Risk control method and system for core contact based on graph network calculation

Abstract

The application provides a risk control method and system for core contacts based on graph network calculation, and belongs to the technical field of data processing. Specifically, based on the construction processing result, when the change condition of the graph network does not meet the requirements, the determination of the update processing scheme of different graph networks is performed based on the contact data and the change condition of the credit application user of different graph networks. Based on the update processing scheme and the credit application information of the new incoming user, the determination of the credit risk management strategy of the new incoming user is performed, and the efficiency of the identification and processing of the aggregation fraud risk is improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of data processing, and particularly relates to a core contact risk control method and system based on a graph network. BACKGROUND

[0002] In the field of credit risk control, collective fraud behavior occurs from time to time. In the existing technical solutions, the association between the credit application information of different applicants or the image recognition data is often used to identify and process the collective fraud risk. Specifically, a similar technical solution is given in the invention patent application CN202311000939.3 "Credit risk control method and system". However, the above technical solution has the following technical problems:

[0003] The contact information of users with collective fraud risk often has certain association. Therefore, by using the contact information to construct a graph network, the efficiency and pertinence of the identification and processing of the collective fraud risk can be improved. Therefore, how to determine the credit risk control strategy for new users based on the update processing scheme of the existing graph network and the similarity of the credit application information of the credit application users in the graph network and the new users has become a technical problem to be solved.

[0004] Therefore, there is an urgent need for a core contact risk control method and system based on a graph network. SUMMARY

[0005] To achieve the purpose of the application, the application adopts the following technical solutions:

[0006] Specifically, the application provides a core contact risk control method based on a graph network, which specifically includes:

[0007] S1 determines the contact data of the credit application users in different graph networks based on the construction result of the graph network of the core contacts. Based on the contact data, a construction processing strategy of the graph network of the contacts of the new users is determined. Based on the construction processing strategy, the construction processing of the graph network of the core contacts is performed. Based on the construction processing result, when the change of the graph network does not meet the requirements, the next step is entered.

[0008] S2 determines the update processing scheme of different graph networks based on the contact data of the credit application users in different graph networks and the change. Based on the update processing scheme and the credit application information of the new users, the credit risk control management strategy of the new users is determined.

[0009] The application has the following beneficial effects:

[0010] Based on the contact data of credit applicants in different graph networks, a strategy for constructing and processing the graph network of contacts for new applicants is determined. This achieves the goal of constructing and processing the graph network of contacts for new applicants from the perspective of the completeness of the existing graph network of credit applicants, thereby avoiding the technical problem of excessive data processing pressure on the server caused by frequent graph network construction. At the same time, timely graph network construction and processing improves the reliability of identifying and processing clustered fraud risks.

[0011] Based on the updated processing scheme and the credit application information of newly added users, the credit risk control management strategy for newly added users is determined. This takes into account the correlation between the credit application information of newly added users and the credit application information of existing credit application users in the graph network, thereby identifying potential clustered fraud risks. Furthermore, by combining the updated processing scheme of the graph network of related credit application users, the system can identify and process credit application users for clustered fraud risks among newly added users, thus improving the efficiency of clustered fraud risk identification and processing.

[0012] Furthermore, the method for determining the construction result of the graph network of the core contact is as follows:

[0013] By collecting data on credit applicants and their contacts, a heterogeneous graph network is constructed with credit applicants and their contacts as nodes and relationships as edges.

[0014] A graph neural network model is used to embed nodes and aggregate the feature information of neighboring nodes to discover the center of influence of credit applicants in the network for each applicant node.

[0015] Furthermore, the contact data of the credit applicant includes the number of credit applicants in the graph network and the number of contacts of the credit applicant.

[0016] Furthermore, the method for determining the construction strategy of the graph network of the contacts of the newly added user is as follows:

[0017] Based on the contact data, determine the composition of credit application users in the graph network among the existing credit application users;

[0018] Based on the aforementioned configuration, identify the credit application users who have not undergone graph network construction processing;

[0019] Based on the credit application user data that has not undergone graph network construction processing, determine the graph network construction processing strategy for the contacts of the newly applied user.

