, 9, 6, 4, 3, 5], representing the number of incidents over those 12 months. Subsequent values ​​represent the quantified severity of the incidents.

[0040] The extraction of the topological eigenvector of the safety network is obtained by analyzing the structural characteristics of the nodes in the associated network. It mainly includes network characteristics such as the degree centrality, betweenness centrality, closeness centrality, and eigenvector centrality of the node. In the specific implementation, for a node v in the network, the number of connections between it and all other nodes is calculated as the degree centrality; the ratio of the number of shortest paths passing through the node v to the number of all shortest paths in the network is calculated as the betweenness centrality; the inverse of the average distance from node v to all other nodes is calculated as the closeness centrality; and the eigenvector centrality is obtained by iteratively calculating the eigenvalues ​​of the adjacency matrix. For a node of a public health management department, its topological eigenvector is [23, 0.15, 0.42, 0.38], which respectively means that the node is connected to 23 other nodes, the betweenness centrality is 0.15, the closeness centrality is 0.42, and the eigenvector centrality is 0.38.

[0041] The three types of feature vectors are combined in series to construct a multidimensional feature vector. For a node with m-dimensional static attribute features, n-dimensional temporal features, and p-dimensional topological features, the length of its multidimensional feature vector is m+n+p. In practical applications, different weights can be assigned to different types of features to enhance the influence of key features. For example, for the node in the above example, its multidimensional feature vector is [3, 4, 2, 5, 5, 7, 3, 4, 8, 10, 12, 9, 6, 4, 3, 5...23, 0.15, 0.42, 0.38].

[0042] During the similarity calculation phase, a sliding window adaptive clustering method is used to calculate the similarity between nodes. This method first constructs a feature window of a preset size. The window size is determined based on the feature vector dimension and data characteristics, typically set to 20%-30% of the feature vector dimension. Within the feature window, the features of the public safety risk factor nodes are dynamically grouped, and the similarity between nodes is calculated based on this dynamic grouping.

[0043] In the implementation, for the multidimensional feature vectors Va and Vb of two nodes a and b, the feature window size is set to w. Starting from the first dimension of the feature vector, w dimensions are taken at a time to form a window. The feature similarity between the two nodes is calculated within the window. The feature similarity within the window is calculated using the cosine similarity method, which is the dot product of the two feature vectors within the window divided by the product of their norms. The weighted average of the calculation results for all windows is taken as the basic similarity value for nodes a and b.

[0044] For example, for the multi-dimensional feature vectors Va = [3, 4, 2, 5, 7, 9, 4, 2] and Vb = [2, 5, 3, 4, 8, 7, 3, 3] of nodes a and b, and the window size w = 3, the features in the first window are Va1 = [3, 4, 2] and Vb1 = [2, 5, 3], and the calculated similarity within the window is 0.85. The features in the second window are Va2 = [4, 2, 5] and Vb2 = [5, 3, 4], and the calculated similarity within the window is 0.92, and so on. The final weighted average gives a basic similarity of 0.88.

[0045] The time series similarity value is calculated using the same sliding window adaptive clustering method, but only the public safety time series feature vector is used. For the time series feature vectors Vat and Vbt of nodes a and b, the time window size is set to wt (typically 3-6 months). The similarity of the node activity patterns is calculated within each time window, and the results of each time window are combined to obtain the time series similarity value.

[0046] The topological similarity value is calculated using the same sliding window adaptive clustering method, but only the safety network topology feature vector is used. For the topological feature vectors Vag and Vbg of nodes a and b, the similarity of the nodes' positions in the network structure is calculated within the feature window, focusing on the similarity of the nodes' connection patterns and centrality characteristics.

[0047] The initial public safety risk factor association network is optimized and reconstructed by weighted fusion of basic similarity, temporal similarity, and topological similarity, resulting in a similarity matrix between public safety risk factor nodes. In practical applications, the weights for basic similarity, temporal similarity, and topological similarity can be set to 0.3, 0.4, and 0.3, respectively. The specific weights can be adjusted based on the application scenario and data characteristics.

[0048] For a network containing N nodes, the similarity matrix S is an N×N matrix, where the element S[i][j] represents the comprehensive similarity value between node i and node j. For example, in a public safety risk association network of a certain city, which contains 50 nodes, in the similarity matrix S calculated by the above method, the similarity value between the fire department and the emergency management department is S

[12]

[25] =0.78, indicating that the two departments have a high similarity in risk characteristics; while the similarity value between the fire department and the education department is S

[12]

[38] =0.32, indicating that the two departments have a low similarity in risk characteristics.

[0049] The construction of a similarity matrix provides a foundation for subsequent community discovery and multi-level network construction. It can effectively identify closely related subgroups of public safety risk factors and provide support for public safety risk early warning. This method can further explore the potential connections between different security management units and risk factors, improving the accuracy and timeliness of risk warnings.

