Identification method, device and equipment for key nodes in network and storage medium

Abstract

The invention relates to the technical field of artificial intelligence, is applied to the field of digital financial services and the field of digital medical services, and particularly relates to an identification method and device for key nodes in a network, computer equipment and a storage medium. According to the method and the device, the key nodes are identified from the network nodes of the target network structure based on the key parameters such as the degree value data, the matrix value data and the path data of the network nodes, and compared with the key nodes identified from the network through a single importance index in the prior art, the key nodes have stronger comprehensive influence in the network; after the key node identified in the application is protected, the purpose of improving the survivability of a financial network or a medical information network can be achieved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and is applied to the fields of digital financial services and digital medical services, and in particular to a method, device, computer equipment and storage medium for identifying key nodes in a network. Background Art

[0002] In the digital financial and healthcare sectors, complex network structures have become a crucial technological feature. For example, in digital finance, diverse actors such as financial institutions, investors, and markets are closely intertwined through transaction relationships and investment chains, forming a dynamic and complex financial network. In digital healthcare, core elements such as medical institutions, patients, and medical data are interconnected through information technology, forming a vast medical information network. A prominent feature of these networks is their heterogeneous topology. In this network structure, a small number of key nodes play a decisive role in the overall structure and function of the network, while the influence of most nodes is relatively weak. These key nodes are not only the core hubs for information dissemination, resource allocation, and energy flow, but also the key to network robustness and vulnerability. With the dual objectives of network protection and disruption, identifying and protecting these key nodes has become a crucial strategy for enhancing network resilience.

[0003] However, existing technologies mostly rely on a single parameter (such as degree centrality) to identify key nodes from the network. However, the key nodes identified based on a single parameter are difficult to fully reflect the comprehensive influence of the key nodes in the network. Therefore, protecting the key nodes identified from the network based on a single parameter will lead to the problem of weak anti-destruction capabilities in financial networks or medical information networks. Summary of the Invention

[0004] An embodiment of the present invention provides a method for identifying key nodes in a network to solve the problem of weak anti-destruction capability of a financial network or a medical information network when protecting key nodes identified from the network based on a single importance indicator.

[0005] A method for identifying key nodes in a network, comprising: Acquire a target network structure, where the target network structure is constructed based on node data and edge data in a target information network; Respectively obtaining key parameters of each network node in the target network structure, the key parameters including degree data, matrix value data, and path data of the network node; the path data including the number of shortest paths passing through the network node in the target network structure; Inputting the key parameters corresponding to each of the network nodes into a key node prediction model to obtain an importance index of each of the network nodes; According to the numerical value of the importance index, key nodes in the target network structure are screened out from the network nodes.

[0006] In the above method for identifying key nodes in the network, optionally, the matrix value data is obtained by: Constructing a target matrix corresponding to the target network structure; Normalizing the target matrix to obtain a standard matrix; Matrix value data corresponding to each of the network nodes is extracted from the standard matrix.

[0007] In the above method for identifying key nodes in the network, optionally, the path data is obtained by: dividing all the network nodes in the target network structure into node combinations including two network nodes, wherein different node combinations include different network nodes; respectively obtaining the shortest path between two network nodes in each node combination; All intermediate nodes in the shortest paths are extracted, and the number of times the same intermediate node is extracted is recorded as the path data of the network node.

[0008] In the above method for identifying key nodes in a network, optionally, the degree value includes the in-degree value, out-degree value and node degree value of the network node; The degree value data is obtained in the following manner: Obtaining the number of out-degree edges from the network node to other network nodes in the target network structure as the out-degree value; Obtaining the number of in-degree edges pointing to the network node as the in-degree value; The node degree value is calculated according to the in-degree value and the out-degree value.

[0009] In the above method for identifying key nodes in a network, optionally, the degree value further includes a key degree value; The critical value is obtained as follows: Obtaining the weights of all the in-degree edges and the weights of all the corresponding out-degree edges corresponding to the network node; The critical degree value is calculated based on the weight and the node degree value.

