Consumption resource violation scoring method, device and electronic equipment

Abstract

The present application discloses a method, device and electronic device for scoring consumer resource violations, which belongs to the field of financial technology. The method includes: obtaining resource lending information of consumer resources; extracting a topological feature set from a pre-configured resource flow graph, wherein the resource flow graph is a topological graph generated in reverse from at least one illegal merchant, and the nodes in the resource flow graph include: each illegal merchant, and a resource account having a resource flow relationship with each illegal merchant; the topological feature set includes at least one topological feature associated with the resource lending information; the topological feature set is input into a pre-trained resource violation scoring model to obtain a resource violation score output by the resource violation scoring model, and the resource violation score is used to characterize the probability of the consumer resource flowing into the at least one illegal merchant.

Description

Technical Field

[0001] The present application belongs to the field of financial technology, and specifically relates to a method, device and electronic device for scoring consumer resource violations. Background Art

[0002] With the rapid development of social economy, people's consumption level is constantly improving, and people's demand for consumer resources is increasing. Among them, more and more financial service platforms are gradually opening resource lending functions, that is, financial service platforms lend a certain amount of consumer resources to users who meet the lending conditions, and users only need to repay the loan interest and principal within the loan period, which can not only meet people's demand for consumer resources, but also bring benefits to financial service platforms.

[0003] However, after users borrow consumer resources from financial service platforms, they may input the consumer resources into illegal merchants (such as real estate companies and securities companies, etc.), resulting in large losses when using consumer resources and being unable to fulfill repayment obligations within the loan period. This will create the risk that the resources borrowed by the financial service platform cannot be recovered, which will in turn cause losses to the financial service platform. Therefore, how financial service platforms can evaluate whether the consumer resources borrowed by users have flowed into illegal merchants has become an issue that needs to be solved urgently. Summary of the invention

[0004] The purpose of the embodiments of the present application is to provide a method, device and electronic device for scoring consumer resource violations, which can provide a reference for evaluating whether the consumer resources borrowed by a user flow into illegal merchants.

[0005] In order to solve the above technical problems, this application is implemented as follows:

[0006] In a first aspect, an embodiment of the present application provides a method for scoring consumption resource violations, including:

[0007] Get resource borrowing information of consumption resources;

[0008] A topological feature set is extracted from a pre-configured resource flow graph, wherein the resource flow graph is a topological graph generated in reverse based on at least one illegal merchant, and the nodes in the resource flow graph include: each illegal merchant, and a resource account having a resource flow relationship with each illegal merchant; the topological feature set includes at least one topological feature associated with the resource lending information;

[0009] The topological feature set is input into a pre-trained resource violation scoring model to obtain a resource violation score output by the resource violation scoring model, wherein the resource violation score is used to characterize the probability that the consumer resource flows into the at least one illegal merchant.

[0010] In a second aspect, an embodiment of the present application provides a consumption resource violation scoring device, the device comprising:

[0011] A loan information acquisition module is used to obtain resource loan information of consumption resources;

[0012] A feature extraction module is used to extract a topological feature set from a pre-configured resource flow graph, wherein the resource flow graph is a topological graph generated in reverse based on at least one illegal merchant, and the nodes in the resource flow graph include: each illegal merchant, and a resource account having a resource flow relationship with each illegal merchant; the topological feature set includes at least one topological feature associated with the resource lending information;

[0013] The scoring module is used to input the topological feature set into a pre-trained resource violation scoring model to obtain a resource violation score output by the resource violation scoring model, wherein the resource violation score is used to characterize the probability that the consumer resources flow into the at least one illegal merchant.

[0014] In a third aspect, an embodiment of the present application provides an electronic device, which includes a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction, when executed by the processor, implements the steps of the method described in the first aspect.

[0015] In a fourth aspect, an embodiment of the present application provides a readable storage medium, on which a program or instruction is stored, and when the program or instruction is executed by a processor, the steps of the method described in the first aspect are implemented.

[0016] In the embodiment of the present application, by pre-configuring a resource flow graph that is reversely generated from the illegal merchant and reflects the resource flow relationship with the illegal merchant, when the resource lending information of the consumer resource is obtained, a topological feature set can be extracted from the resource flow graph according to the resource lending information, and the topological feature set can be input into the pre-trained resource violation scoring model, and the resource violation scoring model outputs a resource violation score that characterizes the probability of the above-mentioned consumer resources flowing into the illegal merchant. In this way, the resource violation score can provide a reference for the financial service platform to evaluate whether the consumer resources flow into the illegal merchant, and the accuracy of the evaluation can be guaranteed. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flowchart of an embodiment of a consumption resource violation scoring method provided by the present application;

[0018] Figure 2 It is a partial structural diagram of the resource flow diagram provided by this application;

[0019] Figure 3It is another structural schematic diagram of the resource flow diagram provided by this application;

[0020] Figure 4 It is another structural schematic diagram of the resource flow diagram provided by this application;

[0021] Figure 5 It is a structural diagram of an embodiment of a consumption resource violation scoring device provided by the present application;

[0022] Figure 6 It is a structural diagram of an embodiment of a computing device provided by the present application. DETAILED DESCRIPTION

[0023] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments.

[0024] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are generally of one type, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0025] The following is a detailed description of the consumption resource violation scoring method provided in the embodiment of the present application through specific embodiments and application scenarios in conjunction with the accompanying drawings.

[0026] See also Figure 1 , is a flow chart of a method for scoring consumption resource violations provided in an embodiment of the present application, which is applied to electronic devices including servers, etc. Figure 1 As shown, the consumption resource violation scoring method at least includes the following steps 101 to 103.

[0027] Step 101: Obtain resource lending information of consumption resources.

[0028] Step 102: extract a topological feature set from a preconfigured resource flow graph, wherein the resource flow graph is a topological graph generated in reverse based on at least one illegal merchant, and the nodes in the resource flow graph include: each illegal merchant, and a resource account having a resource flow relationship with each illegal merchant; the topological feature set includes at least one topological feature associated with the resource lending information.

[0029] Step 103: input the topological feature set into a pre-trained resource violation scoring model to obtain a resource violation score output by the resource violation scoring model, wherein the resource violation score is used to characterize the probability that the consumer resources flow into the at least one violating merchant.