[0050] Figure 2 This is a schematic diagram comparing the performance of feature vector similarity calculations in an embodiment of the present invention. The figure shows the performance trends of three different feature similarity calculation methods as the number of nodes changes. The present technical solution is a new network feature extraction algorithm. The static feature similarity method is a traditional method for similarity calculation based on static network features. The traditional network feature method is the most basic network feature extraction method. As can be seen from the performance curve, the present technical solution maintains optimal performance across the entire range of node numbers, with a starting value of 0.88 and a steady upward trend as the number of nodes increases. When the number of nodes reaches 100, the similarity reaches 0.99. The static feature similarity method performs second best, with an initial similarity of 0.72, which slowly increases with the number of nodes, eventually reaching 0.82 when the number of nodes reaches 100. The traditional network feature method performs the worst, with an initial similarity of only 0.50. Although it also increases with the number of nodes, the increase is smaller, reaching only 0.61 when the number of nodes reaches 100. Notably, the performance gap between the three methods remains relatively stable as the number of nodes increases, with our proposed solution maintaining a significant lead. This demonstrates its stability and reliability when extracting features from networks of varying sizes. Experimental results demonstrate that our proposed solution has a significant performance advantage in network feature similarity calculation, particularly when dealing with large-scale networks.

[0051] In an optional embodiment, based on the public safety time series feature vector, using the sliding window adaptive clustering to calculate the time series similarity value between the public safety risk factor nodes; based on the safety network topology feature vector, using the sliding window adaptive clustering to calculate the topology similarity value between the public safety risk factor nodes includes: Constructing a sliding window of a preset size, wherein the sliding window is used to extract the feature vectors of the public safety risk factor nodes in segments; Based on the sliding window, the temporal feature vectors of the public safety risk factor nodes are segmented and extracted to obtain public safety temporal local features, and the temporal similarity values ​​between the public safety risk factor nodes are calculated using adaptive clustering; based on the sliding window, the topological feature vectors of the public safety risk factor nodes are segmented and extracted to obtain security network topology local features, and the topological similarity values ​​between the public safety risk factor nodes are calculated using the adaptive clustering.

[0052] A sliding window of a preset size is constructed to segmentally extract the feature vectors of public safety risk factor nodes. The size of the sliding window can be set based on the actual application scenario. In this embodiment, the sliding window size is set to 24 hours, and the window sliding step is set to 1 hour. The sliding window moves along the time axis, and local features within the window are extracted and calculated each time, effectively handling the dynamic changes of time series data.

[0053] To calculate the time series similarity value, the system extracts the time series feature vectors of public safety risk factor nodes in segments based on a sliding window. Suppose there are two public safety risk factor nodes A and B in a certain city area. The time series feature vector of node A is [12, 15, 18, 25, 30, 28, 22, 18, 15, 20, 25, 28], and the time series feature vector of node B is [10, 14, 17, 23, 29, 27, 20, 17, 14, 19, 24, 27], representing the safety risk index of these two nodes over a continuous 12-hour period. The sliding window size is set to 6. As the window slides across the time series feature vector, the first six time points of node A [12, 15, 18, 25, 30, 28] and the first six time points of node B [10, 14, 17, 23, 29, 27] are first extracted as local features.

[0054] Adaptive clustering is used to calculate the temporal similarity between the two local features. Adaptive clustering first calculates the Euclidean distance between the two local features, which is 3.74. Then the distance is converted into similarity using the Gaussian kernel function. The calculation formula is exp(-distance 2 / 2σ 2 ), where σ is an adaptive parameter that is dynamically adjusted based on the current data distribution and is initially set to the average standard deviation of the local features. In this example, σ is calculated to be 5.2, so the time series similarity between nodes A and B in the current window is 0.87, indicating that the change trends of the two nodes during this time period are highly similar.

[0055] Slide the window backward one step, extracting [15, 18, 25, 30, 28, 22] for node A and [14, 17, 23, 29, 27, 20] for node B, and recalculate the similarity value to 0.89. By continuously sliding the window, we calculate the similarity values ​​within all time windows, ultimately obtaining the sequence of time series similarity values ​​for nodes A and B: [0.87, 0.89, 0.92, 0.90, 0.88, 0.91, 0.85]. The average of these values, 0.89, is taken as the overall time series similarity value for nodes A and B.

[0056] To calculate topological similarity values, the system extracts the topological feature vectors of public safety risk factor nodes in segments based on a sliding window. Assume that the topological feature vectors of node A are [5, 8, 12, 15, 18, 20, 16, 12, 10, 14, 16, 18] and the topological feature vectors of node B are [6, 9, 13, 16, 19, 21, 17, 13, 11, 15, 17, 19], representing the connectivity metrics of these two nodes with other safety risk nodes over a consecutive 12-hour period. The sliding window size is also set to 6. First, the first six topological features of node A ([5, 8, 12, 15, 18, 20]) and the first six topological features of node B ([6, 9, 13, 16, 19, 21]) are extracted as local topological features.

[0057] Adaptive clustering was used to calculate the similarity between these two local topological features. The Euclidean distance was first calculated as 2.45, which was then converted to a similarity value using a Gaussian kernel function. Based on the current topological data distribution, the adaptive parameter σ was adjusted to 4.8. The calculated topological similarity between nodes A and B in the current window was 0.93, indicating a high degree of similarity between the two nodes in terms of network topology.