[0010] In the above method for identifying key nodes in the network, optionally, the key node prediction model is trained in the following manner: Obtain historical degree value data, historical matrix value data, and historical path data to construct training samples, and mark each of the training samples with an importance label; Inputting the training samples marked with the importance labels into the key node prediction model for training to obtain a predicted importance result of the training samples; Calculating a training loss value according to the importance label and the predicted importance result; When the training loss value meets a preset end condition, determining that the key node prediction model training is completed; When the training loss value does not meet the termination condition, the model parameters of the key node prediction model are adjusted, and the key node prediction model is iteratively trained based on the training samples until the training loss value meets the termination condition, and it is determined that the training of the key node prediction model is completed.

[0011] A device for determining key nodes in a network, comprising: A network structure construction module is used to obtain a target network structure, where the target network structure is constructed based on node data and edge data in a target information network; a key parameter acquisition module, configured to respectively acquire key parameters of each network node in the target network structure, wherein the key parameters include degree value data, matrix value data, and path data of the network node; the path data includes the number of shortest paths passing through the network node in the target network structure; An importance index prediction module, configured to input the key parameters corresponding to each of the network nodes into a key node prediction model to obtain an importance index of each of the network nodes; The key node screening module is used to screen out the key nodes in the target network structure from the network nodes according to the numerical value of the importance index.

[0012] In the above-mentioned device for identifying key nodes in the network, optionally, the matrix value data is obtained by: Constructing a target matrix corresponding to the target network structure; Normalizing the target matrix to obtain a standard matrix; Matrix value data corresponding to each of the network nodes is extracted from the standard matrix.

[0013] A computer device comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements any of the above methods for identifying key nodes in a network when executing the computer program.

[0014] A computer-readable storage medium stores a computer program, wherein the computer program, when executed by a processor, implements any of the above methods for identifying key nodes in a network.

[0015] The present invention provides a method, device, computer equipment and storage medium for identifying key nodes in a network. By obtaining a target network structure, the target network structure is constructed based on node data and edge data in a target information network, and the key parameters of each network node in the target network structure are obtained respectively. The key parameters include degree data, matrix value data and path data of the network node. The path data includes the number of shortest paths passing through the network node in the target network structure. The key parameters corresponding to each network node are input into a key node prediction model to obtain an importance index of each network node. According to the numerical value of the importance index, the key nodes in the target network structure are screened out from the network nodes. It can be seen that in this application, key nodes are identified from the network nodes of the target network structure based on key parameters such as degree data, matrix value data and path data of the network nodes. Compared with the prior art, the key nodes identified from the network by a single importance index have a stronger comprehensive influence in the network. After protecting the key nodes identified in this application, the purpose of improving the anti-destruction capability of the financial network or medical information network can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0017] Figure 1 This is a flowchart of a method for identifying key nodes in a network disclosed in one embodiment of the present invention; Figure 2 is a schematic structural diagram of a target network structure disclosed in one embodiment of the present invention; Figure 3 This is a partial implementation flow chart of a method for identifying key nodes in a network disclosed in one embodiment of the present invention; Figure 4 This is a partial implementation flow chart of a method for identifying key nodes in a network disclosed in one embodiment of the present invention; Figure 5 This is a partial implementation flow chart of a method for identifying key nodes in a network disclosed in one embodiment of the present invention; Figure 6 This is a partial implementation flow chart of a method for identifying key nodes in a network disclosed in one embodiment of the present invention; Figure 7 This is a partial implementation flow chart of a method for identifying key nodes in a network disclosed in one embodiment of the present invention; Figure 8It is a schematic structural diagram of a device for identifying key nodes in a network disclosed in one embodiment of the present invention; Figure 9 It is a structural diagram of a computer device disclosed in one embodiment of the present invention. DETAILED DESCRIPTION

[0018] In order to make the technical problems, technical solutions and beneficial effects solved by the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0019] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof.

[0020] It will also be understood that the term "and / or" used in the present description and appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0021] As used in the present specification and the appended claims, the term "if" may be interpreted as "when" or "upon" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" may be interpreted as meaning "upon determination" or "in response to determining" or "upon detection of [described condition or event]" or "in response to detecting [described condition or event]," depending on the context.

[0022] In addition, in the description of the present specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0023] References to "one embodiment" or "some embodiments" in the present specification mean that a particular feature, structure, or characteristic described in connection with that embodiment is included in one or more embodiments of the present invention. Thus, phrases such as "in one embodiment," "in some embodiments," "in other embodiments," and "in yet other embodiments" appearing in various places in this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically stated. The terms "including," "comprising," "having," and variations thereof mean "including but not limited to," unless otherwise specifically stated.