[0020] Specifically, the method for determining the credit risk management strategy of the new user is:

[0021] Based on the credit application information of the new user, the association of the credit application information of the new user with the credit application information of the credit application users in different graph networks is determined;

[0022] Based on the association of the credit application information of the new user with the credit application information of the credit application users in different graph networks, the associated application users in different graph networks are determined, and the graph network in which the associated application users exist is determined;

[0023] According to the update processing scheme and the credit application user data of the graph network in which the associated application users exist, the credit risk management strategy of the new user is determined.

[0024] In a second aspect, the present application provides a computer system, comprising a memory and a processor connected in communication, and a computer program stored on the memory and capable of running on the processor, wherein the processor executes the computer program to perform the above-mentioned method for calculating core contacts based on graph networks.

[0025] Other features and advantages will be set forth in the following description, and the objectives and other advantages of the present application will be achieved and obtained by the structures particularly pointed out in the description and the accompanying drawings.

[0026] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the following preferred embodiments are specifically described, and the accompanying drawings are described in detail as follows. BRIEF DESCRIPTION OF DRAWINGS

[0027] The above-mentioned and other features and advantages of the present application will become more apparent through the detailed description of example embodiments with reference to the accompanying drawings.

[0028] Figure 1 is a flowchart of a method for calculating core contacts based on graph networks;

[0029] Figure 2 is a flowchart of a method for determining the construction result of the graph network of core contacts;

[0030] Figure 3 is a flowchart of a method for determining the construction processing strategy of the graph network of contacts of the new user;

[0031] Figure 4 is a flowchart of a method for determining the update processing scheme of the graph network;

[0032] Figure 5 is a flowchart of a method for determining the credit risk management strategy of the new user. DETAILED DESCRIPTION

[0033] In order for those skilled in the art to better understand the technical solutions in the specification, the technical solutions in the specification will be clearly and completely described below in combination with the drawings in the specification. Obviously, the described embodiments are only part of the embodiments of the specification, not all. Based on the embodiments of the specification, all other embodiments obtained by those skilled in the art without creative labor should be within the protection scope of the specification.

[0034] In the present application, according to the association of the credit application information of the new application user and the credit application user in the graph network, and the update processing scheme of the graph network, the determination of the range of the recognition processing of the aggregation fraud risk of the new application user is realized, so as to improve the efficiency and reliability of the recognition processing of the aggregation fraud risk.

[0035] Embodiment 1

[0036] As shown in the Figure 1 The present application provides a risk control method for core contact persons based on graph network, which specifically includes:

[0037] S1 determines the contact data of the credit application user in different graph networks based on the construction result of the graph network of the core contact person, determines the construction processing strategy of the graph network of the contact person of the new application user based on the contact data, and performs the construction processing of the graph network of the core contact person based on the construction processing strategy. Based on the construction processing result, when the change of the graph network does not meet the requirements, the next step is entered;

[0038] Specifically, as shown in the Figure 2 The method for determining the construction result of the graph network of the core contact person is:

[0039] By collecting the credit application user and the contact data of the credit application user, a heterogeneous graph network is constructed with the credit application user and the contact person of the credit application user as nodes and the association relationship as edges.

[0040] The graph neural network model is used for embedding calculation of the nodes, and the influence center of the credit application user in the network is mined for each application person node by aggregating the neighbor node feature information.

[0041] Further, the contact data of the credit application user includes the number of credit application users in the graph network and the number of contacts of the credit application person.

[0042] Specifically, as shown in the Figure 3 The method for determining the construction processing strategy of the graph network of the contact person of the new application user is:

[0043] Based on the contact data, determine the composition of the credit application users in the graph network among the existing credit application users;

[0044] Determine the credit application users that do not perform graph network construction processing using the composition;

[0045] Determine the construction processing strategy of the graph network of the new user's contacts using the credit application user data that does not perform graph network construction processing.

