Resource Feature Extraction Method, Device, Equipment and Medium Based on Complex Network

Abstract

The present application discloses a method, apparatus, device and medium for resource feature extraction based on complex networks, which relates to the field of complex networks. The method includes: obtaining resource transfer relationships; constructing a resource transfer network according to the resource transfer relationships, where the resource transfer network includes at least two nodes, and the nodes are used to represent user accounts that perform resource transfers, and there are directed edges corresponding between the nodes, and the directed edges are used to represent the resource flow directions in resource transfers; segmenting the resource transfer network according to a preset sub-network structure to obtain at least two segmented sub-network structures, and different preset sub-network structures are used to represent resource transfer relationships of different topological types; obtaining the resource features corresponding to the resource transfer network according to the topological types to which the segmented sub-network structures belong. It is possible to extract comprehensive resource features from the resource transfer network, thereby improving the efficiency and accuracy of identifying abnormal resource transfers.

Description

Technical Field

[0001] The present application relates to the field of complex networks, and particularly to a method, apparatus, device and medium for extracting resource features based on complex networks. Background Art

[0002] A complex network refers to a network that exhibits high complexity, characterized by a huge number of nodes and a variety of different features in the network structure.

[0003] Taking the resource transfer network as an example of a complex network, in the resource transfer network, each node represents a transfer user who conducts resource transfer, and the connection line between nodes represents the resource transfer relationship. Some nodes and some connection lines form a sub-network structure (Motif), and the resource transfer network includes multiple sub-network structures that appear repeatedly. In the related art, using the sub-network structure as a resource feature, by calculating the frequencies of various sub-network structures appearing in the resource transfer network, it is possible to determine whether the resource transfer represented by the sub-network structure is an abnormal resource transfer.

[0004] In the above technical solution, since nodes in different positions may have different topological features, the judgment of whether the resource transfer is abnormal by the computer device is not accurate enough. Summary of the Invention

[0005] Embodiments of the present application provide a method, apparatus, device and medium for extracting resource features based on complex networks. By using a preset sub-network representing resource transfer relationships of different transaction types to divide the resource transfer network, comprehensive resource features can be extracted from the resource transfer network, thereby improving the efficiency and accuracy of identifying abnormal resource transfers. The technical solution includes:

[0006] According to one aspect of the present application, a method for extracting resource features based on complex networks is provided, the method including:

[0007] Obtain resource transfer relationships;

[0008] Construct a resource transfer network according to the resource transfer relationships, the resource transfer network including at least two nodes, the nodes being used to represent user accounts that conduct resource transfer, and there being directed edges corresponding between the nodes, the directed edges being used to represent the resource flow direction in the resource transfer;

[0009] Partition the resource transfer network according to a preset sub-network structure to obtain at least two partitioned sub-network structures, different preset sub-network structures being used to represent resource transfer relationships of different topological types;

[0010] Obtain the resource features corresponding to the resource transfer network according to the topological type to which the partitioned sub-network structure belongs.

[0011] According to another aspect of the present application, there is provided a resource feature extraction device based on a complex network, and the device includes:

[0012] An acquisition module, configured to acquire resource transfer relationships;

[0013] A construction module, configured to construct a resource transfer network according to the resource transfer relationships, where the resource transfer network includes at least two nodes, the nodes are used to represent user accounts for resource transfer, and there are directed edges corresponding between the nodes, and the directed edges are used to represent the fund flow direction in the resource transfer;

[0014] A segmentation module, configured to segment the resource transfer network according to a preset sub-network structure to obtain at least two segmented sub-network structures, and different preset sub-network structures are used to represent resource transfer relationships of different topological types;

[0015] A feature extraction module, configured to obtain the resource features corresponding to the resource transfer network according to the topological type to which the segmented sub-network structure belongs.

[0016] In an optional embodiment, the feature extraction module is configured to calculate the cumulative number of sub-network structures belonging to each topological type among the at least two segmented sub-network structures; and obtain the resource features corresponding to the resource transfer network according to the cumulative number of the sub-network structures belonging to each topological type.

[0017] In an optional embodiment, the resource transfer network includes n nodes, where n is a positive integer;

[0018] The acquisition module is configured to acquire a first set corresponding to the i-th node, where the first set includes a first neighbor node set and a first directed edge set, the first neighbor node set represents a set of first neighbor nodes connected to the i-th node, and the node identifier of the first neighbor node is greater than the node identifier of the i-th node, the first directed edge set represents a set of directed edges corresponding to the directed edges between the i-th node and the first neighbor nodes, i ≤ n, and i is a positive integer; acquire a second set corresponding to the first neighbor node, where the second set includes a second neighbor node set and a second directed edge set, the second neighbor node set represents a set of second neighbor nodes connected to the first neighbor node, and the second directed edge set represents a set of directed edges corresponding to the directed edges between the first neighbor node and the second neighbor nodes; the feature extraction module is configured to repeat the above steps of acquiring the first set and the second set until all the n nodes are traversed, and calculate the cumulative number of sub-network structures belonging to each topological type according to the first set and the second set respectively corresponding to the n nodes.

[0019] In an optional embodiment, the obtaining module is configured to, in response to the intersection set of the first set and the second set being a non-empty set, and the node identifier of the second neighbor node being greater than the maximum node identifier, obtain the directed edges between the ith node, the first neighbor node, and the second neighbor node, where the maximum node identifier is the maximum of the node identifier of the ith node and the node identifier of the first neighbor node; and obtain the sub-network structure formed by the three according to the directed edges between the ith node, the first neighbor node, and the second neighbor node; the feature extraction module is configured to calculate the cumulative number of sub-network structures belonging to each topological type in the sub-network structure formed by the three.

[0020] In an optional embodiment, the obtaining module is configured to, in response to the intersection set of the first set and the second set being an empty set, and the second neighbor node being connected to the first neighbor node and not connected to the ith node, obtain the second node identifier of the second neighbor node and the third node identifier of the ith node; the feature extraction module is configured to, in response to the second node identifier being greater than the third node identifier, obtain the first directed edge between the ith node and the first neighbor node, and the second directed edge between the first neighbor node and the second neighbor node; and obtain the first sub-network structure corresponding to the ith node, the first neighbor node, and the second neighbor node according to the first directed edge and the second directed edge; the feature extraction module is configured to calculate the cumulative number of sub-network structures belonging to each topological type in the first sub-network structure with the ith node as the target node; and calculate the cumulative number of sub-network structures belonging to each topological type in the first sub-network structure with the first neighbor node as the target node.

[0021] In an optional embodiment, the obtaining module is configured to, in response to the second node identifier being less than the third node identifier, obtain the first directed edge between the ith node and the first neighbor node, and the second directed edge between the first neighbor node and the second neighbor node; and obtain the first sub-network structure corresponding to the ith node, the first neighbor node, and the second neighbor node according to the first directed edge and the second directed edge; the feature extraction module is configured to calculate the cumulative number of sub-network structures belonging to each topological type in the first sub-network structure with the ith node as the target node.

[0022] In an alternative embodiment, the obtaining module is configured to, in response to the intersection set of the first set and the second set being an empty set, and the second neighbor node being connected to the ith node and not connected to the first neighbor node, obtain the second node identifier of the second neighbor node and the first node identifier of the first neighbor node; in response to the second node identifier being greater than the first node identifier, obtain a first directed edge between the ith node and the first neighbor node, and a third directed edge between the ith node and the second neighbor node; and obtain a second sub-network structure corresponding to the ith node, the first neighbor node, and the second neighbor node according to the first directed edge and the third directed edge; the feature extraction module is configured to, in the second sub-network structure, calculate the cumulative number of sub-network structures belonging to each topological type with the ith node as the target node; and calculate the cumulative number of sub-network structures belonging to each topological type with the first neighbor node as the target node.

[0023] In an alternative embodiment, the obtaining module is configured to, in response to the second node identifier being less than the first node identifier, obtain a first directed edge between the ith node and the first neighbor node, and a third directed edge between the ith node and the second neighbor node; and obtain a second sub-network structure corresponding to the ith node, the first neighbor node, and the second neighbor node according to the first directed edge and the third directed edge; the feature extraction module is configured to, in the second sub-network structure, calculate the cumulative number of sub-network structures belonging to each topological type with the first neighbor node as the target node.

[0024] In an alternative embodiment, the directed edge between the nodes corresponds to weight information.

[0025] The feature extraction module is configured to obtain the resource features corresponding to the resource transfer network according to the type of the segmented sub-network structure and the weight information corresponding to the directed edge.

[0026] In an alternative embodiment, the feature extraction module is configured to, in the at least two segmented sub-network structures, calculate the cumulative number of sub-network structures belonging to each topological type; calculate the weight value corresponding to the sub-network structures of the same type according to the weight information corresponding to the directed edge; and obtain the resource features corresponding to the resource transfer network according to the cumulative number and the weight value.

[0027] In an alternative embodiment, the resource transfer network includes n nodes, where n is a positive integer.

[0028] The obtaining module is configured to obtain a first adjacent triple set corresponding to the i-th node. The first adjacent triple set includes a first weight set, and the first weight set represents the weight value corresponding to the directed edge between the i-th node and the first neighbor node. Here, i ≤ n and i is a positive integer. The obtaining module is further configured to obtain a second adjacent triple set corresponding to the first neighbor node. The second adjacent triple set includes a second weight set, and the second weight set represents the weight value corresponding to the directed edge between the first neighbor node and the second neighbor node, where the second neighbor node is connected to the first neighbor node. The steps of obtaining the first adjacent triple set and the second adjacent triple set are repeated until all the n nodes are traversed, and the weight values corresponding to the sub-network structures of the same type are calculated according to the weight values corresponding to the directed edges between the n nodes.

[0029] In an optional embodiment, the apparatus includes a calling module.

[0030] The calling module is configured to call a resource classification model to classify the resource features, so as to obtain a prediction probability that the resource transfer belongs to an abnormal resource transfer; and output the type to which the resource transfer belongs according to the prediction probability.

[0031] According to another aspect of the present application, there is provided a computer device, which includes a processor and a memory. At least one instruction, at least one program, a code set or an instruction set is stored in the memory, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement the method for extracting resource features based on a complex network as described in the above aspect.

[0032] According to another aspect of the present application, there is provided a computer-readable storage medium, in which at least one instruction, at least one program, a code set or an instruction set is stored, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by a processor to implement the method for extracting resource features based on a complex network as described in the above aspect.

[0033] According to another aspect of the present application, there is provided a computer program product or a computer program, which includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the method for extracting resources based on a complex network as described in the above aspect.

[0034] The beneficial effects brought by the technical solutions provided in the embodiments of the present application at least include:

[0035] By constructing a resource transfer network, resource features are extracted based on the types of preset sub-network structures to which the sub-network structures in the resource transfer network belong. The resource transfer network is divided by preset sub-network structures representing different types of resource transfer relationships, so that the extracted resource features can accurately represent various resource transfer relationships, thereby improving the accuracy of using the extracted resource features to determine whether there are abnormal resource transfers in the resource transfer network, and at the same time improving the efficiency of identifying abnormal resource transfers. Description of the Drawings

[0036] To more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0037] Figure 1 It is a schematic diagram of a preset sub-network structure provided by an exemplary embodiment of the present application;

[0038] Figure 2 It is a framework diagram of a computer system provided by an exemplary embodiment of the present application;

[0039] Figure 3 It is a flowchart of a method for extracting resource features based on a complex network provided by an exemplary embodiment of the present application;

[0040] Figure 4 It is a system framework diagram for extracting resource features provided by an exemplary embodiment of the present application;

[0041] Figure 5 It is a flowchart of a method for extracting resource features based on a complex network provided by another exemplary embodiment of the present application;

[0042] Figure 6 It is a schematic diagram of resource features provided by an exemplary embodiment of the present application;

[0043] Figure 7 It is a flowchart of a method for extracting resource features based on a complex network provided by another exemplary embodiment of the present application;

[0044] Figure 8 It is a flowchart of a method for extracting resource features based on a complex network provided by another exemplary embodiment of the present application;

[0045] Figure 9 It is a flowchart of a method for extracting resource features based on a complex network provided by another exemplary embodiment of the present application;

[0046] Figure 10It is a flowchart of a resource feature extraction method based on complex network provided by another exemplary embodiment of the present application;

[0047] Figure 11 It is a flowchart of a resource feature extraction method based on complex network provided by another exemplary embodiment of the present application;

[0048] Figure 12 It is a flowchart of a resource feature extraction method based on complex network provided by another exemplary embodiment of the present application;

[0049] Figure 13 It is a flowchart of a resource feature extraction method based on complex network provided by another exemplary embodiment of the present application;

[0050] Figure 14 It is a schematic diagram of resource features provided by another exemplary embodiment of the present application;

[0051] Figure 15 It is a schematic structural diagram of a resource feature extraction device based on complex network provided by an exemplary embodiment of the present application;

[0052] Figure 16 It is a schematic structural diagram of a server provided by an exemplary embodiment of the present application. Detailed implementation manners

[0053] To make the objectives, technical solutions and advantages of the present application clearer, the embodiments of the present application will be further described in detail below with reference to the accompanying drawings.