[0058] The window continues sliding, extracting subsequent local topological features and calculating similarity values, resulting in a topological similarity value sequence of [0.93, 0.94, 0.92, 0.91, 0.93, 0.95, 0.92]. The average of these values, 0.93, is taken as the overall topological similarity value of nodes A and B.

[0059] To improve the accuracy of adaptive clustering, the system dynamically adjusts clustering parameters during the calculation process. When the change in the feature distribution between two adjacent sliding windows exceeds a preset threshold (set to 15% in this example), the clustering parameter σ is automatically recalculated. For example, when sliding from the third window to the fourth window, the change rate of the time series feature distribution of node A is 17%, exceeding the threshold. The system adjusts σ from 5.2 to 4.8 to better adapt to the current data distribution characteristics.

[0060] In practical applications, temporal and topological similarity can be considered together. For example, a weighted average of 0.4 × 0.89 + 0.6 × 0.93 = 0.914 is used as the combined similarity between nodes A and B. The weights can be adjusted based on the specific application scenario. This combined similarity can be used for subsequent risk propagation path prediction, security event correlation analysis, and risk control strategy formulation.

[0061] This implementation utilizes sliding window adaptive clustering technology to effectively capture the similarities between public safety risk factor nodes in both temporal and topological dimensions, providing strong support for building an accurate public safety risk evolution model. This method is highly adaptable, computationally efficient, and capable of handling security risk networks of varying scales, offering significant application value.

[0062] In an optional embodiment, performing community discovery on the similarity matrix to identify subgroups of public safety risk factors with closely associated relationships; and constructing a multi-level public safety risk factor association network based on the public safety risk factor subgroups includes: Calculating, for each security management node in the similarity matrix, a community affiliation strength based on edge weights between adjacent security management nodes, iteratively calculating and continuously updating the community labels of the security management nodes based on the community affiliation strength until the distribution of the community labels tends to stabilize and converge, thereby obtaining a community division result including multiple initial security management communities; Calculating the cohesion of each security management community in the community division result. The cohesion is quantitatively calculated based on the edge weight relationship between security management nodes within the community. Security management communities with a cohesion greater than a preset cohesion threshold are marked as stable communities. Security management communities with a cohesion less than the preset cohesion threshold are recursively decomposed into multiple security management sub-communities with higher cohesion. The association strength of the security management sub-communities is calculated based on the distribution characteristics of the internal nodes of the security management sub-communities and the connection relationship between the sub-communities. Based on the stable community, the security management sub-community and the association strength, the inter-layer edge weights are dynamically updated and optimized to obtain a multi-level public security risk factor association network with a complete hierarchical structure.

[0063] When calculating the community affiliation strength for each security management node in the similarity matrix, a community label propagation method based on edge weights is used. For node i in the network, let its set of neighboring nodes be N(i), and the edge weight between node i and its neighboring node j be w(i, j), which is the corresponding element value in the similarity matrix. The community affiliation strength of node i is calculated by summing the weights of the community labels of each of node i's neighboring nodes j, where the weight value is equal to the edge weight w(i, j) between node i and node j. For example, in a city's emergency management network, node A is adjacent to nodes B, C, and D, with edge weights of 0.8, 0.6, and 0.3, respectively. Neighboring nodes B and C belong to community 1, and node D belongs to community 2. Therefore, node A's affiliation strength to community 1 is 0.8 + 0.6 = 1.4, and its affiliation strength to community 2 is 0.3.

[0064] During the iterative calculation of community labels, the processing order for each node is randomly selected. Based on the strength of a node's affiliation to each community, the node is assigned to the community with the strongest affiliation. If a node has equal affiliation to multiple communities, one of the communities is randomly selected as the node's label. This iterative process continues until the community labels of more than 95% of the nodes in the network remain unchanged for five consecutive iterations, indicating that the distribution of community labels has stabilized.

[0065] In practical applications, a city's public safety risk association network, for example, consists of 100 safety management nodes. Initially, each node is assigned a unique community label. After iterative calculations, the network converges to eight initial safety management communities, corresponding to different types of safety management functional areas. For example, Community 1 includes nodes such as fire protection and safety supervision, while Community 2 includes nodes such as public security and transportation.

[0066] When calculating the cohesion of each security management community in the resulting community partitioning, we use an edge-weighted approach. The cohesion of community C is calculated as the sum of the edge weights between all pairs of nodes within the community divided by the product of the maximum number of edges within the community and the maximum edge weight. Specifically, for a community with n nodes, the maximum number of internal edges is n × (n-1) / 2, and the maximum edge weight is 1.0. For example, if a community contains 5 nodes and the sum of its internal edge weights is 8.2, its cohesion is 8.2 ÷ (5 × 4 / 2 × 1.0) = 0.82.