[0046] It can be understood that the credit application user that does not perform graph network construction processing is a credit application user that is not in the graph network.

[0047] It should be noted that the credit application user is all users who have performed credit application in history.

[0048] Further, the construction processing strategy of the graph network of the new user's contacts is determined using the credit application user data that does not perform graph network construction processing, specifically including:

[0049] Determine the proportion of credit application users that do not perform graph network construction processing among all credit application users using the credit application user data that does not perform graph network construction processing, and use it as the non-construction proportion.

[0050] Determine the construction processing strategy of the graph network of the new user's contacts based on the non-construction proportion.

[0051] It can be understood that the construction processing strategy of the graph network of the new user's contacts is determined based on the non-construction proportion, specifically including:

[0052] When the non-construction proportion is greater than a preset proportion threshold, for example, 0.3, and the new user or the contacts of the new user overlap with the contacts of the credit application users, then the graph network of the contacts of the new user is constructed.

[0053] When the non-construction proportion is not greater than the preset proportion threshold, and the new user or the contacts of the new user overlap with the contacts of the credit application users and the number of credit application users whose contacts overlap meets the requirements, then the graph network of the contacts of the new user is constructed.

[0054] Specifically, when the number of the credit application users of the new user or the contact person of the new user is not less than a preset number threshold, for example, not less than 3, the graph network construction processing of the contact person of the new user is performed, so that in the case that the graph network construction processing is relatively perfect, the technical problem of unstable credit risk control strategy caused by frequent graph network construction is avoided.

[0055] Optionally, the method for determining the graph network construction processing strategy of the contact person of the new user includes:

[0056] Based on the contact person data, the composition of the credit application users in the graph network of the existing credit application users is determined, and the credit application users for which the graph network construction processing is performed are determined based on the composition.

[0057] Based on the data of the credit application users for which the graph network construction processing is performed, the credit application users involved in different graph networks are determined.

[0058] According to the credit application users involved in different graph networks, the graph network construction processing strategy of the contact person of the new user is determined.

[0059] It should be noted that when the number of the graph networks does not meet the requirement, that is, less than the threshold, at this time, since the number of the graph networks is small, when the contact person of the new user or the contact person of the new user coincides with the contact person of the credit application user, the graph network construction processing of the contact person of the new user is performed.

[0060] In addition, it can be understood that when the number of the graph networks meets the requirement, the credit application users involved in different graph networks are determined, and if the number of the credit application users involved in different graph networks is less than a preset number threshold, when the contact person of the new user or the contact person of the new user coincides with the contact person of the credit application user, the graph network construction processing of the contact person of the new user is performed.

[0061] In other cases, if the number of the credit application users involved in different graph networks is not less than the preset number threshold, when the contact person of the new user or the contact person of the new user coincides with the contact person of the credit application user and the number of the credit application users with which the contact person coincides meets the requirement, the graph network construction processing of the contact person of the new user is performed.

[0062] Specifically, it is determined that the variation of the graph network does not meet the requirement, and specifically includes:

[0063] Based on the construction processing result, determine the variation of the number of credit application users of the graph network;

[0064] Determine the graph network in which the number of credit application users varies based on the variation of the number of credit application users of the graph network;

[0065] According to the number of credit application users of the graph network in which the number of credit application users varies, determine whether the variation of the graph network meets the requirements.

[0066] It can be understood that when the number of graph networks in which the number of credit application users varies is greater than the preset variation number threshold, it is determined that the variation of the graph network does not meet the requirements.

[0067] In addition, it can be understood that when the number of graph networks in which the number of credit application users varies is not greater than the preset variation number threshold, the increase in the number of credit application users of the graph network in which the number of credit application users varies is determined based on the proportion of the number of credit application users of the graph network in which the number of credit application users varies, and when the comprehensive variation factor of different graph networks in which the number of credit application users varies is greater than the preset variation factor threshold, it is determined that the variation of the graph network does not meet the requirements.

