Financial product recommendation method and device, storage medium and electronic equipment

Abstract

The invention discloses a financial product recommendation method and device, a storage medium and electronic equipment, and relates to the field of financial science and technology. The method comprises the steps of collecting a user image of a target user under the condition that the target user is authorized; querying a historical purchase record of the target user for the financial product according to the user image of the target user; determining at least one target financial product to be recommended to the target user according to the user feature portrait of the target user when it is detected that the number of the historical purchase records of the target user is smaller than a preset threshold value; and when it is detected that the number of the historical purchase records of the target user is greater than or equal to a preset threshold value, determining at least one target financial product to be recommended to the target user according to the product association degree between the financial products purchased by the target user and the financial products which are not purchased. The technical problem that in the prior art, the recommendation accuracy of financial products is poor is solved.

Description

Technical Field

[0001] The present application relates to the field of financial technology, and specifically, to a method, device, storage medium and electronic device for recommending financial products. Background Art

[0002] With the rapid development of science and technology, smart banking has become an important driving force for the transformation and upgrading of the banking industry. In smart banking, customer experience is one of the core elements, and personalized product recommendations can effectively improve customer satisfaction and loyalty, which is a key function of smart banking. However, traditional bank product recommendation methods rely on manual inquiries by customers or recommendations by bank employees, ignoring the personalized needs of customers and failing to provide accurate and timely recommendations. To solve this problem, many banks have begun to adopt face recognition technology based on image recognition to achieve more accurate, fast and personalized services.

[0003] At present, the application of facial recognition technology in the banking field is relatively mature, mainly in terms of customer identity verification and security control. Using facial recognition technology, banks can quickly and accurately identify customer identities, improve service efficiency, and reduce the time required for identity verification. In addition, facial recognition technology can also be used for monitoring and security prevention. Through real-time monitoring, banks can detect abnormal behavior in a timely manner and take corresponding security measures. However, the use of facial recognition technology alone for personalized product recommendations still has the problem of poor accuracy in recommending financial products.

[0004] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0005] The main purpose of the present application is to provide a method, device, storage medium and electronic device for recommending financial products, so as to at least solve the technical problem of poor accuracy in recommending financial products in the prior art.

[0006] In order to achieve the above-mentioned purpose, according to one aspect of an embodiment of the present application, a method for recommending financial products is provided, including: with the authorization of the target user, collecting a user image of the target user; querying the target user's historical purchase records for financial products based on the user image of the target user; when it is detected that the number of historical purchase records of the target user is less than a preset threshold, determining at least one target financial product to be recommended to the target user based on the user feature portrait of the target user; when it is detected that the number of historical purchase records of the target user is greater than or equal to the preset threshold, determining at least one target financial product to be recommended to the target user based on the product correlation between the financial products purchased by the target user and the financial products not purchased by the target user, wherein the product correlation represents the probability that different financial products are purchased simultaneously by the same group of people.

[0007] Optionally, querying the historical purchase record of the target user according to the user image of the target user includes: determining the unique user identifier of the target user according to the user image of the target user; and querying the historical purchase record of the target user in a database according to the unique user identifier.

[0008] Optionally, before determining at least one target financial product to be recommended to the target user based on the user feature portrait of the target user, the financial product recommendation method also includes: when it is detected that the number of historical purchase records of the target user is less than a preset threshold, preprocessing the user image of the target user, wherein the preprocessing is used to remove noise information in the user portrait and adjust the size and / or format of the user portrait; determining first-category user features and second-category user features of the target user based on the preprocessed user image, wherein the first-category user features belong to the character features of the target user shown in the image, and the second-category user features belong to user features indirectly inferred based on the first-category user features; and generating a user feature portrait of the target user based on the first-category user features and the second-category user features of the target user.

[0009] Optionally, determining at least one target financial product to be recommended to the target user based on the user feature profile of the target user includes: when the user feature profile includes N user features, obtaining the probability that a reference user with the ith user feature purchases the jth financial product, and using the probability as the recommendation score corresponding to the jth financial product and the ith user feature, wherein N is an integer greater than or equal to 1, i is a positive integer less than or equal to N, and the jth financial product is any financial product to be sold; summing up the N recommendation scores corresponding to the jth financial product and the N user features, respectively, and determining the target recommendation score of the jth financial product based on the calculation result; and determining at least one target financial product to be recommended to the target user from all the financial products to be sold based on the target recommendation score of each financial product to be sold.

[0010] Optionally, N recommendation scores corresponding to the j-th financial product and N user characteristics are summed up, and a target recommendation score of the j-th financial product is determined based on the calculation results, including: summing up the N recommendation scores corresponding to the j-th financial product and N user characteristics to obtain a total score; obtaining the historical rate of return of the j-th financial product, wherein the historical rate of return represents the historical rate of return of the j-th financial product for the financial institution or the historical rate of return for the product holder; and calculating the product of the historical rate of return of the j-th financial product and the total score to obtain the target recommendation score of the j-th financial product.

[0011] Optionally, obtaining the probability that a reference user with the i-th user characteristic purchases the j-th financial product includes: determining the total number of reference users with the i-th user characteristic from a database to obtain a first number; determining the number of users with the i-th user characteristic who have purchased the j-th financial product from the database to obtain a second number; and calculating the ratio of the second number to the first number to obtain the probability that the reference user with the i-th user characteristic purchases the j-th financial product.

[0012] Optionally, when it is detected that the number of historical purchase records of the target user is greater than or equal to a preset threshold, determining at least one target financial product to be recommended to the target user according to product associations between financial products purchased by the target user and financial products not purchased by the target user, including: when the target user has purchased M financial products, obtaining a probability that a user who has purchased the kth financial product also purchased the hth financial product, and using the probability as an association score corresponding to the hth financial product and the kth financial product, wherein M is an integer greater than or equal to 1, the kth financial product is any one of the M financial products, and the hth financial product is any one to-be-sold financial product that the target user does not hold; summing up the association scores corresponding to the hth financial product and the M financial products, respectively, and determining a target association score for the hth financial product according to the calculation result; and determining at least one target financial product to be recommended to the target user from all financial products to-be-sold that the target user does not hold, according to the target association score of each to-be-sold financial product that the target user does not hold.

[0013] Optionally, the association scores corresponding to the h-th financial product and the M financial products are summed up, and the target association score of the h-th financial product is determined according to the calculation result, including: summing up the association scores corresponding to the h-th financial product and the M financial products to obtain a summed result; obtaining the historical rate of return of the h-th financial product, wherein the historical rate of return of the h-th financial product represents the historical rate of return of the h-th financial product for the financial institution or the historical rate of return for the product holder; calculating the product of the historical rate of return of the h-th financial product and the summed result to obtain the target association score of the h-th financial product.

[0014] Optionally, obtaining the probability that a user who purchased the kth financial product also purchased the hth financial product includes: determining from a database the number of users who purchased both the kth financial product and the hth financial product to obtain a third number; determining from a database the number of users who purchased the kth financial product to obtain a fourth number; and calculating a ratio of the third number to the fourth number to obtain the probability that a user who purchased the kth financial product also purchased the hth financial product.

[0015] In order to achieve the above-mentioned purpose, according to another aspect of an embodiment of the present application, a financial product recommendation device is further provided, comprising: a collection unit, which collects a user image of the target user with the authorization of the target user; a query unit, which queries the target user's historical purchase record for financial products based on the user image of the target user; a first determination unit, which determines at least one target financial product to be recommended to the target user based on the user feature portrait of the target user when it is detected that the number of historical purchase records of the target user is less than a preset threshold; and a second determination unit, which determines at least one target financial product to be recommended to the target user based on the product correlation between the financial products purchased by the target user and the financial products not purchased by the target user when it is detected that the number of historical purchase records of the target user is greater than or equal to the preset threshold, wherein the product correlation represents the probability that different financial products are purchased simultaneously by the same group of people.

[0016] Optionally, the query unit includes: a first determining subunit and a first query subunit. The first determining subunit is used to determine the target user's unique user identifier according to the target user's user image; and the first query subunit is used to query the target user's historical purchase record in the database according to the unique user identifier.

