Customer risk identification method and device, server and storage medium

Abstract

The invention relates to the technical field of artificial intelligence, and discloses a customer risk identification method, which comprises the following steps: acquiring to-be-audited data which is submitted by a to-be-audited customer and comprises a customer photo and customer information; obtaining photos corresponding to a plurality of associated customers according to preset type information in the customer information and a risk control relationship graph; generating corresponding text information for describing photo backgrounds for the customer photo and the photos corresponding to the plurality of associated customers through a multi-modal image model, and respectively calculating background similarity values between the customer photo and the photo of each associated customer based on the customer photo and the text information thereof and the photo of each associated customer and the text information thereof; and screening out the photos of the associated clients with the background similarity values greater than a preset threshold value, and determining the risk level of the to-be-audited client according to the number of the screened photos. According to the customer risk identification method, the customers with high similarity between the background of the photo image and the background of the associated customer photo can be identified, and the gang fraud risk is reduced.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular, to a method, device, server, and storage medium for customer risk identification. Background Art

[0002] Artificial Intelligence (AI) technology can be used to simulate, extend, and expand human intelligence using digital computers or machines controlled by digital computers, and to perceive the environment, acquire knowledge, and use knowledge to obtain the best results, including theories, methods, technologies, and application systems, such as in the field of image processing. With the development of computer vision technology, artificial intelligence-based image recognition technology has been widely applied to identity verification and approval work in the financial field.

[0003] In financial operations, when high-risk gang fraud customers conduct photo reviews, the backgrounds of the photos uploaded by the same gang are usually taken in the same scene and have many similarities. However, when the prior art conducts customer photo risk identification, it usually only uses artificial intelligence technology to identify whether the photo image background of a single customer is a high-risk scene, and cannot identify customer photos with similar image backgrounds, making it difficult to judge group fraud risks. Summary of the Invention

[0004] In view of the above, it is necessary to provide a customer risk identification method for identifying a to-be-reviewed customer with a high background similarity value between the photo image background and the background of the associated customer photo and performing fraud risk identification to reduce group fraud risks.

[0005] To achieve the above object, the present invention provides a customer risk identification method, including:

[0006] Obtain the to-be-reviewed materials submitted by the to-be-reviewed customer in the business system, where the to-be-reviewed materials include customer photos and customer information;

[0007] Extract the preset type information in the customer information, add the to-be-reviewed customer to the preset risk control relationship graph based on the preset type information, obtain multiple associated customers of the to-be-reviewed customer based on the risk control relationship graph, and obtain the photos corresponding to the multiple associated customers;

[0008] Generate corresponding text information describing the photo background for the customer photo and the photos corresponding to the multiple associated customers through a pre-constructed multimodal image model, and calculate the background similarity values between the customer photo and the photos of each associated customer based on the customer photo and its text information and the photo and its text information of each associated customer;

[0009] Filter out the photos of associated customers corresponding to the background similarity values greater than the preset threshold, and determine the risk level of the customer to be audited according to the number of filtered photos.

[0010] Optionally, before adding the customer to be audited to the preset risk control relationship graph based on the preset type information, it includes:

[0011] Obtain the historical customer profiles of multiple historical customers from the database of the business system, where the historical customer profiles include historical customer information and historical customer identifiers;

[0012] Extract the preset type information of the historical customers from the historical customer information;

[0013] Set the association relationships between the historical customers according to the preset type information of the historical customers;

[0014] Generate a risk control relationship graph according to the association relationships.

[0015] Optionally, the setting of the association relationships between the historical customers according to the preset type information and the generation of the risk control relationship graph according to the association relationships include:

[0016] Set the corresponding association relationships for historical customers with the same preset type information;

[0017] Configure the relationship graph nodes corresponding to each historical customer according to the historical customer identifier corresponding to each historical customer;

[0018] Configure the association relationships between historical customers as the edges of the relationship graph, and connect the relationship graph nodes of the historical customers corresponding to the association relationships through the edges of the relationship graph to obtain a risk control relationship graph.

[0019] Optionally, adding the customer to be audited to the preset risk control relationship graph based on the preset type information includes:

[0020] Obtain the customer identifier of the customer to be audited, and determine the relationship graph node corresponding to the customer to be audited in the risk control relationship graph based on the customer identifier;

[0021] Determine the association relationship between the customer to be audited and the historical customers corresponding to other relationship graph nodes in the risk control relationship graph according to the preset type information;

[0022] Generate an edge connecting the relationship graph node corresponding to the customer to be audited and other relationship graph nodes in the risk control relationship graph according to the association relationship.