[0024] The present invention provides a method, device, computer equipment and storage medium for identifying key nodes in a network. The method obtains a target network structure, which is constructed based on node data and edge data in a target information network. The key parameters of each network node in the target network structure are obtained respectively. The key parameters include degree data, matrix value data and path data of the network node. The path data includes the number of shortest paths passing through the network node in the target network structure. The key parameters corresponding to each network node are input into a key node prediction model to obtain an importance index of each network node. According to the numerical value of the importance index, the key nodes in the target network structure are screened out from the network nodes. It can be seen that the present application identifies key nodes from the network nodes of the target network structure based on key parameters such as degree data, matrix value data and path data of the network node. Compared with the prior art, the key nodes identified from the network by a single importance index have a stronger comprehensive influence in the network. After protecting the key nodes identified in the present application, the purpose of improving the anti-destruction capability of the financial network or medical information network can be achieved. The following is an explanation through specific embodiments.

[0025] In one embodiment, Figure 1 FIG. 1 is a flowchart of a method for identifying key nodes in a network disclosed in the present invention. The method is applicable to electronic devices with data processing capabilities, such as mobile phones, tablet computers, laptops, personal computers, and servers. In this embodiment, the method specifically includes the following steps: S101: Obtain a target network structure, where the target network structure is constructed based on node data and edge data in a target information network.

[0026] In this embodiment, the target network is a dynamic and complex financial information network in the digital financial business sector, constructed by the close interweaving of multiple entities such as financial institutions, investors, and markets through transaction relationships and investment chains. Alternatively, the target network can be a vast medical information network in the digital healthcare business sector, where core elements such as medical institutions, patients, and medical data are interconnected through information technology. Node and edge data from the medical information network or financial information network is extracted to construct the medical network structure and the financial network structure. The target network is not limited in this embodiment.

[0027] The node data in this embodiment includes at least node identification, node attributes, and status information, among which the node identification can be composed of characters such as numbers and / or letters, and is used to distinguish different network nodes in the target network structure. The node attributes can be the asset scale, business scope, etc. of the financial institution in the financial network, and the status information can be the risk level of the financial institution in the financial network, etc.

[0028] The edge data in this embodiment includes data such as the starting node, the ending node, the weight, and the directionality, among which the connection relationship between the starting node and the ending node is called an edge, the starting node and the ending node are the two endpoints of the edge, the size of the weight is used to represent the strength or frequency of the relationship between the starting node and the ending node, and the directionality is used to characterize the unidirectionality from the starting node to the ending node.

[0029] In a specific implementation, in this embodiment, after obtaining the node data and edge data in the target information network, the target network structure can be constructed based on the node data and edge data. Figure 2 The figure shows a schematic diagram of a target network structure in the present application, wherein A, B, C, D, E, and F are different network nodes, and the connection between two network nodes is an edge. The edge has directionality. Taking network nodes A and D as an example, according to the directionality of the edge, it can be known that network node A is the starting node and network node D is the ending node. For network node A, the edge is an out-degree edge, and for network node B, the edge is an in-degree edge. The weight of the edge between network nodes A and network nodes D is 10. Therefore, in this embodiment, after obtaining the node data and edge data in the target information network, the target network structure can be constructed based on the node data and edge data.

[0030] S102: Acquire key parameters of each network node in the target network structure respectively, where the key parameters include degree value data, matrix value data, and path data of the network node; the path data includes the number of shortest paths passing through the network node in the target network structure.

[0031] The degree data in this embodiment is an important indicator for measuring the connection characteristics of network nodes. The degree data at least includes the out-degree value, in-degree value and node degree value of the corresponding network node, where the node degree value refers to the sum of the out-degree value and the in-degree value.

[0032] The matrix value data in this embodiment is the characteristic values ​​of the network nodes extracted from the network matrix after the network matrix is ​​extracted from the target network structure.

[0033] The path data in this embodiment is the number of shortest paths between any two network nodes other than the current network node passing through the current network node. That is, the greater the number of shortest paths passing through the current network node, the greater the value of the path data of the current network node.

[0034] S103: Inputting the key parameters corresponding to each network node into the key node prediction model to obtain the importance index of each network node.