[0054] First, the nouns related to the embodiments of the present application are introduced.

[0055] Complex Network: It refers to a network presenting high complexity, and this complexity is mainly manifested as follows: 1. The network structure is complex, the number of nodes in the network is huge, and the network structure presents various different characteristics; 2. The connections between nodes may be directional (the connections between nodes are named edges, and the edges with directions are directed edges), the edges between nodes may correspond to weights, and the weights corresponding to the edges of different nodes are different; 3. Nodes can represent anything, for example, a node represents an individual, and the complex network composed of nodes is a human relationship network.

[0056] In the embodiment of the present application, taking the complex network as a resource transfer network as an example, the nodes in the resource transfer network represent user accounts for resource transfer, and there are directed edges between the nodes, where the directed edges represent the flow direction of resources when resource transfer occurs between user accounts. For example, if user account a transfers funds to user account b, the directed edge points from user account a to user account b. Schematically, the directed edges between nodes correspond to weight information, which includes the resource transfer object, the resource transfer type (such as giving virtual gifts or virtual red envelopes representing a certain resource value, transferring resources from one user account to another user account, etc.), the resource transfer value, the resource transfer frequency, the resource transfer time, and the resource transfer message (i.e., the note information filled in by any one of the two parties participating in the resource transfer during the resource transfer process).

[0057] Artificial Intelligence (AI): It is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science. It attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a way similar to human intelligence. Artificial intelligence also studies the design principles and implementation methods of various intelligent machines, enabling the machines to have the functions of perception, reasoning, and decision-making.

[0058] Artificial intelligence technology is an interdisciplinary subject with a wide range of fields, including both hardware-level technologies and software-level technologies. The basic technologies of artificial intelligence generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, and mechatronics. The software technologies of artificial intelligence mainly include several major directions such as computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0059] Machine Learning (ML): It is an interdisciplinary subject involving multiple fields such as probability theory, statistics, approximation theory, convex analysis, and algorithm complexity theory. It specifically studies how computers simulate or implement human learning behaviors to acquire new knowledge or skills and reorganize the existing knowledge structure to continuously improve their own performance. Machine learning is the core of artificial intelligence and the fundamental way to make computers intelligent, and its applications cover all fields of artificial intelligence. Machine learning and deep learning usually include technologies such as artificial neural networks, belief networks, complex networks, reinforcement learning, transfer learning, inductive learning, and learning from demonstration.

[0060] With the research and progress of artificial intelligence technology, artificial intelligence technology has been studied and applied in multiple fields, such as common smart homes, smart wearable devices, virtual assistants, smart speakers, smart marketing, driverless, autonomous driving, drones, robots, smart healthcare, smart customer service, transaction risk monitoring, etc. It is believed that with the development of technology, artificial intelligence technology will be applied in more fields and play an increasingly important role.

[0061] The solution provided in the embodiments of this application relates to a method for extracting resource features based on complex networks, which will be specifically described through the following embodiments.

[0062] In view of the characteristics of resource transfer, the embodiments of this application list 33 preset sub-network structures (Motif), as Figure 1 shown. These 33 preset sub-network structures represent different types of resource transfer. By dividing the resource transfer network through these 33 sub-network structures, the cumulative quantity of sub-network structures belonging to each topological type is calculated to obtain the resource features corresponding to the resource transfer network.

[0063] As Figure 1 shown, the 33 preset sub-network structures (Motif) are numbered. Among them, the 1st, 2nd, and 4th preset sub-network structures each contain two nodes, representing the resource transfer between two user accounts. The black nodes represent the target nodes being analyzed, and the white nodes represent the neighbor nodes connected to the target nodes. The 1st and 2nd preset sub-network structures both represent unidirectional (different directions) resource transfer, and the 4th preset sub-network structure is bidirectional resource transfer.

[0064] The remaining numbered sub-network structures each contain three nodes, that is, the resource transfer among three user accounts. The 3rd to 9th, 15th to 22nd preset sub-network structures contain two directed edges, and the 10th to 13th, 23rd to 33rd preset sub-network structures contain three directed edges.

[0065] The method for extracting resource features based on complex networks provided in the embodiments of this application can be applied to computer devices with strong data processing capabilities. In a possible implementation manner, the method for extracting resource features based on complex networks provided in the embodiments of this application can be applied to personal computers, workstations, or servers, that is, resource features can be extracted from the resource transfer network through personal computers, workstations, or servers. Schematically, the method for extracting resource features based on complex networks is applied to a distributed server.

[0066] Figure 2 Shows a block diagram of a computer system provided by an exemplary embodiment of this application. The computer system 100 includes a first computer device 110 and a second computer device 120.

[0067] The first computer device 110 and the second computer device 120 are two node servers in a distributed server. The network structure diagram 101 corresponding to the resource transfer network is input into the first computer device 110, and the first computer device 110 stores the node information in the resource transfer network. The resource transfer network includes at least two nodes, and the nodes represent user accounts for resource transfer; there are directed edges between the nodes, and the directed edges represent the resource flow direction during resource transfer. Schematically, the first computer device 110 includes a Parameter Server (PS).

[0068] The first computer device 110 divides the input network structure diagram 101 according to a preset sub-network structure, and the preset sub-network structure is 33 representative sub-network structures obtained according to common resource transfer relationship types, that is, 33 types of sub-network structures.

[0069] The second computer device 120 is used to pull node information in a preset batch size (Batch Size), and the batch size refers to the size of the node information and directed edge information pulled by the worker server in each batch. The second computer device 120 divides the pulled node information according to a sub-network structure (Motif) and calculates the cumulative quantity belonging to each type of sub-network structure. Taking the node v i in the resource transfer network as the target node, obtain the adjacency triple set of the target node v i , obtain the sub-network structure corresponding to the target node according to the adjacency triple set, and calculate the cumulative quantity of the sub-network structure corresponding to the target node belonging to each preset sub-network structure of each topology type. And so on, taking all the nodes in the resource transfer network as target nodes to perform the above cumulative quantity calculation, so as to count the cumulative quantity of the sub-network structures that meet each type in the resource transfer network, and thus determine the resource characteristics of the resource transfer network. Schematically, the second computer device 120 includes a Worker.

[0070] The above-mentioned first computer device 110 and second computer device 120 generally refer to one or more computer devices. In the embodiments of the present application, only the first computer device 110 and the second computer device 120 are taken as examples for illustration. The types of the above-mentioned computer devices can be at least one of a laptop computer, a desktop computer, a tablet computer, a server computer, and a workstation computer. The present application does not limit the types of computer devices.

[0071] For the convenience of description, in the following embodiments, the resource feature extraction method based on complex networks is taken as an example to be executed by a parameter server and a worker server.

[0072] Figure 3The flowchart of the resource feature extraction method based on complex network provided by an exemplary embodiment of the present application is shown. In this embodiment, it is described by taking the method being used in the computer system 100 as shown in Figure 2 as an example. The method includes the following steps:

[0073] Step 301, obtain the resource transfer relationship.

[0074] The resource transfer relationship refers to the relationship between the two parties involved in the resource transfer. The resources can be funds (cash), game equipment (such as virtual props), game materials, game pets, game coins, icons (or logos), members, titles, value-added services, points, ingots, golden beans, gift vouchers, exchange vouchers, coupons, greeting cards, etc. The present application does not limit the resource types.

[0075] In the resource transfer network, the user accounts corresponding to the two parties involved in the resource transfer are represented by nodes, and the directed edges between the nodes represent the resource transfer relationship. The resource transfer relationship includes the resource transfer relationship between users and users, the resource transfer relationship between merchants and consumers, and the resource transfer relationship between enterprises and enterprises. In this embodiment of the present application, it is described by taking the resource transfer relationship including the resource flow direction during the resource transfer as an example. For example, if the resource transfer relationship is a transfer relationship between users, and user account a transfers money to user account b, then the directed edge points from the node corresponding to user account a to the node corresponding to user account b.

[0076] In some embodiments, the resource transfer relationship also includes information such as the resource transfer value, resource transfer time, resource transfer message, etc. The resource transfer message refers to the note information filled in by any one of the two parties involved in the resource transfer during the resource transfer. Usually, the resource transfer message is the note information filled in by the resource payer during the resource transfer process.

[0077] Step 302, construct a resource transfer network according to the resource transfer relationship. The resource transfer network includes at least two nodes. The nodes are used to represent the user accounts involved in the resource transfer, and there are corresponding directed edges between the nodes. The directed edges are used to represent the resource flow direction in the resource transfer.

[0078] Let G=(V, E) represent the resource transfer network. Let v i ∈V represent the node corresponding to the user account, and let e ij ∈E represent the directed edge between the nodes, that is, the resource transfer relationship. e ij represents the directed edge between node v i and node v j . If the resource flow direction between node v i and node v j is two-way, that is, the resource flows from node v i to node vj and resources also flow from node v j to node v i , then e ij represents a bidirectional edge. If the resource flow between node v i and node v j is unidirectional, that is, the resources flow from node v i to node v j (or node v j flows to node v i ), then e ij represents a unidirectional edge.

[0079] As Figure 2 shown, the network structure diagram 101 corresponding to the resource transfer network includes 10 nodes, and there are directed edges between the 10 nodes. Schematically, the directed edge between node 1 and node 2 points from node 1 to node 2, that is, the resource flow is from node 1 to node 2, and the directed edge between node 2 and node 4 is bidirectional, that is, the resources flow from node 2 to node 4, and the resources also flow from node 4 to node 2. Schematically, there is no resource transfer relationship between node 6 and node 8, that is, there is no directed edge.

[0080] Step 303, slice the resource transfer network according to a preset sub-network structure to obtain at least two sliced sub-network structures, and different preset sub-network structures are used to represent resource transfer relationships of different topological types.

[0081] As Figure 1 shown, different preset sub-networks represent resource transfer relationships of different topological types. For example, the 33rd preset sub-network represents that there are resource transfer relationships among three nodes, and resources flow between any two nodes.

[0082] Schematically, input the resource transfer network into the first computer device 110, and the first computer device 110 pre-stores as Figure 1For the 33 preset sub-network structures shown, when the first computer device 110 receives an input resource transfer network, it stores the nodes and resource transfer relationships in the resource transfer network. The first computer device 110 also stores the preset sub-network structures. The second computer device 120 pulls a certain amount of node information from the first computer device 110 in a preset batch size (BatchSize) and splits the resource transfer network according to the preset sub-network structures. That is, the second computer device 120 determines the preset sub-network structures included in the resource transfer network. Schematically, taking the network structure diagram 101 as an example, among them, taking the sub-network structure composed of nodes 1, 2, 3, and 4 as an example for illustration. Taking node 1 as the target node, the network structures corresponding to node 1 and node 2 conform to the No. 1 preset sub-network structure, the network structures corresponding to node 1 and node 3 conform to the No. 1 preset sub-network structure, and the network structures corresponding to node 1 and node 4 also conform to the No. 1 preset sub-network structure. The network structures corresponding to node 1, node 2, and node 3 conform to the No. 25 preset sub-network structure, and the network structures corresponding to node 1, node 2, and node 4 conform to the No. 25 preset sub-network structure. In this sub-network structure, there are two topological types of sub-network structures (No. 1 and No. 25) with node 1 as the target node.

[0083] Schematically, taking node 2 as the target node, the network structures corresponding to node 1 and node 2 conform to the No. 2 preset sub-network structure, the network structures corresponding to node 2 and node 3 conform to the No. 14 preset sub-network structure, and the network structures corresponding to node 2 and node 4 also conform to the No. 14 preset sub-network structure. In this sub-network structure, there are two topological types of sub-network structures (No. 2 and No. 14) with node 2 as the target node.