[0023] Optionally, obtaining multiple associated customers of the customer to be audited based on the risk control relationship graph and obtaining photos corresponding to the multiple associated customers includes:

[0024] Obtaining multiple associated nodes connected to the relationship graph node corresponding to the customer to be audited in the risk control relationship graph;

[0025] Obtaining the associated customer identifiers corresponding to each associated node, and obtaining the corresponding customer profiles from the database of the business system according to the associated customer identifiers;

[0026] Obtaining the customer photos stored at the time point closest to the current time from the obtained customer profiles to obtain the associated customer photos corresponding to the associated customer identifiers.

[0027] Optionally, calculating the background similarity values between the customer photo and its text information and the photos and their text information of each associated customer respectively includes:

[0028] Inputting the customer photo and the associated customer photo into the image encoder of the multi-modal image model respectively to obtain a first image feature vector corresponding to the customer photo and a second image feature vector corresponding to the associated customer photo;

[0029] Inputting the text information corresponding to the customer photo and the text information corresponding to the associated customer photo into the text encoder of the multi-modal image model respectively to obtain a first text feature vector corresponding to the customer photo and a second text feature vector corresponding to the associated customer photo;

[0030] Calculating the similarity between the first image feature vector and the first text feature vector and the second image feature vector and the second text feature vector to obtain the background similarity values between the customer photo and the photos of each associated customer.

[0031] Optionally, determining the risk level of the customer to be audited according to the number of photos screened out includes:

[0032] Obtaining the associated data table of the preset number of background similar photos and the risk level from the database of the business system;

[0033] Searching for the corresponding risk level in the associated data table according to the number of photos screened out to obtain the risk level corresponding to the customer information to be audited.

[0034] In addition, to achieve the above object, the present invention also provides a customer risk identification device, and the customer risk identification device includes:

[0035] A data acquisition module, configured to acquire the to-be-reviewed materials submitted by the customers to be reviewed in the business system, where the to-be-reviewed materials include customer photos and customer information;

[0036] A photo acquisition module, configured to extract the preset type information in the customer information, add the customer to be reviewed to a preset risk control relationship graph based on the preset type information, obtain multiple associated customers of the customer to be reviewed based on the risk control relationship graph, and obtain the photos corresponding to the multiple associated customers;

[0037] A similarity calculation module, configured to generate corresponding text information describing the photo background for the customer photo and the photos corresponding to the multiple associated customers through a pre-constructed multi-modal image model, and calculate the background similarity values between the customer photo and the photos of each associated customer respectively based on the customer photo and its text information and the photo and its text information of each associated customer;

[0038] A risk identification module, configured to screen out the photos of the associated customers corresponding to the background similarity values greater than a preset threshold, and determine the risk level of the customer to be reviewed according to the number of the screened photos.

[0039] In addition, to achieve the above object, the present invention further provides a server, where the server includes:

[0040] At least one processor; and,

[0041] A memory communicatively connected to the at least one processor; wherein,

[0042] The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above-mentioned customer risk identification method.

[0043] In addition, to achieve the above object, the present invention further provides a computer-readable storage medium, storing a computer program, characterized in that the computer program realizes the above-mentioned customer risk identification method when being executed by a processor.

[0044] The present invention obtains the photos of the associated customers of the customer to be reviewed through a risk control relationship graph, calculates the background similarity values between the photo of the customer to be reviewed and the photos of the associated customers through a multi-modal image model, and determines the risk level of the materials of the customer to be reviewed according to the number of the photos of the associated customers with the background similarity values greater than a preset similarity threshold, which can effectively identify the customers to be reviewed with high background similarity values between the photo image background and the photos of the associated customers and perform fraud risk identification, reducing the risk of group fraud. Description of the Drawings

[0045] Figure 1It is an exemplary system architecture diagram to which the present invention can be applied;

[0046] Figure 2 It is a flowchart of an embodiment of the method for identifying customer risks of the present invention;

[0047] Figure 3 It is a schematic diagram of an embodiment of the server of the present invention;

[0048] Figure 4 It is a module schematic diagram of an embodiment of the customer risk identification device of the present invention.

[0049] The realization of the purpose of this application, functional features and advantages will be further described with reference to the embodiments and the accompanying drawings. Detailed Embodiments

[0050] In order to make the purpose, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention and are not used to limit the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0051] It should be noted that the descriptions involving "first", "second", etc. in the present invention are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one of these features. In addition, the technical solutions between various embodiments can be combined with each other, but it must be based on the ability of those of ordinary skill in the art to implement. When the combination of technical solutions appears to be contradictory or unable to be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the protection scope required by the present invention.