[0030] Based on this, by pre-configuring a resource flow graph that is reverse-generated from illegal merchants and reflects the resource flow relationship with illegal merchants, when the resource lending information of consumer resources is obtained, a topological feature set can be extracted from the resource flow graph according to the resource lending information, and the topological feature set can be input into a pre-trained resource violation scoring model, and the resource violation scoring model outputs a resource violation score that characterizes the probability of the above-mentioned consumer resources flowing into illegal merchants. In this way, the resource violation score can provide a reference for the financial service platform to evaluate whether consumer resources flow into illegal merchants, and the accuracy of the evaluation can be guaranteed.

[0031] In the above step 101, when the financial service platform needs to monitor whether the consumer resources it borrows are in violation of regulations, the financial service platform can input the resource borrowing information of the consumer resources into the electronic device, and the electronic device obtains the resource borrowing information.

[0032] The resource lending information may be any information that can be used to indicate the consumption resource. For example, the resource lending information may include the unique code of the consumption resource; or the resource lending information may include the resource account of the user who borrows the consumption resource on the financial service platform, and the borrowing date of the consumption resource, etc.

[0033] It should be noted that the resource lending information of the consumption resource can be obtained by the financial service platform directly providing the resource lending information of the consumption resource to the electronic device; or, the electronic device can search for the resource lending information of the consumption resource based on the information provided by the financial service platform. For example, the financial service platform can input the identity information (including at least one of the identity card number and mobile phone number, etc.) and the borrowing date of the user who borrows the consumption resource to the electronic device, and the electronic device first searches for the user's resource account based on the identity information, and generates resource lending information including the found financial account and the borrowing date.

[0034] In the above step 102, after the electronic device obtains the resource lending information of the above-mentioned consumption resource, the electronic device may extract a topological feature set from a pre-configured resource flow graph according to the resource lending information of the consumption resource.

[0035] The resource flow graph is a topological graph generated in reverse based on at least one illegal merchant, and the nodes in the resource flow graph include each illegal merchant and resource accounts having a resource flow relationship with each illegal merchant.

[0036] In addition, the resource flow diagram may be a topological diagram constructed in advance based on resource flow information in a period of time before the current moment, that is, before the step 102, the method may further include:

[0037] According to the resource flow information within a preset period, determining a target resource account having a resource flow relationship with each of the illegal merchants within the preset period;

[0038] The at least one illegal merchant and the determined target resource account are taken as nodes, and based on the resource flow information between any two of the nodes within the preset time period, a directed edge between the two nodes is defined to generate the resource flow graph.

[0039] Based on this, a resource flow diagram is generated through resource flow information within a preset time period, so that the generated resource flow diagram is more accurate and comprehensive.

[0040] In the implementation manner of the present application, the preset time period may be a time period set according to actual needs. For example, the preset time period may be set to a time period within half a year from the current time, and so on.

[0041] It should be noted that the resource flow information within the preset time period may include all resource flow information generated by at least one financial service platform within the preset time period. Specifically, the at least one financial service platform may include all financial service platforms in the financial service industry, so that the constructed resource flow diagram has more comprehensive coverage and improves the accuracy of the assessment of whether consumer resources flow into illegal merchants.

[0042] In addition, the resource flow information may be information used to characterize resource flows between two resource accounts, which may include resource accounts associated with the flowing resources, resource quantities, resource flow dates, etc. Of course, the two resource accounts may include any two user accounts, or may include a user account and a merchant account, which is not limited here.

[0043] In the embodiment of the present application, each of the at least one illegal merchant can be understood as a resource account of the illegal merchant, and each illegal merchant can be a merchant with a pre-set illegal label, and when consumption resources flow into each illegal merchant, it may cause a greater risk of loss to the consumption resources. For example, the at least one illegal merchant can include a resource account of a real estate company and a resource account of a securities company, etc.

[0044] In addition, the target resource account may be any resource account that has a resource flow relationship with the illegal merchant, and may include a resource account that has a direct resource flow relationship with the illegal merchant. For example, if resource account A1 has flowed resources to illegal merchant B during the preset period, the resource account A1 is the target resource account that has a direct resource flow relationship with illegal merchant B.

[0045] Of course, the target resource account may also include a resource account that indirectly has a resource flow relationship with the illegal merchant, and the number of steps between the resource account and the illegal merchant is less than or equal to N steps, where N is an integer greater than 1, and the number of steps represents the number of times the resource flow from the target resource account to the illegal merchant needs to occur. Specifically, the above N can be set to 6.

[0046] Exemplarily, when N is 4, if resources have flowed into resource account A1 from resource account A2 within a preset period of time, resources have flowed into resource account A2 from resource account A3 within a preset period of time, and resources have flowed into resource account A3 from resource account A4 within a preset period of time, then resource accounts A2, A3 and A4 are all target resource accounts that indirectly have a resource flow relationship with the illegal merchant B, and the number of steps between resource account A4 and the illegal merchant is 4, the number of steps between resource account A3 and the illegal merchant is 3, and the number of steps between resource account A2 and the illegal merchant is 2.

[0047] In an embodiment of the present application, after determining the above-mentioned target resource account, the above-mentioned electronic device can use the above-mentioned illegal merchants and each target resource account as nodes, and based on the resource flow information between any two nodes within the above-mentioned preset time period, define a directed edge between two of the nodes to generate the resource flow graph.

[0048] Among them, the above-mentioned construction of a directed edge between two nodes based on the resource flow information between any two nodes within the above-mentioned preset time period may be that when there is resource flow between the two nodes, a directed edge is generated between the two nodes based on the resource flow information between the two nodes, and the edge attributes of the generated directed edge may include the resource flow information of the two nodes within the preset time period.