[0067] A preset cohesion threshold of 0.7 is set, and security management communities with a cohesion greater than 0.7 are marked as stable communities. Communities with a cohesion less than 0.7 are recursively decomposed. This decomposition process uses a hierarchical clustering method, starting with the edge with the smallest weight within the community, dividing the community into two subcommunities. The cohesion of each subcommunity is then recursively calculated until the cohesion of all subcommunities exceeds the preset threshold or the subcommunity size is reduced to an indivisible level (e.g., containing only three nodes).

[0068] For example, consider community 3 in a city's public safety network. This community consists of 15 nodes and has a cohesion of 0.65, which is less than the preset threshold of 0.7. Through recursive decomposition, community 3 is split into two subcommunities, 3a and 3b, containing 8 and 7 nodes, respectively. Their cohesions are 0.75 and 0.72, respectively, both exceeding the preset threshold. Therefore, the decomposition process ends.

[0069] For the decomposed security management subcommunities, we need to calculate the strength of the connections between them. This strength is calculated based on the edge weights between the subcommunities: the sum of all edge weights between the two subcommunities divided by the product of the number of nodes in the two subcommunities. For example, if there are 12 edges between subcommunities 3a and 3b, and the sum of the edge weights is 5.4, the strength of the connection is 5.4 ÷ (8 × 7) = 0.096.

[0070] When constructing a multi-level public safety risk factor association network, stable communities serve as nodes in the upper network, the decomposed subcommunities serve as nodes in the middle network, and the original security management nodes form the bottom network. The weights of the inter-layer edges are set based on the affiliation and association strength between communities and subcommunities, and between subcommunities and original nodes.

[0071] An adaptive weight adjustment mechanism is used to dynamically update and optimize inter-layer edge weights. For community node A in the upper network and sub-community node B in the middle network, if sub-community B is decomposed from community A, the inter-layer edge weight from A to B is initially set to the cohesion value of B. For sub-community node B in the middle network and original node C in the bottom network, if node C belongs to sub-community B, the inter-layer edge weight from B to C is initially set to the centrality value of node C in sub-community B, which is calculated based on the edge weights connecting node C to other nodes in sub-community B.

[0072] As the risk network evolves dynamically, inter-layer edge weights require regular updates. The update cycle can be set to every 7 or 30 days, depending on the real-time requirements of the application scenario. During the update process, the cohesion of sub-communities and the strength of connections between communities are recalculated based on changes in the relationships between underlying nodes. Edge weights at each layer are then updated from the bottom up.

[0073] For example, the hierarchical structure of a city's public safety risk network includes four stable communities at the top (corresponding to the four major safety management areas), 12 sub-communities at the middle level (corresponding to subdivided safety management functional groups), and 100 original safety management nodes at the bottom level. After a month of data updates, the relationships between nodes in the bottom-level network changed, causing the cohesion of one sub-community to drop from 0.75 to 0.68, falling below the preset threshold and requiring further decomposition. At the same time, the correlation strength between two sub-communities, which originally had a correlation strength of 0.1, increased to 0.25, indicating a strengthening of their connection, which needed to be reflected in the upper-level network.

[0074] By dynamically updating and optimizing the weights of inter-layer edges, a multi-level public safety risk factor association network with a complete hierarchical structure is ultimately constructed. This network can clearly reflect the relationships between security management units at different levels and provide a basis for subsequent risk transmission analysis.

[0075] Figure 3 This is a bar chart comparing the performance of the multi-level public safety risk factor association network of an embodiment of the present invention. The figure shows the performance comparison of three different community analysis methods. Among them, the single-layer community division method is a traditional basic algorithm for community identification based on a single level, the hierarchical community division method is an algorithm that considers the hierarchical relationship of groups, and the present technical solution is a new type of comprehensive community analysis method. From the perspective of three key performance indicators, in terms of community discovery accuracy, the present technical solution performed the best, reaching 83.2%, which is significantly better than the 67.8% of the hierarchical community division method and the 52.5% of the single-layer community division method; in terms of cohesion recognition efficiency, the present technical solution also maintained a leading advantage, reaching 82.5%, while the hierarchical method was 59.7% and the single-layer method was only 46.3%; in terms of hierarchical association performance, the present technical solution still performed the best, reaching 76.7%, the hierarchical method was 53.2%, and the single-layer method was relatively weak, at only 31.9%. The data comparison clearly demonstrates that this technical solution significantly outperforms traditional methods across all three evaluation dimensions, particularly in addressing hierarchical relationships within communities. This demonstrates the solution's strong practical value and potential for application in community structure analysis. This data fully demonstrates the importance and effectiveness of this new comprehensive analysis method in community structure research.