[0084] Schematically, taking node 3 as the target node, the network structures corresponding to node 1 and node 3 conform to the No. 2 preset sub-network structure, and the network structures corresponding to node 2 and node 3 conform to the No. 14 preset sub-network structure. In this sub-network structure, there are two topological types of sub-network structures (No. 2 and No. 14) with node 3 as the target node.

[0085] Similarly, in the above manner, when taking node 4 as the target node, the first computer device 110 determines the topological types existing in this sub-network structure. And so on, taking each node in the resource transfer network as the target node in the above manner, the topological types in the sub-network structure corresponding to each target node are determined, so as to obtain the topological types existing in the resource transfer network.

[0086] Step 304, obtain the resource characteristics corresponding to the resource transfer network according to the topological types to which at least two sub-network structures after splitting belong.

[0087] Schematically, according to the implementation manner shown in step 303, when node 4 is the target node, the network structures corresponding to node 1 and node 4 conform to the 2nd preset sub-network structure, and the network structures corresponding to node 2 and node 4 conform to the 14th preset sub-network structure. In this sub-network structure, there are two topological types of sub-network structures (the 2nd and the 14th) with node 4 as the target node.

[0088] It can be seen from this that in this sub-network structure, the sub-network structures belonging to the 2nd topological type and the 14th type account for a relatively high proportion. Therefore, the second computer device 120 determines that both the 2nd and the 14th are resource characteristics in this sub-network structure. Similarly, the second computer device 120 determines the topological types corresponding to other segmented sub-network structures in the resource transfer network, so as to determine the resource characteristics corresponding to the entire resource transfer network.

[0089] Resource characteristics refer to the characteristics used to describe the resource transfer process, and the resource characteristics include at least one of the following characteristics: the characteristics corresponding to the user account for resource transfer, the characteristics corresponding to the resource transfer value in the resource transfer process, the characteristics corresponding to the remarks information in the resource transfer process, and the characteristics corresponding to the resource transfer type.

[0090] Schematically, taking the resource transfer as a fund transaction as an example, transfer characteristics can be extracted according to the topological types to which at least two segmented sub-network structures belong. The resource classification model classifies the transfer characteristics and outputs that the transaction type is a transfer transaction between users.

[0091] In summary, the method provided in this embodiment extracts resource characteristics by constructing a resource transfer network and using the types of preset sub-network structures to which the sub-network structures in the resource transfer network belong. The resource transfer network is divided by preset sub-network structures representing different types of resource transfer relationships, so that the extracted resource characteristics can accurately represent various resource transfer relationships, thereby improving the accuracy of using the extracted resource characteristics to determine whether there are abnormal resource transfers in the resource transfer network, and at the same time improving the efficiency of identifying abnormal resource transfers.

[0092] Figure 4 Fig. shows the system framework diagram for extracting resource characteristics provided by an exemplary embodiment of the present application. The network structure diagram 101 corresponding to the resource transfer network is input into the parameter server 130. The nodes in the resource transfer network represent the user accounts for resource transfer, and the directed edges between the nodes represent the resource flow directions in the resource transfer process, and the node identifiers are represented by node numbers.

[0093] The parameter server (ParameterServer, PS) 130 stores the node information and directed edge information in the resource transfer network. The parameter server 130 also pre-stores, for exampleFigure 1 The preset sub-network structure shown. Different preset sub-network structures represent different types of resource transfer relationships, and any node in the resource transfer network can be used as the target node. Schematically, the parameter server 130 is a server node in a distributed machine learning framework (Angel), which is used to reasonably split a high-dimensional model into multiple parameter server nodes to implement an efficient machine learning algorithm. When inputting a relatively large-scale resource transfer network, the resource transfer network is split, and the split sub-networks are respectively stored on different parameter server nodes.

[0094] In the parameter server 130, there are several grids near each node, and each grid includes numbers. Taking the target node as 1 for illustration. Among the 6 numbers in the 6 grids near node 1, the first three numbers represent the node identifiers of the neighbor nodes of node 1, and the last three numbers represent the indication directions of the directed edges between the target node and the neighbor nodes. Taking 0 to represent the directed edge from the target node to the neighbor node, 1 to represent the directed edge from the neighbor node to the target node, and 2 to represent the bidirectional edge between the target node and the neighbor node. The neighbor nodes connected to the target node include node 2, node 3, and node 4. The directed edges between node 1 and these three neighbor nodes are all from node 1 to the neighbor nodes. Thus, it can be known that the directed edges corresponding to node 2, node 3, and node 4 are all represented by 0.

[0095] Schematically, the parameter server 130 sends the node information and directed edge information in the resource transfer network to the computing layer logic 102. The worker server 140 obtains the node information and directed edge information from the computing logic layer and extracts resource features based on this information. Schematically, the worker server 140 is a server node (Spark) in a distributed machine learning framework.

[0096] Schematically, the parameter server 130 directly sends the node information and directed edge information in the resource transfer network to the worker server 140, and the worker server 140 extracts resource features based on the node information and directed edge information.

[0097] In some embodiments, the resource transfer network is input into both the parameter server 130 and the worker server 140 at the same time. The parameter server 130 stores the node information and directed edge information of the resource transfer network, and the worker server 140 determines the neighbor node information and directed edge information corresponding to the target node and stores this information.

[0098] The working server 140 is used to pull the required node information and directed edge information in a preset batch size (BatchSize) to complete the resource feature extraction process. The batch size refers to the size of the node information and directed edge information pulled by the working server in each batch. Schematically, the BatchSize is determined comprehensively by balancing factors such as the computing performance, data scale, and cluster resources among the servers. Schematically, the working server 140 adopts a unified BatchSize for the entire resource transfer network; Schematically, the working server 140 adopts different BatchSizes for different types of structures according to the structure of the resource transfer network; Schematically, the working server 140 adopts different BatchSizes for resource transfer relationships of different importance according to the importance of the resource transfer relationships in the resource transfer network. For example, in the resource transfer network, the importance of the resource transfer relationship between enterprises > the importance of the resource transfer relationship between merchants and consumers > the importance of the resource transfer relationship between users and users. Therefore, when the working server 140 reads the resource transfer relationship between enterprises, it can pull the node information through BatchSize1; when the working server 140 reads the resource transfer relationship between users, it can pull the node information through BatchSize2, and BatchSize1 is greater than BatchSize2, that is, the working server 140 pulls more node information through BatchSize1 than through BatchSize2.

[0099] In the working server 140, there are several grids near each node, and each grid includes a number or does not include a number. Taking node 4 as the target node for illustration. The neighbor nodes of node 4 are node 1, node 2, and node 6. The working server 140 determines the neighbor node whose node identifier is greater than the node identifier of node 4 (using the serial number of the node to represent the node identifier), and this node is node 6. The resource transfer relationship between node 4 and node 6 is that the resource flow direction is from node 4 to node 6. Therefore, a directed edge between node 4 and node 6 is represented by 0. Then, the working server 140 will take node 4 as the target node, and in the sub-network structure corresponding to node 4 and node 6, the cumulative quantity of the sub-network structure belonging to the 1st topological type is incremented by 1.

[0100] And so on, until all nodes in the resource transfer network are traversed, and the cumulative quantity of the sub-network structure belonging to each topological type is counted to obtain the subnet (Motif) calculation result 103 corresponding to the resource transfer network.

[0101] Figure 5 Shows a resource feature extraction method based on a complex network provided by another exemplary embodiment of the present application. In this embodiment, this method is used as Figure 2Taking the computer system 100 shown as an example, the method includes the following steps:

[0102] Step 501, obtain the resource transfer relationship.

[0103] Schematically, as Figure 4 shown, the system framework is connected to the background server of the application program. The application program is an application program that supports resource transfer, and the application program includes at least one of an instant messaging program, a payment application program, a shopping application program, and a food delivery application program. The resource transfer relationship is obtained through the background server of this type of application program. The resource transfer relationship includes the resource transfer relationship between users, the resource transfer relationship between merchants and consumers, and the resource transfer relationship between enterprises. Among them, the resource transfer relationship between users includes transfer relationships and gift relationships, such as sending virtual item packages or virtual red envelopes to give funds to others.

[0104] Step 502, construct a resource transfer network according to the resource transfer relationship. The resource transfer network includes at least two nodes. The nodes are used to represent user accounts for resource transfer, and there are directed edges corresponding between the nodes. The directed edges are used to represent the resource flow direction in resource transfer.

[0105] Schematically, the background server of this type of application program sends the resource transfer relationship to the server for constructing the resource transfer network, and the resource transfer network is constructed through this server.

[0106] Schematically, the background server of the application program constructs a resource transfer network according to the obtained resource transfer relationship.

[0107] The resource transfer network constructed according to the resource transfer relationship is as Figure 4 shown in the network structure diagram 101 in

[0108] Step 503, segment the resource transfer network according to a preset sub-network structure to obtain at least two segmented sub-network structures. Different preset sub-network structures are used to represent different topological types of resource transfer relationships.

[0109] In some embodiments, as Figure 4 shown, the system framework includes multiple parameter server nodes and multiple worker server nodes. Due to the large scale of the resource transfer network, before inputting the resource transfer network into the parameter server and the worker server, it is necessary to first segment the resource transfer network. Each parameter server node stores a part of the segmented sub-network structure, and each worker server node processes a part of the segmented sub-network structure. For example, the parameter server node stores the node information of nodes 1 to 1000, and the worker server processes the nodes of nodes 1500 to 2000.

[0110] Each working server node further divides the split sub-network structure according to the preset sub-network structure, that is, determines the sub-network structures belonging to each topology type in the split sub-network structure.

[0111] Step 504: Calculate the cumulative quantity of the sub-network structures belonging to each topology type among at least two divided sub-network structures.

[0112] Taking the working process of one working server node as an example for illustration, as Figure 6 shown. Figure 6 The sub-network structure in Figure 1 is a partial sub-network structure in the resource transfer network. Divide the sub-network structure 21 according to the preset sub-network structure as shown in Figure 6 The cumulative quantity of the sub-network structures belonging to each topology type after division is shown in the table in

[0113] Schematically, taking the target node as node 1, the nodes connected to node 1 include node 2, node 3, and node 4. In the sub-network structures corresponding to node 1, node 2, node 3, and node 4, it can be determined that the sub-network structure contains the topology type of the 10th preset sub-network structure, and there is no other topology type that conforms to the preset sub-network structure. Therefore, the quantity corresponding to the 10th preset sub-network structure is 1.

[0114] And so on, taking each node as the target node, determine the cumulative quantity of the sub-network structures that conform to the topology type of the preset sub-network structure. It can be seen that in the sub-network structure composed of nodes 1 to 6, the cumulative quantity of the sub-network structures belonging to the 10th topology type is 3, and the cumulative quantity of the sub-network structures of this topology type is the largest.

[0115] It can be known from step 503 that there are still many nodes in the split sub-network structure. Therefore, when the working server determines the cumulative quantity of the sub-network structures belonging to each topology type, it needs to batch pull the required node information.

[0116] Schematically, the resource transfer network includes n nodes, where n is a positive integer. Taking the i-th node in the resource transfer network as the target node, calculate the cumulative quantity of the sub-network structures belonging to the topology type through the i-th node and the neighbor nodes connected to the i-th node.

[0117] Step 504 can be replaced by the following steps:

[0118] Step 5041: Obtain the first set corresponding to the i-th node. The first set includes a first neighbor node set and a first directed edge set. The first neighbor node set represents the set of first neighbor nodes connected to the i-th node, and the node identifier of the first neighbor node is greater than that of the i-th node. The first directed edge set represents the set of directed edges between the i-th node and the first neighbor nodes. Here, i ≤ n and i is a positive integer.

[0119] Let G = (V, E) represent the resource transfer network, which includes all nodes v i ∈V. The working server obtains the first set i (corresponding to the i-th node) This first set includes a first neighbor node set and a first directed edge set Among them, the first neighbor node set is represented by In the first directed edge set d ij represents the directed edge between node v i and neighbor node v j (the first neighbor node). When d ij = 0, it means the directed edge points from node v i to neighbor node v j When d ij = 1, it means the directed edge points from node v i to neighbor node v j When d ij = 2, it means a two-way edge between node v i and neighbor node v j .

[0120] Step 5042: Obtain the second set corresponding to the first neighbor node. The second set includes a second neighbor node set and a second directed edge set. The second neighbor node set represents the set of second neighbor nodes connected to the first neighbor node, and the second directed edge set represents the set of directed edges between the first neighbor node and the second neighbor nodes.