[0017] Optionally, the first determination unit includes: a first processing subunit, a second determination subunit, and a first generation subunit. The first processing subunit is used to preprocess the user image of the target user when it is detected that the number of historical purchase records of the target user is less than a preset threshold, wherein the preprocessing is used to remove noise information in the user portrait and adjust the size and / or format of the user portrait; the second determination subunit is used to determine the first type of user features and the second type of user features of the target user based on the preprocessed user image, wherein the first type of user features belong to the character features of the target user shown in the image, and the second type of user features belong to the user features indirectly inferred based on the first type of user features; the first generation subunit is used to generate a user feature portrait of the target user based on the first type of user features and the second type of user features of the target user.

[0018] Optionally, the first determination unit includes: a first acquisition subunit, a third determination subunit, and a fourth determination subunit. The first acquisition subunit is used to obtain the probability of a reference user with the i-th user feature purchasing the j-th financial product when the user feature profile includes N user features, and use the probability as the recommendation score corresponding to the j-th financial product and the i-th user feature, wherein N is an integer greater than or equal to 1, i is a positive integer less than or equal to N, and the j-th financial product is any financial product to be sold; the third determination subunit is used to sum and calculate the N recommendation scores corresponding to the j-th financial product and the N user features, and determine the target recommendation score of the j-th financial product according to the calculation result; the fourth determination subunit is used to determine at least one target financial product to be recommended to the target user from all the financial products to be sold according to the target recommendation score of each financial product to be sold.

[0019] Optionally, the third determination subunit includes: a first calculation module, a first acquisition module, and a second calculation module. The first calculation module is used to sum up the N recommendation scores corresponding to the j-th financial product and the N user characteristics to obtain the summed total score; the first acquisition module is used to obtain the historical rate of return of the j-th financial product, wherein the historical rate of return represents the historical rate of return of the j-th financial product for the financial institution or the historical rate of return for the product holder; the second calculation module is used to calculate the product of the historical rate of return of the j-th financial product and the summed total score to obtain the target recommendation score of the j-th financial product.

[0020] Optionally, the first acquisition subunit includes: a first determination module, a second determination module, and a third calculation module. The first determination module is used to determine the total number of reference users with the i-th user feature from the database to obtain a first number; the second determination module is used to determine the number of users with the i-th user feature who have purchased the j-th financial product from the database to obtain a second number; and the third calculation module is used to calculate the ratio of the second number to the first number to obtain the probability that the reference user with the i-th user feature purchases the j-th financial product.

[0021] Optionally, the second determination unit includes: a second acquisition subunit, a first calculation subunit, and a fifth determination subunit. The second acquisition subunit is used to obtain the probability that the user who has purchased the kth financial product also purchased the hth financial product when the target user has purchased M financial products, and use the probability as the correlation score corresponding to the hth financial product and the kth financial product, wherein M is an integer greater than or equal to 1, the kth financial product is any one of the M financial products, and the hth financial product is any one of the to-be-sold financial products that the target user does not hold; the first calculation subunit is used to sum the correlation scores corresponding to the hth financial product and the M financial products, respectively, and determine the target correlation score of the hth financial product according to the calculation result; the fifth determination subunit is used to determine at least one target financial product to be recommended to the target user from all the to-be-sold financial products that the target user does not hold, according to the target correlation score of each to-be-sold financial product that the target user does not hold.

[0022] Optionally, the first calculation subunit includes: a fourth calculation module, a second acquisition module, and a fifth calculation module. The fourth calculation module is used to sum the association scores corresponding to the h-th financial product and the M financial products to obtain a sum result; the second acquisition module is used to obtain the historical yield of the h-th financial product, wherein the historical yield of the h-th financial product represents the historical yield of the h-th financial product for the financial institution or the historical yield for the product holder; the fifth calculation module is used to calculate the product of the historical yield of the h-th financial product and the sum result to obtain the target association score of the h-th financial product.

[0023] Optionally, the second acquisition subunit includes: a third determination module, a fourth determination module, and a sixth calculation module. The third determination module is used to determine the number of users who purchased the kth financial product and the hth financial product from the database to obtain a third number; the fourth determination module is used to determine the number of users who purchased the kth financial product from the database to obtain a fourth number; and the sixth calculation module is used to calculate the ratio of the third number to the fourth number to obtain the probability that the user who purchased the kth financial product also purchased the hth financial product.

[0024] According to another aspect of an embodiment of the present application, there is also provided an electronic device, including: a memory storing an executable program; and a processor for running the program, wherein the above-mentioned method for recommending financial products is executed when the program is running.

[0025] According to another aspect of an embodiment of the present application, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored executable program, wherein when the executable program is running, the device where the computer-readable storage medium is located is controlled to execute the above-mentioned financial product recommendation method.

[0026] According to another aspect of an embodiment of the present application, a computer program product is also provided, including computer instructions, which implement the steps of the above-mentioned financial product recommendation method when executed by a processor.

[0027] In an embodiment of the present application, first, with the authorization of the target user, a user image of the target user is collected, then the target user's historical purchase record for financial products is queried based on the target user's user image, and then, when it is detected that the number of historical purchase records of the target user is less than a preset threshold, at least one target financial product to be recommended to the target user is determined based on the user feature portrait of the target user, and finally, when it is detected that the number of historical purchase records of the target user is greater than or equal to the preset threshold, at least one target financial product to be recommended to the target user is determined based on the product correlation between the financial products purchased by the target user and the financial products not purchased by the target user, wherein the product correlation represents the probability of different financial products being purchased simultaneously by the same group of people.

[0028] From the above content, it can be seen that this application first collects customer image information, including facial images and full body features, through a high-precision camera installed on the self-service number-taking machine with the customer's explicit authorization. This step ensures that the customer's data is collected and used with their knowledge and consent, in compliance with laws and regulations, and protects customer privacy.

[0029] Then, using advanced image analysis algorithms, the customer's unique identity is identified and determined from the collected user images. Subsequently, the customer's previous financial product purchase history is queried in the bank's database based on this identity to provide a basis for subsequent recommendation strategies.

[0030] Secondly, for customers with insufficient historical transaction records (new customers or customers with a small number of transaction records), the method of this application preprocesses the user image, removes unnecessary noise, adjusts the size and format, and uses the model to extract the key features of the customer. Based on the features inferred directly and indirectly, a user feature profile is generated, based on which financial products that may be of interest are recommended.

[0031] In addition, for customers with rich historical transaction records, this application adopts a product recommendation algorithm based on association rules. Through in-depth mining of customers' historical purchasing behaviors, the potential correlation between products is identified, that is, which products are more likely to be selected by customers in the same group at the same time. Based on this analysis, financial products that are highly correlated with their previous purchasing behaviors are recommended to target customers, thereby improving the accuracy and acceptance of recommendations.

[0032] It can be seen that through the technical solution of this application, the purpose of using image recognition technology to make personalized financial product recommendations to customers visiting bank branches is achieved, thereby achieving accurate product recommendations, improving the efficiency of bank branches, and enhancing customer satisfaction and loyalty, thereby solving the technical problem of poor accuracy in recommending financial products in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0034] Figure 1 A hardware structure block diagram of a computer terminal for implementing a financial product recommendation method is shown;

[0035] Figure 2 is a flowchart of an optional financial product recommendation method according to an embodiment of the present application;

[0036] Figure 3 It is a flowchart of an optional technical solution of a financial product recommendation method according to an embodiment of the present application;

[0037] Figure 4 is an optional recommendation flow chart of a feature-weighted recommendation algorithm according to an embodiment of the present application;

[0038] Figure 5 is an optional recommendation flow chart of a recommendation algorithm based on association rules according to an embodiment of the present application;

[0039] Figure 6 is a recommendation overall flow chart of an optional financial product recommendation method according to an embodiment of the present application;

[0040] Figure 7 is a schematic diagram of an optional recommendation method according to an embodiment of the present application;

[0041] Figure 8 is a schematic diagram of a financial product recommendation device provided according to an embodiment of the present application;

[0042] Fig. 9It is a structural block diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0043] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0044] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0045] It should also be noted that the collected information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for display, data for analysis, etc.) involved in this application are information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of relevant data are in compliance with relevant laws, regulations and standards, necessary confidentiality measures are taken, and public order and good customs are not violated, and corresponding operation entrances are provided for users to choose to authorize or refuse. For example, an interface is set up between this system and relevant users or institutions to provide users with corresponding operation entrances for users to choose to agree or refuse the results of automated decision-making; if the user chooses to refuse, the expert decision-making process will be entered.

