Financial product recommendation method and device, equipment and medium
By generating virtual interaction records for inactive customers and building a knowledge graph, the problem of inaccurate financial product recommendations caused by sparse customer interaction records is solved, and higher recommendation accuracy and credibility are achieved.
Patent Information
- Application Number
- CN202510635885.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-16
- Publication Date
- 2025-08-22
AI Technical Summary
In the prior art, the interaction records between customers and bank products are sparse, resulting in low accuracy in financial product recommendations and prone to overfitting problems.
By generating virtual interaction records of inactive customers, building a product recommendation knowledge graph, combining actual interaction records, and recommending target financial products.
It improves the accuracy of financial product recommendations, avoids overfitting problems when data is sparse, and enhances the credibility of recommendations.
Smart Images

Figure CN120525612A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of data analysis technology, and in particular to a method, device, equipment and medium for recommending financial products. Background Art
[0002] Analyzing and tapping into customer potential to recommend suitable financial products is a common marketing strategy for banks, often achieved through the construction of a recommendation system. Based on a customer's interaction history with financial products, the recommendation system infers and outputs financial products that the customer might be interested in. The accuracy of the recommendation system depends largely on the density of interaction records between customers and financial products in the database. The denser the interaction records, the more accurate the financial products recommended by the system.
[0003] However, in real-world scenarios, customer interaction records with banking products are often very sparse, especially since many customers prefer to purchase financial products online rather than through traditional banking channels. For example, a customer may have only made a few simple savings transactions at a bank, or may only have monthly salary deposit and withdrawal records. Using these few interaction records to predict a large amount of unknown information increases the risk of overfitting, resulting in low usability of recommendation results. Summary of the Invention
[0004] The present invention provides a financial product recommendation method, apparatus, device, and medium to improve the accuracy of financial product recommendations and avoid overfitting problems.
[0005] According to one aspect of the present invention, a method for recommending financial products is provided, comprising:
[0006] generating a virtual interaction record of the inactive customer based on the customer profile of the inactive customer;
[0007] generating a product recommendation knowledge graph based on the virtual interaction records and actual interaction records of the inactive customers;
[0008] Determine a target financial product to recommend to the inactive customer based on the product recommendation knowledge graph.
[0009] According to another aspect of the present invention, a device for recommending financial products is provided, comprising:
[0010] A record generation module, configured to generate a virtual interaction record of an inactive customer based on the customer profile of the inactive customer;
[0011] A graph generation module, configured to generate a product recommendation knowledge graph based on the virtual interaction records and actual interaction records of the inactive customers;
[0012] A product recommendation module is used to determine a target financial product to recommend to the inactive customer based on the product recommendation knowledge graph.
[0013] According to another aspect of the present invention, a computer program product is provided, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements the method for recommending financial products according to any embodiment of the present invention.
[0014] According to another aspect of the present invention, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the method for recommending financial products described in any embodiment of the present invention.
[0015] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the financial product recommendation method according to any embodiment of the present invention when executed.
[0016] The embodiment of the present invention completes the interaction records of inactive customers so that dense rather than sparse interaction records can be used to recommend financial products in the future, thereby avoiding the overfitting problem when data is sparse and improving the accuracy of financial product recommendations.
[0017] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present invention, nor is it intended to limit the scope of the present invention. Other features of the present invention will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0019] Figure 1 is a flowchart of a method for recommending financial products according to an embodiment of the present invention;
[0020] Figure 2 is a flowchart of a method for recommending financial products according to another embodiment of the present invention;
[0021] Figure 3 is a schematic structural diagram of a financial product recommendation device provided according to another embodiment of the present invention;
[0022] Figure 4 It is a schematic structural diagram of an electronic device implementing an embodiment of the present invention. DETAILED DESCRIPTION
[0023] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.
[0024] It should be noted that the terms "first," "second," and the like in the present invention are used to distinguish similar objects and are not necessarily used to describe a specific order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including," "having," and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to these processes, methods, products, or apparatuses.
[0025] Figure 1 This is a flowchart of a method for recommending financial products provided by an embodiment of the present invention. This embodiment is applicable to situations where financial products need to be recommended to customers, but the interaction information between customers and bank products is very sparse. The method can be executed by a financial product recommendation device, which can be implemented in the form of hardware and / or software. The device can be configured in an electronic device with corresponding data processing capabilities, such as the recommendation system of the Agricultural Bank of China. Figure 1 As shown, the method includes:
[0026] S110 : Generate a virtual interaction record of the inactive customer based on the customer portrait of the inactive customer.