[0121] Taking the first neighbor node v j as the target node, obtain the second set j corresponding to the first neighbor node v This second set includes a second neighbor node set and a second directed edge set The second neighbor nodes are the nodes connected to the first neighbor node. In the second directed edge set

[0122] ​​It should be noted that step 5041 can be executed prior to step 5042 or simultaneously with step 5042.

[0123] For step 5043, repeat the above steps of obtaining the first set and the second set until all n nodes are traversed. Based on the first set and the second set corresponding to each of the n nodes, calculate the cumulative quantity of sub-network structures belonging to each topological type.

[0124] Based on the relationship between the first set and the second set, the cumulative quantity of sub-network structures belonging to each topological type can be calculated. Repeat step 5041 and step 5042 to traverse all nodes in the resource transfer network, and calculate the cumulative quantity of sub-network structures belonging to each topological type through the first set and the second set corresponding to all nodes.

[0125] Step 505: Obtain the resource characteristics corresponding to the resource transfer network based on the cumulative quantity of sub-network structures belonging to each topological type.

[0126] Schematically, from Figure 6 it can be seen that the cumulative quantity of sub-network structures belonging to the 10th topological type is the largest. Then, the 10th topological type is taken as the resource characteristic corresponding to the sub-network structure 21. Use this method to calculate the cumulative quantity of the topological types to which the sub-network structures 21 corresponding to all nodes in the entire resource transfer network belong, so as to obtain the resource characteristics corresponding to the resource transfer network.

[0127] Schematically, the cumulative quantities of sub-network structures belonging to the 3rd and 5th topological types are both 2. Then, the resource characteristic corresponding to the sub-network structure 21 is determined by combining the 3rd and 5th topological types. Use this method to obtain the resource characteristics corresponding to the sub-network structures 21 corresponding to all nodes, and thus obtain the resource characteristics corresponding to the resource transfer network by combining the resource characteristics corresponding to all sub-network structures.

[0128] Step 506: Invoke a resource classification model to classify the resource characteristics to obtain the predicted probability that the resource transfer transaction belongs to an abnormal resource transfer.

[0129] The resource classification model is a machine learning model with the ability to classify resource transfer types. The resource classification model is a pre-trained model. Schematically, the resource classification model can be trained by using the resource characteristics extracted in the above embodiments.

[0130] Input the resource features into the resource classification model, and output the predicted probability that the resource transfer belongs to an abnormal resource transfer. Schematically, call the resource classification model to classify the resource gift features. The resource classification model determines that the resource transfer is that user account A sends a certain amount of resources to user account B through a virtual red envelope according to the resource gift features, and outputs the predicted probability that this gift resource behavior belongs to an abnormal resource transfer.

[0131] Step 507, output the resource type of the resource transfer according to the predicted probability.

[0132] Schematically, set a probability threshold for the predicted probability. For example, the probability threshold is 0.6. When the predicted probability output by the resource classification model is 0.75, then this resource transfer belongs to an abnormal resource transfer; when the predicted probability output by the resource classification model is 0.4, then this resource transfer belongs to a normal resource transfer. An abnormal resource transfer refers to a resource transfer with risks, such as a fraud transaction in a financial scenario.

[0133] In summary, the method of this embodiment, by constructing a resource transfer network, extracts resource features based on the type of the preset sub-network structure to which the sub-network structure in the resource transfer network belongs, and divides the resource transfer network through the preset sub-network structures representing different types of resource transaction relationships, so that the extracted transaction resource features can accurately represent various resource transfer relationships, thereby improving the accuracy of using the extracted resource features to judge whether there are abnormal resource transfers in the resource transfer network, and at the same time improving the efficiency of identifying abnormal resource transfers.

[0134] By calculating the cumulative quantity of the sub-network structures belonging to each topological type in the segmented sub-network structure, and determining the resource features based on the cumulative quantity of the sub-network structures distributed under different types of preset sub-network structures, so as to ensure that the resource features extracted from the resource transfer network are accurate and comprehensive.

[0135] By obtaining the first set corresponding to the i-th node and the second set corresponding to the first neighbor node, and calculating the cumulative quantity belonging to each topological type in the sub-network structure according to the first set and the second set, so as to extract comprehensive resource features from the resource transfer network according to the cumulative quantity of the qualified sub-network structures.

[0136] The following describes the process of calculating the cumulative quantity of each topological type in the sub-network structure.

[0137] Based on Figure 5 In an optional embodiment of, when the intersection set of the first set and the second set is a non-empty set, the process for the working server to calculate the cumulative quantity includes the following steps, as Figure 7 shown:

[0138] With vi denotes the i-th node. For node v i ∈V, for each node in V, initialize the feature vectors corresponding to the sub-network structures of 33 dimensions (corresponding to 33 preset sub-network structures) respectively (zero matrix).

[0139] The working server determines each neighbor node of node v i Taking the first neighbor node as v as an example, the working server pulls the second set of node v from the parameter server j Obtain the common neighbor nodes existing between node v j and node v That is, take the intersection of the first set and the second set i and node v j (that is, both connected to node v i and node v j ), that is, take the intersection of the first set and the second set

[0140] Step 701, in response to the intersection set of the first set and the second set being a non-empty set, and the node identifier of the second neighbor node being greater than the maximum node identifier, obtain the directed edge between the i-th node, the first neighbor node, and the second neighbor node. The maximum node identifier is the maximum node identifier among the node identifier of the i-th node and the node identifier of the first neighbor node

[0141] Step 702, according to the directed edge between the i-th node, the first neighbor node, and the second neighbor node, obtain the sub-network structure formed by the three

[0142] Step 703, in the sub-network structure formed by the three, calculate the cumulative quantity of the sub-network structures belonging to each topological type

[0143] Input the node identifiers of the i-th node, the first neighbor node, and the second neighbor node. The working server obtains the first directed edge d between the i-th node and the first neighbor node ij , the second directed edge d between the first neighbor node and the second neighbor node jk , and the third directed edge d between the i-th node and the third neighbor node ik .

[0144] Taking Figure 4 the network structure diagram 101 corresponding to the resource transfer network in as an example, taking node 1 as node v i , then the first neighbor node v j (that is, the nodes in the first neighbor node set ) includes node 2, node 3, and node 4. When taking node 2 as node v j , the second neighbor node v kincluding Node 1, Node 3, and Node 4 (i.e., the second neighbor node set ), when taking Node 3 as node v j , the second neighbor node v k includes Node 2 and Node 8. When taking Node 4 as node v j , the second neighbor node v k includes Node 1, Node 2, and Node 6.

[0145] Use Table 1 to represent the corresponding relationship between node v i and the common neighbor nodes.

[0146] Table 1

[0147]

[0148] It can be seen from Table 1 that the intersection set of the first set and the second set is a non-empty set, that is, there are common neighbor nodes between the i-th node and the first neighbor nodes.

[0149] If commmonNeighbors! = φ, then execute the following logic:

[0150] For node v k ∈commonNeighbors ij , if v k > max(v i , v j ), node v k is the second neighbor node, and execute the following logic:

[0151]

[0152] When node v k is Node 3 or Node 4 and satisfies the above logic, the working server outputs the cumulative number of topological types that conform to the preset sub-network structure topology type in the sub-network structure formed by these three nodes according to the directed edges between node v i , node v j , and node v k . In the embodiments of the present application, the serial numbers of the nodes are used to represent the node identifiers, and the larger the serial number, the larger the node identifier.

[0153] When node v k is Node 3, the working server obtains the directed edges between Node 1, Node 2, and Node 3. Taking Node 1 as the target node, the sub-network structure formed by Node 1, Node 2, and Node 3 conforms to Figure 1 the topological type of the 25th preset sub-network structure in Figure 1If the topological type of the 26th preset sub-network structure in [], then the cumulative quantity for the 26th topological type is incremented by 1; with node 3 as the target node, the sub-network structure formed by nodes 1, 2, and 3 conforms to Figure 1 If the topological type of the 26th preset sub-network structure in [], then the cumulative quantity for the 26th topological type is incremented by 1 again.

[0154] When node v k is node 4, the working server obtains the directed edges between nodes 1, 2, and 4. With node 1 as the target node, the sub-network structure formed by nodes 1, 2, and 4 conforms to Figure 1 If the topological type of the 25th preset sub-network structure in [], then the cumulative quantity for the 25th topological type is incremented by 1; with node 2 as the target node, the sub-network structure formed by nodes 1, 2, and 4 conforms to Figure 1 If the topological type of the 26th preset sub-network structure in [], then the cumulative quantity for the 26th topological type is incremented by 1; with node 4 as the target node, the sub-network structure formed by nodes 1, 2, and 4 conforms to Figure 1 If the topological type of the 26th preset sub-network structure in [], then the cumulative quantity for the 26th topological type is incremented by 1 again.

[0155] From this, it can be seen that when node v k is node 3 or node 4, the cumulative quantity belonging to the 25th topological type is 2, and the cumulative quantity belonging to the 26th topological type is 4.

[0156] According to the above method, when node v i is the target node and node v j is other nodes, in the corresponding sub-network structure, the cumulative quantity of the sub-network structures belonging to each topological type is calculated. Furthermore, the cumulative quantity of the sub-network structures belonging to each topological type in the sub-network structure corresponding to each node in the resource transfer network is calculated, so as to obtain the resource characteristics corresponding to the resource transfer network.

[0157] In summary, for the method of this embodiment, when the intersection set of the first set and the second set is a non-empty set, and the node identifier of the second neighbor node is greater than the maximum node identifier among the i-th node and the first neighbor nodes, in this case, with a certain node in the resource transfer network as the target node, the target node, the first neighbor nodes connected to the target node, and the second neighbor nodes connected to the first neighbor nodes, according to the topological type to which the sub-network structure formed by the three belongs, the cumulative quantity of the sub-network structures under this topological type is calculated, so as to accurately calculate the cumulative quantity of the sub-network structures belonging to each topological type in the resource transfer network.

[0158] As can be seen from the above embodiments, when the intersection set of the first set and the second set is an empty set, that is, there are no common neighbor nodes between node v i and node v j , that is, commonNeighbors ij = φ. There are two cases: 1. Node v k is a neighbor node of node v j , but not a neighbor node of node v i ; 2. Node v k is a neighbor node of node v i , but not a neighbor node of node v j .

[0159] 1. Node v k is a neighbor node of node v j , but not a neighbor node of node v i . The process of the working server calculating the cumulative quantity includes the following steps, as Figure 8 shown:

[0160] Step 801a, in response to the intersection set of the first set and the second set being an empty set, and the second neighbor node being connected to the first neighbor node and not connected to the i-th node, obtain the second node identifier of the second neighbor node and the third node identifier of the i-th node.

[0161] Step 802a, in response to the second node identifier being greater than the third node identifier, obtain the first directed edge between the i-th node and the first neighbor node, and the second directed edge between the first neighbor node and the second neighbor node.

[0162] Step 803a, according to the first directed edge and the second directed edge, obtain the first sub-network structure corresponding to the i-th node, the first neighbor node, and the second neighbor node.

[0163] Step 804a, in the first sub-network structure, taking the i-th node as the target node, calculate the cumulative quantity of the sub-network structures belonging to each topological type; taking the first neighbor node as the target node, calculate the cumulative quantity of the sub-network structures belonging to each topological type.

[0164] Input the node identifiers of the i-th node, the first neighbor node, and the second neighbor node. Among them, the second neighbor node v k is connected to the first neighbor node v j , but not connected to the i-th node v i . The working server obtains the first directed edge d i between node v ij and the first neighbor node, and the second directed edge d jk between the first neighbor node and the second neighbor node.

[0165] Taking Figure 4 the network structure diagram 101 corresponding to the resource transfer network in i as an example, taking node 7 as node v j , the first neighbor node v k is node 9, and the second neighbor node v

[0166] For If v k > v i , execute the following logic:

[0167] index i index j = motifTwoEdgesLookupTable(d ij , d jk )

[0168]

[0169] According to the above logic, the working server obtains the node identifiers of node 7, node 8, and node 9, as well as the first directed edge between node 7 and node 9, and the second directed edge between node 8 and node 9, and outputs the cumulative quantity of the topological types that conform to the preset sub-network structure in the sub-network structure formed by these three nodes.

[0170] Taking the target node as node 7, the sub-network structure formed by node 7, node 8, and node 9 conforms to Figure 1 the topological type of the 3rd preset sub-network structure in Figure 1 , then the cumulative quantity for the 3rd topological type is incremented by 1; taking the target node as node 9, the sub-network structure formed by node 7, node 8, and node 9 conforms to

[0171] Step 801b, in response to the second node identifier being less than the third node identifier, obtain the first directed edge between the i-th node and the first neighbor node, and the second directed edge between the first neighbor node and the second neighbor node.