[0046] According to an embodiment of the present application, a method embodiment of a method for recommending a financial product is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0047] It should be noted that a financial product recommendation system can be used as the execution subject of the financial product recommendation method of the embodiment of the present application. It is understandable that the financial product recommendation method provided in the embodiment of the present application can also be executed by other systems or devices, and the embodiment of the present application does not specifically limit this.

[0048] According to an embodiment of the present application, a method embodiment of a method for recommending a financial product is also provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0049] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1 FIG. 1 shows a hardware structure block diagram of a computer terminal (or mobile device) for implementing a method for recommending financial products. Figure 1 As shown, the computer terminal 10 (or mobile device) may include one or more (102a, 102b, ..., 102n are used to illustrate) processors 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 104 for storing data, and a transmission device 106 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0050] It should be noted that the above-mentioned one or more processors 102 and / or other data processing circuits may generally be referred to as "data processing circuits" in this article. The data processing circuit may be embodied in whole or in part as software, hardware, firmware or any other combination. In addition, the data processing circuit may be a single independent processing module, or may be fully or partially integrated into any of the other components in the computer terminal 10 (or mobile device). As in the financial product recommendation method involved in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0051] The memory 104 can be used to store software programs and modules of application software, such as program instructions / data storage devices corresponding to the method for recommending financial products in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, realizing the method for recommending financial products mentioned above. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the computer terminal 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0052] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the computer terminal 10. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0053] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the computer terminal 10 (or mobile device).

[0054] Under the above operating environment, this application provides Figure 2 Recommended methods for the financial products shown. Figure 2 FIG. 1 is a flow chart of an optional method for recommending financial products according to an embodiment of the present application. Figure 2 As shown, the method comprises the following steps:

[0055] Step S201: With authorization from the target user, collect the user image of the target user.

[0056] Optionally, when the customer approaches the self-service number collection machine, the system will inform the customer of the purpose of the upcoming image collection and the potential benefits of the financial product recommendation through clear prompt information, and clearly point out that this system strictly complies with data protection regulations to ensure the security and privacy of personal data. The user only needs to click the "Agree" button on the screen on the authorization page to grant the system permission to collect images, thereby completing the target user authorization.

[0057] Optionally, the financial product recommendation system will start the high-precision image acquisition module to capture the target user's facial features and full-body images in a non-invasive manner. By using advanced image sensors and optimized acquisition algorithms, it can ensure that high-quality image data can be obtained under any lighting conditions, laying a solid foundation for subsequent feature analysis and product recommendations.

[0058] Step S202: query the target user's historical purchase records of financial products based on the target user's user image.

[0059] Optionally, the image processing module in the financial product recommendation system will conduct in-depth analysis of the collected user images. Through advanced image recognition technology, it can not only identify the customer's facial features, but also capture subtle non-explicit features such as age, gender, emotional state, etc. This information, together with the customer's identity, will be seamlessly integrated into the bank's customer information system for rapid location and retrieval of the user's historical transaction data.

[0060] Optionally, the financial product recommendation system will first accurately match the customer's identity information obtained through image analysis in the database. This matching process is like unlocking the customer's personal financial timeline, revealing the customer's footprint in the banking service field. Subsequently, the algorithm will dig deeper into this data to extract the customer's purchase history of various financial products, including but not limited to product types such as deposits, loans, insurance, investments, and key details such as purchase time, frequency and amount.

[0061] Step S203, when it is detected that the number of historical purchase records of the target user is less than a preset threshold, at least one target financial product to be recommended to the target user is determined according to the user feature portrait of the target user.

[0062] Optionally, the preset threshold is used to measure the richness of the customer's historical purchase records to determine the direction of the recommendation strategy. This threshold is not fixed, but is dynamically adjusted based on the bank's business needs, market trends, and the latest analysis of customer behavior patterns. When the historical transaction data is lower than the preset threshold, it indicates that the depth of customer information is not enough to support accurate recommendations based on association rules. At this time, the system will turn to another innovative recommendation path.

[0063] Optionally, after the financial product recommendation system has identified the customer through image recognition technology and retrieved his financial product purchase history from a large database, the algorithm will carefully analyze the quantity and quality of these records. If it is detected that the number of historical purchase records of the target user is less than a preset threshold, this usually means that the customer is a new user or has limited previous transaction activities at the bank. The recommendation algorithm when the number is less than the preset threshold is used for product recommendation.

[0064] Optionally, if the target user's historical purchase record is confirmed to be below a preset threshold, the financial product recommendation system will mobilize its image processing and analysis module to extract a series of key features from the user's image, including but not limited to age, gender, expression, whether the user is carrying children, etc., to build a vivid user feature portrait. Then, the recommendation algorithm is used to make product recommendations.

[0065] Step S204, when it is detected that the number of historical purchase records of the target user is greater than or equal to a preset threshold, at least one target financial product to be recommended to the target user is determined according to the product association between the financial products purchased by the target user and the financial products not purchased by the target user.

[0066] In step S204, the product association represents the probability that different financial products are purchased by the same group of people at the same time.

[0067] Alternatively, product relevance indicates the close connection between financial products that customers have purchased and potential products that they have not yet explored. Through deep learning and big data analysis technology, the financial product recommendation system can dig out the hidden relationship between different financial products and reveal the inherent probability that products are simultaneously selected by similar customer groups. This relationship reflects the complex logic of the financial market and also provides a solid data foundation for personalized recommendations.

[0068] Optionally, the financial product recommendation system calculates and characterizes the probability of different financial products being purchased by the same group of people by analyzing historical transaction data. This probability not only covers the direct relationship between products, but also reflects the potential needs of customers at a specific life stage, income level or risk preference. Through this quantitative method, the financial product recommendation system can provide accurate decision-making basis for product recommendations.

[0069] Optionally, the recommendation algorithm will comprehensively consider the relevance of each financial product and the overall benefits of the bank, intelligently sort the products, and ultimately lock in the target financial products that best meet the customer's current needs and future plans.

[0070] From the contents of step S201 to step S204, it can be seen that in the embodiment of the present application, first, with the explicit authorization of the customer, the customer's image information, including facial image and whole body features, is collected through a high-precision camera installed on the self-service number-taking machine. This step ensures that the customer's data is collected and used with their knowledge and consent, complies with the requirements of laws and regulations, and protects the privacy and security of the customer.

[0071] Then, using advanced image analysis algorithms, the customer's unique identity is identified and determined from the collected user images. Subsequently, the customer's previous financial product purchase history is queried in the bank's database based on this identity to provide a basis for subsequent recommendation strategies.

[0072] Secondly, for customers with insufficient historical transaction records (new customers or customers with a small number of transaction records), the method of this application preprocesses the user image, removes unnecessary noise, adjusts the size and format, and uses the model to extract the key features of the customer. Based on the features inferred directly and indirectly, a user feature profile is generated, based on which financial products that may be of interest are recommended.

[0073] In addition, for customers with rich historical transaction records, this application adopts a product recommendation algorithm based on association rules. Through in-depth mining of customers' historical purchasing behaviors, the potential correlation between products is identified, that is, which products are more likely to be selected by customers in the same group at the same time. Based on this analysis, financial products that are highly correlated with their previous purchasing behaviors are recommended to target customers, thereby improving the accuracy and acceptance of recommendations.

[0074] It can be seen that through the technical solution of this application, the purpose of using image recognition technology to make personalized financial product recommendations to customers visiting bank branches is achieved, thereby achieving accurate product recommendations, improving the efficiency of bank branches, and enhancing customer satisfaction and loyalty, thereby solving the technical problem of poor accuracy in recommending financial products in the prior art.

[0075] In an optional embodiment, the financial product recommendation system first determines the target user's unique user identifier based on the target user's user image, and then searches the database for the target user's historical purchase record based on the unique user identifier.

[0076] Optionally, the financial product recommendation system collects user images through the camera in front of the bank's self-service number-taking machine. The system extracts unique features from the user images through high-precision image acquisition and recognition technology. These features represent a unique user identifier and are the basis for the subsequent personalized recommendation process.