[0052] As Figure 1 shown, the system architecture 100 may include terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, etc.

[0053] Users can use the terminal devices 101, 102, 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications may be installed on the terminal devices 101, 102, 103, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0054] The terminal devices 101, 102, and 103 can be various electronic devices with a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer III), MP4 (Moving Picture Experts Group Audio Layer IV) players, laptop computers, desktop computers, and so on.

[0055] The server 105 can be a server that provides various services, such as a background server that supports the pages displayed on the terminal devices 101, 102, and 103.

[0056] It should be noted that the customer risk identification method provided by the embodiments of the present application is generally executed by the server / terminal device. Correspondingly, the computer-readable storage medium is generally set in the server / terminal device.

[0057] It should be understood that Figure 1 the numbers of the terminal devices, networks, and servers in are merely illustrative. According to the implementation requirements, there can be any number of terminal devices, networks, and servers.

[0058] As Figure 2 shown, it is a flowchart of an embodiment of the customer risk identification method of the present invention, including steps S1 - S4.

[0059] S1. Obtain the to-be-reviewed materials submitted by the customer to be reviewed in the business system. The to-be-reviewed materials include customer photos and customer information.

[0060] In an embodiment, the to-be-reviewed materials are the incoming materials for risk review when the customer handles business in the business system. The business system can be a financial business system, such as the loan business system and insurance business system of a bank. The customer photo is a real-time self-taken photo of the customer when handling the current business in the business system. The customer information includes the terminal information of the customer terminal that submits the to-be-reviewed materials, as well as the basic information of the customer, such as GPS location, IP address, Wi-Fi BSSID, terminal device ID, terminal device SIM card information, the name of the customer's unit, the customer contact number, customer identification, etc.

[0061] S2. Extract the preset type information in the customer information, add the to-be-reviewed customer to the preset risk control relationship graph based on the preset type information, obtain multiple associated customers of the to-be-reviewed customer based on the risk control relationship graph, and obtain the photos corresponding to the multiple associated customers.

[0062] In one embodiment, extracting the preset type information from the customer information includes:

[0063] Extracting the corresponding preset type information from the customer information, where the preset type information is pre-set customer basic information and customer terminal information, including but not limited to GPS positioning, IP address, Wi-Fi BSSID, terminal device ID, terminal device SIM card information, customer's company name, and customer contact number.

[0064] In one embodiment, before adding the customer to be audited to the preset risk control relationship graph based on the preset type information, it includes:

[0065] Obtaining historical customer profiles of multiple historical customers from the database of the business system, where the historical customer profiles include historical customer information and historical customer identifiers;

[0066] Extracting the preset type information of the historical customers from the historical customer information;

[0067] Setting the association relationships between the historical customers according to the preset type information of the historical customers;

[0068] Generating a risk control relationship graph according to the association relationships.

[0069] For example: if the preset type information is GPS positioning and the GPS positions of historical customers M and N are the same, then set an association relationship with the same GPS positioning between M and N; if the preset type information is contact number and the contact numbers of historical customers M and O are the same, then set an association relationship with the same contact number between M and N; no association relationship of this preset type information is set between historical customers with different preset type information.

[0070] In one embodiment, setting the association relationships between the historical customers according to the preset type information and generating a risk control relationship graph according to the association relationships includes:

[0071] Setting corresponding association relationships for historical customers with the same preset type information;

[0072] Configuring relationship graph nodes corresponding to each historical customer according to the historical customer identifiers corresponding to each historical customer;

[0073] Configuring the association relationships between historical customers as the edges of the relationship graph, and connecting the relationship graph nodes of the historical customers corresponding to the association relationships through the edges of the relationship graph to obtain a risk control relationship graph.

[0074] Specifically, the risk control relationship graph includes nodes and edges connecting the nodes. The nodes are used to represent customers, and the edges are used to represent the association relationships between customers. In this embodiment, a corresponding relationship graph node is generated for each historical customer, and the node is named with the unique identifier of the historical customer. Edges connecting the corresponding relationship graph nodes are generated according to the association relationships between the historical customers.

[0075] For example: a corresponding node M is generated for the historical customer M, a corresponding node N is generated for the historical customer N, and an association relationship of GPS positioning is set for the historical customers M and N with the same GPS positioning. Based on this association relationship, an edge connecting node M and node N is generated in the relationship graph.