[0172] Step 802b, according to the first directed edge and the second directed edge, obtain the first sub-network structure corresponding to the i-th node, the first neighbor node, and the second neighbor node.

[0173] Step 803b, in the first sub-network structure, taking the i-th node as the target node, calculate the cumulative quantity belonging to each topological type.

[0174] Taking Figure 4Taking the network structure diagram 101 corresponding to the resource transfer network in [as an example], with node 8 as node v i and node 9 as node v j and node 7 as node v k .

[0175] If v k < v i , execute the following logic:

[0176] index i , index j = motifTwoEdgesLookupTable(d ij , d jk )

[0177]

[0178] The working server obtains the node identifiers of node 7, node 8, and node 9, as well as the first directed edge between node 8 and node 9 and the second directed edge between node 7 and node 9. Taking node 8 as the target node, the sub-network structure formed by node 7, node 8, and node 9 conforms to Figure 1 the 5th preset sub-network structure in [, and the cumulative quantity for the 5th topological type is incremented by 1.

[0179] When taking node 7 and node 9 as the target nodes respectively, it is repeated with the topological types calculated with node 7 and node 9 as the target nodes in steps 801a to 804a above. Therefore, to avoid double counting, the cumulative quantities of the topological types corresponding to taking node 7 and node 9 as the target nodes are no longer calculated.

[0180] To sum up, when the intersection of the first set and the second set is an empty set, and the second neighbor node is connected to the first neighbor node and not connected to the i-th node, in this case, obtain the second node identifier of the second neighbor node and the third node identifier of the i-th node, as well as the directed edge between the connected nodes to determine the sub-network structure formed by the three nodes. Calculate the cumulative quantities of the sub-network structures belonging to each topological type with the i-th node as the target node and the first neighbor node as the target node respectively, so that the results of the cumulative quantities of each topological type are accurate, ensuring that comprehensive and accurate resource characteristics can be extracted from the resource transfer network based on the accurate cumulative quantities.

[0181] When the second node identifier is less than the third node identifier, by calculating the cumulative quantities of the sub-network structures belonging to each topological type in the sub-network structure corresponding to taking the i-th node as the target node, double counting is avoided, so that the results of the cumulative quantities of each topological type are accurate, ensuring that comprehensive and accurate resource characteristics can be extracted from the resource transfer network based on the accurate cumulative quantities.

[0182] 2. Node v k is the neighbor node of node v i but not the neighbor node of node v j . The method for calculating the cumulative quantity includes the following steps, as Figure 9 shown:

[0183] Step 901a: In response to the intersection set of the first set and the second set being an empty set, and the second neighbor node being connected to the i-th node and not connected to the first neighbor node, obtain the second node identifier of the second neighbor node and the first node identifier of the first neighbor node.

[0184] Step 902a: In response to the second node identifier being greater than the first node identifier, obtain the first directed edge between the i-th node and the first neighbor node, and the third directed edge between the i-th node and the second neighbor node.

[0185] Step 903a: According to the first directed edge and the third directed edge, obtain the second sub-network structure corresponding to the i-th node, the first neighbor node, and the second neighbor node.

[0186] Step 904a: In the second sub-network structure, with the i-th node as the target node, calculate the cumulative quantity of the sub-network structures belonging to each topological type; with the first neighbor node as the target node, calculate the cumulative quantity of the sub-network structures belonging to each topological type.

[0187] Input the node identifiers of the i-th node, the first neighbor node, and the second neighbor node. Among them, the second neighbor node v k is connected to the i-th node v i but not connected to the first neighbor node v j . The working server obtains the first directed edge d i between node v ij and the first neighbor node, and the third directed edge d ki between the i-th node and the second neighbor node.

[0188] Taking Figure 4 the network structure diagram 101 corresponding to the resource transfer network in i as an example, with the 7-bit node v j as node 7, the first neighbor node v k as node 9, and the second neighbor node v

[0189] For if v k > v j , then execute the following logic:

[0190] index i index j= motifTwoEdgesLookupTable(d ij , d ki )

[0191]

[0192] According to the above logic, the working server obtains the node identifiers of node 7, node 9, and node 10, as well as the first directed edge between node 7 and node 9 and the third directed edge between node 7 and node 10, and outputs the cumulative number of topological types that conform to the preset sub-network structure in the sub-network structure formed by these three nodes.

[0193] Taking the target node as node 7, the sub-network structure formed by node 7, node 9, and node 10 conforms to Figure 1 the topological type of the 4th preset sub-network structure in Figure 1 , then the cumulative number for the 4th topological type is incremented by 1; taking the target node as node 9, the sub-network structure formed by node 7, node 9, and node 10 conforms to

[0194] the topological type of the 5th preset sub-network structure in k When the second node identifier of the second neighbor node v is less than the first node identifier of the first neighbor node v j , the method for calculating the cumulative number includes the following steps. As Figure 9 shown:

[0195] Step 901b, in response to the second node identifier being less than the first node identifier, obtain the first directed edge between the i-th node and the first neighbor node, and the third directed edge between the i-th node and the second neighbor node.

[0196] Step 902b, based on the first directed edge and the third directed edge, obtain the second sub-network structure corresponding to the i-th node, the first neighbor node, and the second neighbor node.

[0197] Step 903b, in the second sub-network structure, taking the first neighbor node as the target node, calculate the cumulative number of sub-network structures belonging to each topological type.

[0198] Taking Figure 4 the network structure diagram 101 corresponding to the resource transfer network in i as an example, taking node 7 as node v j , node 10 as node v k .

[0199] If v k < v j , then execute the following logic:

[0200] index i ,index j = motifTwoEdgesLookupTable(d ij ,d ki )

[0201]

[0202] The working server obtains the node identifiers of Node 7, Node 9, and Node 10, as well as the first directed edge between Node 7 and Node 10, and the third directed edge between Node 7 and Node 9. Taking Node 10 as the target node, the sub-network structure formed by Node 7, Node 9, and Node 10 conforms to Figure 1 the No. 3 preset sub-network structure in, and the cumulative quantity for the No. 3 topology type is incremented by 1.

[0203] When taking Node 7 and Node 9 as the target nodes respectively, it is repeated with the topology types calculated with Node 7 and Node 9 as the target nodes in Steps 901a to 904a above. Therefore, in order to avoid double counting, the cumulative quantities of the topology types corresponding to Node 7 and Node 9 as the target nodes are no longer calculated.

[0204] To sum up, when the intersection of the first set and the second set is an empty set, and the second neighbor node is connected to the i-th node and not connected to the first neighbor node, in this case, the second node identifier of the second neighbor node, the first node identifier of the first neighbor node, and the directed edge between the connected nodes are obtained to determine the sub-network structure formed by the three nodes. Taking the first neighbor node as the target node, the cumulative quantity of the sub-network structure belonging to each topology type is calculated, so that the result of the cumulative quantity of each topology type is accurate, ensuring that comprehensive and accurate resource features are extracted from the resource transfer network based on the accurate cumulative quantity.

[0205] When the second node identifier is less than the first node identifier, double counting is avoided by calculating the cumulative quantity of the sub-network structure belonging to each topology type in the sub-network structure corresponding to the first neighbor node as the target node, so that the result of the cumulative quantity of each topology type is accurate, ensuring that comprehensive and accurate resource features are extracted from the resource transfer network based on the accurate cumulative quantity.

[0206] In some embodiments, the directed edge between nodes corresponds to weight information, and the working server obtains the resource features corresponding to the resource transfer network according to the type of the sub-network structure after segmentation and the weight information corresponding to the directed edge.

[0207] Figure 10 shows a flowchart of a method for extracting resource features based on a complex network provided by another exemplary embodiment of the present application. In this embodiment, the method is used for such asFigure 2 Taking the computer system 100 shown as an example for illustration, the method includes the following steps:

[0208] Step 1001, obtain the resource transfer relationship.

[0209] Step 1002, construct a resource transfer network according to the resource transfer relationship. The resource transfer network includes at least two nodes, where the nodes are used to represent user accounts for resource transfer, and there are directed edges corresponding between the nodes, and the directed edges are used to represent the resource flow direction in the resource transfer.

[0210] Step 1003, partition the resource transfer network according to a preset sub-network structure to obtain at least two partitioned sub-network structures. Different preset sub-network structures are used to represent resource transfer relationships of different topological types.

[0211] Step 1004, in at least two partitioned sub-network structures, calculate the cumulative quantity of the sub-network structures belonging to each topological type.

[0212] The implementation manners of Step 1001 to Step 1004 are the same as those of Step 501 to Step 504 in the Figure 5 embodiment shown, and will not be elaborated here.

[0213] Step 1005, calculate the weight value corresponding to the sub-network structures belonging to the same type according to the weight information corresponding to the directed edges.

[0214] When calculating the weight information corresponding to the directed edges, Step 1005 can be replaced by the following steps:

[0215] S1. Obtain the first adjacency triple set corresponding to the i-th node. The first adjacency triple set includes a first weight set, and the first weight set represents the weight value corresponding to the directed edge between the i-th node and the first neighbor node. i ≤ n, and i is a positive integer.

[0216] The working server pulls the adjacency triple set of the i-th node v i from the parameter server As can be seen from the above embodiments, represents the first neighbor node set, represents the first directed edge set, represents the set composed of the weight values corresponding to the directed edges between the i-th node and the first neighbor node v j therebetween,

[0217] S2. Obtain the second adjacency triple set corresponding to the first neighbor node. The second adjacency triple set includes a second weight set, and the second weight set represents the weight value corresponding to the directed edge between the first neighbor node and the second neighbor node, and the second neighbor node is connected to the first neighbor node.

[0218] For node v i (the i-th node)'s first neighbor node The working server pulls the first neighbor node v from the parameter server j 's adjacency triple set Denote the second neighbor node set, and the second neighbor node set is the set corresponding to the nodes connected to the first neighbor node. Denote the second directed edge set, and the second directed edge set is the first neighbor node v j and the second neighbor node v k The set corresponding to the directed edge between them. Denote the set composed of the weight values corresponding to the directed edges between the first neighbor node and the second neighbor node.

[0219] S3. Repeat the above steps of obtaining the adjacency triple set until all n nodes are traversed, and calculate the weight values corresponding to the sub-network structures of the same type according to the weight values corresponding to the directed edges between the n nodes.

[0220] Repeat steps S1 and S2 to obtain the adjacency triple set corresponding to each node in the resource transfer network. According to the weight values in the adjacency triple set, calculate the weight values corresponding to the directed edges of the sub-network structures of the same type in the sub-network structure corresponding to each node.

[0221] Step 1006. Obtain the resource characteristics corresponding to the resource transfer network according to the cumulative quantity and the weight value.

[0222] The working server outputs the resource characteristics corresponding to the resource transfer network by combining the cumulative quantity and the weight value of the sub-network structures of each topological type.

[0223] In summary, the method of this embodiment obtains the resource characteristics corresponding to the resource transfer network by obtaining the weight information corresponding to the directed edges in the sub-network structure and combining the topological quantity, making the obtained resource characteristics more comprehensive, accurate and representative, and ensuring that the abnormal resource transfer in the resource transfer network can be accurately identified according to the resource characteristics subsequently.

[0224] By calculating the weight values corresponding to the sub-network structures of the same type according to the weight information corresponding to the directed edges, a comprehensive resource feature is obtained in combination with the topological quantity. The obtained resource feature is more comprehensive, accurate and representative, ensuring that subsequent abnormal resource transfer transactions in the resource transfer network can be accurately identified based on this resource feature.

[0225] By obtaining the first adjacent triple set of the i-th node and the second adjacent triple set of the corresponding first neighbor node, the weight value corresponding to the directed edge between the nodes is calculated according to the weight set, and this calculation step is repeated to obtain the weight value corresponding to the directed edge in the sub-network structure of the same type in the resource transfer network, making the calculated weight value in the resource transfer network more accurate and ensuring that a comprehensive and accurate resource feature can be obtained based on the accurate weight value subsequently.

[0226] Based on Figure 10 In the optional embodiment of, when the intersection set of the first set and the second set is a non-empty set, the process for the working server to extract the resource feature includes the following steps, as Figure 11 shown:

[0227] Let v i represent the i-th node. For the node v i ∈V, for each node in V, the feature vector (zero matrix) corresponding to the sub-network structure of 33 dimensions (corresponding to 33 preset sub-network structures) and the weight feature vector (zero matrix) corresponding to the directed edge are respectively initialized.