[0077] Optionally, the unique user identifier of the target user is successfully locked, and the financial product recommendation system performs an accurate and efficient historical purchase record retrieval through the bank's database, so that the system can quickly locate and extract the user's past transaction information.

[0078] Optionally, database query is based on precise matching of user unique identifiers, ensuring high accuracy and relevance of historical purchase records. These records not only include basic transaction types, amounts and times, but also deeply explore the financial preferences and life trajectories hidden behind the transactions, providing detailed data basis for subsequent intelligent recommendations.

[0079] From the above content, we can see that the financial product recommendation system determines the target user's unique user ID based on the target user's user image, ensuring the accurate identification of each user. Then, based on the unique user ID, the target user's historical purchase records are queried in the database, allowing the system to quickly locate and extract the user's past transaction information, and accurately capture the user's needs.

[0080] In an optional embodiment, the financial product recommendation system first preprocesses the user image of the target user when it is detected that the number of historical purchase records of the target user is less than a preset threshold, wherein the preprocessing is used to remove noise information in the user portrait and adjust the size and / or format of the user portrait, and then determines the first type of user features and the second type of user features of the target user based on the preprocessed user image, wherein the first type of user features belong to the character features of the target user shown in the image, and the second type of user features belong to the user features indirectly inferred based on the first type of user features, and then generates a user feature portrait of the target user based on the first type of user features and the second type of user features of the target user.

[0081] Optionally, the preprocessing step first uses advanced algorithms to remove noise information in the image, including background clutter, uneven lighting and other elements that may affect feature recognition, making the user portrait clearer and more vivid. Subsequently, the system will automatically adjust the size and / or format of the user portrait to ensure data standardization and improve the efficiency of feature extraction. This series of operations not only purifies the image, but also lays the foundation for in-depth analysis of user features.

[0082] Optionally, from the preprocessed user images, the financial product recommendation system will extract the first category of user features of the target user, that is, the character attributes directly interpreted from the image, such as age, gender, external expressions, etc., which can intuitively reflect the customer's current physiological state and emotions.

[0083] Optionally, the financial product recommendation system indirectly infers the second type of user characteristics based on the first type of characteristics, such as potential financial status, family role or career inclination, etc. These characteristics are an important supplement to the comprehensive interpretation of customers through artificial intelligence and big data analysis, combined with sociological and psychological principles.

[0084] Optionally, through the comprehensive use of the first and second types of user characteristics, the financial product recommendation system will obtain a user characteristic portrait of the target user. This portrait not only includes the customer's external characteristics, but also deeply analyzes their internal needs and preferences, becoming the basis for personalized financial product recommendations.

[0085] From the above content, we can see that the financial product recommendation system first pre-processes the user image of the target user to eliminate all interference factors to ensure the accuracy and reliability of subsequent analysis. Then, the first and second user characteristics of the target user are determined based on the pre-processed user image, and the customer's internal needs and preferences are deeply analyzed to interpret the customer in an all-round way. Through the comprehensive application of the first and second user characteristics, the financial product recommendation system will obtain a user characteristic portrait of the target user. The user characteristic portrait will better guide the system to provide customers with financial product recommendations that meet their current needs and future expectations.

[0086] In an optional embodiment, the financial product recommendation system first obtains the probability of a reference user with the ith user feature purchasing the jth financial product when the user feature portrait includes N user features, and uses the probability as the recommendation score corresponding to the jth financial product and the ith user feature, wherein N is an integer greater than or equal to 1, i is a positive integer less than or equal to N, and the jth financial product is any financial product to be sold. Then, the N recommendation scores corresponding to the jth financial product and the N user features are summed up and calculated, and a target recommendation score of the jth financial product is determined based on the calculation result. Finally, based on the target recommendation score of each financial product to be sold, at least one target financial product to be recommended to the target user is determined from all the financial products to be sold.

[0087] Optionally, the financial product recommendation system will calculate the probability of a reference user with this feature purchasing the jth financial product for the i-th user feature, and convert this probability into a recommendation score as a recommendation indicator for the j-th financial product corresponding to the i-th user feature. This process not only quantifies the connection between products and features, but also provides data support for subsequent comprehensive evaluation.

[0088] Optionally, the financial product recommendation system will sum up the N recommendation scores corresponding to the j-th financial product and the N user features, and finally obtain a target recommendation score through fusion and weighting.

[0089] Optionally, the target recommendation score not only reflects the comprehensive match between the product and the user's characteristics, but also takes into account the market value of the product itself and the bank's benefits.

[0090] Optionally, the financial product recommendation system will intelligently sort and screen all candidate financial products based on the target recommendation score of each financial product to be sold. The entire decision-making process ensures that the recommendation results are not only based on a deep match of customer feature profiles, but also fully consider product benefits and market trends.

[0091] From the above content, it can be seen that the financial product recommendation system calculates the intelligent recommendation score of financial products based on user feature portraits, which can quantify the relationship between user features and financial products, and can also achieve accurate selection of target financial products through the intelligent integration of comprehensive scores.

[0092] In an optional embodiment, the financial product recommendation system first sums up the N recommendation scores corresponding to the j-th financial product and N user features to obtain the total score, and then obtains the historical rate of return of the j-th financial product, where the historical rate of return represents the historical rate of return of the j-th financial product for the financial institution or the historical rate of return for the product holder, and then calculates the product of the historical rate of return of the j-th financial product and the total score to obtain the target recommendation score of the j-th financial product.

[0093] Optionally, when the j-th financial product corresponds to N recommendation scores for each of the N features in the user feature profile, the financial product recommendation will sum up these N recommendation scores to generate a total score that intuitively reflects the product's attractiveness to the target user group.

[0094] Optionally, the financial product recommendation system obtains the historical rate of return of the j-th financial product, not only to measure the past performance of the product, but also to evaluate its long-term investment value and market stability. This data point represents the historical footprint of the financial product and provides a historical perspective for recommendation decisions.

[0095] Optionally, historical returns are obtained, covering the performance of the product in different market cycles, ensuring that the recommendation results are not only based on current needs, but also consider the long-term potential of the product, providing customers with a more comprehensive product perspective.

[0096] Optionally, the financial product recommendation system calculates the product of the historical rate of return of the j-th financial product and the summed total score to obtain the target recommendation score of the product, generating a comprehensive score that reflects the historical stability of the product and takes into account the personalized needs of users.

[0097] Optionally, the determination of the target recommendation score not only reflects the algorithm's consideration of the comprehensive value of the product, but also the result of a deep understanding of user needs. The target recommendation score ensures that the recommended financial products not only have good returns in the past, but also resonate with the target user group in the current situation, thereby improving the pertinence and effectiveness of product recommendations.

[0098] From the above content, we can see that the financial product recommendation system is based on the comprehensive recommendation score calculation process of financial products and user characteristics. It not only combines the historical performance of the product with the current characteristics of user needs to generate a target recommendation score, but also ensures that every product recommendation is a well-considered decision that can accurately match customer needs.

[0099] In an optional embodiment, the financial product recommendation system first determines from the database the total number of reference users having the i-th user characteristic to obtain a first number, then determines from the database the number of users having the i-th user characteristic who have purchased the j-th financial product to obtain a second number, and then calculates the ratio of the second number to the first number to obtain the probability that the reference user having the i-th user characteristic purchases the j-th financial product.

[0100] Optionally, through precise data mining, the financial product recommendation system can determine the total number of reference users having the i-th user feature, that is, the first number.

[0101] Optionally, the first quantity not only represents the size of the user group with specific characteristics, but also provides important cardinality information for subsequent purchase probability calculation.

[0102] Optionally, through this data matching, the financial product recommendation system can accurately obtain the number of users who have the i-th user feature and have purchased the j-th financial product, and this number is marked as the second number.

[0103] Optionally, the second quantity not only quantifies the direct relationship between user characteristics and financial product purchasing behavior, but also provides key molecular data for subsequent probability calculations.

[0104] Optionally, the financial product recommendation system will calculate the ratio of the second number to the first number, which is directly converted into the probability of a reference user with the i-th user characteristics purchasing the j-th financial product. This not only quantifies the impact of user characteristics on the purchasing tendency of financial products, but also provides solid data support for personalized recommendation strategies.