[0076] In one embodiment, adding the to-be-audited customer to a preset risk control relationship graph based on the preset type information includes:

[0077] Obtain the customer identifier of the to-be-audited customer, and determine the corresponding relationship graph node of the to-be-audited customer in the risk control relationship graph based on the customer identifier;

[0078] Determine the association relationship between the to-be-audited customer and the historical customers corresponding to other relationship graph nodes in the risk control relationship graph according to the preset type information;

[0079] Generate an edge connecting the relationship graph node corresponding to the to-be-audited customer and other relationship graph nodes in the risk control relationship graph according to the association relationship.

[0080] For example: configure node A for the to-be-audited customer A in the risk control relationship graph, match the preset type information of the to-be-audited customer A with the preset type information of other nodes in the risk control relationship graph, set the association relationship between node A and one or more matching nodes with the same preset type information according to the matching result, and generate an edge connecting node A and the matching nodes according to the association relationship.

[0081] In one embodiment, obtaining multiple associated customers of the to-be-audited customer based on the risk control relationship graph and obtaining the corresponding photos of the multiple associated customers includes:

[0082] Obtain multiple associated nodes connected to the relationship graph node corresponding to the to-be-audited customer in the risk control relationship graph;

[0083] Obtain the associated customer identifiers corresponding to the respective associated nodes, and obtain the corresponding customer profiles from the database of the business system according to the associated customer identifiers;

[0084] Obtain the customer photo stored at the time point closest to the current time from the obtained customer profiles to obtain the associated customer photo corresponding to the associated customer identifier.

[0085] In this embodiment, by generating a risk control relationship graph based on the associated relationship of the preset type information of the customers, it is possible to identify customers with the same preset type information, so as to confirm whether the customer to be audited forms a group relationship with other customers.

[0086] S3. Use the pre-constructed multimodal image model to generate corresponding text information describing the photo background for the customer photo and the photos corresponding to the multiple associated customers, and calculate the background similarity values between the customer photo and each associated customer's photo respectively based on the customer photo and its text information and the photo and its text information of each associated customer.

[0087] In one embodiment, the step of using the pre-constructed multimodal image model to generate corresponding text information describing the photo background for the customer photo and the photos corresponding to the multiple associated customers includes:

[0088] The multimodal image model obtains the background similarity calculation instruction, the customer photo, and the photos of the associated customers.

[0089] According to the background similarity calculation instruction, identify the image background of each obtained photo, and generate text information describing the image background of each photo based on artificial intelligence, so as to obtain the text information corresponding to the customer photo and the text information corresponding to the photo of each associated customer.

[0090] For example: The multimodal image model performs background recognition and description text generation on the customer photo, and the obtained text information is "Scene function: Office area, Decoration style: Glass partition, Background color tone: Cold color tone"; for the photo of the associated customer, perform background recognition and description text generation, and the obtained text information is "Scene function: Leisure area, Decoration style: Leisure furniture, Background color tone: Warm color tone".

[0091] In one embodiment, the step of calculating the background similarity values between the customer photo and each associated customer's photo respectively based on the customer photo and its text information and the photo and its text information of each associated customer includes:

[0092] Input the customer photo and the photos of the associated customers into the image encoder of the multimodal image model respectively to obtain the first image feature vector corresponding to the customer photo and the second image feature vector corresponding to the photos of the associated customers.

[0093] Input the text information corresponding to the customer photo and the text information corresponding to the photos of the associated customers into the text encoder of the multimodal image model respectively to obtain the first text feature vector corresponding to the customer photo and the second text feature vector corresponding to the photos of the associated customers.

[0094] Calculate the similarity between the first image feature vector and the first text feature vector and the second image feature vector and the second text feature vector to obtain the background similarity value between the customer photo and the photos of each associated customer.

[0095] In one embodiment, the calculating the similarity between the first image feature vector and the first text feature vector and the second image feature vector and the second text feature vector to obtain the background similarity value between the customer photo and the photos of each associated customer includes:

[0096] The multimodal image model projects the first image feature vector and the first text feature vector into a high-dimensional embedding space for feature vector fusion to obtain a first fused feature vector;

[0097] Project the second image feature vector and the second text feature vector into a high-dimensional embedding space for feature vector fusion to obtain a second fused feature vector;

[0098] By calculating the cosine similarity between the first fused feature vector and the second fused feature vector, obtain the background similarity value between the customer photo and the photos of each associated customer.