[0035] In a specific implementation, the key node prediction model in this embodiment can be any one of a BP neural network model, a convolutional neural network model, and an XGBoost model. After obtaining the key parameters of each network node, the key parameters are respectively input into the key node prediction model for prediction to obtain an importance index for each network node. The importance index can be the probability score of each network node output by the key node prediction model, with the probability score used as the importance index of the network node. Alternatively, the probability score can be adaptively processed according to actual needs to obtain the required importance index in a certain format. For example, a grading rule can be obtained based on the probability score, such as (0.9, 1.0) being the first level, i.e., the highest level, (0.8, 0.9) being the second level, (0.7, 0.8) being the third level, and so on. Based on the grading rule and the probability score corresponding to each network node, the corresponding level of each network node is obtained. It should be noted that the above is only one method of adaptively processing the probability score according to actual needs in this embodiment, and this embodiment does not limit the specific method of adaptively processing the probability score according to actual needs.

[0036] S104: According to the numerical values ​​of the importance indicators, key nodes in the target network structure are screened out from the network nodes.

[0037] In a specific implementation, this embodiment can screen out one or more key nodes from network nodes based on actual needs. When it is necessary to screen out a key node from network nodes, the network node corresponding to the importance index with the largest value is selected as the key node; when it is necessary to screen out multiple key nodes from network nodes, the network nodes whose importance index values ​​are less than a preset index threshold are selected as key nodes, or the network nodes are sorted in descending order according to the values ​​of the importance index, and then a target number of network nodes are selected from large to small according to the values ​​of the importance index as key nodes. Among them, the numerical values ​​of the index threshold and the numerical values ​​of the target number are set according to actual needs and are not specifically limited in this embodiment.

[0038] It is understandable that after filtering out the key nodes in the target network structure from the network nodes, the preset protection strategy is executed to protect the key nodes to improve the invulnerability of the target network. In this embodiment, the protection strategy may include the following strategies: First, the protection strategy in this embodiment may be: by establishing a redundant backup node for each key node, it is ensured that when the primary node fails, the backup node can quickly take over, thereby improving the anti-destruction capability of the target network.

[0039] Secondly, the protection strategy in this embodiment may be: improving the hardware performance of key node devices, such as deploying servers and network devices at key nodes, so as to improve the anti-destruction capability of the target network.

[0040] Thirdly, the protection strategy in this embodiment can be: establishing a complete error detection and automatic recovery mechanism at key nodes, monitoring the traffic and load of key nodes in real time, promptly discovering anomalies and repairing them, thereby improving the target network's anti-destruction capability.

[0041] It should be noted that the above three key node protection strategies are only part of the implementation methods in this embodiment, and this embodiment does not limit the specific implementation methods of the protection strategies.

[0042] In summary, the present invention provides a method for identifying key nodes in a network, by obtaining a target network structure, the target network structure is constructed based on the node data and edge data in the target information network, and the key parameters of each network node in the target network structure are obtained respectively, the key parameters include the degree value data, matrix value data and path data of the network node, the path data includes the number of shortest paths passing through the network node in the target network structure, the key parameters corresponding to each network node are input into the key node prediction model, and the importance index of each network node is obtained. According to the numerical value of the importance index, the key nodes in the target network structure are screened out from the network nodes. It can be seen that in this application, key nodes are identified from the network nodes of the target network structure based on key parameters such as the degree value data, matrix value data and path data of the network nodes. Compared with the prior art, the key nodes identified from the network by a single importance index have a stronger comprehensive influence in the network. After protecting the key nodes identified in this application, the purpose of improving the anti-destruction capability of the financial network or medical information network can be achieved.

[0043] In one embodiment, Figure 3 As shown, the matrix value data in this embodiment can be obtained in the following way: S301: Construct a target matrix corresponding to the target network structure.

[0044] In a specific implementation, this embodiment can extract the in-degree edges and out-degree edges that represent the connection relationship between network nodes from the target network structure, as well as the weights corresponding to the in-degree edges and the weights corresponding to the out-degree edges, and construct a target matrix corresponding to the target network structure based on the in-degree edges and the out-degree edges, as well as the weights corresponding to the in-degree edges and the weights corresponding to the out-degree edges.