[0228] Step 1101, in response to the intersection set of the first set and the second set being a non-empty set and the node identifier of the second neighbor node being greater than the maximum node identifier, obtain the directed edges among the i-th node, the first neighbor node and the second neighbor node. The maximum node identifier is the maximum node identifier among the node identifier of the i-th node and the node identifier of the first neighbor node.

[0229] Step 1102, obtain the sub-network structure formed by the three according to the directed edges among the i-th node, the first neighbor node and the second neighbor node.

[0230] Step 1103, in the sub-network structure formed by the three, calculate the cumulative quantity of the sub-network structures belonging to each topological type and the weight value corresponding to each directed edge.

[0231] In steps 1101 and 1103, the implementation manner of calculating the cumulative quantity of the sub-network structures belonging to each topological type in the sub-network structure is the same as that of Figure 7 the steps 701 to 703 shown, and will not be elaborated here.

[0232] For this case, in the sub-network structure formed by the i-th node, the first neighbor node, and the second neighbor node, there is a directed edge between any two nodes, and the weight value corresponding to the directed edge of the sub-network structure of the same type is calculated. Taking the sub-network structure corresponding to the i-th node as the target node, calculate the weight value corresponding to the directed edge in this sub-network structure; taking the sub-network structure corresponding to the first neighbor node as the target node, calculate the weight value corresponding to the directed edge in this sub-network structure; taking the sub-network structure corresponding to the second neighbor node as the target node, calculate the weight value corresponding to the directed edge in this sub-network structure.

[0233] If commmonNeighbors!= φ, for node v k ∈commonNeighbors ij and v k > max(v i , v j ), node v k is the second neighbor node, and the following logic is executed:

[0234]

[0235] Based on Figure 10 the optional embodiment of, when the intersection set of the first set and the second set is an empty set, the process for the working server to extract resource features includes the following steps, as Figure 12 shown:

[0236] 1. Node v k is a neighbor node of node v j , but not a neighbor node of node v i .

[0237] Step 1201a, in response to the intersection set of the first set and the second set being an empty set, and the second neighbor node being connected to the first neighbor node and not connected to the i-th node, obtain the second node identifier of the second neighbor node and the third node identifier of the i-th node.

[0238] Step 1202a, in response to the second node identifier being greater than the third node identifier, obtain the first directed edge between the i-th node and the first neighbor node, and the second directed edge between the first neighbor node and the second neighbor node.

[0239] Step 1203a, based on the first directed edge and the second directed edge, obtain the first sub-network structure corresponding to the i-th node, the first neighbor node, and the second neighbor node.

[0240] Step 1204a: In the first sub-network structure, taking the i-th node as the target node, calculate the cumulative number of sub-network structures belonging to each topological type, and calculate the weight values corresponding to the sub-network structures of the same type; taking the first neighbor node as the target node, calculate the cumulative number of sub-network structures belonging to each topological type, and calculate the weight values corresponding to the sub-network structures of the same type.

[0241] For If v k > v i , perform the following logic:

[0242] index i , index j = motifTwoEdgesLookupTable(d ij , d jk )

[0243]

[0244] In steps 1201a and 1204a, the implementation manner of calculating the cumulative number of sub-network structures belonging to each topological type in the sub-network structure is the same as that of steps 801a to 804a shown in Figure 8 , and will not be elaborated here.

[0245] For this case, in the sub-network structure formed by the i-th node, the first neighbor node, and the second neighbor node, there is a directed edge between node v i and node v j , and there is a directed edge between node v j and node v k . Calculate the weight values corresponding to the directed edges of the sub-network structures of the same type. For the sub-network structure with the i-th node as the target node, calculate the weight values corresponding to the directed edges in this sub-network structure; for the sub-network structure with the first neighbor node as the target node, calculate the weight values corresponding to the directed edges in this sub-network structure.

[0246] When the second node identifier of the second neighbor node v k is less than the third node identifier of node v i , the resource feature extraction method includes the following steps, as shown in Figure 12 :

[0247] Step 1201b: In response to the second node identifier being less than the third node identifier, obtain the first directed edge between the i-th node and the first neighbor node, and the second directed edge between the first neighbor node and the second neighbor node.

[0248] Step 1202b: Obtain the first sub-network structure corresponding to the i-th node, the first neighbor node, and the second neighbor node according to the first directed edge and the second directed edge.

[0249] Step 1203b: In the first sub-network structure, with the i-th node as the target node, calculate the cumulative number of sub-network structures belonging to each topological type, and calculate the weight values corresponding to the sub-network structures of the same type.

[0250] If v k < v i , execute the following logic:

[0251] index i , index j = motifTwoEdgesLookupTable(d ij , d jk )

[0252]

[0253] In Step 1201b and Step 1203a, the implementation manner of calculating the cumulative number of sub-network structures belonging to each topological type in the sub-network structure is the same as that of Steps 801b to 803b shown in Figure 8 , which will not be elaborated here.

[0254] For this case, in the sub-network structure formed by the i-th node, the first neighbor node, and the second neighbor node, there is a directed edge between node v i and node v j , and there is a directed edge between node v j and node v k . Calculate the weight values corresponding to the directed edges of the sub-network structures of the same type. Calculate the weight values corresponding to the directed edges in the sub-network structure with the i-th node as the target node.

[0255] 2. Node v k is a neighbor node of node v i , but not a neighbor node of node v j . The process for the working server to extract resource features includes the following steps, as shown in Figure 13 :

[0256] Step 1301a: In response to the intersection set of the first set and the second set being an empty set, and the second neighbor node being connected to the i-th node and not connected to the first neighbor node, obtain the second node identifier of the second neighbor node and the first node identifier of the first neighbor node.

[0257] Step 1302a, in response to the second node identifier being greater than the first node identifier, obtain the first directed edge between the i-th node and the first neighbor node, and the third directed edge between the i-th node and the second neighbor node.

[0258] Step 1303a, according to the first directed edge and the third directed edge, obtain the second sub-network structure corresponding to the i-th node, the first neighbor node, and the second neighbor node.

[0259] Step 1304a, in the second sub-network structure, taking the i-th node as the target node, calculate the cumulative number of sub-network structures belonging to each topological type, and calculate the weight value of the sub-network structures belonging to the same type; taking the first neighbor node as the target node, calculate the cumulative number of sub-network structures belonging to each topological type, and calculate the weight value of the sub-network structures belonging to the same type.

[0260] For If v k > v j , then execute the following logic:

[0261] index i , index j = motifTwoEdgesLookupTable(d ij , d ki )

[0262]

[0263] In Step 1301a and Step 1304a, the implementation manner of calculating the cumulative number of sub-network structures belonging to each topological type in the sub-network structure is the same as that of Steps 901a to 904a shown in Figure 9 , and will not be elaborated here.

[0264] For this case, in the sub-network structure formed by the i-th node, the first neighbor node, and the second neighbor node, there is a directed edge between node v i and node v j , and there is a directed edge between node v i and node v k . Calculate the weight value corresponding to the directed edge of the sub-network structures belonging to the same type. For the sub-network structure with the i-th node as the target node, calculate the weight value corresponding to the directed edge in this sub-network structure; for the sub-network structure with the first neighbor node as the target node, calculate the weight value corresponding to the directed edge in this sub-network structure.

[0265] When the second node identifier of the second neighbor node v k is less than that of the first neighbor node v jWhen the first node identifier is used, the method for extracting resource characteristics includes the following steps, as Figure 13 shown:

[0266] Step 1301b, in response to the second node identifier being less than the first node identifier, obtain the first directed edge between the i-th node and the first neighbor node, and the third directed edge between the i-th node and the second neighbor node.

[0267] Step 1302b, according to the first directed edge and the third directed edge, obtain the second sub-network structure corresponding to the i-th node, the first neighbor node, and the second neighbor node.

[0268] Step 1303b, in the second sub-network structure, taking the first neighbor node as the target node, calculate the cumulative number of sub-network structures belonging to each topological type, and calculate the weight value corresponding to the sub-network structures of the same type.

[0269] If v k < v j , then execute the following logic:

[0270] index i index j = motifTwoEdgesLookupTable(d ij d ki )

[0271]

[0272] In steps 1301b and 1303b, the implementation manner of calculating the cumulative number of sub-network structures belonging to each topological type in the sub-network structure is the same as that of steps 901b to 903b shown in Figure 9 , and will not be elaborated here.

[0273] For this case, the sub-network structure formed by the i-th node, the first neighbor node, and the second neighbor node has a directed edge between node v i and node v j , and there is a directed edge between node v i and node v k . Calculate the weight value corresponding to the directed edge of the sub-network structures of the same type. Taking the sub-network structure corresponding to the first neighbor node as the target node, calculate the weight value corresponding to the directed edge in this sub-network structure.

[0274] In summary, the method of this embodiment calculates the weight value corresponding to the sub-network structure composed of the target node (the i-th node), the first neighbor node, and the second neighbor node corresponding to the target node, and combines the cumulative number of sub-network structures of each topological type calculated previously to comprehensively output the resource characteristics corresponding to the resource transfer network, so that subsequent abnormal resource transfers can be accurately identified based on the resource characteristics, and risky resource transfers can be controlled in a timely manner.

[0275] A specific example is used to illustrate the method for calculating the weight value corresponding to the sub-network structure.

[0276] As Figure 14 shown, a sub-network structure 31 is formed among node A, node B, node C, node D, and node E. There is weight information (weight value) corresponding to the directed edges between nodes. By splitting the calculated sub-network structure 31 according to the preset sub-network structure and calculating the corresponding weight values for the split sub-network structures respectively, the weight value corresponding to the sub-network structure 31 is obtained.

[0277] Taking node A as the target node, according to the preset sub-network structure as Figure 1 shown, the sub-network structure 31 is split into a first sub-network structure 32, a second sub-network structure 33, a third sub-network structure 34, and a fourth sub-network structure 35.

[0278] The first sub-network structure 32: It is composed of node A, node B, and node C. The corresponding weight value in the first sub-network structure 32 is: As can be seen from the above embodiment, from the perspective of the direction indicated by the directed edge, the value corresponding to the directed edge AC is 0, the value corresponding to the directed edge BC is 0, and the value corresponding to the directed edge BA is 1.

[0279] The second sub-network structure 33: It is composed of node A, node D, and node C. The corresponding weight value in the second sub-network structure 33 is: As can be seen from the above embodiment, from the perspective of the direction indicated by the directed edge, the value corresponding to the directed edge CD is 0, the value corresponding to the directed edge AC is 0, and the value corresponding to the directed edge DA is 1.

[0280] The third sub-network structure 34: The sub-network structures that satisfy this sub-network structure include the sub-network structure composed of node A, node D, and node E and the sub-network structure composed of node A, node E, and node B. Since the sub-network structures formed by the two cases belong to the same topological type, the weight values calculated for the two cases are added. Since the third sub-network structure 34 includes two directed edges, the weight value corresponding to the third sub-network structure 34 is: As can be seen from the above embodiments, from the perspective of the direction indicated by the directed edge, the value corresponding to the directed edge AE is 2, the value corresponding to the directed edge DA is 1, and the value corresponding to the directed edge BA is 1.

[0281] The fourth sub-network structure 35: It is composed of node A, node C, and node E. Since the fourth sub-network structure 35 includes two directed edges, the corresponding weight value in the fourth sub-network structure 35 is: 12.25 As can be seen from the above embodiments, from the perspective of the direction indicated by the directed edge, the value corresponding to the directed edge AE is 2, and the value corresponding to the directed edge AC is 0.

[0282] It should be noted that in this embodiment, the sub-network structure 31 can also be split into sub-network structures between two nodes. This embodiment only takes the sub-network structure of three nodes as an example for illustration.

[0283] The resource feature extraction method based on complex network provided in the embodiments of the present application is applicable to large-scale resource transfer networks, such as multiple scenarios including transaction fraud, financial credit investigation, merchant profiling, user (consumer) profiling, risk prediction, etc.