[0105] From the above content, we can see that the financial product recommendation system determines the size of the user group with specific characteristics through data mining, and then calculates the number and probability of specific users purchasing specific products. It not only quantifies the impact of user characteristics on the purchase tendency of financial products, but also provides solid data support for personalized recommendation strategies, which can be based on the precise matching of market performance and user characteristics.

[0106] In an optional embodiment, the financial product recommendation system first obtains the probability that the user who has purchased the kth financial product also purchased the hth financial product when the target user has purchased M financial products, and uses the probability as the association score corresponding to the hth financial product and the kth financial product, wherein M is an integer greater than or equal to 1, the kth financial product is any one of the M financial products, and the hth financial product is any one of the to-be-sold financial products that the target user does not hold, then sums up the association scores corresponding to the hth financial product and the M financial products, respectively, determines a target association score for the hth financial product based on the calculation result, and then determines at least one target financial product to be recommended to the target user from all the financial products to be sold that the target user does not hold, based on the target association score of each to-be-sold financial product that the target user does not hold.

[0107] Optionally, the financial product recommendation system will dig deeper into the data to calculate the probability that a user who purchased the kth financial product also purchased the hth financial product, and this probability value serves as the correlation score corresponding to the hth financial product and the kth financial product.

[0108] Optionally, the correlation score not only quantifies the potential correlation between different financial products, but also provides a solid mathematical basis for building recommendation links between products, ensuring that the recommendation strategy can accurately predict the target user's possible future financial needs based on his or her historical behavior.

[0109] Optionally, the financial product recommendation system will perform a summation calculation on the relevance scores corresponding to the hth financial product and the M financial products, respectively, to ensure that the target relevance score of each financial product integrates its multi-dimensional relevance with the products purchased by the target user.

[0110] Optionally, the financial product recommendation system will select at least one of the target financial products to be sold that the target user does not hold based on the target relevance scores of all the financial products to be sold, using a one-time decision algorithm, as the target financial product to be recommended to the target user. This decision process ensures that the recommended product is not only highly relevant to the target user's historical purchase behavior, but also can meet their potential future needs.

[0111] From the above content, we can see that the financial product recommendation system quantifies the potential correlation between financial products through probability calculation, and selects the most relevant and attractive target financial products from the financial products for sale that the target users do not hold through the decision-making algorithm of intelligent summation and personalized recommendation, ensuring that each financial product recommendation is based on a precise match of deep insights and personalized needs.

[0112] In an optional embodiment, the financial product recommendation system first sums up the correlation scores corresponding to the h-th financial product and the M financial products respectively to obtain a sum result, and then obtains the historical rate of return of the h-th financial product, wherein the historical rate of return of the h-th financial product represents the historical rate of return of the h-th financial product for the financial institution or the historical rate of return for the product holder, and then calculates the product of the historical rate of return of the h-th financial product and the sum result to obtain the target correlation score of the h-th financial product.

[0113] Optionally, the financial product recommendation system will sum the correlation scores corresponding to the h-th financial product and the M financial products that the target user has purchased in the past. The summed result not only quantifies the general appeal of the h-th financial product among the target user group, but also provides a solid foundation for subsequent target score optimization.

[0114] Optionally, the financial product recommendation system obtains the historical yield of the hth financial product. This data not only reveals the product's performance in the historical cycle, but also reflects the long-term value of the product to the financial institution or product holder. The exploration of historical yield provides the recommendation strategy with a product value assessment that goes beyond the surface, ensuring that the recommended content not only meets the user's preferences, but also takes into account the product's real performance and long-term potential.

[0115] Optionally, the financial product recommendation system will calculate the product of the historical rate of return of the h-th financial product and the sum result to obtain the target relevance score of the h-th financial product, ensuring that the recommended financial product is not only highly relevant to the historical purchasing behavior of the target user, but also has good historical performance and long-term investment value.

[0116] From the above content, we can see that the financial product recommendation system constructs a quantitative indicator of the comprehensive attractiveness of the product through summation calculation, and also obtains historical yields, which is important data reflecting the true value of the product. Finally, by calculating the target relevance score, the depth, breadth and accuracy of the recommendation strategy are ensured.

[0117] In an optional embodiment, the financial product recommendation system first determines from the database the number of users who purchased both the kth financial product and the hth financial product to obtain a third number, then determines from the database the number of users who purchased the kth financial product to obtain a fourth number, and then calculates the ratio of the third number to the fourth number to obtain the probability that the user who purchased the kth financial product also purchased the hth financial product.

[0118] Optionally, when the financial product recommendation system focuses on the combination of the kth financial product and the hth financial product, the financial product recommendation system will go deep into the data core to determine the number of users who purchased the kth financial product and the hth financial product at the same time, and this number is marked as the third number.

[0119] Optionally, a third quantity is determined, and the financial product recommendation system quantifies the potential combination purchase demand in the market, injecting deep market insights into the construction of personalized recommendation strategies.

[0120] Optionally, the financial product recommendation system determines the number of users who have purchased the kth financial product from the database, and this number is marked as a fourth number.

[0121] Optionally, by determining the fourth quantity, the financial product recommendation system not only focuses on the deep understanding of the purchase trajectory of a single product, but also quantifies the size of the user group with a specific purchase history, providing solid data support for subsequent probability calculations.

[0122] Optionally, the financial product recommendation system converts the ratio of the third quantity to the fourth quantity into the probability that the user who purchased the kth financial product also purchased the hth financial product. The calculation of this probability reveals a potential market connection, which not only quantifies the possibility of a specific combination of purchases, but also provides intelligent guidance based on data science for personalized recommendation strategies.

[0123] From the above content, we can see that the financial product recommendation system draws a market map of product combination purchase behavior and determines the size of the user group based on the probability analysis algorithm of financial product combination purchase behavior. The financial product recommendation system reveals the possibility of combination purchase between products through intelligent probability calculation, ensuring the depth, breadth and accuracy of the recommendation strategy.

[0124] It can be seen from the above content that according to the technical solution of this application, at least the following technical effects can be achieved:

[0125] 1. It realizes automatic identification of people who go to bank branches to handle business, helping them understand the bank's various products. Even if the customer does not have identity information retained in the bank, they can still choose recommended products through facial recognition.

[0126] 2. Reduced the pressure on branch employees.

[0127] 3. It avoids the problem of customers being wary when talking to others.

[0128] 4. Since customers are required to read the number-picking slip, the problem of customers not reading advertisements placed on other channels is avoided, thus improving efficiency.

[0129] In an optional embodiment, Figure 3A financial product recommendation method according to an embodiment of the present application is shown. Figure 3 The flowchart of the technical solution of the financial product recommendation method shown is: an intelligent recommendation system based on image recognition for self-service number-taking machines. It includes: the front-end transaction module is an existing banking business transaction system, which passes the information in each transaction, including customer information, product information, whether the recommendation is successful or not, etc. to the data storage module as historical information, and also passes the relevant information of this transaction to the data storage module. The data storage module is used to store all the data information required for the generation model; the image acquisition module is used to obtain the data information of the user's face and full-body image; the image processing module performs regional positioning, size adjustment, denoising and standardization on the collected images, and uses an external ResNet model to extract the visual features of the customer, including age, gender, mood, whether to bring children, possible occupations, etc., to generate a user portrait, and pass the user portrait to the mapping management module; the mapping management module generates a mapping relationship between the user portrait and the product suitable for recommendation based on successful historical transactions, and matches it with the customer portrait of this transaction, generates the recommendation result of this transaction and passes it to the intelligent output module, and saves the existing large model; the intelligent output module outputs the recommended content externally based on the recommendation result obtained by the mapping management module, mainly referring to the printing function of the automatic number-taking machine.

[0130] Among them, the recommendation model uses a combination of feature weighted recommendation algorithm and association recommendation algorithm.