[0076] In an optional embodiment, performing a transmission path analysis on the multi-level public safety risk factor association network, calculating the transmission weights between public safety risk factor nodes, and generating public safety risk warning information based on the transmission weights and the analysis results of the multi-level public safety risk factor association network includes: Obtaining a conduction path between pairs of public safety risk factor nodes in the multi-level public safety risk factor association network, obtaining historical conduction data of the public safety risk factor nodes at different times based on the conduction path to obtain a conduction feature vector, and calculating a temporal dependency relationship between the pairs of public safety risk factor nodes based on the conduction feature vector; Based on the temporal dependency, a temporal attenuation factor is calculated in combination with the time difference between the current moment and the historical moment, and the temporal dependency and the temporal attenuation factor are multiplied to obtain an initial transmission weight between the pairs of public safety risk factor nodes; the initial transmission weight is adjusted according to the connection relationship of the public safety risk factor nodes in the transmission path to obtain a dynamic transmission weight; The dynamic conduction weight is combined with the risk degree of the public safety risk element node along the conduction path to obtain the public safety risk conduction intensity, and the public safety risk conduction intensity is accumulated to obtain a public safety risk warning index. When the public safety risk warning index exceeds a preset warning threshold, public safety risk warning information is generated.

[0077] like Figure 4 As shown, the method includes: Obtain the transmission paths between pairs of public safety risk factor nodes in the network. For example, in a network containing multi-level risk factors such as natural disasters, social security, and public health, it can be determined that there is a transmission path between earthquake (A) and the secondary disaster fire (B), and there is also a transmission path between fire (B) and casualties (C), thus forming a transmission link A--B--C.

[0078] Based on the determined transmission paths, the system collects historical transmission data from these risk factor nodes at different times. For example, historical data on fires caused by earthquakes over the past five years is collected, including indicators such as earthquake intensity, fire probability, and fire size. Specifically, it can be determined that when the earthquake intensity is 6, the probability of causing a fire is 0.3; when the earthquake intensity is 7, the probability of causing a fire is 0.5; and when the earthquake intensity is 8, the probability of causing a fire is 0.7. Based on this historical data, a transmission characteristic vector is formed. For example, the transmission characteristic vector between earthquakes and fires can be expressed as [6, 0.3; 7, 0.5; 8, 0.7].

[0079] Based on the transmission eigenvectors, the temporal dependencies between pairs of risk factor nodes are calculated. By analyzing the time delay and probability relationship between earthquake and fire occurrence in historical data, the strength of their temporal dependencies is determined. For example, the analysis found that the dependency strength for fire occurrence within 24 hours after an earthquake is 0.6, 0.2 within 24-48 hours, and 0.1 within 48-72 hours.

[0080] Based on the calculated temporal dependency and the time difference between the current moment and the historical moment, a temporal decay factor is calculated. Assuming a magnitude 6 earthquake has occurred 12 hours after it occurred, the temporal decay factor can be calculated as 0.8 according to the time decay model. Multiplying the temporal dependency of 0.6 by the temporal decay factor of 0.8 yields an initial transmission weight of 0.48 between the earthquake and the fire.

[0081] Based on the connections between risk factor nodes in the transmission path, the initial transmission weight is adjusted to obtain a dynamic transmission weight. Considering the multiple transmission paths between earthquakes and fires, such as earthquake-building damage-fire, earthquake-power system damage-fire, etc., a comprehensive assessment of the weights of these paths is performed. Assuming the adjustment coefficient after comprehensive assessment is 1.2, the dynamic transmission weight is 0.48 × 1.2 = 0.576.

[0082] The dynamic transmission weights along the transmission path are combined with the risk levels of the risk factor nodes to obtain the public safety risk transmission intensity. Assuming the current risk level of a magnitude 6 earthquake is assessed as 0.7 (out of a maximum score of 1), the earthquake-fire risk transmission intensity is 0.576 × 0.7 = 0.4032.

[0083] The public safety risk transmission intensity is accumulated to generate a public safety risk warning indicator. In transmission link A--B--C, if the fire-casualty risk transmission intensity is 0.5, the cumulative risk warning indicator from earthquake to casualties can be calculated as 0.4032 × 0.5 = 0.2016. The system's preset warning threshold is 0.2. When the calculated risk warning indicator of 0.2016 exceeds the warning threshold, the system generates a public safety risk warning message: "After a magnitude 6 earthquake in the current area, the risk of fire-related casualties is high. Relevant departments are requested to prepare for emergency response." In practical applications, the system can be expanded to more complex, multi-layered risk networks. For example, different types of public safety risk factors, such as environmental pollution and traffic accidents, can be incorporated into the network to analyze the transmission relationships between them. The system can identify non-obvious risk transmission paths, such as the transmission chain from environmental pollution to public health incidents to social stability issues and economic impacts.

[0084] By adjusting the parameters of the time-degradation model, the system can adapt to the time-degradation characteristics of different risk types. For example, the risk transmission of tsunamis caused by earthquakes has a rapid decay characteristic, so the time-degradation factor can be set to an exponential decay model; while the time-degradation factor of public health problems caused by environmental pollution has a long-term cumulative effect, so the time-degradation factor can be set to a slow linear decay model.

[0085] The risk assessment of risk factor nodes is dynamically updated. For example, if a region's rainfall is continuously increasing, the flood risk level in that region will increase accordingly, which in turn affects the calculation results of all flood-related risk transmission paths.

[0086] The public safety risk warning method of this embodiment can timely discover potential risk chains by analyzing the transmission relationship between multi-level risk factors, generate warning information in advance, and provide a scientific basis for emergency decision-making of relevant departments.