[0284] Figure 15 The structural block diagram of an extraction device for resource features based on complex network provided in an exemplary embodiment of the present application is shown. The device includes the following parts:

[0285] The acquisition module 1510 is used to acquire resource transfer relationships;

[0286] The construction module 1520 is used to construct a resource transfer network according to the resource transfer relationships. The resource transfer network includes at least two nodes, where the nodes are used to represent user accounts for resource transfer, and there are corresponding directed edges between the nodes, and the directed edges are used to represent the resource flow direction in the resource transfer;

[0287] The splitting module 1530 is used to split the resource transfer network according to a preset sub-network structure to obtain at least two split sub-network structures. Different preset sub-network structures are used to represent different topological types of resource transfer relationships;

[0288] The feature extraction module 1540 is used to obtain the resource features corresponding to the resource transfer network according to the topological type to which the split sub-network structure belongs.

[0289] In an optional embodiment, the feature extraction module 1540 is used to calculate the cumulative number of sub-network structures belonging to each topological type among at least two split sub-network structures; according to the cumulative number of sub-network structures belonging to each topological type, the resource features corresponding to the resource transfer network are obtained.

[0290] In an alternative embodiment, the resource transfer network includes n nodes, where n is a positive integer;

[0291] The obtaining module 1510 is configured to obtain a first set corresponding to the i-th node. The first set includes a first neighbor node set and a first directed edge set. The first neighbor node set represents a set of first neighbor nodes connected to the i-th node, and the node identifier of the first neighbor node is greater than the node identifier of the i-th node. The first directed edge set represents a set of directed edges corresponding to the directed edges between the i-th node and the first neighbor nodes. i ≤ n, and i is a positive integer. Obtain a second set corresponding to the first neighbor nodes. The second set includes a second neighbor node set and a second directed edge set. The second neighbor node set represents a set of second neighbor nodes connected to the first neighbor nodes. The second directed edge set represents a set of directed edges corresponding to the directed edges between the first neighbor nodes and the second neighbor nodes. The feature extraction module 1540 is configured to repeat the steps of obtaining the first set and the second set until all n nodes are traversed, and calculate the cumulative number of sub-network structures belonging to each topological type according to the first set and the second set corresponding to each of the n nodes.

[0292] In an alternative embodiment, the obtaining module 1510 is configured to, in response to the intersection set of the first set and the second set being a non-empty set, and the node identifier of the second neighbor node being greater than the maximum node identifier, obtain the directed edges between the i-th node, the first neighbor node, and the second neighbor node. The maximum node identifier is the maximum of the node identifier of the i-th node and the node identifier of the first neighbor node. Obtain the sub-network structure formed by the i-th node, the first neighbor node, and the second neighbor node according to the directed edges between the i-th node, the first neighbor node, and the second neighbor node. The feature extraction module 1540 is configured to calculate the cumulative number of sub-network structures belonging to each topological type in the sub-network structure formed by the three.

[0293] In an alternative embodiment, the obtaining module 1510 is configured to, in response to the intersection set of the first set and the second set being an empty set, and the second neighbor node being connected to the first neighbor node and not connected to the i-th node, obtain the second node identifier of the second neighbor node and the third node identifier of the i-th node. In response to the second node identifier being greater than the third node identifier, obtain the first directed edge between the i-th node and the first neighbor node, and the second directed edge between the first neighbor node and the second neighbor node. Obtain the first sub-network structure corresponding to the i-th node, the first neighbor node, and the second neighbor node according to the first directed edge and the second directed edge. The feature extraction module 1540 is configured to calculate the cumulative number of sub-network structures belonging to each topological type in the first sub-network structure with the i-th node as the target node, and calculate the cumulative number of sub-network structures belonging to each topological type with the first neighbor node as the target node.

[0294] In an optional embodiment, the obtaining module 1510 is configured to obtain a first directed edge between the i-th node and a first neighbor node, and a second directed edge between the first neighbor node and a second neighbor node in response to the second node identifier being less than the third node identifier; and obtain a first sub-network structure corresponding to the i-th node, the first neighbor node, and the second neighbor node according to the first directed edge and the second directed edge; the feature extraction module 1540 is configured to calculate the cumulative number of sub-network structures belonging to each topological type with the i-th node as the target node in the first sub-network structure.

[0295] In an optional embodiment, the obtaining module 1510 is configured to obtain a second node identifier of the second neighbor node and a first node identifier of the first neighbor node in response to the intersection set of the first set and the second set being an empty set, and the second neighbor node being connected to the i-th node and not connected to the first neighbor node; obtain a first directed edge between the i-th node and the first neighbor node, and a third directed edge between the i-th node and the second neighbor node in response to the second node identifier being greater than the first node identifier; obtain a second sub-network structure corresponding to the i-th node, the first neighbor node, and the second neighbor node according to the first directed edge and the third directed edge; the feature extraction module 1540 is configured to calculate the cumulative number of sub-network structures belonging to each topological type with the i-th node as the target node in the second sub-network structure; and calculate the cumulative number of sub-network structures belonging to each topological type with the first neighbor node as the target node.

[0296] In an optional embodiment, the obtaining module 1510 is configured to obtain a first directed edge between the i-th node and a first neighbor node, and a third directed edge between the i-th node and a second neighbor node in response to the second node identifier being less than the first node identifier; obtain a second sub-network structure corresponding to the i-th node, the first neighbor node, and the second neighbor node according to the first directed edge and the third directed edge; the feature extraction module 1540 is configured to calculate the cumulative number of sub-network structures belonging to each topological type with the first neighbor node as the target node in the second sub-network structure.

[0297] In an optional embodiment, the directed edges between nodes correspond to weight information.

[0298] The feature extraction module 1540 is configured to obtain resource features corresponding to the resource transfer network according to the type of the segmented sub-network structure and the weight information corresponding to the directed edges.

[0299] In an optional embodiment, the feature extraction module 1540 is configured to calculate the cumulative number of sub-network structures belonging to each topological type in at least two segmented sub-network structures; calculate the weight values corresponding to the sub-network structures of the same type according to the weight information corresponding to the directed edges; and obtain the resource features corresponding to the resource transfer network according to the cumulative number and the weight values.

[0300] In an optional embodiment, the resource transfer network includes n nodes, where n is a positive integer;

[0301] The obtaining module 1510 is configured to obtain a first adjacent triple set corresponding to the i-th node. The first adjacent triple set includes a weight set, and the first weight set represents the weight value corresponding to the directed edge between the i-th node and the first neighbor node, where i ≤ n and i is a positive integer; obtain a second adjacent triple set corresponding to the first neighbor node. The second adjacent triple set includes a second weight set, and the second weight set represents the weight value corresponding to the directed edge between the first neighbor node and the second neighbor node, and the second neighbor node is connected to the first neighbor node; the feature extraction module 1540 is configured to repeat the above steps of obtaining the first adjacent triple set and the second adjacent triple set until all n nodes are traversed, and calculate the weight values corresponding to the sub-network structures of the same type according to the weight values corresponding to the directed edges between the n nodes.

[0302] In an optional embodiment, the apparatus includes a calling module 1550;

[0303] The calling module 1550 is configured to call a resource classification model to classify the resource features, obtain the prediction probability that the resource transfer belongs to an abnormal resource transfer; and output the type to which the resource transfer belongs according to the prediction probability.

[0304] In summary, the apparatus provided in this embodiment constructs a resource transfer network, extracts resource features based on the types of preset sub-network structures to which the sub-network structures in the resource transfer network belong, divides the resource transfer network through preset sub-networks representing different types of resource transfer relationships, so that the extracted resource features can accurately represent various resource transfer relationships, thereby improving the accuracy of using the extracted resource features to determine whether there is an abnormal resource transfer in the resource transfer network, and at the same time improving the efficiency of identifying abnormal resource transfers.

[0305] By calculating the cumulative number of sub-network structures belonging to each topological type in the segmented sub-network structures, and determining the resource transfer features based on the cumulative number of sub-network structures distributed under different types of preset sub-network structures, it is ensured that the resource features extracted from the resource transfer network are accurate and comprehensive.

[0306] By obtaining the first set corresponding to the \(i\)-th node and the second set corresponding to the first neighbor node, and calculating the cumulative quantity belonging to each topological type in the sub-network structure according to the first set and the second set, so as to extract comprehensive resource features from the resource transfer network according to the cumulative quantity of the qualified sub-network structure.

[0307] When the intersection set of the first set and the second set is a non-empty set, and the node identifier of the second neighbor node is greater than the maximum node identifier among the \(i\)-th node and the first neighbor node, in this case, taking a certain node in the resource transfer network as the target node, and based on the sub-network structure formed by the target node, the first neighbor node connected to the target node, and the second neighbor node connected to the first neighbor node, calculate the cumulative quantity of the sub-network structure belonging to each topological type in this topological type, so as to accurately calculate the cumulative quantity of the sub-network structure belonging to each topological type in the resource transfer network.

[0308] When the intersection of the first set and the second set is an empty set, and the second neighbor node is connected to the first neighbor node but not to the \(i\)-th node, in this case, obtain the second node identifier of the second neighbor node and the third node identifier of the \(i\)-th node. When the third node identifier is less than the second node identifier, determine the sub-network structure formed by the \(i\)-th node, the first neighbor node, and the second neighbor node by obtaining the first directed edge and the second directed edge, and then calculate the cumulative quantity of the sub-network structure belonging to each type in this sub-network structure, so as to extract comprehensive resource features from the resource transfer network according to the accurate cumulative quantity.

[0309] When the third node identifier is greater than the second node identifier, in order to avoid double counting, determine the sub-network structure formed by the first neighbor node and the second neighbor node by obtaining the second directed edge, and then calculate the cumulative quantity of the sub-network structure belonging to each topological type in this sub-network structure, so as to extract comprehensive resource features from the resource transfer network according to the accurate cumulative quantity.

[0310] When the intersection of the first set and the second set is an empty set, and the second neighbor node is connected to the \(i\)-th node but not to the first neighbor node, in this case, obtain the second node identifier of the second neighbor node and the first node identifier of the first neighbor node. When the second node identifier is greater than the first node identifier, determine the sub-network structure formed by the \(i\)-th node, the first neighbor node, and the second neighbor node by obtaining the first directed edge and the third directed edge, and then calculate the cumulative quantity of the sub-network structure belonging to each type in this sub-network structure, so as to extract comprehensive resource features from the resource transfer network according to the accurate cumulative quantity.

[0311] When the second node identifier is less than the first node identifier, to avoid double counting, the third directed edge is obtained to determine the sub-network structure formed by the i-th node and the second neighbor node, and then the cumulative quantity of the sub-network structures belonging to each topological type in this sub-network structure is calculated, so that comprehensive resource features can be extracted from the resource transfer network based on the accurate cumulative quantity.

[0312] By obtaining the weight information corresponding to the directed edges in the sub-network structure and combining it with the topological quantity, the resource features corresponding to the resource transfer network are obtained, making the obtained resource features more comprehensive, accurate and representative, and ensuring that subsequent abnormal resource transfers in the resource transfer network can be accurately identified based on these resource features.

[0313] By calculating the weight values corresponding to the sub-network structures of the same type according to the weight information corresponding to the directed edges, comprehensive resource features are obtained in combination with the topological quantity. The obtained resource features are more comprehensive, accurate and representative, and it is ensured that subsequent abnormal resource transfers in the resource transfer network can be accurately identified based on these resource features.

[0314] By obtaining the adjacency triple set of the i-th node, which includes a weight set, the weight value corresponding to the directed edge between the i-th node and its neighbor nodes is calculated according to the weight set. Repeating this step, the weight values corresponding to the directed edges in the sub-network structures of the same type in the resource transfer network are obtained, making the calculated weight values in the resource transfer network more accurate and ensuring that comprehensive resource features can be obtained based on the accurate weight values subsequently.

[0315] It should be noted that: for the resource feature extraction device based on complex network provided in the above embodiments, only the division of the above functional modules is used for illustration. In practical applications, the above functions can be allocated to different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the resource feature extraction device based on complex network provided in the above embodiments and the method embodiments of resource feature extraction based on complex network belong to the same concept, and the specific implementation process is detailed in the method embodiments and will not be repeated here.

[0316] Figure 16 The structural schematic diagram of a server provided by an exemplary embodiment of the present application is shown. The server can be the first computer device 110 and the second computer device 120 of the server in the computer system 100 as Figure 1 shown.

[0317] The server 1600 includes a central processing unit (CPU) 1601, a system memory 1604 including a random access memory (RAM) 1602 and a read only memory (ROM) 1603, and a system bus 1605 connecting the system memory 1604 and the central processing unit 1601. The server 1600 also includes a basic input / output system (I / O system) 1606 for facilitating information transfer between various components within the computer, and a mass storage device 1607 for storing an operating system 1613, application programs 1614, and other program modules 1615.