[0131] like Figure 4 An optional recommendation flow chart using a feature-weighted recommendation algorithm is shown in FIG: For customers who lack sufficient historical data in the banking system, the recommendation model uses a feature-weighted recommendation algorithm to calculate the sum of the success rate of recommending a certain product based on different user characteristics in the historical data and the weight of the comprehensive benefits of the product, and sort out the top few data, that is, the most qualified data, and recommend them to the user. In essence, it recommends customers without historical purchase behavior to the preferred products of customers with historical purchase behavior in the same customer group. For example, a customer profile is ||a, b, c, d,·······||, where a, b, c, d are all customer characteristics. Let a1 = the probability that a customer with feature a buys product 1 = the total number of customers with feature a in the historical information who buy product 1 / the total number of customers with feature a in the historical information, where a1 is updated as the historical information is updated. β1 is the comprehensive benefit of product 1 to the bank, which is a fixed parameter and is manually updated at fixed intervals. Then for a certain customer, the recommendation score of product 1 is α1 = (a1+b1+c1······)β1. If α1>α1*, then recommend product 1. α1* is dynamically adjusted by the data model according to the changes in the success rate of product recommendations in historical data. If the historical success rate increases within a month, α1* will be reduced, and if the historical success rate decreases, α1* will be increased.

[0132] like Figure 5 An optional recommendation flow chart using a recommendation algorithm based on association rules is shown: For customers with sufficient historical transaction information in the banking system, the recommendation model uses a recommendation algorithm based on association rules. The historical data of all customers is mined and analyzed to find out the association between commodities and commodities. Products with a greater probability of being purchased by the same customer have associations, and these commodities can be combined into a combination for recommendation. For the current customer, the product information purchased by the customer is identified, and the association result of a purchased product is obtained by multiplying the probability of a purchased product to other products by the comprehensive benefit, and the association results of all purchased products are accumulated to obtain the final result. If a customer profile is ||A, B, C, D, ·······||, where A, B, C, D are all customer characteristics, let AB = the probability of a customer who purchased product A purchasing product B = the number of customers who purchased products A and B at the same time in the historical information / the number of customers who purchased product A in the historical information, and βB is the comprehensive benefit of product B, then according to the historical transaction information of the customer, the association score of product B αB = (AB+BB+CB+······)βB. If αB>αB*, then recommend product 1. αB* is dynamically adjusted by the data model according to the changes in the success rate of product recommendations in historical data. If the historical success rate increases within a month, αB* will be reduced, and if the historical success rate decreases, αB* will be increased.

[0133] Overall, Figure 6 The overall recommendation flow chart of an optional financial product recommendation method is shown: First, when the customer takes a number at the automatic number-taking machine, the customer's image information is automatically collected and the identity is identified. Then, different recommendation algorithms are used according to whether the customer has sufficient product purchase records. That is, if the customer has no historical transaction information, a new entry for the customer is created, and a feature-weighted recommendation algorithm is used; if there are not enough product purchase records, a feature-weighted recommendation algorithm is also used, and then the image data is analyzed to obtain customer features and customer portraits, and then the customer portraits are input into the large model to obtain products suitable for recommending to the customer. Finally, the customer handles the business, and the result of the success or failure of the recommendation is used to continue to improve the model. If the customer has sufficient product purchase records, a recommendation algorithm based on association rules is used, and the customer's historical product purchase information is first analyzed, and then the appropriate recommended products are obtained through the product association recommendation, and finally the result of the success or failure of the recommendation is used to continue to improve the model.

[0134] In an optional embodiment, Figure 7A schematic diagram of an optional recommendation method according to an embodiment of the present application is also shown, which can more clearly understand the recommendation ideas of financial products: first, collect and obtain customer portraits, then input the customer portraits into the large model, obtain the given recommended products, and then feedback the success or failure to the model, and the recommendation success rate in the past year is adjusted to αi*, and finally αi* is additionally increased for the customer. The mapping method of this scheme uses a method combining feature weighted recommendation with product association recommendation. For different customer groups, a feature weighted recommendation method is used for new customers or customers who lack transaction information. The core idea is that customers with the same characteristics are more likely to have similar financial business needs. For example, customers with the characteristics of "middle-aged women with young children" may be interested in life insurance; "elderly people" may be interested in serious illness insurance; "middle-aged men" may be interested in housing mortgage loans. For customers with rich historical transaction information, an association recommendation method is used to select recommended products based on historical product association information. For example, customers who have applied for time deposits are more inclined to continue to apply for time deposits, and customers who have applied for credit cards are more inclined to apply for loans. This combined recommendation method is suitable for business scenarios with strong real-time business and fast product updates.

[0135] According to another aspect of the embodiment of the present application, a financial product recommendation device is also provided, wherein: Figure 8 is a schematic diagram of an optional financial product recommendation device according to an embodiment of the present application, such as Figure 8 As shown, the financial product recommendation system device includes: a collection unit 801, a query unit 802, a first determination unit 803, and a second determination unit 804.

[0136] The collection unit 801 collects the user image of the target user with the authorization of the target user; the query unit 802 queries the target user's historical purchase record of financial products based on the user image of the target user; the first determination unit 803 determines at least one target financial product to be recommended to the target user based on the user feature portrait of the target user when it is detected that the number of historical purchase records of the target user is less than a preset threshold; the second determination unit 804 determines at least one target financial product to be recommended to the target user based on the product correlation between the financial products purchased by the target user and the financial products not purchased by the target user when it is detected that the number of historical purchase records of the target user is greater than or equal to the preset threshold, wherein the product correlation represents the probability that different financial products are purchased by the same group of people at the same time.

[0137] Optionally, the query unit 802 includes: a first determining subunit and a first query subunit. The first determining subunit is used to determine the user unique identifier of the target user according to the user image of the target user; and the first query subunit is used to query the historical purchase record of the target user in the database according to the user unique identifier.

[0138] Optionally, the first determination unit 803 includes: a first processing subunit, a second determination subunit, and a first generation subunit. The first processing subunit is used to preprocess the user image of the target user when it is detected that the number of historical purchase records of the target user is less than a preset threshold, wherein the preprocessing is used to remove noise information in the user portrait and adjust the size and / or format of the user portrait; the second determination subunit is used to determine the first type of user features and the second type of user features of the target user based on the preprocessed user image, wherein the first type of user features belong to the character features of the target user shown in the image, and the second type of user features belong to the user features indirectly inferred based on the first type of user features; the first generation subunit is used to generate a user feature portrait of the target user based on the first type of user features and the second type of user features of the target user.

[0139] Optionally, the first determination unit 803 includes: a first acquisition subunit, a third determination subunit, and a fourth determination subunit. The first acquisition subunit is used to obtain the probability of a reference user with the i-th user feature purchasing the j-th financial product when the user feature profile includes N user features, and use the probability as the recommendation score corresponding to the j-th financial product and the i-th user feature, wherein N is an integer greater than or equal to 1, i is a positive integer less than or equal to N, and the j-th financial product is any financial product to be sold; the third determination subunit is used to sum and calculate the N recommendation scores corresponding to the j-th financial product and the N user features, and determine the target recommendation score of the j-th financial product according to the calculation result; the fourth determination subunit is used to determine at least one target financial product to be recommended to the target user from all the financial products to be sold according to the target recommendation score of each financial product to be sold.

[0140] Optionally, the third determination subunit includes: a first calculation module, a first acquisition module, and a second calculation module. The first calculation module is used to sum up the N recommendation scores corresponding to the j-th financial product and the N user characteristics to obtain the summed total score; the first acquisition module is used to obtain the historical rate of return of the j-th financial product, wherein the historical rate of return represents the historical rate of return of the j-th financial product for the financial institution or the historical rate of return for the product holder; the second calculation module is used to calculate the product of the historical rate of return of the j-th financial product and the summed total score to obtain the target recommendation score of the j-th financial product.

[0141] Optionally, the first acquisition subunit includes: a first determination module, a second determination module, and a third calculation module. The first determination module is used to determine the total number of reference users with the i-th user feature from the database to obtain a first number; the second determination module is used to determine the number of users with the i-th user feature who have purchased the j-th financial product from the database to obtain a second number; and the third calculation module is used to calculate the ratio of the second number to the first number to obtain the probability that the reference user with the i-th user feature purchases the j-th financial product.