[0087] In an optional embodiment, the public safety risk transmission intensity is obtained by combining the dynamic transmission weight with the risk level of the public safety risk factor node along the transmission path, and the public safety risk early warning index is obtained by accumulating the public safety risk transmission intensity. Based on the dynamic transmission weight and risk level of the public safety risk factor node, the PageRank algorithm is used to calculate the centrality score of the public safety risk factor node, and the centrality score is combined with the dynamic transmission weight to obtain the carrying capacity of the public safety risk factor node; the structural importance of the public safety risk factor node is calculated based on the PageRank algorithm, and the structural importance is combined with the carrying capacity to obtain the failure threshold of the public safety risk factor node; Combining the dynamic transmission weight with the risk level of the public safety risk factor node along the transmission path to obtain a public safety risk transmission intensity; determining whether the public safety risk factor node has reached the failure threshold based on the public safety risk transmission intensity, and marking the public safety risk factor node that has reached the failure threshold as a trigger node; The PageRank algorithm is used to calculate the degree of influence of the trigger node on the adjacent public safety risk factor nodes on the transmission path, the propagation impact is determined based on the degree of influence, and the propagation impact is accumulated to obtain the public safety risk cumulative effect. The public safety risk warning index is calculated based on the public safety risk cumulative effect and the public safety risk transmission intensity.

[0088] Based on the established public safety risk factor node network and transmission path, the centrality score, carrying capacity and failure threshold of the node are calculated. Then, it is determined whether the node has reached the failure threshold and the trigger node is marked. Finally, the cumulative effect of public safety risk and early warning indicators are calculated.

[0089] The public safety risk factor node network can include multiple different types of risk factor nodes, such as natural disaster nodes, accident and disaster nodes, public health nodes, and social security nodes. There are specific transmission paths between these nodes, along which risks can be transmitted from one node to another.

[0090] The PageRank algorithm is used to calculate the centrality scores of public safety risk factor nodes. For example, for a network containing 10 public safety risk factor nodes, the centrality score of each node can be calculated iteratively. Assume that node A has a centrality score of 0.15, node B has a centrality score of 0.12, node C has a centrality score of 0.09, and so on. These centrality scores reflect the importance and influence of each node in the network.

[0091] The node's centrality score is combined with its dynamic transmission weight to calculate its carrying capacity. Assuming node A's dynamic transmission weight is 0.8, its carrying capacity can be calculated as 0.15 × 0.8 = 0.12; node B's dynamic transmission weight is 0.7, its carrying capacity is 0.12 × 0.7 = 0.084; and node C's dynamic transmission weight is 0.6, its carrying capacity is 0.09 × 0.6 = 0.054. Carrying capacity represents the maximum risk intensity a node can withstand.

[0092] The PageRank algorithm is used to calculate the structural importance of public safety risk factor nodes. Structural importance reflects the criticality of a node in the network structure. Assume that the structural importance of node A is 0.25, the structural importance of node B is 0.2, and the structural importance of node C is 0.18. Combining structural importance with carrying capacity can determine the node failure threshold. For example, the failure threshold of node A can be calculated as 0.25 + 0.12 = 0.37; the failure threshold of node B is 0.2 + 0.084 = 0.284; and the failure threshold of node C is 0.18 + 0.054 = 0.234. The failure threshold represents the critical point at which a node transitions from a normal state to a risky state.

[0093] When transmitting risks along the transmission path, the dynamic transmission weight is combined with the risk level of the public safety risk factor node to calculate the public safety risk transmission intensity. Assuming the risk level of node A is 0.5, the risk level of node B is 0.4, and the risk level of node C is 0.3, the risk transmission intensity from node A to node B can be calculated as 0.8 × 0.5 = 0.4; the risk transmission intensity from node B to node C can be calculated as 0.7 × 0.4 = 0.28.

[0094] Based on the calculated public safety risk transmission intensity, each node is judged to have reached its failure threshold. For example, if the risk transmission intensity received by node B is 0.4, which is greater than its failure threshold of 0.284, node B is marked as a trigger node. If the risk transmission intensity received by node C is 0.28, which is greater than its failure threshold of 0.234, node C is also marked as a trigger node.

[0095] For each marked trigger node, the PageRank algorithm is used to calculate its influence on adjacent nodes along the transmission path. For example, trigger node B has an influence of 0.35 on its neighboring node D and 0.3 on its neighboring node E. Based on these influences, the propagation impact can be determined. Assuming the risk level of node D is 0.25, the propagation impact of node B on D can be calculated as 0.35 × 0.25 = 0.0875. Assuming the risk level of node E is 0.2, the propagation impact of node B on E can be calculated as 0.3 × 0.2 = 0.06.

[0096] All propagation effects are accumulated to obtain the cumulative effect of public safety risk. For example, if there are multiple triggering nodes (B and C) in the system, and the total propagation effect of B is 0.15 and the total propagation effect of C is 0.12, then the cumulative effect of public safety risk is 0.15 + 0.12 = 0.27.