[0027] S120: Generate a product recommendation knowledge graph based on the virtual interaction records and actual interaction records of the inactive customers.
[0028] S130. Determine a target financial product to recommend to the inactive customer based on the product recommendation knowledge graph.
[0029] Inactive customers are those with sparse interaction records with financial products (e.g., fewer than a preset number of interaction records). Active customers are those with frequent interaction records (e.g., more than a preset number of interaction records). Interaction records include browsing history, inquiry history, and purchase history of financial products.
[0030] Specifically, for active customers, the system can directly determine recommended financial products based on their interaction records. For inactive customers, whose interaction records are relatively few, if the system determines recommended financial products directly based on these records, it is likely to encounter the overfitting problem analyzed above. Therefore, for inactive customers, their interaction records need to be expanded before recommendations are made. Based on the static data of the inactive customers in the database, a customer profile is generated. Based on this customer profile, active customers with similar static data to the inactive customers are searched in the database as reference objects. The interaction records of active customers are retrieved from the database and used as a reference to generate virtual interaction records for the inactive customers.
[0031] Virtual and actual interaction records are collated to generate target interaction records. Entities and entity relationships within these target interaction records are defined and stored using a graph database to generate a product recommendation knowledge graph. When recommending financial products, the entities and relationships within the product recommendation knowledge graph are used to identify target financial products that inactive customers may be interested in and then recommend them.
[0032] The embodiment of the present invention completes the interaction records of inactive customers so that dense rather than sparse interaction records can be used to recommend financial products in the future, thereby avoiding the overfitting problem when data is sparse and improving the accuracy of financial product recommendations.
[0033] Based on the above embodiment, optionally, before generating the product recommendation knowledge graph based on the virtual interaction records and actual interaction records of the inactive customers, the method further includes:
[0034] For each virtual interaction record, determining whether the virtual interaction record complies with the domain expert rules;
[0035] If the virtual interaction record does not comply with the domain expert rules, the virtual interaction record is deleted;
[0036] If the virtual interaction record complies with the domain expert rules, the virtual interaction record is retained.
[0037] Specifically, domain expert rules are manually formulated by experts (for example, customers aged > 60 will not purchase high-risk financial products) and input into the recommendation system as the primary means of verifying the rationality of virtual interaction records. For each generated virtual interaction record, the virtual interaction record is compared one by one with the domain expert rules in the database. If the virtual interaction record meets all the domain expert rules, the virtual interaction record is retained; if the virtual interaction record does not meet any domain expert rules, the virtual interaction record is deleted. By referring to the domain expert rules, it can be ensured that the generated virtual transaction records comply with business logic and risk control requirements, ensuring that only reasonable virtual interaction records are adopted.
[0038] Figure 2 This is a flowchart of a method for recommending financial products provided by another embodiment of the present invention. This embodiment is optimized and improved on the basis of the above embodiment. Figure 2 As shown, the method includes:
[0039] S210: Generate a customer profile of the inactive customer based on static attributes of the inactive customer; and determine active customers similar to the inactive customer based on the customer profile.
[0040] S220: Generate virtual interaction records of the inactive customers according to the high-frequency interaction records of the active customers.
[0041] Among them, static attributes refer to the basic information of customers that does not change, usually including demographic characteristics (such as age, gender, and region), device attributes (such as mobile phone model and operating system), account attributes (such as registration channel and membership level), and socioeconomic attributes (such as occupation and income level).
[0042] Specifically, active and inactive customers are screened from the entire customer base, and customer profiles are generated for each based on static attributes. Based on their respective customer profiles, the similarity between inactive and active customers is calculated. Active customers whose similarity with inactive customers exceeds a threshold are identified as active customers similar to the inactive customers. For these active customers, their interaction records are extracted from the database and statistically analyzed to identify high-frequency interaction records. Based on these interaction records, virtual interaction records are generated for these inactive customers. By comparing customer profiles to identify active customers similar to inactive customers, efficient screening of active customers is achieved, ensuring the relative reliability of the generated virtual interaction records.
[0043] Based on the above embodiment, optionally, generating the virtual interaction record of the inactive customer according to the high-frequency interaction record of the active customer includes:
[0044] Determining a consumption habit model of the active customer based on the high-frequency interaction records of the active customer;
[0045] Based on the consumption habit model, the values of the high-frequency interaction records are randomly adjusted to obtain virtual interaction records of the inactive customers.