[0045] In one embodiment, based on Figure 2 The target matrix constructed by the target network structure shown can be shown as follows: Among them, each column from left to right in the target matrix corresponds to the out-degree edges and corresponding weights of network nodes A, B, C, D, E, and F, and each column from top to bottom in the target matrix corresponds to the in-degree edges and corresponding weights of network nodes A, B, C, D, E, and F.

[0046] For example, taking the first column of the target matrix as an example, three zeros represent that there is no connection between network node A and itself, network node D, and network node E. That is, two network nodes without a connection relationship are recorded as 0 in the target matrix; 5, 8, and 10 respectively represent that there is a connection between network node A and network node B, network node C, and network node F, with network node A as the starting node and network node B, network node C, and network node F as the ending nodes, and the weights are 5, 8, and 10, respectively. Similarly, the target matrix can be constructed based on the connection relationships and weights between network nodes in the target network structure.

[0047] For example, taking the first row in the target matrix as an example, since in the target network structure, network node A is only a starting node and is not the end node of any network node, all values ​​in the first row are zero.

[0048] For example, taking the second row of the target matrix as an example, the three zeros represent that network node B has no connection relationship with itself, network node E, and network node F. That is, two network nodes without a connection relationship are recorded as 0 in the target matrix. 5, 9, and 15 represent that network node B has a connection relationship with network node A, network node C, and network node D, respectively, with network node B as the ending node and network node A, network node C, and network node D as the starting nodes, with weights of 5, 9, and 15, respectively. Similarly, a target matrix can be constructed based on the connection relationships and weights between network nodes in the target network structure.

[0049] S302: Normalize the target matrix to obtain a standard matrix.

[0050] In a specific implementation, in this embodiment, the target matrix can be normalized by any one of Min-Max Normalization and Z-score normalization to obtain a standard matrix. The specific method of normalizing the target matrix is ​​not limited in this embodiment.

[0051] Among them, the target matrix is ​​normalized by minimum-maximum normalization, and the normalization formula can be shown as follows: in, Indicates the target matrix Row, No. Elements of the column, Indicates the standard matrix Row, No. Elements of the column, Indicates the target matrix The minimum value of the column, Indicates the target matrix The maximum value of the column.

[0052] S303: Extracting matrix value data corresponding to each network node from the standard matrix.

[0053] The matrix value data corresponding to the network node may be eigenvalues ​​extracted from a standard matrix.

[0054] In a specific implementation, in this embodiment, the matrix value data corresponding to each network node can be extracted from the standard matrix through linear algebra methods, such as power iteration method, QR algorithm, etc.

[0055] In one embodiment, Figure 4 As shown, the path data in this embodiment can be obtained in the following way: S401: Divide all network nodes in the target network structure into node combinations including two network nodes, where different node combinations include different network nodes.

[0056] All network nodes in the target network structure are arranged and combined to form multiple node combinations. Different node combinations contain different network nodes, that is, any two node combinations contain different network nodes, and the two network nodes in the same node combination are different.

[0057] For example, taking six network nodes A, B, C, D, E, and F as an example, after permutations and combinations of the six network nodes, we can get 15 permutations and combinations: (A, B), (A, C), (A, D), (A, E), (A, F), (B, C), (B, D), (B, E), (B, F), (C, D), (C, E), (C, F), (D, E), (D, F), and (E, F).

[0058] S402: Obtain the shortest path between two network nodes in each node combination respectively.

[0059] Obtain all paths between two network nodes in each node combination and select the path with the least intermediate nodes, that is, the shortest path.

[0060] refer to Figure 2As shown, taking the node combination (A, B) as an example, the paths from network node A to network node B include (A→B) and (A→C→B). In the path (A→B), network nodes A and network node B are directly connected. Therefore, path (A→B) is the shortest path from network node A to network node B. Taking the node combination (B, E) as an example, network nodes B and network nodes E only have an in-degree edge. Therefore, network nodes B and network nodes E cannot be connected, that is, there is no path between network nodes B and network nodes E. Similarly, the shortest path between two network nodes in each node combination can be obtained separately.

[0061] S403: extracting intermediate nodes from all shortest paths, and recording the number of times the same intermediate node is extracted as path data of the network node.

[0062] The intermediate nodes are network nodes other than the starting node and the ending node in the path.