[0049] For example, assume that there is a resource flow between the resource account A1 and the resource account A2 within half a year, such as Figure 2As shown, a directed edge between resource account A1 and resource account A2 can be constructed, and the edge attribute of the directed edge from resource account A2 to resource account A1 can indicate that: the amount of resources flowing from A1 to A2 on the day one day ago was 10,000, the amount of resources flowing from A1 to A2 on the day one day ago was 20,000, there was no resource flow from A1 to A2 on the day three days ago and four days ago, and there was a resource flow from A1 to A2 except for the most recent month; and the edge attribute of the directed edge from A1 to A2 indicates that there was no resource flow from A2 to A1 on the day one day ago and two days ago, the amount of resources flowing from A2 to A1 was 10,000 on the day three days ago, the amount of resources flowing from A2 to A1 was 20,000 four days ago, and there was no resource flow from A2 to A1 except for the most recent month;

[0050] Assume that within half a year, there is a resource flow between the above resource account A1 and the illegal merchant B, such as Figure 3 As shown, a directed edge can be constructed between resource account A1 and illegal merchant B, and the edge attribute of the directed edge from resource account A1 to illegal merchant B can be expressed as follows: the number of resources flowing from A1 to B on the day one day ago was 30,000, and there was no resource flow from A1 to B on the day two days ago, three days ago, and four days ago, and there was no resource flow from A1 to B except in the most recent month;

[0051] In summary, resource accounts A1, A2 and illegal merchant B can form a simple resource flow graph, such as Figure 4 shown.

[0052] Optionally, in the resource flow graph generated above, the illegal merchant node may include attributes of overlapping illegal merchant types; and the target resource account node may include the identity information of the corresponding user (such as ID number or mobile phone number, etc.) and attributes of statistical values ​​such as the number of resources and flow times of the flowing resources over a period of time.

[0053] Optionally, in the resource flow graph generated above, the illegal merchant node may be omitted, and the graph may be simplified by adding the attribute {illegal merchant type: resource quantity of liquid resources} in the target resource account node.

[0054] Optionally, the edge attributes of each directed edge in the above resource flow graph can be simplified to the time of the last resource flow between nodes within a period of time (such as one month) from the current time and the number of resources flowing on that day, the total amount of flowing resources from before the last resource flow to one month, and whether there is any resource flow except for the most recent month.

[0055] For example, Figure 4 As shown, the attributes of the directed edge from A1 to A2 can be simplified to:

[0056] {day-1:10000,day_between:20000,mon-1:”yes”},

[0057] Among them, day-1:10000 indicates the time of the last resource flow and the number of resources flowing on that day;

[0058] day_between:20000 indicates the total amount of flowing resources from one month before the last resource flow;

[0059] mon-1: "yes" means there is resource flow except for the most recent month.

[0060] Optionally, when setting the edge attributes of a directed edge, a resource threshold can be set in advance. The resource threshold is used to filter the resource quantity of flow resources that is smaller than the resource threshold. That is, in the above resource flow graph, the resource quantity of flow resources in the edge attributes of each directed edge is greater than or equal to the above resource threshold.

[0061] In some implementations, after generating the resource flow graph, the following steps may also be included:

[0062] When a preset update period arrives, obtaining resource flow information within the preset update period;

[0063] Based on the resource flow information within the preset update period, at least one of the nodes and directed edges of the resource flow graph is updated.

[0064] Based on this, when a preset update cycle arrives, the electronic device can update at least one of the nodes and directed edges of the resource flow graph according to the resource flow information within the preset update cycle, so that the resource flow graph is updated in a timely manner.

[0065] In an embodiment of the present application, the preset update period may be a duration set according to actual needs. For example, the preset update period may be set to a daily, weekly or custom duration, and so on.

[0066] In addition, the above-mentioned updating of at least one item of the nodes and directed edges of the resource flow graph based on the resource flow information between nodes may be at least one item of adding, deleting nodes in the resource flow graph and adding, deleting, modifying edge attributes of directed edges, etc.

[0067] For example, it is assumed that the edge attribute of the directed edge from resource account A1 to resource account A2 is:

[0068] {day-1:10000,day-2:0,day-3:10000,…,day-29:5000,day-30:10000,mon-1:”no”},

[0069] On that day, the amount of resources flowing from resource account A to resource account A2 is 20,000. At 24:00 on that day, the edge attribute of the directed edge from resource account A1 to resource account A2 can be updated to:

[0070] {day-1:20000,day-2:10000,day-3:0,day-4:10000,……,day-30:5000,mon-1:”yes”}.

[0071] In the embodiment of the present application, the above-mentioned extraction of the topological feature set from the pre-configured resource flow graph may be to extract the topological features between the nodes in the resource flow graph that are associated with the above-mentioned resource lending information, and form the above-mentioned topological feature set with the extracted topological features.

[0072] In some implementations, before step 102, the method may further include:

[0073] The resource flow graph is clustered to obtain a community weak relationship graph corresponding to the resource flow graph, wherein the community weak relationship graph includes multiple communities (also referred to as "clusters" or "communities", etc.), each of the communities includes multiple nodes in the resource flow graph and edges between the multiple nodes, and the multiple communities are connected by edges.

[0074] The above step 102 may include:

[0075] A first topological feature subset and a second topological feature subset are extracted from the community weak relationship graph, wherein the first topological feature subset includes at least one first topological feature, and the first topological feature is a topological feature between the communities associated with the resource lending information in the multiple communities; the second topological feature subset includes at least one second topological feature, and the second topological feature is a topological feature between the nodes associated with the resource lending information in each of the communities.

[0076] The topological feature set may include the first topological feature subset and the second topological feature subset.

[0077] Based on this, by generating a community weak relationship graph of the resource flow graph, and extracting a first topological feature subset including topological features between communities from the community weak relationship graph, and also extracting a second topological feature subset including topological features between nodes within the community, the topological feature set formed by the first topological feature subset and the second topological feature subset can better reflect the topological characteristics associated with the above-mentioned resource lending information in the resource flow graph, thereby making the resource violation score predicted by the topological feature set more accurate, further improving the accuracy of assessing whether the consumer resources borrowed by users flow into illegal merchants, and can improve the efficiency of extracting the above-mentioned topological feature set.

[0078] In an embodiment of the present application, the above-mentioned clustering of the resource flow graph to obtain a community weak relationship graph corresponding to the resource flow graph can be implemented by any model that can realize node clustering in the topology graph. Specifically, the above-mentioned clustering of the resource flow graph can be implemented by a greedy model based on modularity. Furthermore, it can be implemented by a community generation model based on node aggregation. Since the process of clustering the resource flow graph based on various models is well known to those skilled in the art, it will not be repeated here.

[0079] In addition, the first topological feature can be any topological feature between communities associated with the resource lending information in the multiple communities of the community weak relationship graph. Specifically, the first topological feature subset can include at least one of the in-and-out degrees of the community, the edge attribute vectors between communities, the number of resources calculated by the edge attribute vectors, the total number of resources vector of directed edges pointing to illegal merchants in the community, and whether the graph contains a ring structure.