[0068] It can be understood that the comprehensive variation factor is determined according to the sum of the variation factors of different graph networks in which the number of credit application users varies.

[0069] Optionally, determining that the variation of the graph network does not meet the requirements specifically includes:

[0070] Based on the construction processing result, determine the variation of the number of credit application users of the graph network;

[0071] Determine the graph network in which the number of credit application users varies based on the variation of the number of credit application users of the graph network;

[0072] According to the number of credit application users of the graph network in which the number of credit application users varies, determine the variation factor of the graph network, and determine whether the variation of the graph network meets the requirements according to the variation factors of different graph networks.

[0073] It should be noted that determining whether the variation of the graph network meets the requirements according to the variation factors of different graph networks specifically includes:

[0074] Determine that the variation of the graph network meets the requirements when the variation factors of different graph networks are all less than the preset variation factor threshold, for example, less than 0.05.

[0075] Further, when the variation factor of different graph networks is not less than the preset variation factor threshold, the graph network with the variation factor not less than the preset variation factor threshold is determined as a user variation network, and when the number of the user variation network does not meet the requirement, for example, more than 10, it is determined that the variation of the graph network does not meet the requirement.

[0076] In addition, it needs to be further explained that when the number of the user variation network meets the requirement, it is determined that the variation of the graph network meets the requirement.

[0077] S2 determines the update processing scheme of different graph networks based on the contact data and the variation of the credit application users of different graph networks, determines the credit risk management strategy of the new application user based on the update processing scheme and the credit application information of the new application user.

[0078] Specifically, the method for determining the update processing scheme of the graph network is:

[0079] Determine the number of credit application users in the graph network based on the contact data of the credit application users of the graph network.

[0080] Determine the variation factor of the graph network according to the variation of the credit application users of the graph network, and determine the graph network with the update processing scheme being the preset update scheme by using the variation factor.

[0081] Determine the update processing scheme of the graph network based on the graph network with the update processing scheme being the preset update scheme and the number of credit application users in the graph network.

[0082] Specifically, if the graph network data with the preset update scheme meets the requirement, for example, the number of the graph network with the preset update scheme is greater than 20, it is determined that the update processing scheme of the graph network is other update scheme.

[0083] Further, if the graph network data with the preset update scheme does not meet the requirement, the determination of the update processing scheme of the graph network is performed according to the number of credit application users in the graph network at this time.

[0084] It should be noted that the preset update scheme is that when the contact information of the credit application user or the contact person of the credit application user of the graph network coincides with the contact information of the new application user or the contact person of the new application user, the graph network is updated. The other update scheme is that when the contact information of the credit application user or the contact person of the credit application user of the graph network coincides with the contact information of the new application user or the contact person of the new application user, the coincident new application user is regarded as a coincident new application user. If the number of coincident new application users is greater than a preset number of users, for example, greater than 2, the graph network is updated.

[0085] Optionally, the method for determining the credit risk management strategy of the new application user is:

[0086] Based on the credit application information of the new application user, the association of the credit application information of the new application user with the credit application information of the credit application users in different graph networks is determined.

[0087] Based on the association of the credit application information of the new application user with the credit application information of the credit application users in different graph networks, the associated application users in different graph networks are determined, and the graph network in which the associated application users exist is determined.

[0088] According to the graph network update processing scheme and the credit application user data, the credit risk management strategy of the new application user is determined.

[0089] If the number of graph networks with the update processing scheme of the associated application users does not meet the requirement, for example, is not less than 2, the credit risk management strategy of the new application user is determined to further consider the credit application users other than the graph network. By determining the similarity of the background information in the credit image of the associated application user in the credit application users other than the graph network and the background information in the credit image of the new application user, it is determined whether the new application user has an aggregation fraud risk. In addition, it is necessary to determine the similarity of the background information in the credit image of the new application user and the background information in the credit image of the new application user, and determine whether the new application user has an aggregation fraud risk.