[0318] The basic input / output system 1606 includes a display 1608 for displaying information and input devices 1609 such as a mouse and a keyboard for user input of information. Both the display 1608 and the input devices 1609 are connected to the central processing unit 1601 through an input / output controller 1610 connected to the system bus 1605. The basic input / output system 1606 may also include the input / output controller 1610 for receiving and processing inputs from a plurality of other devices such as a keyboard, a mouse, or an electronic stylus. Similarly, the input / output controller 1610 also provides outputs to a display screen, a printer, or other types of output devices.

[0319] The mass storage device 1607 is connected to the central processing unit 1601 through a mass storage controller (not shown) connected to the system bus 1605. The mass storage device 1607 and its associated computer-readable medium provide non-volatile storage for the server 1600. That is, the mass storage device 1607 may include a computer-readable medium (not shown) such as a hard disk or a compact disc read only memory (CD-ROM) drive.

[0320] A computer-readable medium can include a computer storage medium and a communication medium. The computer storage medium includes volatile and non-volatile, removable and non-removable media implemented by any method or technology for storing information such as computer-readable instructions, data structures, program modules, or other data. The computer storage medium includes RAM, ROM, erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other solid-state storage technologies, CD-ROM, digital versatile disc (DVD), or solid-state drives (SSD), other optical storage, magnetic tape cartridges, tapes, disk storage, or other magnetic storage devices. Among them, random access memory can include resistive random access memory (ReRAM) and dynamic random access memory (DRAM). Of course, those skilled in the art know that the computer storage medium is not limited to the above several types. The above-mentioned system memory 1604 and mass storage device 1607 can be collectively referred to as memory.

[0321] According to various embodiments of the present application, the server 1600 can also operate by connecting to a remote computer on the network through a network such as the Internet. That is, the server 1600 can be connected to the network 1612 through the network interface unit 1611 connected to the system bus 1605. Or rather, the network interface unit 1611 can also be used to connect to other types of networks or remote computer systems (not shown).

[0322] The above-mentioned memory further includes one or more programs, and one or more programs are stored in the memory and are configured to be executed by the CPU.

[0323] In an alternative embodiment, a computer device is provided. The computer device includes a processor and a memory. At least one instruction, at least one program segment, a code set, or an instruction set is stored in the memory, and at least one instruction, at least one program segment, a code set, or an instruction set is loaded and executed by the processor to implement the resource feature extraction method based on complex networks as described above.

[0324] In an alternative embodiment, a computer-readable storage medium is provided, in which at least one instruction, at least one program, a code set or an instruction set is stored, and the at least one instruction, at least one program, the code set or the instruction set is loaded and executed by a processor to implement the resource feature extraction method based on complex network as described above.

[0325] Optionally, the computer-readable storage medium may include: read-only memory (ROM), random access memory (RAM), solid state drives (SSD), optical discs, etc. Among them, the random access memory may include resistive random access memory (ReRAM) and dynamic random access memory (DRAM). The serial numbers of the embodiments of the present application are only for description and do not represent the advantages and disadvantages of the embodiments.

[0326] The embodiments of the present application further provide a computer program product or a computer program. The computer program product or the computer program includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the resource feature extraction method based on complex network as described in the above aspect.

[0327] Those of ordinary skill in the art can understand that all or part of the steps to implement the above embodiments can be completed by hardware, or can be completed by a program instructing relevant hardware. The program can be stored in a computer-readable storage medium, and the storage medium mentioned above can be read-only memory, a magnetic disk or an optical disc, etc.

[0328] The above are only alternative embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for extracting resource features based on complex networks, characterized in that, the method includes: Obtaining resource transfer relationships; Constructing a resource transfer network according to the resource transfer relationships, the resource transfer network includes at least two nodes, the nodes are used to represent user accounts for resource transfer, and there are directed edges corresponding between the nodes, and the directed edges are used to represent the resource flow direction in the resource transfer; Segmenting the resource transfer network according to a preset sub-network structure to obtain at least two segmented sub-network structures, and different preset sub-network structures are used to represent resource transfer relationships of different topological types; the resource transfer network includes n nodes, and n is a positive integer; In the at least two segmented sub-network structures, through the i-th node and the neighbor nodes connected to the i-th node, calculate the cumulative number of sub-network structures belonging to each topological type, i ≤ n, and i is a positive integer; According to the cumulative number of the sub-network structures belonging to each topological type, obtain the resource features corresponding to the resource transfer network.

2. The method according to claim 1, characterized in that, in the at least two segmented sub-network structures, through the i-th node and the neighbor nodes connected to the i-th node, calculating the cumulative number of sub-network structures belonging to each topological type includes: Obtaining a first set corresponding to the i-th node, the first set includes a first neighbor node set and a first directed edge set, the first neighbor node set represents a set of first neighbor nodes connected to the i-th node, and the node identifier of the first neighbor node is greater than the node identifier of the i-th node, and the first directed edge set represents a set of directed edges corresponding to the directed edges between the i-th node and the first neighbor node; Obtaining a second set corresponding to the first neighbor node, the second set includes a second neighbor node set and a second directed edge set, the second neighbor node set represents a set of second neighbor nodes connected to the first neighbor node, and the second directed edge set represents a set of directed edges corresponding to the directed edges between the first neighbor node and the second neighbor node; Repeat the above steps of obtaining the first set and the second set until all the n nodes are traversed, and calculate the cumulative number of sub-network structures belonging to each topological type according to the first set and the second set respectively corresponding to the n nodes.

3. The method according to claim 2, characterized in that, calculating the cumulative number of sub-network structures belonging to each topological type according to the first set and the second set respectively corresponding to the n nodes includes: In response to the intersection set of the first set and the second set being a non-empty set, and the node identifier of the second neighbor node being greater than the maximum node identifier, obtain the directed edges between the i-th node, the first neighbor node and the second neighbor node, and the maximum node identifier is the maximum of the node identifier of the i-th node and the node identifier of the first neighbor node; Obtain the sub-network structure formed by the i-th node, the first neighbor node, and the second neighbor node according to the directed edges among them. In the sub-network structure formed by the three, calculate the cumulative number of sub-network structures belonging to each topological type.

4. The method according to claim 2, wherein, the calculating the cumulative number of sub-network structures belonging to each topological type according to the first set and the second set respectively corresponding to the n nodes includes: In response to the intersection set of the first set and the second set being an empty set, and the second neighbor node being connected to the first neighbor node and not connected to the i-th node, obtain the second node identifier of the second neighbor node and the third node identifier of the i-th node; In response to the second node identifier being greater than the third node identifier, obtain the first directed edge between the i-th node and the first neighbor node, and the second directed edge between the first neighbor node and the second neighbor node; According to the first directed edge and the second directed edge, obtain the first sub-network structure corresponding to the i-th node, the first neighbor node, and the second neighbor node; In the first sub-network structure, taking the i-th node as the target node, calculate the cumulative number of sub-network structures belonging to each topological type; taking the first neighbor node as the target node, calculate the cumulative number of sub-network structures belonging to each topological type.

5. The method according to claim 4, wherein, the method further includes: In response to the second node identifier being less than the third node identifier, obtain the first directed edge between the i-th node and the first neighbor node, and the second directed edge between the first neighbor node and the second neighbor node; According to the first directed edge and the second directed edge, obtain the first sub-network structure corresponding to the i-th node, the first neighbor node, and the second neighbor node; In the first sub-network structure, taking the i-th node as the target node, calculate the cumulative number of sub-network structures belonging to each topological type.

6. The method according to claim 2, wherein, the calculating the cumulative number of sub-network structures belonging to each topological type according to the first set and the second set respectively corresponding to the n nodes includes: In response to the intersection set of the first set and the second set being an empty set, and the second neighbor node being connected to the i-th node and not connected to the first neighbor node, obtain the second node identifier of the second neighbor node and the first node identifier of the first neighbor node; In response to the second node identifier being greater than the first node identifier, obtain the first directed edge between the i-th node and the first neighbor node, and the third directed edge between the i-th node and the second neighbor node; According to the first directed edge and the third directed edge, obtain the second sub-network structure corresponding to the i-th node, the first neighbor node, and the second neighbor node; In the second sub-network structure, taking the \(i\)-th node as the target node, calculate the cumulative number of sub-network structures belonging to each topological type; taking the first neighbor node as the target node, calculate the cumulative number of sub-network structures belonging to each topological type.

7. The method according to claim 6, wherein, the method further includes: in response to the second node identifier being less than the first node identifier, obtain the first directed edge between the \(i\)-th node and the first neighbor node, and the third directed edge between the \(i\)-th node and the second neighbor node; obtain the second sub-network structure corresponding to the \(i\)-th node, the first neighbor node, and the second neighbor node according to the first directed edge and the third directed edge; in the second sub-network structure, taking the first neighbor node as the target node, calculate the cumulative number of sub-network structures belonging to each topological type.

8. The method according to any one of claims 1 to 7, wherein, the directed edges between the nodes correspond to weight information; the method further includes: obtain the resource characteristics corresponding to the resource transfer network according to the type of the segmented sub-network structure and the weight information corresponding to the directed edges.

9. The method according to claim 8, wherein, the obtaining the resource characteristics corresponding to the resource transfer network according to the type of the segmented sub-network structure and the weight information corresponding to the directed edges includes: in the at least two segmented sub-network structures, calculate the cumulative number of sub-network structures belonging to each topological type; calculate the weight value corresponding to the sub-network structures of the same type according to the weight information corresponding to the directed edges; obtain the resource characteristics corresponding to the resource transfer network according to the cumulative number and the weight value.

10. The method according to claim 9, wherein, the resource transfer network includes \(n\) nodes, \(n\) being a positive integer; the calculating the weight value corresponding to the sub-network structures of the same type according to the weight information corresponding to the directed edges includes: obtain the first adjacent triple set corresponding to the \(i\)-th node, the first adjacent triple set including a first weight set, the first weight set characterizing the weight value corresponding to the directed edge between the \(i\)-th node and the first neighbor node, \(i\leq n\) and \(i\) being a positive integer; obtain the second adjacent triple set corresponding to the first neighbor node, the second adjacent triple set including a second weight set, the second weight set characterizing the weight value corresponding to the directed edge between the first neighbor node and the second neighbor node, the second neighbor node being connected to the first neighbor node; repeat the steps of obtaining the first adjacent triple set and the second adjacent triple set until all the \(n\) nodes are traversed, and calculate the weight value corresponding to the sub-network structures of the same type according to the weight values corresponding to the directed edges between the \(n\) nodes.

11. The method according to any one of claims 1 to 7, wherein, the method further includes: Call a resource classification model to classify the resource features, and obtain the predicted probability that the resource transfer belongs to an abnormal resource transfer; Output the type to which the resource transfer belongs according to the predicted probability.

12. A resource feature extraction device based on a complex network, characterized in that, the device includes: an acquisition module for acquiring resource transfer relationships; a construction module for constructing a resource transfer network according to the resource transfer relationships, the resource transfer network including at least two nodes, the nodes being used to represent user accounts for resource transfer, and there being directed edges corresponding between the nodes, the directed edges being used to represent the resource flow direction in the resource transfer; a splitting module for splitting the resource transfer network according to a preset sub-network structure to obtain at least two split sub-network structures, different preset sub-network structures being used to represent resource transfer relationships of different topological types; the resource transfer network includes n nodes, and n is a positive integer; a feature extraction module for calculating, in the at least two split sub-network structures, the cumulative number of sub-network structures belonging to each topological type through the i-th node and the neighbor nodes connected to the i-th node, where i ≤ n and i is a positive integer; Obtain the resource features corresponding to the resource transfer network according to the cumulative number of the sub-network structures belonging to each topological type.

13. A computer device, characterized in that, the computer device includes a processor and a memory, and at least one instruction, at least one program, a code set or an instruction set is stored in the memory, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement the resource feature extraction method based on a complex network according to any one of claims 1 to 11.

14. A computer-readable storage medium, characterized in that, at least one instruction, at least one program, a code set or an instruction set is stored in the readable storage medium, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by a processor to implement the resource feature extraction method based on a complex network according to any one of claims 1 to 11.

15. A computer program product, characterized in that, the computer program product includes computer instructions, the computer instructions are stored in a computer-readable storage medium, and the processor reads and executes the computer instructions from the computer-readable storage medium to implement the resource feature extraction method based on a complex network according to any one of claims 1 to 11 above.

Images