[0080] Of course, the second topological feature can be any topological feature between nodes associated with the resource lending information in each community in multiple communities of the community weak relationship graph. Specifically, the second topological feature subset can include at least one of the node in-and-out degree, the minimum number of layers required for biased walks to illegal merchants, the number of paths for biased walks to illegal merchants, the average number of steps required for biased random walks to propagate to illegal merchants, the average number of steps required for biased random walks to propagate to illegal merchants under date restrictions, the path summary vector for biased random walks to propagate to illegal merchants under date restrictions, whether the graph contains a ring structure, the number of second-order adjacent cards, and the number of resources propagated at each layer. Among them, the biased random walk means that when summarizing the attribute vectors involved in the random walk path, the time order cannot be violated.

[0081] It should be noted that when generating a biased random walk sequence (i.e., topological features under a biased random walk), it can be generated through a model such as singular value decomposition, thereby avoiding the transition probability matrix from being too large. Furthermore, based on the singular value decomposition model, the biased random walk sequence can be generated in combination with the attribute vector of the edge, thereby being able to more accurately restore the flow of consumption resources. Furthermore, when calculating the transition probability matrix, the borrowing and lending dates of consumption resources can also be combined to generate truncated transition probability matrices containing dates.

[0082] In addition, since there can only be a one-way edge between the target resource account and the illegal merchant (i.e., the flow from the target resource account to the illegal merchant), there is no possibility of wandering to the target resource account node in other communities. Therefore, when calculating the transfer probability, only the target resource account node in the community and the directly related illegal merchant node can be considered, and the illegal merchant node can be used as an absorption wall.

[0083] In some implementations, after clustering the resource flow graph to obtain a community weak relationship graph corresponding to the resource flow graph, the method may further include:

[0084] After a preset update period arrives and the resource flow graph is updated, the community weak relationship graph is updated based on the updated resource flow graph.

[0085] Based on this, by timely updating the community weak relationship graph of the updated resource flow graph, the topological feature set extracted based on the community weak relationship graph can be made more accurate after the resource flow graph is updated, thereby further improving the accuracy of assessing whether the consumer resources borrowed by users flow into illegal merchants.

[0086] In the embodiment of the present application, the community weak relationship graph is updated based on the updated resource flow graph, and the community weak relationship graph can be updated periodically. For example, the resource flow graph and the community weak relationship graph can be updated simultaneously when the preset update cycle arrives; or, an update cycle greater than the preset update cycle can be set. For example, when the preset update cycle is daily, the update cycle of the community weak relationship graph can be set to weekly or monthly, and so on.

[0087] Alternatively, the updating of the community weak relationship graph based on the updated resource flow graph may include:

[0088] Based on the updated resource flow graph, obtaining a relationship graph update evaluation index corresponding to the updated resource flow graph, wherein the relationship graph update evaluation index is used to determine whether the community weak relationship graph of the updated resource flow graph is updated;

[0089] When the relationship graph update evaluation index determines that the community weak relationship graph of the updated resource flow graph is updated, the updated resource flow graph is re-clustered to obtain an updated community weak relationship graph.

[0090] Based on this, the relationship graph update evaluation index corresponding to the updated resource flow graph can be obtained according to the updated resource flow graph, and based on the relationship graph update evaluation index, it is determined whether to update the community weak relationship graph of the updated resource flow graph, so as to make the update more reasonable and reduce the computing resource consumption of the electronic device.

[0091] In the embodiment of the present application, the above-mentioned relationship graph update evaluation index can be any community weak relationship graph that can be used to determine whether the updated resource flow graph is updated. For example, in the process of re-clustering the updated resource flow graph using the above-mentioned greedy model of modularity, the community modularity or node flip rate (i.e., relationship graph update evaluation index) of each community can be calculated first, and when the community modularity or node flip rate does not meet the preset conditions (such as the amount of change exceeds a certain threshold, etc.), the resource flow graph is re-clustered, and so on.

[0092] Of course, after the above-mentioned resource flow graph is updated, the electronic device may also use a label propagation model to update the community weak relationship graph of the updated resource flow graph. Specifically, the above-mentioned acquisition of the relationship graph update evaluation index corresponding to the updated resource flow graph may include: obtaining the relationship graph update evaluation index corresponding to the updated resource flow graph through the label propagation model, the relationship graph update evaluation index includes at least one of the existing node flipping rate, community modularity and difference ratio, the difference ratio is the percentage of the first score Score value to the second Score value in the newly added node, the first Score value is the maximum Score value, and the second Score value is only less than the first Score value. In this way, the judgment method for whether to update the community weak relationship graph can be made more flexible.

[0093] It should be noted that in the process of updating the community weak relationship graph of the updated resource flow graph through the above-mentioned label propagation model, if it is determined that the above-mentioned relationship graph update evaluation index meets the requirements, the updated resource flow graph is re-clustered to complete the update of the community weak relationship graph; otherwise, it is not re-clustered, that is, the community weak relationship graph is not updated.

[0094] Exemplarily, when the above-mentioned difference ratio is greater than or equal to the preset ratio, the updated resource flow graph is re-clustered to complete the update of the community weak relationship graph; when the above-mentioned difference ratio is less than the preset ratio, the updated resource flow graph is not re-clustered, that is, the community weak relationship graph is not updated.

[0095] In addition, the label propagation model used to update the community weak relationship graph of the updated resource flow graph may only consider adjacent nodes.

[0096] Of course, in the process of updating the community weak relationship graph of the updated resource flow graph through the above-mentioned label propagation model, the above-mentioned relationship graph update evaluation index may not only consider adjacent nodes, but also consider high-order adjacent nodes, dynamic indicators for evaluating the influence of node labels, and at least one of the static hyperparameters for controlling the influence of high-order adjacent node labels.

[0097] Specifically, the relationship graph update evaluation index corresponding to the updated resource flow graph obtained through the label propagation model may include:

[0098] Adjusting the number of label propagation layers and the attenuation of propagation parameters of each layer of the label propagation model;

[0099] The relationship graph update evaluation index corresponding to the updated resource flow graph is obtained through the adjusted label propagation model.