[0097] The public safety risk early warning index is calculated based on the cumulative effect of public safety risks and the intensity of public safety risk transmission. For example, the early warning index can be calculated by adding the average value of the cumulative effect of public safety risks and the average value of the risk transmission intensity. Assuming the average value of the risk transmission intensity is 0.34, the public safety risk early warning index is 0.27 + 0.34 = 0.61. This index can be used to guide public safety management departments in taking appropriate preventive and response measures.

[0098] Through the above steps, dynamic monitoring and early warning of public safety risks can be achieved, providing a scientific basis for public safety management decisions. This method considers key characteristics such as the centrality, carrying capacity, and failure threshold of risk factor nodes. It can more accurately reflect the transmission process and cumulative effects of public safety risks, helping to improve the efficiency and accuracy of public safety risk management.

[0099] Figure 5This is a bar chart comparing the performance of the public safety risk transmission model of an embodiment of the present invention. This comparison chart shows the performance comparison of three different risk assessment models. The basic PageRank model is a basic risk assessment algorithm based on the calculation of web page importance, the dynamic transmission weight model is a risk transmission assessment method that considers the dynamic relationship between network nodes, and the multi-level risk accumulation model is a comprehensive risk assessment model that combines the characteristics of multi-layer network structure. From the specific data, in terms of warning accuracy, the multi-level risk accumulation model performed best, reaching 89.5%, the dynamic transmission weight model reached 78.2%, and the basic PageRank model was 65.8%. In terms of risk transmission recognition rate, all three models showed high recognition ability, with the multi-level model reaching 91.2%, the dynamic transmission model 83.7%, and the basic model 72.3%. In terms of computational efficiency, the basic PageRank model performed best, reaching 93.5%, followed by the dynamic transmission weight model at 87.1%, and the multi-level model slightly lower at 84.3%. Overall, the multi-level risk accumulation model performed best in the two core indicators of early warning accuracy and risk transmission identification rate, but was slightly inferior to the other two models in computational efficiency, which reflects that there is a certain trade-off between model performance and computational complexity.

[0100] According to a second aspect of an embodiment of the present invention, an electronic device is provided, including: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0101] According to a third aspect of an embodiment of the present invention, a computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the method described above is implemented.

[0102] The present invention may be a method, an apparatus, a system and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for executing various aspects of the present invention.

[0103] 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 above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A knowledge graph-based method for identifying and analyzing the association of multi-source risk factors in public safety, characterized by: include: Acquiring public safety multi-source risk factor data, wherein the public safety multi-source risk factor data includes structured public safety risk factor data and unstructured public safety risk factor data; Processing the unstructured public safety risk factor data to extract public safety risk factor information; Integrating the structured public safety risk factor data and the processed unstructured public safety risk factor data to construct an initial public safety risk factor association network; Performing feature analysis on the public safety risk factor nodes in the initial public safety risk factor association network to obtain feature vectors of the public safety risk factor nodes; and calculating a similarity matrix based on the feature vectors. Performing community discovery on the similarity matrix to identify subgroups of public safety risk factors with closely related relationships; Constructing a multi-level public safety risk factor association network based on the public safety risk factor subgroups; Conducting path analysis on the multi-level public safety risk factor association network, calculating the conduction weights between public safety risk factor nodes, and generating public safety risk warning information based on the conduction weights and the analysis results of the multi-level public safety risk factor association network.

2. The method according to claim 1, characterized in that Integrating the structured public safety risk factor data and the processed unstructured public safety risk factor data to construct an initial public safety risk factor association network includes: Processing the unstructured public safety risk factor data, extracting security management unit entities, public safety attributes, and public safety relationship information from the unstructured public safety risk factor data, and converting the security management unit entities, the public safety attributes, and the public safety relationship information into a data structure that matches the structured public safety risk factor data according to a preset mapping rule; The structured public safety risk factor data and the converted unstructured public safety risk factor data are integrated according to a preset public safety system to construct an initial public safety risk factor association network.

3. The method according to claim 1, characterized in that Performing feature analysis on the public safety risk factor nodes in the initial public safety risk factor association network to obtain feature vectors of the public safety risk factor nodes; and calculating a similarity matrix between the public safety risk factor nodes based on the feature vectors includes: Performing feature analysis on public safety risk factor nodes in the initial public safety risk factor association network, extracting a security management unit static attribute feature vector, a public safety time series feature vector, and a security network topology feature vector of the public safety risk factor node, and combining the security management unit static attribute feature vector, the public safety time series feature vector, and the security network topology feature vector to construct a multi-dimensional feature vector of the public safety risk factor node; Based on the multidimensional feature vector, a sliding window adaptive clustering method is used to calculate a basic similarity value between the public safety risk factor nodes. The sliding window adaptive clustering method includes constructing a feature window of a preset size, dynamically grouping the features of the public safety risk factor nodes within the feature window, and calculating the similarity between the public safety risk factor nodes based on the dynamic grouping. Based on the public safety time series feature vector, the sliding window adaptive clustering is used to calculate the time series similarity value between the public safety risk factor nodes; Based on the safety network topology feature vector, the sliding window adaptive clustering is used to calculate the topological similarity value between the public safety risk factor nodes; The basic similarity value, the temporal similarity value and the topological similarity value are used to optimize and reconstruct the initial public safety risk factor association network to obtain a similarity matrix between public safety risk factor nodes.