[0046] Specifically, multi-dimensional cluster analysis and association rule mining are performed on high-frequency interaction records to identify customer behavior patterns in different scenarios, which serve as models of their consumption habits. Once the consumption habit model is defined, virtual interaction records for inactive customers can be generated by randomly adjusting the numerical features in their high-frequency interaction records. This "virtualization" process aims to construct behavioral samples that meet business requirements but are distinct from those of active customers, facilitating subsequent financial product recommendations. Specific operations include: introducing noise perturbations to key fields such as time interval, interaction frequency, and spending amount in high-frequency interaction records; randomly shifting values using Gaussian or uniform distributions; performing random replacement or resampling on categorical variables; and even introducing a certain percentage of missing data to simulate the low-frequency and erratic behavior of inactive customers to generate virtual interaction records. It should be noted that the generation of virtual interaction records is not completely random; rather, it should adhere to specific business logic and statistical rules (i.e., the previously generated consumption habit model) to ensure that the overall distribution of the generated data is consistent with the real data. For example, if the average visit frequency of an active customer is five times per week, the virtual records for inactive customers can be set to one to two times per week, with random fluctuations within this range. By analyzing the high-frequency interaction records of active customers, a consumption habit model is established, and parameters are reasonably adjusted to generate virtual behavior data of inactive customers, thereby improving the depth and breadth of virtual interaction records.
[0047] Based on the above embodiment, optionally, generating the virtual interaction record of the inactive customer based on the high-frequency interaction record of the active customer includes:
[0048] Perform feature engineering preprocessing on the high-frequency interaction records of the jump customers;
[0049] The high-frequency interaction records preprocessed by the feature engineering are input into a generator in a generative adversarial network to obtain virtual interaction records of inactive customers generated by the generator.
[0050] Generative Adversarial Networks (GANs) are a type of deep learning model. The core idea of GANs is to train two neural networks through a game of mutual competition: one is the generator and the other is the discriminator.
[0051] Specifically, the generative adversarial network is pre-trained to train the generator's ability to generate new interaction records based on existing interaction records. The discriminator compares existing interaction records with new interaction records, outputs the probability that the new interaction record is true, and optimizes the generator's network structure and parameters based on the probability until the new interaction record generated by the generator can pass the judgment of the discriminator, and the training is completed. When using a generative adversarial network, the high-frequency interaction records of active customers are first pre-processed with feature engineering, including desensitization and deduplication. The high-frequency interaction records that have undergone feature engineering pre-processing are input into the generator of the generative adversarial network to obtain the interaction records output by the generator, which are determined to be virtual interaction records. By training the generative adversarial network to generate virtual interaction records, the efficiency of generating virtual interaction records is improved.
[0052] S230: Generate a product recommendation knowledge graph based on the virtual interaction records and actual interaction records of the inactive customers.
[0053] S240: Determine a target financial product to be recommended to the inactive customer based on the product recommendation knowledge graph.
[0054] S250: Update the virtual interaction record according to the consumption behavior data of the inactive customer on the target financial product.
[0055] Among them, consumer behavior data includes conversion behavior data (direct conversion behaviors such as product purchase, trial application, online consultation, etc.), behavior trajectory data (page jump path, functional module usage timing, page exit node), feedback data (product ratings, complaint records, satisfaction survey results), and transaction data (actual purchase amount, payment method, capital flow, etc.).
[0056] Specifically, since virtual interaction records can be generated, they can also be modified, as opposed to actual interaction records. After obtaining the inactive customer's consumption behavior data for the target financial product, the recommended consumption behavior data is analyzed to determine if it is relevant to the previously generated virtual interaction records. If so, the virtual interaction records are updated accordingly. If not, the update to the virtual interaction records is abandoned. By optimizing and updating the virtual interaction records, the credibility of the virtual interaction records and the target financial product is further enhanced.
[0057] Based on the above embodiment, optionally, updating the virtual interaction record according to the consumption behavior data of the inactive customer on the target financial product includes:
[0058] determining, according to the product type of the target financial product, a target virtual interaction record related to the target financial product;
[0059] Based on the consumption behavior data of the inactive customer on the target financial product, the weight of the target virtual interaction record when generating the product recommendation knowledge graph is updated.