[0100] Based on this, the acquired relationship graph update evaluation index can be made more accurate by adjusting the number of label propagation layers of the label propagation model (i.e., considering high-order adjacent nodes) and the attenuation degree of propagation parameters of each layer (i.e., considering dynamic indicators).

[0101] For example, the Score value of a node can be obtained by the following formula (1):

[0102]

[0103] Where n represents the target node;

[0104] m is the number of pre-set label propagation layers;

[0105] ρ represents the attenuation of the propagation parameters of each layer;

[0106] N represents the number of nodes that can point to the target node n and can only be reached in j steps;

[0107] A ij The attribute vectors of all edges on the path from node i of the jth order neighbor to the target node n are summed up by the number of resources (if there are multiple paths, the calculation can be repeated);

[0108] B i It represents the total number of resources of the attribute vectors of all edges on the path from node i to all nodes reachable in j steps (if there are multiple paths, the calculation should be repeated);

[0109] Score niis a vector. Its business meaning is the proportion of resources that flow from node i to target node n every day. The maximum element value, maximum sum value, or maximum average value can be used to mark node n as being in the same community as node i.

[0110] In the above step 103, after the electronic device extracts the above topological feature set, the electronic device can input the topological feature set into a pre-trained resource violation scoring model to obtain a resource violation score output by the resource violation scoring model, so that the financial service platform can evaluate whether the consumer resources it borrows flow into illegal merchants through the resource violation score.

[0111] The electronic device is configured with a pre-trained resource violation scoring model, that is, before step 103, the method may further include: training a neural network model using a training set including a plurality of training samples to obtain the resource violation scoring model. Each training sample may include a topological feature set corresponding to a historical consumption resource and a violation label, the violation label being used to indicate whether the historical consumption resource has flowed into a violating merchant.

[0112] It should be noted that since the process of training a neural network model through a training set is well known to those skilled in the art, it will not be described in detail here.

[0113] In some implementations, after step 103, the method may further include:

[0114] Obtain resource violation scores of the plurality of consumer resources, and determine resource violation results of each consumer resource based on the resource violation scores of each consumer resource;

[0115] Based on the resource violation results of the multiple consumption resources, obtaining the prediction accuracy of the resource violation scoring model;

[0116] When the prediction accuracy does not meet the preset condition, the resource violation scoring model is updated.

[0117] Based on this, the resource violation results are determined by the resource violation scores of multiple consumer resources to obtain the prediction accuracy of the resource violation scoring model. If the prediction accuracy does not meet the preset conditions, the above resource violation scoring model is retrained to ensure the prediction accuracy of the resource violation scoring model.

[0118] Exemplarily, after obtaining the resource violation scores of the above-mentioned multiple consumer resources through the above-mentioned resource violation scoring model prediction, the financial service platform can investigate the actual resource violation results of the multiple consumer resources, that is, whether each consumer resource has flowed into a violating merchant, and compare the resource violation results determined by the resource violation scores of the multiple consumer resources with the corresponding actual resource violation results, determine the proportion of consumer resources whose determined resource violation results among the multiple consumer resources are consistent with the actual resource violation results (that is, the prediction accuracy), and when the proportion is less than the preset proportion, re-acquire the most recent training set to train the neural network model to update the above-mentioned resource violation scoring model.

[0119] It should be noted that the resource violation result of each consumer resource is determined based on the resource violation score of each consumer resource, which can be a comparison of the resource violation score of each consumer resource with a preset score, and the resource violation result of the consumer resource is determined according to the comparison result. For example, if the resource violation score of the consumer resource is greater than or equal to the preset score, it is determined that the consumer resource has not flowed into the illegal merchant; otherwise, it has flowed into the illegal merchant.

[0120] See also Figure 5 , the present application embodiment provides a consumption resource violation scoring device. Figure 5 As shown, the device 500 includes:

[0121] The loan information acquisition module 501 is used to acquire resource loan information of consumption resources;

[0122] The feature extraction module 502 is used to extract a topological feature set from a pre-configured resource flow graph, wherein the resource flow graph is a topological graph generated in reverse based on at least one illegal merchant, and the nodes in the resource flow graph include: each illegal merchant, and a resource account having a resource flow relationship with each illegal merchant; the topological feature set includes at least one topological feature associated with the resource lending information;

[0123] The scoring module 503 is used to input the topological feature set into a pre-trained resource violation scoring model to obtain a resource violation score output by the resource violation scoring model, wherein the resource violation score is used to characterize the probability that the consumption resources flow into the at least one violating merchant.

[0124] In some embodiments, the apparatus 500 further comprises:

[0125] A resource account determination module, used to determine, based on the resource flow information within a preset period, a target resource account that has a resource flow relationship with each of the illegal merchants within the preset period;

[0126] A flow graph generation module is used to take the at least one illegal merchant and the determined target resource account as nodes, and based on the resource flow information between any two of the nodes within the preset time period, define a directed edge between two of the nodes to generate the resource flow graph.

[0127] In some embodiments, the apparatus 500 further comprises:

[0128] A flow information acquisition module, used to acquire resource flow information within a preset update period when a preset update period arrives;

[0129] A flow graph updating module is used to update at least one of the nodes and directed edges of the resource flow graph based on the resource flow information within the preset update period.

[0130] In some embodiments, the apparatus 500 further comprises:

[0131] A clustering module is used to cluster the resource flow graph to obtain a community weak relationship graph corresponding to the resource flow graph, wherein the community weak relationship graph includes multiple communities, each of the communities includes multiple nodes in the resource flow graph and edges between the multiple nodes, and the multiple communities are connected by edges.

[0132] The feature extraction module can be specifically used for:

[0133] A first topological feature subset and a second topological feature subset are extracted from the community weak relationship graph, wherein the first topological feature subset includes at least one first topological feature, and the first topological feature is a topological feature between the communities associated with the resource lending information in the multiple communities; the second topological feature subset includes at least one second topological feature, and the second topological feature is a topological feature between the nodes associated with the resource lending information in each of the communities.

[0134] The topological feature set may include the first topological feature subset and the second topological feature subset.

[0135] In some embodiments, the apparatus 500 further comprises:

[0136] The community weak relationship graph updating module is used to update the community weak relationship graph based on the updated resource flow graph after a preset update cycle arrives and the resource flow graph is updated.