[0099] In one embodiment, the multimodal image model is a multimodal pre-training model architecture that aligns images and texts through contrastive learning. Its core idea is to train an image encoder and a text encoder simultaneously, so that the matching image-text pairs have similar vector representations in the high-dimensional embedding space, so that this cross-modal feature alignment ability can be utilized in downstream tasks. The text encoder is responsible for converting text into feature vectors. After the input text is processed by the embedding layer, the feature vectors of the text are generated through the self-attention mechanism. The image encoder is responsible for converting images into feature vectors, and the input images are processed into feature vectors of a fixed length. The text encoder and the image encoder achieve cross-modal information interaction and fusion by sharing a vector space (high-dimensional embedding space).

[0100] Specifically, during the pre-training phase, the multimodal image model learns the matching relationship between images and texts by comparing their vectors. The multimodal image model receives a batch of image-text pairs as input, and attempts to bring the matching image and text vectors closer in the embedding space, while pushing the mismatched vectors further away. This learning method enables the model to capture the deep semantic connection between images and texts and achieve cross-modal understanding. In addition, during the training process, the multimodal image model adopts a symmetric contrast loss function, including contrast loss (training the model by maximizing the similarity of correct image-text pairs and minimizing the similarity of incorrect image-text pairs) and classification loss (used to train the model to perform multi-task classification of images and texts), which means that for each image-text pair, the model calculates losses in two directions: image to text and text to image. This symmetry ensures that the model can effectively learn matching relationships in both directions.

[0101] In one embodiment, the customer risk identification method further includes:

[0102] Based on the background similarity value between the customer photo and each associated customer photo, as well as the text information corresponding to the customer photo and each associated customer photo, an artificial intelligence template is used to output the background similarity report between the customer photo and the associated customer photos in a natural language format.

[0103] For example, the similarity between customer photo A and related customer photo B is calculated using a pre-built multimodal image model, and the output in a natural language format using an AI template is as follows:

[0104] "similarity_score":30, / / background similarity value

[0105] "threshold":60, / / preset similarity threshold

[0106] "is_same_location":false, / / Is it a photo with similar background?

[0107] "reason":[

[0108] Different scene types: A is a leisure scene, B is an office scene;

[0109] Different background decorations: A is open leisure seating, B is a glass partition;

[0110] Different lighting sources: A is soft natural lighting, B is uniform indoor lighting;

[0111] Different color tones: A is warm and B is cool.]

[0112] This embodiment utilizes the recognition ability of the multimodal large model in the field of image analysis to achieve efficient and accurate identification of potential customer associations, thereby effectively evaluating the risk of gang fraud.

[0113] S4. Screen out the photos of associated customers corresponding to the background similarity values greater than the preset threshold, and determine the risk level of the to-be-reviewed customer according to the number of screened photos.

[0114] In one embodiment, the customer risk identification method further includes:

[0115] After obtaining the to-be-reviewed materials, extract the customer identifier in the to-be-reviewed materials;

[0116] Match the customer identifier in the historical customer materials in the database of the business system, and determine whether the customer corresponding to the to-be-reviewed materials is a historical customer;

[0117] If so, use the risk level of the matched historical customer as the risk level corresponding to the to-be-reviewed materials;

[0118] If not, execute the above steps S2 - S4 to determine the risk level corresponding to the to-be-reviewed materials.

[0119] In one embodiment, the determining the risk level of the to-be-reviewed customer according to the number of screened photos includes:

[0120] Obtain the associated data table of the preset number of background similar photos and the risk level from the database of the business system;

[0121] Find the corresponding risk level in the associated data table according to the number of screened photos, and obtain the risk level corresponding to the to-be-reviewed customer materials.

[0122] For example: The scoring range of the preset background similarity value is 0 - 100 points, and the preset similarity threshold is 60 points. Then, count the number of photos of associated customers with background similarity values greater than 60 points as the number of background similar photos. Match according to the associated data table. When the number of background similar photos is 0 - 2, the preset risk level is 0; when the number of background similar photos is 3 - 5, the preset risk level is 1; when the number of background similar photos is greater than 5, the preset risk level is 2.

[0123] In one embodiment, after determining the risk level corresponding to the to-be-reviewed materials based on the association relationship between the number of background similar photos and the preset risk level, it further includes:

[0124] When the risk level corresponding to the to-be-reviewed materials is higher than the warning level, generate a warning notice and send it to the background of the business system for manual review and processing.