[0060] Specifically, based on the target financial product's product type, its key functions and service features are analyzed. Based on this information, target virtual interaction records directly related to the target financial product are screened from all virtual interaction records. For example, if the target financial product is a specific investment fund product, the target virtual interaction record would be the customer's virtual browsing history of similar investment fund products. While both actual and virtual interaction records are referenced when generating the product recommendation knowledge graph, the weights assigned to them differ. Actual interaction records have a higher reference value; that is, when generating the product recommendation knowledge graph, actual interaction records have a higher initial weight (e.g., 0.8) while virtual interaction records have a lower initial weight (e.g., 0.2). Based on consumer behavior data, the customer's propensity to consume the target financial product is determined. If the propensity is positive, the weight of the target virtual interaction record in generating the product recommendation knowledge graph is increased (e.g., 0.3); if the propensity is negative, the weight of the target virtual interaction record in generating the product recommendation knowledge graph is decreased (e.g., 0.1). Subsequent generation of new product recommendation knowledge graphs uses the updated weights instead of the original weights. By setting different weights for actual interaction records and virtual interaction records and dynamically updating them, the reference value of the product recommendation knowledge graph is improved.
[0061] This application further improves the credibility of virtual interaction records and target financial products by optimizing and updating virtual interaction records.
[0062] Figure 3 This is a schematic diagram of a financial product recommendation device provided by another embodiment of the present invention. Figure 3 As shown, the device includes:
[0063] A record generation module 310 is configured to generate a virtual interaction record of an inactive customer based on the customer profile of the inactive customer;
[0064] A graph generation module 320 is configured to generate a product recommendation knowledge graph based on the virtual interaction records and actual interaction records of the inactive customers;
[0065] The product recommendation module 330 is configured to determine a target financial product to recommend to the inactive customer based on the product recommendation knowledge graph.
[0066] The financial product recommendation device provided in the embodiment of the present invention can execute the financial product recommendation method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.
[0067] Optionally, the record generation module 310 includes:
[0068] A portrait generating unit, configured to generate a customer portrait of an inactive customer based on static attributes of the inactive customer;
[0069] a customer determining unit, configured to determine an active customer similar to the inactive customer based on the customer portrait;
[0070] A record generating unit is configured to generate a virtual interaction record of the inactive customer based on the high-frequency interaction record of the active customer.
[0071] Optionally, the record generation unit is specifically used to: determine the consumption habit model of the active customer based on the high-frequency interaction records of the active customer; based on the consumption habit model, randomly adjust the numerical value of the high-frequency interaction record to obtain the virtual interaction record of the inactive customer.
[0072] Optionally, the record generation unit is specifically used to: perform feature engineering preprocessing on the high-frequency interaction records of the active customers; input the high-frequency interaction records after feature engineering preprocessing into the generator in the generative adversarial network to obtain virtual interaction records of inactive customers generated by the generator.
[0073] Optionally, the device further includes:
[0074] An audit module, configured to determine, for each virtual interaction record, whether the virtual interaction record complies with domain expert rules;
[0075] A deletion module, configured to delete the virtual interaction record if the virtual interaction record does not comply with the domain expert rule;
[0076] The retaining module is configured to retain the virtual interaction record if the virtual interaction record complies with the domain expert rule.
[0077] Optionally, the device further includes:
[0078] An updating module is used to update the virtual interaction record according to the consumption behavior data of the inactive customer on the target financial product.
[0079] Optionally, the update module is specifically used to: determine the target virtual interaction records related to the target financial product based on the product type of the target financial product; and update the weight of the target virtual interaction records when generating the product recommendation knowledge graph based on the consumption behavior data of the inactive customers on the target financial product.
[0080] The financial product recommendation device further described can also execute the financial product recommendation method provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.
[0081] Figure 4 A schematic diagram of the structure of an electronic device 40 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or claimed herein.
[0082] like Figure 4 As shown, the electronic device 40 includes at least one processor 41 and a memory, such as a read-only memory (ROM) 42, a random access memory (RAM) 43, etc., which is communicatively connected to the at least one processor 41. The memory stores a computer program that can be executed by the at least one processor, and the processor 41 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 42 or the computer program loaded from the storage unit 48 into the random access memory (RAM) 43. Various programs and data required for the operation of the electronic device 40 can also be stored in the RAM 43. The processor 41, ROM 42, and RAM 43 are connected to each other via a bus 44. An input / output (I / O) interface 45 is also connected to the bus 44.
[0083] Multiple components in the electronic device 40 are connected to the I / O interface 45, including an input unit 46, such as a keyboard, a mouse, etc.; an output unit 47, such as various types of displays, speakers, etc.; a storage unit 48, such as a magnetic disk, an optical disk, etc.; and a communication unit 49, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 49 allows the electronic device 40 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.
[0084] Processor 41 can be any general-purpose and / or specialized processing component with processing and computing capabilities. Some examples of processor 41 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any other suitable processor, controller, microcontroller, etc. Processor 41 executes the various methods and processes described above, such as the method for recommending financial products.