[0137] In some implementations, the base community weak relationship graph updating module includes:

[0138] An update evaluation index acquisition unit is used to acquire a relationship graph update evaluation index corresponding to the updated resource flow graph based on the updated resource flow graph, wherein the relationship graph update evaluation index is used to determine whether the community weak relationship graph of the updated resource flow graph is updated;

[0139] The community weak relationship graph updating unit is used to re-cluster the updated resource flow graph to obtain an updated community weak relationship graph when the community weak relationship graph of the updated resource flow graph is determined to be updated by the relationship graph update evaluation index.

[0140] In some implementations, the updating evaluation index obtaining unit is specifically used to:

[0141] The relationship graph update evaluation index corresponding to the updated resource flow graph is obtained through the label propagation model, and the relationship graph update evaluation index includes at least one of the existing node flipping rate, community modularity and difference ratio. The difference ratio is the percentage of the first score value and the second score value in the newly added node. The first score value is the maximum score value, and the second score value is only less than the first score value.

[0142] In some implementations, the community weak relationship graph updating unit includes:

[0143] A parameter adjustment subunit, used to adjust the number of label propagation layers of the label propagation model and the attenuation degree of propagation parameters of each layer;

[0144] The community weak relationship graph updating subunit is used to obtain a relationship graph updating evaluation index corresponding to the updated resource flow graph through the adjusted label propagation model.

[0145] In some embodiments, the apparatus 500 further comprises:

[0146] A resource violation result determination module, used to obtain resource violation scores of the plurality of consumption resources, and determine the resource violation result of each consumption resource based on the resource violation score of each consumption resource;

[0147] A prediction accuracy acquisition module, used to acquire the prediction accuracy of the resource violation scoring model based on the resource violation results of the multiple consumption resources;

[0148] The model updating module is used to update the resource violation scoring model when the prediction accuracy does not meet the preset conditions.

[0149] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and the effects that can be achieved by each implementation have also been described in detail in the embodiment of the above method, and will not be elaborated here.

[0150] Based on the same inventive concept, the present disclosure also provides a computing device, specifically in combination with Figure 6 Provide detailed explanation.

[0151] Figure 6 It is a structural block diagram of a computing device according to an exemplary embodiment.

[0152] like Figure 6 As shown, the computing device 600 is capable of implementing the consumption resource violation scoring method and consumption resource violation scoring apparatus according to the embodiment of the present disclosure. The computing device may refer to the electronic device in the embodiment of the present disclosure, and the computing device 600 includes the electronic device in the embodiment of the present disclosure.

[0153] Specifically, each of the above-mentioned processors may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0154] Each of the above memories may include a large capacity memory for information or instructions. By way of example and not limitation, the memory may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a tape, or a universal serial bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory may include a removable or non-removable (or fixed) medium. Where appropriate, the memory may be inside or outside the integrated gateway device. In a particular embodiment, the memory is a non-volatile solid-state memory. In a particular embodiment, the memory includes a read-only memory (ROM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically rewritable ROM (EAROM) or a flash memory, or a combination of two or more of these.

[0155] The processor 601 reads and executes the computer program instructions stored in the memory 602 to perform the following steps:

[0156] Processor 601 executes obtaining resource borrowing information of consumption resources;

[0157] A topological feature set is extracted from a pre-configured resource flow graph, wherein the resource flow graph is a topological graph generated in reverse based on at least one illegal merchant, and the nodes in the resource flow graph include: each illegal merchant, and a resource account having a resource flow relationship with each illegal merchant; the topological feature set includes at least one topological feature associated with the resource lending information;

[0158] The topological feature set is input into a pre-trained resource violation scoring model to obtain a resource violation score output by the resource violation scoring model, wherein the resource violation score is used to characterize the probability that the consumer resource flows into the at least one illegal merchant.

[0159] In some implementations, the processor 601 further determines, based on the resource flow information within a preset period, a target resource account that has a resource flow relationship with each of the illegal merchants within the preset period;

[0160] The at least one illegal merchant and the determined target resource account are taken as nodes, and based on the resource flow information between any two of the nodes within the preset time period, a directed edge between the two nodes is defined to generate the resource flow graph.

[0161] In some implementations, the processor 601 further executes, when a preset update period arrives, obtaining resource flow information within the preset update period;

[0162] Based on the resource flow information within the preset update period, at least one of the nodes and directed edges of the resource flow graph is updated.

[0163] In some implementations, the processor 601 further performs clustering on the resource flow graph to obtain a community weak relationship graph corresponding to the resource flow graph, wherein the community weak relationship graph includes multiple communities, each of the communities includes multiple nodes in the resource flow graph and edges between the multiple nodes, and the multiple communities are connected by edges;

[0164] A first topological feature subset and a second topological feature subset are extracted from the community weak relationship graph, wherein the first topological feature subset includes at least one first topological feature, and the first topological feature is a topological feature between communities associated with the resource lending information in the multiple communities; and the second topological feature subset includes at least one second topological feature, and the second topological feature is a topological feature between nodes associated with the resource lending information in each of the communities.

[0165] The topological feature set includes the first topological feature subset and the second topological feature subset.

[0166] In some implementations, the processor 601 further executes, after a preset update period arrives and the resource flow graph is updated, updating the community weak relationship graph based on the updated resource flow graph.

[0167] In some implementations, the processor 601 further executes, based on the updated resource flow graph, obtaining a relationship graph update evaluation index corresponding to the updated resource flow graph, wherein the relationship graph update evaluation index is used to determine whether a community weak relationship graph of the updated resource flow graph is updated;

[0168] When the relationship graph update evaluation index determines that the community weak relationship graph of the updated resource flow graph is updated, the updated resource flow graph is re-clustered to obtain an updated community weak relationship graph.

[0169] In some embodiments, the processor 601 also executes a relationship graph update evaluation index corresponding to the updated resource flow graph obtained through a label propagation model, wherein the relationship graph update evaluation index includes at least one of an existing node flipping rate, a community modularity, and a difference ratio, wherein the difference ratio is a percentage of a first score value to a second score value in a newly added node, wherein the first score value is a maximum score value, and the second score value is only less than the first score value.