[0142] Optionally, the second determination unit 804 includes: a second acquisition subunit, a first calculation subunit, and a fifth determination subunit. The second acquisition subunit is used to obtain the probability that the user who has purchased the kth financial product also purchased the hth financial product when the target user has purchased M financial products, and use the probability as the correlation score corresponding to the hth financial product and the kth financial product, wherein M is an integer greater than or equal to 1, the kth financial product is any one of the M financial products, and the hth financial product is any one of the to-be-sold financial products that the target user does not hold; the first calculation subunit is used to sum the correlation scores corresponding to the hth financial product and the M financial products, respectively, and determine the target correlation score of the hth financial product according to the calculation result; the fifth determination subunit is used to determine at least one target financial product to be recommended to the target user from all the to-be-sold financial products that the target user does not hold, according to the target correlation score of each to-be-sold financial product that the target user does not hold.

[0143] Optionally, the first calculation subunit includes: a fourth calculation module, a second acquisition module, and a fifth calculation module. The fourth calculation module is used to sum the association scores corresponding to the h-th financial product and the M financial products to obtain a sum result; the second acquisition module is used to obtain the historical yield of the h-th financial product, wherein the historical yield of the h-th financial product represents the historical yield of the h-th financial product for the financial institution or the historical yield for the product holder; the fifth calculation module is used to calculate the product of the historical yield of the h-th financial product and the sum result to obtain the target association score of the h-th financial product.

[0144] Optionally, the second acquisition subunit includes: a third determination module, a fourth determination module, and a sixth calculation module. The third determination module is used to determine the number of users who purchased the kth financial product and the hth financial product from the database to obtain a third number; the fourth determination module is used to determine the number of users who purchased the kth financial product from the database to obtain a fourth number; and the sixth calculation module is used to calculate the ratio of the third number to the fourth number to obtain the probability that the user who purchased the kth financial product also purchased the hth financial product.

[0145] An embodiment of the present application may provide an electronic device, Fig. 9 is a structural block diagram of an electronic device according to an embodiment of the present application. Fig. 9 As shown, the electronic device may include: one or more ( Fig. 9 Only one is shown) a processor, a memory, a storage controller, and a peripheral interface, wherein the peripheral interface is connected to a radio frequency module, an audio module, and a display.

[0146] Among them, the memory can be used to store software programs and modules, such as program instructions / modules corresponding to the methods and devices in the embodiments of the present application, and the processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, realizing the above-mentioned method. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to the terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0147] The processor can call the information and application stored in the memory through the transmission device to perform the following steps: with the authorization of the target user, collect the user image of the target user; query the target user's historical purchase record of financial products based on the user image of the target user; when it is detected that the number of historical purchase records of the target user is less than a preset threshold, determine at least one target financial product to be recommended to the target user based on the user feature portrait of the target user; when it is detected that the number of historical purchase records of the target user is greater than or equal to the preset threshold, determine at least one target financial product to be recommended to the target user based on the product correlation between the financial products purchased by the target user and the financial products not purchased by the target user, wherein the product correlation represents the probability of different financial products being purchased by the same group of people at the same time.

[0148] The processor can also call the information and application programs stored in the memory through the transmission device to perform the following steps: determine the target user's unique user identifier based on the target user's user image; and query the target user's historical purchase records in the database based on the user unique identifier.

[0149] The processor can also call the information and application stored in the memory through the transmission device to perform the following steps: when it is detected that the number of historical purchase records of the target user is less than a preset threshold, preprocess the user image of the target user, wherein the preprocessing is used to remove noise information in the user portrait and adjust the size and / or format of the user portrait; determine the first type of user features and the second type of user features of the target user based on the preprocessed user image, wherein the first type of user features belong to the character features of the target user shown in the image, and the second type of user features belong to the user features indirectly inferred based on the first type of user features; generate a user feature portrait of the target user based on the first type of user features and the second type of user features of the target user.

[0150] The processor can also call the information and application stored in the memory through the transmission device to perform the following steps: when the user feature portrait includes N user features, obtain the probability that the reference user with the i-th user feature purchases the j-th financial product, and use the probability as the recommendation score corresponding to the j-th financial product and the i-th user feature, where N is an integer greater than or equal to 1, i is a positive integer less than or equal to N, and the j-th financial product is any financial product to be sold; summing up the N recommendation scores corresponding to the j-th financial product and the N user features, and determining the target recommendation score of the j-th financial product according to the calculation result; and determining at least one target financial product to be recommended to the target user from all the financial products to be sold according to the target recommendation score of each financial product to be sold.

[0151] The processor can also call the information and application programs stored in the memory through the transmission device to perform the following steps: summing up the N recommendation scores corresponding to the j-th financial product and the N user characteristics to obtain the total score; obtaining the historical rate of return of the j-th financial product, wherein the historical rate of return represents the historical rate of return of the j-th financial product for the financial institution or the historical rate of return for the product holder; calculating the product of the historical rate of return of the j-th financial product and the total score to obtain the target recommendation score of the j-th financial product.

[0152] The processor can also call the information and application stored in the memory through the transmission device to perform the following steps: determine the total number of reference users with the i-th user characteristics from the database to obtain a first number; determine the number of users with the i-th user characteristics who have purchased the j-th financial product from the database to obtain a second number; calculate the ratio of the second number to the first number to obtain the probability that the reference user with the i-th user characteristics purchases the j-th financial product.

[0153] The processor may also call the information and application program stored in the memory through the transmission device to perform the following steps: when the target user has purchased M financial products, obtain the probability that the user who has purchased the kth financial product has also purchased the hth financial product, and use the probability as the association score corresponding to the hth financial product and the kth financial product, wherein M is an integer greater than or equal to 1, the kth financial product is any one of the M financial products, and the hth financial product is any one of the to-be-sold financial products that the target user does not hold; summing up the association scores corresponding to the hth financial product and the M financial products, respectively, and determining a target association score for the hth financial product according to the calculation result; and determining at least one target financial product to be recommended to the target user from all the financial products to be sold that the target user does not hold, according to the target association score of each to-be-sold financial product that the target user does not hold.

[0154] The processor can also call the information and application stored in the memory through the transmission device to perform the following steps: summing the correlation scores corresponding to the h-th financial product and the M financial products respectively to obtain a sum result; obtaining the historical rate of return of the h-th financial product, wherein the historical rate of return of the h-th financial product represents the historical rate of return of the h-th financial product for the financial institution or the historical rate of return for the product holder; calculating the product of the historical rate of return of the h-th financial product and the sum result to obtain the target correlation score of the h-th financial product.

[0155] The processor can also call the information and application stored in the memory through the transmission device to perform the following steps: determine the number of users who purchased the kth financial product and the hth financial product at the same time from the database to obtain a third number; determine the number of users who purchased the kth financial product from the database to obtain a fourth number; calculate the ratio of the third number to the fourth number to obtain the probability that the user who purchased the kth financial product also purchased the hth financial product.

[0156] In the embodiment of the present application, first, with the explicit authorization of the customer, the customer's image information, including facial image and whole body features, is collected through a high-precision camera installed on the self-service number-taking machine. This step ensures that the customer's data is collected and used with their knowledge and consent, complies with laws and regulations, and protects the privacy and security of the customer.

[0157] Then, using advanced image analysis algorithms, the customer's unique identity is identified and determined from the collected user images. Subsequently, the customer's previous financial product purchase history is queried in the bank's database based on this identity to provide a basis for subsequent recommendation strategies.

[0158] Secondly, for customers with insufficient historical transaction records (new customers or customers with a small number of transaction records), the method of this application preprocesses the user image, removes unnecessary noise, adjusts the size and format, and uses the model to extract the key features of the customer. Based on the features inferred directly and indirectly, a user feature profile is generated, based on which financial products that may be of interest are recommended.

[0159] In addition, for customers with rich historical transaction records, this application adopts a product recommendation algorithm based on association rules. Through in-depth mining of customers' historical purchasing behaviors, the potential correlation between products is identified, that is, which products are more likely to be selected by customers in the same group at the same time. Based on this analysis, financial products that are highly correlated with their previous purchasing behaviors are recommended to target customers, thereby improving the accuracy and acceptance of recommendations.

[0160] It can be seen that through the technical solution of this application, the purpose of using image recognition technology to make personalized financial product recommendations to customers visiting bank branches is achieved, thereby achieving accurate product recommendations, improving the efficiency of bank branches, and enhancing customer satisfaction and loyalty, thereby solving the technical problem of poor accuracy in recommending financial products in the prior art.