[0125] As can be seen from the above embodiments, the customer risk identification method proposed by the present invention obtains associated customer photos of the customer photo to be reviewed through the risk control relationship graph, calculates the background similarity value between the customer photo to be reviewed and the associated customer photos through the multi-modal image model, and determines the risk level of the customer information to be reviewed based on the number of associated customer photos whose background similarity value is greater than the preset similarity threshold. It can effectively identify the customers to be reviewed with a high background similarity value between the photo image background and the associated customer photos and conduct fraud risk identification, reducing the risk of gang fraud.

[0126] As Figure 3 shown, it is a schematic diagram of an embodiment of the server of the present invention. The server 1 is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions. The server 1 can be a single network server, a server group composed of multiple network servers, or a cloud composed of a large number of hosts or network servers based on cloud computing, where cloud computing is a type of distributed computing and consists of a super virtual computer formed by a group of loosely coupled computer sets.

[0127] In this embodiment, the server 1 includes, but is not limited to, a memory 11, a processor 12, and a network interface 13 that can communicate with each other through a system bus. The memory 11 stores a customer risk identification program 10, and the customer risk identification program 10 can be executed by the processor 12. Figure 4 Only the server 1 with components 11-13 and the customer risk identification program 10 is shown. Those skilled in the art can understand that Figure 4 the shown structure does not constitute a limitation on the server 1, and it can include fewer or more components than shown, or combine some components, or have different component arrangements.

[0128] Among them, the memory 11 includes a memory and at least one type of readable storage medium. The memory provides a cache for the operation of the server 1; the readable storage medium can be volatile or non-volatile. Specifically, the readable storage medium can be a storage medium such as flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the readable storage medium can be an internal storage unit of the server 1, such as the hard disk of the server 1; in other embodiments, the storage medium can also be an external storage device of the server 1, such as a plug-in hard disk equipped on the server 1, a SmartMedia Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. In this embodiment, the readable storage medium of the memory 11 mainly includes a program storage area and a data storage area. Among them, the program storage area is usually used to store the operating system installed on the server 1 and various application software, such as storing the code of the customer risk identification program 10 in an embodiment of the present invention, etc.; the data storage area can store data created according to the use of the blockchain node, such as various data that have been output or will be output.

[0129] In some embodiments, the processor 12 can be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips. The processor 12 is generally used to control the overall operation of the server 1, such as performing control and processing related to data interaction or communication with other devices. In this embodiment, the processor 12 is used to run the program code stored in the memory 11 or process data, such as running the customer risk identification program 10, etc.

[0130] The network interface 13 can include a wireless network interface or a wired network interface, and this network interface 13 is used to establish a communication connection between the server 1 and a client terminal (not shown in the figure).

[0131] Optionally, the server 1 may further include a user interface, which may include a display, an input unit such as a keyboard, and optionally, the user interface may further include a standard wired interface and a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch liquid crystal display, and an organic light-emitting diode (OLED) touch device, etc. Among them, the display may also be appropriately referred to as a display screen or a display unit, which is used to display the information processed in the server 1 and to display a visual user interface.

[0132] In an embodiment of the present invention, when the customer risk identification program 10 is executed by the processor 12, the following steps S1 - S4 are implemented.

[0133] S1. Obtain the materials to be reviewed submitted by the customer to be reviewed in the business system, where the materials to be reviewed include customer photos and customer information;

[0134] S2. Extract the preset type information in the customer information, add the customer to be reviewed to the preset risk control relationship graph based on the preset type information, obtain multiple associated customers of the customer to be reviewed based on the risk control relationship graph, and obtain the photos corresponding to the multiple associated customers;

[0135] S3. Use an artificial intelligence template to extract the image backgrounds of the customer photo and the photos of the associated customers, and generate text information describing the image backgrounds. Calculate the image background similarity between the customer photo and the corresponding text information and each associated customer photo and the corresponding text information through a pre-constructed multimodal image model respectively, and obtain the background similarity values between the customer photo and each associated customer photo;

[0136] S4. Screen out the photos of the associated customers corresponding to the background similarity values greater than the preset threshold, and determine the risk level of the customer to be reviewed according to the number of the screened photos.

[0137] The specific operation steps implemented by the above steps S1 - S4 are substantially the same as those in an embodiment of the above customer risk identification method of the present invention, and will not be elaborated herein.

[0138] In other embodiments, the customer risk identification program 10 may also be divided into one or more modules, and the one or more modules are stored in the memory 11 and executed by one or more processors (in this embodiment, the processor 12) to complete the present invention. The module referred to in the present invention refers to a series of computer program instruction segments that can complete specific functions and is used to describe the execution process of the customer risk identification program 10 in the server 1.