4. The method according to claim 3, characterized in that Based on the public safety time series feature vector, the sliding window adaptive clustering is used to calculate the time series similarity value between the public safety risk factor nodes; based on the safety network topology feature vector, the sliding window adaptive clustering is used to calculate the topology similarity value between the public safety risk factor nodes, including: Constructing a sliding window of a preset size, wherein the sliding window is used to extract the feature vectors of the public safety risk factor nodes in segments; Based on the sliding window, the temporal feature vectors of the public safety risk factor nodes are segmented and extracted to obtain public safety temporal local features, and the temporal similarity values ​​between the public safety risk factor nodes are calculated using adaptive clustering; based on the sliding window, the topological feature vectors of the public safety risk factor nodes are segmented and extracted to obtain security network topology local features, and the topological similarity values ​​between the public safety risk factor nodes are calculated using the adaptive clustering.

5. The method according to claim 1, wherein Performing community discovery on the similarity matrix to identify subgroups of public safety risk factors with close associations; and constructing a multi-level public safety risk factor association network based on the subgroups of public safety risk factors includes: Calculating, for each security management node in the similarity matrix, a community affiliation strength based on edge weights between adjacent security management nodes, iteratively calculating and continuously updating the community labels of the security management nodes based on the community affiliation strength until the distribution of the community labels tends to stabilize and converge, thereby obtaining a community division result including multiple initial security management communities; Calculating the cohesion of each security management community in the community division result. The cohesion is quantitatively calculated based on the edge weight relationship between security management nodes within the community. Security management communities with a cohesion greater than a preset cohesion threshold are marked as stable communities. Security management communities with a cohesion less than the preset cohesion threshold are recursively decomposed into multiple security management sub-communities with higher cohesion. The association strength of the security management sub-communities is calculated based on the distribution characteristics of the internal nodes of the security management sub-communities and the connection relationship between the sub-communities. Based on the stable community, the security management sub-community and the association strength, the inter-layer edge weights are dynamically updated and optimized to obtain a multi-level public security risk factor association network with a complete hierarchical structure.

6. The method according to claim 1, characterized in that Conducting a transmission path analysis on the multi-level public safety risk factor association network, calculating the transmission weights between public safety risk factor nodes, and generating public safety risk warning information based on the transmission weights and the analysis results of the multi-level public safety risk factor association network include: Obtaining a conduction path between pairs of public safety risk factor nodes in the multi-level public safety risk factor association network, obtaining historical conduction data of the public safety risk factor nodes at different times based on the conduction path to obtain a conduction feature vector, and calculating a temporal dependency relationship between the pairs of public safety risk factor nodes based on the conduction feature vector; Based on the temporal dependency, a temporal decay factor is calculated in combination with the time difference between the current moment and the historical moment, and the temporal dependency and the temporal decay factor are multiplied to obtain the initial transmission weight between the public safety risk factor node pairs; Adjusting the initial transmission weight according to the connection relationship of the public safety risk factor node in the transmission path to obtain a dynamic transmission weight; The dynamic conduction weight is combined with the risk degree of the public safety risk element node along the conduction path to obtain the public safety risk conduction intensity, and the public safety risk conduction intensity is accumulated to obtain a public safety risk warning index. When the public safety risk warning index exceeds a preset warning threshold, public safety risk warning information is generated.

7. The method according to claim 6, characterized in that The public safety risk transmission intensity is obtained by combining the dynamic transmission weight with the risk level of the public safety risk factor node along the transmission path, and the public safety risk early warning index is obtained by accumulating the public safety risk transmission intensity, including: Based on the dynamic transmission weight and risk level of the public safety risk factor node, the PageRank algorithm is used to calculate the centrality score of the public safety risk factor node, and the centrality score is combined with the dynamic transmission weight to obtain the carrying capacity of the public safety risk factor node; Calculating the structural importance of the public safety risk factor node based on the PageRank algorithm, and combining the structural importance with the carrying capacity to obtain a failure threshold of the public safety risk factor node; Combining the dynamic transmission weight with the risk level of the public safety risk factor node along the transmission path to obtain the public safety risk transmission intensity; Determining whether the public safety risk factor node has reached the failure threshold based on the public safety risk transmission intensity, and marking the public safety risk factor node that has reached the failure threshold as a trigger node; The PageRank algorithm is used to calculate the degree of influence of the trigger node on the adjacent public safety risk factor nodes on the transmission path, the propagation impact is determined based on the degree of influence, and the propagation impact is accumulated to obtain the public safety risk cumulative effect. The public safety risk warning index is calculated based on the public safety risk cumulative effect and the public safety risk transmission intensity.

8. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the method according to any one of claims 1 to 7.

9. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images