[0161] It can be understood by those skilled in the art that Fig. 9 The structure shown is for illustration only, and the electronic device may also be a smart phone (such as an Android phone, an iOS phone, etc.), a tablet computer, a PDA, a mobile Internet device (Mobile Internet Devices, MID), a PAD, or other terminal devices. Fig. 9 The structure of the electronic device is not limited. Fig. 9 More or fewer components (such as network interfaces, display devices, etc.) shown in, or having Fig. 9 Different configurations are shown.

[0162] A person of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the hardware related to the terminal device through a program, and the program can be stored in a computer-readable storage medium, and the storage medium may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0163] According to another aspect of the present application, a computer-readable storage medium is provided, wherein a computer program is stored in the computer-readable storage medium, wherein when the computer program is executed, the device where the computer-readable storage medium is located executes the above-mentioned method for recommending financial products.

[0164] According to another aspect of the present application, a computer program product is also provided, wherein the computer program product includes computer instructions, wherein when the computer instructions are executed, the device where the computer program product is located executes the above-mentioned financial product recommendation method.

[0165] The above-mentioned embodiments or examples disclosed in the present application are not exhaustive, but are only illustrative of some embodiments or examples, and are not intended to be specific limitations on the scope of protection disclosed in the present application. In the absence of contradiction, each step in a certain embodiment or example in the present application can be implemented as an independent example, and the steps can be combined arbitrarily. For example, the scheme after removing some steps in a certain embodiment or example can also be implemented as an independent example, and the order of the steps in a certain embodiment or example can be arbitrarily exchanged. In addition, the optional methods or optional examples in a certain embodiment or example can be combined arbitrarily; in addition, the various embodiments or examples can be combined arbitrarily, for example, some or all steps of different embodiments or examples can be combined arbitrarily, and a certain embodiment or example can be combined arbitrarily with the optional methods or optional examples of other embodiments or examples.

[0166] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0167] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0168] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0169] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0170] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0171] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk and other media that can store program codes.

[0172] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for recommending financial products, characterized in that: include: With authorization from the target user, collecting a user image of the target user; querying the target user's historical purchase records of financial products according to the target user's user image; When it is detected that the number of historical purchase records of the target user is less than a preset threshold, determining at least one target financial product to be recommended to the target user according to the user feature profile of the target user; When it is detected that the number of historical purchase records of the target user is greater than or equal to the preset threshold, at least one target financial product to be recommended to the target user is determined based on the product association between the financial products purchased by the target user and the financial products not purchased by the target user, wherein the product association represents the probability of different financial products being purchased simultaneously by the same group of people.

2. The method for recommending financial products according to claim 1, characterized in that: Querying the historical purchase record of the target user according to the user image of the target user includes: Determine a unique user identifier of the target user according to the user image of the target user; The historical purchase records of the target user are queried in a database according to the unique identifier of the user.

3. The method for recommending financial products according to claim 1, characterized in that: Before determining at least one target financial product to be recommended to the target user according to the user feature portrait of the target user, the financial product recommendation method further includes: When it is detected that the number of historical purchase records of the target user is less than the preset threshold, preprocessing the user image of the target user, wherein the preprocessing is used to remove noise information in the user portrait and adjust the size and / or format of the user portrait; Determining first-category user features and second-category user features of the target user according to the preprocessed user image, wherein the first-category user features belong to the character features of the target user in the image, and the second-category user features belong to the user features indirectly inferred based on the first-category user features; A user feature portrait of the target user is generated according to the first type of user features and the second type of user features of the target user.

4. The method for recommending financial products according to claim 1, characterized in that: Determining at least one target financial product to be recommended to the target user according to the user feature portrait of the target user includes: In the case where the user feature portrait includes N user features, the probability of a reference user having the ith user feature purchasing the jth financial product is obtained, and the probability is used as a recommendation score of the jth financial product corresponding to the ith user feature, wherein N is an integer greater than or equal to 1, i is a positive integer less than or equal to N, and the jth financial product is any financial product to be sold; Summing up the N recommendation scores corresponding to the j-th financial product and the N user features, and determining the target recommendation score of the j-th financial product according to the calculation result; At least one target financial product to be recommended to the target user is determined from all the financial products to be sold according to the target recommendation score of each financial product to be sold.

5. The method for recommending financial products according to claim 4, characterized in that: Summing up the N recommendation scores corresponding to the j-th financial product and the N user features, and determining the target recommendation score of the j-th financial product according to the calculation result, including: Summing up the N recommendation scores corresponding to the j-th financial product and the N user characteristics to obtain a summed total score; Obtaining a historical rate of return of the j-th financial product, wherein the historical rate of return represents a historical rate of return of the j-th financial product to a financial institution or a historical rate of return to a product holder; The product of the historical rate of return of the j-th financial product and the summed total score is calculated to obtain a target recommendation score for the j-th financial product.

6. The method for recommending financial products according to claim 4, characterized in that: Obtain the probability of a reference user with the i-th user feature purchasing the j-th financial product, including: Determine the total number of reference users having the i-th user feature from a database to obtain a first number; Determine from the database the number of users who have the i-th user characteristic and have purchased the j-th financial product, to obtain a second number; The ratio of the second number to the first number is calculated to obtain a probability that a reference user having the i-th user feature purchases the j-th financial product.

7. The method for recommending financial products according to claim 1, characterized in that: When it is detected that the number of historical purchase records of the target user is greater than or equal to the preset threshold, determining at least one target financial product to be recommended to the target user according to the product association between the financial products purchased by the target user and the financial products not purchased by the target user, including: In the case where the target user has purchased M financial products, the probability that the user who has purchased the kth financial product has also purchased the hth financial product is obtained, and the probability is used as the correlation score corresponding to the hth financial product and the kth financial product, wherein M is an integer greater than or equal to 1, the kth financial product is any one of the M financial products, and the hth financial product is any one of the financial products to be sold that the target user does not hold; Sum the association scores corresponding to the h-th financial product and the M financial products, and determine the target association score of the h-th financial product according to the calculation result; At least one target financial product to be recommended to the target user is determined from all the financial products to be sold that the target user does not own, according to the target relevance score of each financial product to be sold that the target user does not own.

8. The method for recommending financial products according to claim 7, characterized in that: Summing the association scores corresponding to the h-th financial product and the M financial products, and determining the target association score of the h-th financial product according to the calculation result, includes: Summing the correlation scores corresponding to the h-th financial product and the M financial products to obtain a summation result; Obtaining a historical rate of return of the h-th financial product, wherein the historical rate of return of the h-th financial product represents a historical rate of return of the h-th financial product to a financial institution or a historical rate of return to a product holder; The product of the historical rate of return of the h-th financial product and the summation result is calculated to obtain the target relevance score of the h-th financial product.

9. The method for recommending financial products according to claim 7, characterized in that: Obtain the probability that a user who purchased the kth financial product also purchased the hth financial product, including: Determine from the database the number of users who have purchased the kth financial product and the hth financial product at the same time, to obtain a third number; Determining the number of users who purchased the k-th financial product from the database to obtain a fourth number; The ratio of the third number to the fourth number is calculated to obtain the probability that the user who purchased the kth financial product also purchased the hth financial product.

10. A device for recommending financial products, characterized in that: include: A collection unit, which collects a user image of the target user with authorization from the target user; A query unit, querying the target user's historical purchase record of financial products according to the user image of the target user; A first determination unit, when detecting that the number of historical purchase records of the target user is less than a preset threshold, determines at least one target financial product to be recommended to the target user according to a user feature profile of the target user; The second determination unit determines at least one target financial product to be recommended to the target user based on a product correlation between financial products purchased by the target user and financial products not purchased by the target user, when it is detected that the number of historical purchase records of the target user is greater than or equal to the preset threshold, wherein the product correlation represents the probability of different financial products being purchased simultaneously by the same group of people.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored executable program, wherein when the executable program is executed, the device where the computer-readable storage medium is located is controlled to execute the financial product recommendation method according to any one of claims 1 to 9.

12. An electronic device, characterized in that: include: A memory storing an executable program; A processor is used to run the program, wherein the program, when running, executes the financial product recommendation method described in any one of claims 1 to 9.

13. A computer program product comprising computer instructions, characterized in that: When the computer instructions are executed by a processor, the steps of the method for recommending financial products described in any one of claims 1 to 9 are implemented.

Images