[0139] As Figure 4 shown, it is a schematic diagram of modules of an embodiment of the customer risk identification device of the present invention.

[0140] In an embodiment of the present invention, the customer risk identification device 1 includes a data acquisition module 110, a photo acquisition module 120, a similarity calculation module 130, and a risk identification module 140. Exemplarily:

[0141] The data acquisition module 110 is used to acquire the to-be-reviewed data submitted by the customer to be reviewed in the business system, and the to-be-reviewed data includes customer photos and customer information;

[0142] The photo acquisition module 120 is used to extract the preset type information in the customer information, add the customer to be reviewed to the preset risk control relationship graph based on the preset type information, obtain multiple associated customers of the customer to be reviewed based on the risk control relationship graph, and obtain the photos corresponding to the multiple associated customers;

[0143] The similarity calculation module 130 is used to use an artificial intelligence template to extract the image backgrounds of the customer photo and the photos of the associated customers, generate text information describing the image backgrounds, and calculate the image background similarity between the customer photo and the corresponding text information and each associated customer photo and the corresponding text information through a pre-constructed multi-modal image model, so as to obtain the background similarity values between the customer photo and the photos of each associated customer;

[0144] The risk identification module 140 is used to screen out the photos of the associated customers corresponding to the background similarity values greater than the preset threshold, and determine the risk level of the customer to be reviewed according to the number of the screened photos.

[0145] The specific operation steps implemented when the above data acquisition module 110, photo acquisition module 120, similarity calculation module 130, and risk identification module 140 are executed are substantially the same as those of an embodiment of the above customer risk identification method, and will not be elaborated here.

[0146] In addition, an embodiment of the present invention also proposes a computer-readable storage medium, which can be volatile or non-volatile. Specifically, the computer-readable storage medium can be any one or any combination of a hard disk, a multimedia card, an SD card, a flash card, an SMC, a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a portable compact disc read-only memory (CD-ROM), a USB memory, etc. The computer-readable storage medium stores a customer risk identification program 10.

[0147] When the customer risk identification program 10 is executed by a processor, the following operations are implemented:

[0148] A1. Obtain the materials to be reviewed submitted by the customers to be reviewed in the business system. The materials to be reviewed include customer photos and customer information;

[0149] A2. Extract the preset type information from the customer information, add the customer to be reviewed to the preset risk control relationship graph based on the preset type information, obtain multiple associated customers of the customer to be reviewed based on the risk control relationship graph, and obtain the photos corresponding to the multiple associated customers;

[0150] A3. Use an artificial intelligence template to extract the image backgrounds of the customer photos and the photos of the associated customers, and generate text information describing the image backgrounds. Calculate the image background similarity between the customer photos and the corresponding text information and each photo of the associated customers and the corresponding text information through a pre-constructed multimodal image model to obtain the background similarity values between the customer photos and the photos of each associated customer;

[0151] A4. Screen out the photos of the associated customers corresponding to the background similarity values greater than the preset threshold, and determine the risk level of the customer to be reviewed according to the number of the screened photos.

[0152] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments.

[0153] It should be noted that in this article, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements not only includes those elements but also includes other elements not expressly listed, or further includes elements inherent to such a process, device, article or method. Without further limitation, an element defined by the phrase "including one..." does not exclude the existence of additional identical elements in the process, device, article or method including the element.

[0154] Through the description of the above embodiments, those skilled in the art can clearly understand that the above embodiment methods can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, in essence, or the part that makes a contribution to the prior art can be embodied in the form of a software product. The computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions for causing a terminal device (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0155] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present invention.

Claims

1. A customer risk identification method, characterized in that, The method includes: Obtain the materials to be reviewed submitted by the customer to be reviewed in the business system, where the materials to be reviewed include customer photos and customer information; Extract the preset type information from the customer information, add the customer to be reviewed to the preset risk control relationship graph based on the preset type information, obtain multiple associated customers of the customer to be reviewed based on the risk control relationship graph, and obtain the photos corresponding to the multiple associated customers; Generate corresponding text information describing the photo background for the customer photo and the photos corresponding to the multiple associated customers through a pre-constructed multi-modal image model, and calculate the background similarity values between the customer photo and the photos of each associated customer respectively based on the customer photo and its text information and the photo and its text information of each associated customer; Screen out the photos of the associated customers corresponding to the background similarity values greater than the preset threshold, and determine the risk level of the customer to be reviewed according to the number of the screened photos.