[0063] For a shortest path, if there are only two network nodes on the shortest path, namely the starting node and the ending node, there is no need to record them and obtain the next shortest path. If there are more than two network nodes on the shortest path, extract the intermediate nodes on the shortest path, and so on, extract the intermediate nodes in all shortest paths. It should be noted that in the final extraction result, the number of times the same intermediate node is extracted is recorded. For example, if the same intermediate node is extracted in different shortest paths, the number of times the intermediate node is extracted is recorded. If a network node is not extracted as an intermediate node, the number of times the network node is extracted is recorded as 0, that is, the path data of the network node is 0. If the network node is extracted as an intermediate node more times, the path data value of the network node will be larger.

[0064] In one embodiment, Figure 5 As shown, the degree values ​​in this embodiment include the in-degree value, out-degree value and node degree value of the network node. The degree value data in this embodiment is obtained by the following method: S501: Obtain the number of out-degree edges from the network node to other network nodes in the target network structure as the out-degree value.

[0065] S502: Obtain the number of in-degree edges pointing to the network node as the in-degree value.

[0066] S503: Calculate the node degree value according to the in-degree value and the out-degree value.

[0067] That is, for a network node, the number of out-degree edges pointing from that network node to other network nodes in the target network structure is used as the out-degree value of the network node, and the number of in-degree edges pointing to the network node is used as the in-degree value of the network node. Similarly, the in-degree and out-degree values ​​of all network nodes can be obtained. Furthermore, the sum of the in-degree and out-degree values ​​is used as the node degree value of the network node.

[0068] refer to Figure 2 As shown in the figure, taking network node C as an example, there are three out-degree edges in network node C pointing to network node B, network node D, and network node E respectively. Therefore, the out-degree value of network node C is 3. There is one in-degree edge pointing from network node A to network node C. Therefore, the in-degree value of network node C is 1. The sum of the out-degree value and the in-degree value is 4. In this way, the out-degree value, in-degree value, and node degree value of all network nodes can be obtained.

[0069] In one embodiment, Figure 6 As shown, the degree value in this embodiment also includes a key degree value, which is obtained by the following method: S601: Obtain the weights of all in-degree edges and the weights of all corresponding out-degree edges corresponding to the network node.

[0070] S602: Calculate the criticality value based on the weight and node degree value.

[0071] For a network node, the weights of all in-degree edges and out-degree edges of the network node are obtained, and then the weights and node degrees are input into the critical degree value calculation formula to obtain the critical degree value.

[0072] In one embodiment, the criticality value calculation formula may be as follows: in, represents the node degree value, Indicates the weights, Indicates the criticality value.

[0073] In one embodiment, Figure 7 As shown, the key node prediction model in this embodiment is trained in the following way: S701: Obtain historical degree value data, historical matrix value data, and historical path data to construct training samples, and mark each training sample with an importance label.

[0074] In the specific implementation, this embodiment can extract the historical degree value data, historical matrix value data and historical path data of each historical node from the historical network structure, use the historical degree value data, historical matrix value data and historical path data of each historical node as a training sample, and manually annotate an importance label for each training sample.

[0075] S702: Input the training samples marked with importance labels into the key node prediction model for training to obtain the predicted importance results of the training samples.

[0076] The training samples marked with importance labels are input into the key node prediction model for training, so that the key node prediction model can learn the training samples during the training process and output the prediction importance results.

[0077] S703: Calculate a training loss value based on the importance label and the predicted importance result.

[0078] In a specific implementation, in this embodiment, the training loss value can be calculated based on the importance label and the predicted importance result using methods such as mean square error (MSE) or mean absolute error (MAE). This embodiment does not limit the specific method for calculating the training loss value.

[0079] S704: When the training loss value meets the preset end condition, it is determined that the key node prediction model training is completed.

[0080] S705: When the training loss value does not meet the end condition, adjust the model parameters of the key node prediction model, and iteratively train the key node prediction model based on the training samples until the training loss value meets the end condition, and determine that the training of the key node prediction model is completed.

[0081] In one embodiment, the termination condition in this embodiment can be that when the training loss value is less than a preset loss value threshold, it is determined that the key node prediction model training is completed. Among them, the loss value threshold is set according to actual needs and is not limited in this embodiment.

[0082] In another embodiment, the end condition in this embodiment may be that when the training loss value converges, it is determined that the key node prediction model training is completed.