[0090] Further, if the number of the credit application users in the graph network with the associated credit application users meets the requirement, the number of the credit application users in the graph network with the associated credit application users is determined. If the total number of the credit application users in the graph network with the associated credit application users meets the requirement, for example, the total number of the credit application users in the graph network with the associated credit application users is less than 10, the credit risk management strategy of the new user is determined as follows: only the similarity of the background information in the credit image of the credit application users in the graph network with the associated credit application users and the credit image of the new user is considered, and whether the new user has the gathering fraud risk is determined.

[0091] Further, if the total number of the credit application users in the graph network with the associated credit application users does not meet the requirement, the credit risk management strategy of the new user is determined as follows: the credit application users other than the graph network are further considered, the similarity of the background information in the credit image of the associated credit application users other than the graph network and the background information in the credit image of the new user is determined, whether the new user has the gathering fraud risk is determined, and the similarity of the background information in the credit image of the credit application users in the graph network with the associated credit application users and the background information in the credit image of the new user is determined, whether the new user has the gathering fraud risk is determined.

[0092] Embodiment 2

[0093] In a second aspect, the present application provides a computer system, comprising a memory and a processor connected in communication, and a computer program stored on the memory and capable of running on the processor, wherein the processor executes the computer program to perform the above-mentioned method for determining core contacts based on a graph network.

[0094] Specifically, as shown in Figure 4 The method for determining the update processing scheme of the graph network comprises the following steps:

[0095] Determining the number of the credit application users in the graph network based on the contact data of the credit application users in the graph network;

[0096] Determining the variation factor of the graph network according to the variation of the credit application users in the graph network;

[0097] Determining the update processing scheme of the graph network based on the variation factor and the number of the credit application users of different graph networks.

[0098] It can be understood that when the average value of the change factor of different graph networks is greater than the preset change factor threshold, the change of different graph networks is more serious at this time, and therefore the update processing scheme of all graph networks is determined as the preset update scheme.

[0099] Further, when the average value of the change factor of different graph networks is not greater than the preset change factor threshold, the number of trusted application users in the graph network is used as a basis to determine the construction ratio based on the proportion of the number of trusted application users in the total number of trusted application users in different graph networks. When the construction ratio is less than the preset construction ratio threshold, for example, less than 0.5, at this time, because the construction is not comprehensive enough, the update processing scheme of all graph networks is determined as the preset update scheme.

[0100] In addition, it should be noted that when the construction ratio is not less than the preset construction ratio threshold, the change factor of the graph network is used as a basis for determination. If the change factor of the graph network is greater than the preset change factor threshold, the update processing scheme of the graph network is determined as the preset update scheme.

[0101] In addition, it can be understood that when the change factor of the graph network is not greater than the preset change factor threshold, if the graph network data of the preset update scheme meets the requirements, for example, the number of graph networks of the preset update scheme is greater than 20, the update processing scheme of the graph network is determined as other update schemes.

[0102] Further, if the graph network data of the preset update scheme does not meet the requirements, the determination of the update processing scheme of the graph network is made according to the number of trusted application users in the graph network.

[0103] It can be understood that the determination of the update processing scheme of the graph network according to the number of trusted application users in the graph network specifically includes:

[0104] When the number of trusted application users in the graph network does not meet the requirements, that is, the number of trusted application users is small, for example, less than 5, the update processing scheme of the graph network is determined as the preset update scheme.

[0105] When the number of trusted application users in the graph network meets the requirements, the update processing scheme of the graph network is determined as other update schemes.

[0106] Embodiment 3

[0107] Specifically, as shown in Figure 5 The method for determining the credit risk management strategy of the new user is:

[0108] determine the association of the credit application information of the new application user with the credit application information of the credit application users in different graph networks based on the credit application information of the new application user;

[0109] determine the associated application users in different graph networks based on the association of the credit application information of the new application user with the credit application information of the credit application users in different graph networks;

[0110] determine the credit risk management strategy of the new application user according to the associated application user data in different graph networks and the update processing scheme of the graph network.