[0085] In some embodiments, the financial product recommendation method may be implemented as a computer program tangibly embodied in a computer-readable storage medium, such as storage unit 48. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 40 via ROM 42 and / or communication unit 49. When the computer program is loaded into RAM 43 and executed by processor 41, one or more steps of the financial product recommendation method described above may be performed. Alternatively, in other embodiments, processor 41 may be configured to execute the financial product recommendation method in any other suitable manner (e.g., via firmware).
[0086] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.
[0087] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer program is executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer program may be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.
[0088] In the context of the present invention, computer-readable storage media can be tangible media that can contain or store a computer program for use with an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. Computer-readable storage media can include but are not limited to electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, computer-readable storage media can be machine-readable signal media. More specific examples of machine-readable storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.
[0089] To provide interaction with a customer, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the customer; and a keyboard and pointing device (e.g., a mouse or trackball) through which the customer can provide input to the electronic device. Other types of devices can also be used to provide interaction with the customer; for example, the feedback provided to the customer can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and the input from the customer can be received in any form (including acoustic input, voice input, or tactile input).
[0090] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a client computer with a graphical client interface or web browser through which a client can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.
[0091] A computing system may include clients and servers. The clients and servers are typically remote from each other and typically interact via a communication network. This client-server relationship arises through computer programs running on the respective computers, creating a client-server relationship. The server may be a cloud server, also known as a cloud computing server or cloud host. This server is a hosting product within the cloud computing service ecosystem that addresses the management difficulties and limited scalability of traditional physical hosting and VPS services.
[0092] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in the present invention can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved. This is not limited herein.
[0093] The above specific embodiments do not limit the scope of protection of the present invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.
Claims
1. A method for recommending financial products, characterized in that: The method comprises: generating a virtual interaction record of the inactive customer based on the customer profile of the inactive customer; generating a product recommendation knowledge graph based on the virtual interaction records and actual interaction records of the inactive customers; Determine a target financial product to recommend to the inactive customer based on the product recommendation knowledge graph.
2. The method according to claim 1, characterized in that Generating the virtual interaction record of the inactive customer according to the customer portrait of the inactive customer includes: generating a customer profile of the inactive customer based on the static attributes of the inactive customer; determining active customers similar to the inactive customers based on the customer profile; The virtual interaction records of the inactive customers are generated according to the high-frequency interaction records of the active customers.
3. The method according to claim 2, characterized in that Generating the virtual interaction record of the inactive customer according to the high-frequency interaction record of the active customer includes: Determining a consumption habit model of the active customer based on the high-frequency interaction records of the active customer; Based on the consumption habit model, the values of the high-frequency interaction records are randomly adjusted to obtain virtual interaction records of the inactive customers.
4. The method according to claim 2, characterized in that Generating the virtual interaction record of the inactive customer according to the high-frequency interaction record of the active customer includes: Perform feature engineering preprocessing on the high-frequency interaction records of the jump customers; The high-frequency interaction records preprocessed by the feature engineering are input into a generator in a generative adversarial network to obtain virtual interaction records of inactive customers generated by the generator.
5. The method according to claim 1, characterized in that Before generating the product recommendation knowledge graph based on the virtual interaction records and actual interaction records of the inactive customers, the method further includes: For each virtual interaction record, determining whether the virtual interaction record complies with the domain expert rules; If the virtual interaction record does not comply with the domain expert rules, the virtual interaction record is deleted; If the virtual interaction record complies with the domain expert rules, the virtual interaction record is retained.
6. The method according to claim 1, characterized in that After determining the target financial product to be recommended to the inactive customer based on the product recommendation knowledge graph, the method further includes: The virtual interaction record is updated according to the consumption behavior data of the inactive customer on the target financial product.
7. The method according to claim 6, characterized in that The updating of the virtual interaction record based on the consumption behavior data of the inactive customer on the target financial product includes: determining, according to the product type of the target financial product, a target virtual interaction record related to the target financial product; Based on the consumption behavior data of the inactive customer on the target financial product, the weight of the target virtual interaction record when generating the product recommendation knowledge graph is updated.
8. A financial product recommendation device, characterized in that: The device comprises: A record generation module, configured to generate a virtual interaction record of an inactive customer based on the customer profile of the inactive customer; A graph generation module, configured to generate a product recommendation knowledge graph based on the virtual interaction records and actual interaction records of the inactive customers; A product recommendation module is used to determine a target financial product to recommend to the inactive customer based on the product recommendation knowledge graph.
9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to perform the financial product recommendation method according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the financial product recommendation method according to any one of claims 1 to 7 when executed.