[0083] It should be noted that the above are only some implementation methods of the termination conditions in this embodiment, and this embodiment does not specifically limit the termination conditions.

[0084] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0085] In one embodiment, a device for identifying key nodes in a network is provided, and the device for identifying key nodes in a network corresponds one-to-one to the method for identifying key nodes in a network in the above embodiment. Figure 8 As shown, the device for identifying key nodes in the network includes a network structure building module, a key parameter acquisition module, an importance index prediction module and a key node screening module. The detailed description of each functional module is as follows: The network structure construction module 801 is used to obtain a target network structure, where the target network structure is constructed based on node data and edge data in the target information network; A key parameter acquisition module 802 is configured to respectively acquire key parameters of each network node in the target network structure, wherein the key parameters include degree data, matrix value data, and path data of the network node; the path data includes the number of shortest paths passing through the network node in the target network structure; The importance index prediction module 803 is used to input the key parameters corresponding to each of the network nodes into a key node prediction model to obtain the importance index of each of the network nodes; The key node screening module 804 is configured to screen out key nodes in the target network structure from the network nodes according to the numerical values ​​of the importance indicators.

[0086] The present invention provides a device for identifying key nodes in a network, which obtains a target network structure based on node data and edge data in a target information network, obtains key parameters of each network node in the target network structure, and inputs the key parameters corresponding to each network node into a key node prediction model to obtain an importance index of each network node. According to the numerical value of the importance index, the key nodes in the target network structure are screened out from the network nodes. It can be seen that the present application identifies key nodes from the network nodes of the target network structure based on key parameters such as the degree data, matrix value data, and path data of the network nodes. Compared with the prior art, the key nodes identified from the network by a single importance index have a stronger comprehensive influence in the network. After protecting the key nodes identified in the present application, the purpose of improving the anti-destruction capability of the financial network or medical information network can be achieved.

[0087] In one embodiment, the matrix value data is obtained as follows: Constructing a target matrix corresponding to the target network structure; Normalizing the target matrix to obtain a standard matrix; Matrix value data corresponding to each of the network nodes is extracted from the standard matrix.

[0088] In one embodiment, the path data is obtained by: dividing all the network nodes in the target network structure into node combinations including two network nodes, wherein different node combinations include different network nodes; respectively obtaining the shortest path between two network nodes in each node combination; All intermediate nodes in the shortest paths are extracted, and the number of times the same intermediate node is extracted is recorded as the path data of the network node.

[0089] In one embodiment, the degree value includes the in-degree value, out-degree value and node degree value of the network node; The degree value data is obtained in the following manner: Obtaining the number of out-degree edges from the network node to other network nodes in the target network structure as the out-degree value; Obtaining the number of in-degree edges pointing to the network node as the in-degree value; The node degree value is calculated according to the in-degree value and the out-degree value.

[0090] In one embodiment, the degree value also includes a critical degree value; The critical value is obtained as follows: Obtaining the weights of all the in-degree edges and the weights of all the corresponding out-degree edges corresponding to the network node; The critical degree value is calculated based on the weight and the node degree value.

[0091] In one embodiment, the key node prediction model is trained in the following manner: Obtain historical degree value data, historical matrix value data, and historical path data to construct training samples, and mark each of the training samples with an importance label; Inputting the training samples marked with the importance labels into the key node prediction model for training to obtain a predicted importance result of the training samples; Calculating a training loss value according to the importance label and the predicted importance result; When the training loss value meets a preset end condition, it is determined that the key node prediction model training is completed.

[0092] When the training loss value does not meet the termination condition, the model parameters of the key node prediction model are adjusted, and the key node prediction model is iteratively trained based on the training samples until the training loss value meets the termination condition, and it is determined that the training of the key node prediction model is completed.

[0093] For the specific definition of the device for identifying key nodes in the network, please refer to the definition of the method for identifying key nodes in the network above, and will not be repeated here. The various modules in the above-mentioned device for identifying key nodes in the network can be implemented in whole or in part by software, hardware, or a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in hardware form, or can be stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to the above modules.

[0094] In one embodiment, a computer device is provided, such as Figure 9 As shown, it includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor. When the processor executes the computer program, the method for identifying key nodes in the network in the above embodiment is implemented, for example Figure 1 Identification of key nodes in the network shown, or Figures 3 to 7 Alternatively, when the processor executes the computer program, the functions of each module / unit in the embodiment of the device for identifying key nodes in the network are realized, for example, Figure 8 The function of identifying key nodes in the network shown is not described here in detail to avoid repetition.