[0111] Further, the associated application user is a credit application user whose number of the same information items as the credit application information of the new application user meets the requirements, for example, more than 3 credit application users whose number of the same information items as the credit application information of the new application user, as the associated application user of the new application user.

[0112] Further, the credit risk management strategy of the new application user is determined according to the associated application user data in different graph networks and the update processing scheme of the graph network, specifically including:

[0113] determine the graph network with the associated application user based on the associated application user data in different graph networks;

[0114] It should be noted that if there is no graph network with the associated application user, then the credit risk management strategy of the new application user is determined to be further considered the credit application users other than the graph network, and by determining the similarity of the background information in the credit image of the associated application user in the credit application users other than the graph network and the background information in the credit image of the new application user, it is determined whether the new application user has an aggregation fraud risk.

[0115] Further, if there is a graph network with associated application users, when the number of graph networks with associated application users does not meet the requirement, that is, if the number of graph networks with associated application users is less than the preset graph network number threshold, for example, greater than 3, at this time it indicates that the association degree with the existing credit application user is high, so it is determined that the credit risk management strategy of the new incoming user still needs to further consider the credit application users other than the graph network. By determining the similarity of the background information in the credit image of the associated application users in the credit application users other than the graph network and the background information in the credit image of the new incoming user, it is determined whether the new incoming user has a gathering fraud risk. In addition, it is also necessary to determine the similarity of the background information in the credit image of the credit application users of the graph network with associated application users and the background information in the credit image of the new incoming user, to determine whether the new incoming user has a gathering fraud risk.

[0116] In addition, it should be noted that if the number of graph networks with associated application users meets the requirement, and if the number of graph networks with other update scheme update processing scheme in the graph network with associated application users does not meet the requirement, for example, not less than 2, it is determined that the credit risk management strategy of the new incoming user still needs to further consider the credit application users other than the graph network. By determining the similarity of the background information in the credit image of the associated application users in the credit application users other than the graph network and the background information in the credit image of the new incoming user, it is determined whether the new incoming user has a gathering fraud risk. In addition, it is also necessary to determine the similarity of the background information in the credit image of the credit application users of the graph network with associated application users and the background information in the credit image of the new incoming user, to determine whether the new incoming user has a gathering fraud risk.

[0117] Further, if the number of graph networks with other update scheme update processing scheme in the graph network with associated application users meets the requirement, at this time it is determined that the number of credit application users in the graph network with associated application users. If the total number of credit application users in the graph network with associated application users meets the requirement, for example, the total number of credit application users in all graph networks with associated application users is less than 10, it is determined that the credit risk management strategy of the new incoming user only needs to consider the similarity of the background information in the credit image of the credit application users of the graph network with associated application users and the background information in the credit image of the new incoming user, to determine whether the new incoming user has a gathering fraud risk.

[0118] Further, if the total number of the trusted application users in the graph network with the associated application user does not meet the requirement, the credit risk management strategy of the new application user is determined to further consider the trusted application users other than the graph network, by determining the similarity of the background information in the trusted image of the associated application user in the trusted application users other than the graph network and the background information in the trusted image of the new application user, whether the new application user has the gathering fraud risk, and further determining the similarity of the background information in the trusted image of the new application user and the trusted application users in the graph network with the associated application user, whether the new application user has the gathering fraud risk.

[0119] Specifically, the trusted image is an environmental image of the trusted application user in the process of remote risk control interrogation, and if the trusted image of the trusted application user is consistent with the background in the trusted image of the new application user, it is determined that the new application user has the gathering fraud risk. It can be understood that the new application user is a user currently applying for trust.

[0120] Each of the embodiments in the specification is described in a progressive manner, and the same and similar parts of each embodiment can be referred to each other. Each embodiment focuses on the difference from other embodiments. In particular, for the device, equipment and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the part of the method embodiment.