[0170] In some implementations, the processor 601 further performs adjusting the number of label propagation layers and the attenuation degree of propagation parameters of each layer of the label propagation model;

[0171] The relationship graph update evaluation index corresponding to the updated resource flow graph is obtained through the adjusted label propagation model.

[0172] In some implementations, the processor 601 further executes obtaining resource violation scores of the plurality of consumer resources, and determining a resource violation result of each consumer resource based on the resource violation score of each consumer resource;

[0173] Based on the resource violation results of the multiple consumption resources, obtaining the prediction accuracy of the resource violation scoring model;

[0174] When the prediction accuracy does not meet the preset condition, the resource violation scoring model is updated.

[0175] Regarding the computing device in the above embodiment, the specific manner in which the processor performs operations has been described in detail in the embodiment of the method, and the effects that can be achieved by each implementation have also been described in detail in the embodiment of the above method, and will not be elaborated here.

[0176] The embodiment of the present disclosure further provides a computer storage medium, in which computer executable instructions are stored, and the computer executable instructions are used to implement the consumption resource violation scoring method recorded in the embodiment of the present disclosure.

[0177] The embodiment of the present disclosure also provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the method for scoring consumption resource violations as described in the first aspect.

[0178] Wherein, the computer program product may adopt any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to: an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0179] The present disclosure is described with reference to flowcharts and / or block diagrams of methods, devices, and computer program products according to the present disclosure. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable transaction data statistics device to produce a machine, so that the instructions executed by the processor of the computer or other programmable transaction data statistics device generate instructions for implementing the process in the flowchart. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0180] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable transaction data statistical device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0181] These computer program instructions may also be loaded onto a computer or other programmable transaction data statistics device, so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable device to implement the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0182] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. The present disclosure is intended to cover any variations, uses or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The description and examples are to be considered exemplary only, and the true scope and spirit of the present disclosure are indicated by the following claims.

[0183] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A method for scoring consumption resource violations, characterized in that: include: Get resource borrowing information of consumption resources; A topological feature set is extracted from a pre-configured resource flow graph, wherein the resource flow graph is a topological graph generated in reverse based on at least one illegal merchant, and the nodes in the resource flow graph include: each illegal merchant, and a resource account having a resource flow relationship with each illegal merchant; a directed edge between any two of the nodes is determined based on resource flow information of the two nodes within a preset time period; the topological feature set includes at least one topological feature associated with the resource lending information; The topological feature set is input into a pre-trained resource violation scoring model to obtain a resource violation score output by the resource violation scoring model, wherein the resource violation score is used to characterize the probability that the consumption resource flows into the at least one illegal merchant; the resource violation scoring model is trained based on the topological feature set corresponding to the historical consumption resource and the violation label, wherein the violation label is used to indicate whether the historical consumption resource flows into the illegal merchant; Before extracting the topology feature set from the pre-configured resource flow graph, the method further includes: Clustering the resource flow graph to obtain a community weak relationship graph corresponding to the resource flow graph, wherein the community weak relationship graph includes multiple communities, each of the communities includes multiple nodes in the resource flow graph and edges between the multiple nodes, and the multiple communities are connected by edges; The extracting of a topology feature set from a pre-configured resource flow graph includes: Extracting a first topological feature subset and a second topological feature subset from the community weak relationship graph, wherein the first topological feature subset includes at least one first topological feature, and the first topological feature is a topological feature between communities associated with the resource lending information in the multiple communities; the second topological feature subset includes at least one second topological feature, and the second topological feature is a topological feature between nodes associated with the resource lending information in each of the communities; Among them, the topological feature set includes the first topological feature subset and the second topological feature subset; the first topological feature subset includes the community's in-and-out degree, the edge attribute vector between communities, the number of resources calculated by the edge attribute vector, the total resource quantity vector of directed edges pointing to illegal merchants in the community, and whether the graph contains at least one of a ring structure; the second topological feature subset includes the node in-and-out degree, the minimum number of layers required for biased walks to illegal merchants, the number of paths for biased walks to illegal merchants, the average number of steps required for biased random walks to propagate to illegal merchants, the average number of steps required for biased random walks to propagate to illegal merchants under date restrictions, the path summary vector of biased random walks to propagate to illegal merchants under date restrictions, whether the graph contains a ring structure, the number of second-order adjacent cards, and at least one of the number of resources propagated at each layer.

2. The method according to claim 1, characterized in that Before extracting the topology feature set from the pre-configured resource flow graph, the method further includes: According to the resource flow information within a preset period, determining a target resource account having a resource flow relationship with each of the illegal merchants within the preset period; The at least one illegal merchant and the determined target resource account are taken as nodes, and based on the resource flow information between any two of the nodes within the preset time period, a directed edge between the two nodes is defined to generate the resource flow graph.

3. The method according to claim 2, characterized in that After generating the resource flow graph, the method further includes: When a preset update period arrives, obtaining resource flow information within the preset update period; Based on the resource flow information within the preset update period, at least one of the nodes and directed edges of the resource flow graph is updated.

4. The method according to claim 1, characterized in that: After clustering the resource flow graph to obtain a community weak relationship graph corresponding to the resource flow graph, the method further includes: After a preset update period arrives and the resource flow graph is updated, the community weak relationship graph is updated based on the updated resource flow graph.

5. The method according to claim 4, characterized in that The updating of the community weak relationship graph based on the updated resource flow graph includes: Based on the updated resource flow graph, obtaining a relationship graph update evaluation index corresponding to the updated resource flow graph, wherein the relationship graph update evaluation index is used to determine whether the community weak relationship graph of the updated resource flow graph is updated; When the relationship graph update evaluation index determines that the community weak relationship graph of the updated resource flow graph is updated, the updated resource flow graph is re-clustered to obtain an updated community weak relationship graph.

6. The method according to claim 5, characterized in that The obtaining of the relationship graph update evaluation index corresponding to the updated resource flow graph includes: The relationship graph update evaluation index corresponding to the updated resource flow graph is obtained through the label propagation model, and the relationship graph update evaluation index includes at least one of the existing node flipping rate, community modularity and difference ratio. The difference ratio is the percentage of the first score value and the second score value in the newly added node. The first score value is the maximum score value, and the second score value is only less than the first score value.