[0095] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the method for identifying key nodes in a network in the above embodiment is implemented, for example Figure 1 Identification of key nodes in the network shown, or Figures 3 to 7 Alternatively, when the processor executes the computer program, the functions of each module / unit in the embodiment of the device for identifying key nodes in the network are realized, for example, Figure 8 The function of identifying key nodes in the network shown is not described here in detail to avoid repetition.

[0096] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A method for identifying key nodes in a network, characterized in that: include: Acquire a target network structure, where the target network structure is constructed based on node data and edge data in a target information network; Respectively obtaining key parameters of each network node in the target network structure, the key parameters including degree data, matrix value data, and path data of the network node; the path data including the number of shortest paths passing through the network node in the target network structure; Inputting the key parameters corresponding to each of the network nodes into a key node prediction model to obtain an importance index of each of the network nodes; According to the numerical value of the importance index, key nodes in the target network structure are screened out from the network nodes.

2. The method for identifying key nodes in a network according to claim 1, characterized in that: The matrix value data is obtained as follows: Constructing a target matrix corresponding to the target network structure; Normalizing the target matrix to obtain a standard matrix; Matrix value data corresponding to each of the network nodes is extracted from the standard matrix.

3. The method for identifying key nodes in a network according to claim 1, characterized in that: The path data is obtained in the following manner: dividing all the network nodes in the target network structure into node combinations including two network nodes, wherein different node combinations include different network nodes; respectively obtaining the shortest path between two network nodes in each node combination; All intermediate nodes in the shortest paths are extracted, and the number of times the same intermediate node is extracted is recorded as the path data of the network node.

4. The method for identifying key nodes in a network according to claim 1, wherein: The degree value includes the in-degree value, out-degree value and node degree value of the network node; The degree value data is obtained in the following manner: Obtaining the number of out-degree edges from the network node to other network nodes in the target network structure as the out-degree value; Obtaining the number of in-degree edges pointing to the network node as the in-degree value; The node degree value is calculated according to the in-degree value and the out-degree value.

5. The method for identifying key nodes in a network according to claim 4, characterized in that: The degree value also includes a key degree value; The critical value is obtained as follows: Obtaining the weights of all the in-degree edges and the weights of all the corresponding out-degree edges corresponding to the network node; The critical degree value is calculated based on the weight and the node degree value.

6. The method for identifying key nodes in a network according to claim 1, characterized in that: The key node prediction model is trained in the following way: Obtain historical degree value data, historical matrix value data, and historical path data to construct training samples, and mark each of the training samples with an importance label; Inputting the training samples marked with the importance labels into the key node prediction model for training to obtain a predicted importance result of the training samples; Calculating a training loss value according to the importance label and the predicted importance result; When the training loss value meets a preset end condition, determining that the key node prediction model training is completed; When the training loss value does not meet the termination condition, the model parameters of the key node prediction model are adjusted, and the key node prediction model is iteratively trained based on the training samples until the training loss value meets the termination condition, and it is determined that the training of the key node prediction model is completed.

7. A device for determining key nodes in a network, characterized in that: include: A network structure construction module is used to obtain a target network structure, where the target network structure is constructed based on node data and edge data in a target information network; a key parameter acquisition module, configured to respectively acquire key parameters of each network node in the target network structure, wherein the key parameters include degree value data, matrix value data, and path data of the network node; the path data includes the number of shortest paths passing through the network node in the target network structure; An importance index prediction module, configured to input the key parameters corresponding to each of the network nodes into a key node prediction model to obtain an importance index of each of the network nodes; The key node screening module is used to screen out the key nodes in the target network structure from the network nodes according to the numerical value of the importance index.

8. The device for determining a key node in a network according to claim 7, wherein: The matrix value data is obtained as follows: Constructing a target matrix corresponding to the target network structure; Normalizing the target matrix to obtain a standard matrix; Matrix value data corresponding to each of the network nodes is extracted from the standard matrix.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method for identifying key nodes in a network according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for identifying key nodes in a network according to any one of claims 1 to 7 is implemented.

Images