2. The customer risk identification method according to claim 1, wherein, Before adding the customer to be reviewed to the preset risk control relationship graph based on the preset type information, it includes: Obtain the historical customer materials of multiple historical customers from the database of the business system, where the historical customer materials include historical customer information and historical customer identifiers; Extract the preset type information of the historical customers from the historical customer information; Set the association relationships between the historical customers according to the preset type information of the historical customers, and generate a risk control relationship graph according to the association relationships.

3. The customer risk identification method according to claim 2, characterized in that: The setting of the association relationships between the historical customers according to the preset type information and the generation of a risk control relationship graph according to the association relationships include: Set the corresponding association relationships for the historical customers with the same preset type information; Configure the relationship graph nodes corresponding to each historical customer according to the historical customer identifier corresponding to each historical customer; Configure the association relationships between the historical customers as the edges of the relationship graph, and connect the relationship graph nodes of the historical customers corresponding to the association relationships through the edges of the relationship graph to obtain a risk control relationship graph.

4. The customer risk identification method according to claim 1, characterized in that, The adding of the customer to be reviewed to the preset risk control relationship graph based on the preset type information includes: Obtain the customer identifier of the customer to be reviewed, and determine the relationship graph node corresponding to the customer to be reviewed in the risk control relationship graph based on the customer identifier; Determine the association relationship between the customer to be reviewed and the historical customers corresponding to the other relationship graph nodes in the risk control relationship graph according to the preset type information; Generate an edge connecting the relationship graph node corresponding to the customer to be reviewed and the other relationship graph nodes in the risk control relationship graph according to the association relationship.

5. The customer risk identification method according to claim 1, characterized in that, The obtaining of multiple associated customers of the customer to be reviewed based on the risk control relationship graph and the obtaining of the photos corresponding to the multiple associated customers include: Obtain multiple associated nodes connected to the relationship graph node corresponding to the customer to be reviewed in the risk control relationship graph; Obtain the associated customer identifiers corresponding to each associated node, and obtain the corresponding customer materials from the database of the business system according to the associated customer identifiers; Obtain the customer photo stored at the time point closest to the current time from the acquired customer data to obtain the photo corresponding to the associated customer.

6. The customer risk identification method according to claim 1, characterized in that, Based on the customer photo and its text information, and the photos and text information of each associated customer, calculate the background similarity values between the customer photo and the photos of each associated customer respectively, including: Input the customer photo and the associated customer photo into the image encoder of the multimodal image model respectively to obtain the first image feature vector corresponding to the customer photo and the second image feature vector corresponding to the associated customer photo; Input the text information corresponding to the customer photo and the text information corresponding to the associated customer photo into the text encoder of the multimodal image model respectively to obtain the first text feature vector corresponding to the customer photo and the second text feature vector corresponding to the associated customer photo; Calculate the similarity between the first image feature vector and the first text feature vector and the second image feature vector and the second text feature vector to obtain the background similarity values between the customer photo and the photos of each associated customer.

7. The customer risk identification method according to claim 1, wherein, Determine the risk level of the customer to be audited according to the number of selected photos, including: Obtain the associated data table of the preset number of background similar photos and the risk level from the database of the business system; Search for the corresponding risk level in the associated data table according to the number of selected photos to obtain the risk level corresponding to the data to be audited.

8. A customer risk identification device, characterized in that, The customer risk identification device includes: A data acquisition module, configured to acquire the data to be audited submitted by the customer to be audited in the business system, where the data to be audited includes a customer photo and customer information; A photo acquisition module, configured to extract the preset type information in the customer information, add the customer to be audited to the preset risk control relationship graph based on the preset type information, obtain multiple associated customers of the customer to be audited based on the risk control relationship graph, and obtain the photos corresponding to the multiple associated customers; A similarity calculation module, configured to generate the text information describing the photo background corresponding to the customer photo and the photos corresponding to the multiple associated customers through a pre-constructed multimodal image model, and based on the customer photo and its text information, and the photos and text information of each associated customer, calculate the background similarity values between the customer photo and the photos of each associated customer respectively; A risk identification module, configured to screen out the photos of the associated customers corresponding to the background similarity values greater than the preset threshold, and determine the risk level of the customer to be audited according to the number of selected photos.

9. A server, characterized in that, The server includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the customer risk identification method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, The computer program, when executed by a processor, implements the customer risk identification method according to any one of claims 1 to 7.

Images