7. The method according to claim 6, characterized in that The step of obtaining the relationship graph update evaluation index corresponding to the updated resource flow graph through the label propagation model includes: Adjusting the number of label propagation layers and the attenuation of propagation parameters of each layer of the label propagation model; The relationship graph update evaluation index corresponding to the updated resource flow graph is obtained through the adjusted label propagation model.

8. The method according to claim 1, characterized in that After obtaining the resource violation score output by the resource violation scoring model, the method further includes: Obtain resource violation scores of the plurality of consumer resources, and determine resource violation results of each consumer resource based on the resource violation scores of each consumer resource; Based on the resource violation results of the plurality of consumption resources, obtaining the prediction accuracy of the resource violation scoring model; When the prediction accuracy does not meet the preset condition, the resource violation scoring model is updated.

9. A consumption resource violation scoring device, characterized in that: include: A loan information acquisition module is used to obtain resource loan information of consumption resources; A feature extraction module is used to extract a topological feature set from a pre-configured resource flow graph, wherein the resource flow graph is a topological graph generated in reverse based on at least one illegal merchant, and the nodes in the resource flow graph include: each illegal merchant, and a resource account having a resource flow relationship with each illegal merchant; a directed edge between any two of the nodes is determined based on resource flow information of the two nodes within a preset time period; the topological feature set includes at least one topological feature associated with the resource lending information; A scoring module, used for inputting the topological feature set into a pre-trained resource violation scoring model to obtain a resource violation score output by the resource violation scoring model, wherein the resource violation score is used to characterize the probability that the consumption resource flows into the at least one illegal merchant; the resource violation scoring model is trained based on the topological feature set corresponding to the historical consumption resource and the violation label, wherein the violation label is used to indicate whether the historical consumption resource flows into the illegal merchant; A clustering module, used for clustering the resource flow graph before extracting the topological feature set from the pre-configured resource flow graph to obtain a community weak relationship graph corresponding to the resource flow graph, wherein the community weak relationship graph includes a plurality of communities, each of the communities includes a plurality of nodes in the resource flow graph and edges between the plurality of nodes, and the plurality of communities are connected by edges; The feature extraction module is specifically used for: A first topological feature subset and a second topological feature subset are extracted from the community weak relationship graph, wherein the first topological feature subset includes at least one first topological feature, and the first topological feature is a topological feature between communities associated with the resource lending information in the multiple communities; and the second topological feature subset includes at least one second topological feature, and the second topological feature is a topological feature between nodes associated with the resource lending information in each of the communities. Among them, the topological feature set includes the first topological feature subset and the second topological feature subset; the first topological feature subset includes the community's in-and-out degree, the edge attribute vector between communities, the number of resources calculated by the edge attribute vector, the total resource quantity vector of directed edges pointing to illegal merchants in the community, and whether the graph contains at least one of a ring structure; the second topological feature subset includes the node in-and-out degree, the minimum number of layers required for biased walks to illegal merchants, the number of paths for biased walks to illegal merchants, the average number of steps required for biased random walks to propagate to illegal merchants, the average number of steps required for biased random walks to propagate to illegal merchants under date restrictions, the path summary vector of biased random walks to propagate to illegal merchants under date restrictions, whether the graph contains a ring structure, the number of second-order adjacent cards, and at least one of the number of resources propagated at each layer.

10. The device according to claim 9, characterized in that Also includes: A resource account determination module, used to determine, based on the resource flow information within a preset period, a target resource account that has a resource flow relationship with each of the illegal merchants within the preset period; A flow graph generation module is used to take the at least one illegal merchant and the determined target resource account as nodes, and based on the resource flow information between any two of the nodes within the preset time period, define a directed edge between two of the nodes to generate the resource flow graph.

11. The device according to claim 10, characterized in that Also includes: A flow information acquisition module, used to acquire resource flow information within a preset update period when a preset update period arrives; A flow graph updating module is used to update at least one of the nodes and directed edges of the resource flow graph based on the resource flow information within the preset update period.

12. The device according to claim 9, characterized in that Also includes: The community weak relationship graph updating module is used to update the community weak relationship graph based on the updated resource flow graph after a preset update cycle arrives and the resource flow graph is updated.

13. The device according to claim 12, characterized in that The community weak relationship graph updating module includes: An update evaluation index acquisition unit is used to acquire a relationship graph update evaluation index corresponding to the updated resource flow graph based on the updated resource flow graph, wherein the relationship graph update evaluation index is used to determine whether the community weak relationship graph of the updated resource flow graph is updated; The community weak relationship graph updating unit is used to re-cluster the updated resource flow graph to obtain an updated community weak relationship graph when the community weak relationship graph of the updated resource flow graph is determined to be updated by the relationship graph update evaluation index.

14. The device according to claim 13, characterized in that The updating evaluation index obtaining unit is specifically used for: The relationship graph update evaluation index corresponding to the updated resource flow graph is obtained through the label propagation model, and the relationship graph update evaluation index includes at least one of the existing node flipping rate, community modularity and difference ratio. The difference ratio is the percentage of the first score value and the second score value in the newly added node. The first score value is the maximum score value, and the second score value is only less than the first score value.

15. The device according to claim 14, characterized in that The community weak relationship graph updating unit includes: A parameter adjustment subunit, used to adjust the number of label propagation layers of the label propagation model and the attenuation degree of propagation parameters of each layer; The community weak relationship graph updating subunit is used to obtain a relationship graph updating evaluation index corresponding to the updated resource flow graph through the adjusted label propagation model.

16. The device according to claim 9, characterized in that Also includes: A resource violation result determination module, used to obtain resource violation scores of the plurality of consumption resources, and determine the resource violation result of each consumption resource based on the resource violation score of each consumption resource; A prediction accuracy acquisition module, used to acquire the prediction accuracy of the resource violation scoring model based on the resource violation results of the plurality of consumption resources; The model updating module is used to update the resource violation scoring model when the prediction accuracy does not meet the preset conditions.

17. An electronic device, characterized in that: The electronic device includes a processor, a memory, and a program or instruction stored in the memory and executable on the processor. When the program or instruction is executed by the processor, the steps of the consumption resource violation scoring method as described in any one of claims 1 to 8 are implemented.

18. A readable storage medium, characterized in that: The readable storage medium stores a program or an instruction, and when the program or the instruction is executed by the processor, the steps of the consumption resource violation scoring method according to any one of claims 1 to 8 are implemented.

Images