[0121] The above describes specific embodiments of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be executed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are possible or can be advantageous.

[0122] The above only describes one or more embodiments of the specification and does not limit the specification. One or more embodiments of the specification can have various changes and variations for those skilled in the art. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of one or more embodiments of the specification shall be included in the scope of the claims of the specification.

Claims

1. A risk control method based on graph network computation of core contacts, specifically including: Based on the construction results of the graph network of the core contact, the contact data of the credit application users in different graph networks are determined. Based on the contact data, the construction processing strategy of the graph network of the contacts of the new applicants is determined. Based on the construction processing strategy, the graph network of the core contact is constructed. Based on the construction processing results, if the changes in the graph network do not meet the requirements, proceed to the next step. Based on the contact data and changes of credit application users in different graph networks, different graph network update processing schemes are determined. Based on the update processing schemes and the credit application information of newly added users, the credit risk control management strategy for newly added users is determined. The method for determining the construction strategy of the graph network of contacts for newly added users is as follows: Based on the contact data, determine the composition of credit application users in the graph network among the existing credit application users; Based on the aforementioned configuration, identify the credit application users who have not undergone graph network construction processing; Based on the credit application user data that has not undergone graph network construction processing, determine the graph network construction processing strategy for the contacts of the newly applied user.

2. The risk control method for calculating core contacts based on graph network computation as described in claim 1, characterized in that, The method for determining the construction result of the graph network of the core contact is as follows: By collecting data on credit applicants and their contacts, a heterogeneous graph network is constructed with credit applicants and their contacts as nodes and relationships as edges. A graph neural network model is used to embed nodes and aggregate the feature information of neighboring nodes to discover the center of influence of credit applicants in the network for each applicant node.

3. The risk control method for calculating core contacts based on graph network computation as described in claim 1, characterized in that, The contact data of the credit applicant includes the number of credit applicants in the graph network and the number of contacts of the credit applicant.

4. The risk control method for calculating core contacts based on graph network computation as described in claim 1, characterized in that, The credit application users who have not undergone graph network construction processing are those who are not in the graph network.

5. The risk control method for calculating core contacts based on graph network computation as described in claim 1, characterized in that, The credit granting users are all users who have made credit granting applications in the past.

6. The risk control method for calculating core contacts based on graph network computation as described in claim 1, characterized in that, Based on the credit application user data that has not undergone graph network construction processing, determine the graph network construction processing strategy for the contacts of the newly applied user, specifically including: Using the credit application user data that has not undergone graph network construction processing, determine the proportion of credit application users that have not undergone graph network construction processing among all credit application users, and use this proportion as the non-construction proportion; Based on the unbuilt proportion, a construction and processing strategy for the graph network of the contacts of the newly added user is determined.

7. The risk control method for calculating core contacts based on graph network computation as described in claim 1, characterized in that, The changes to the graph network do not meet the requirements, specifically including: Based on the construction and processing results, determine the changes in the number of users applying for credit in the graph network; Based on the changes in the number of credit application users in the graph network, the graph network in which the number of credit application users changes is determined; Based on the data on the changes in the number of credit applicants in the graph network, determine whether the changes in the graph network meet the requirements.

8. The risk control method for calculating core contacts based on graph network computation as described in claim 1, characterized in that, The method for determining the credit risk control management strategy for newly applied users is as follows: Based on the credit application information of the newly admitted user, determine the association between the credit application information of the newly admitted user and the credit application information of users in different graph networks; Based on the association of credit application information with credit application users in different graph networks, identify the associated application users in different graph networks; Based on the associated application user data in different graph networks and the graph network update processing scheme, the credit risk control management strategy for the newly applied users is determined.

9. A computer system, comprising: A memory and processor with a communication connection, and a computer program stored on the memory and capable of running on the processor, characterized in that, when the processor runs the computer program, it executes a risk control method for calculating core contacts based on graph network computing as described in any one of claims 1-8.

Images