Recommendation method and device of pension financial products, electronic equipment and storage medium

Abstract

The invention discloses a recommendation method and device for an old-age care financial product, electronic equipment and a storage medium, and relates to the field of artificial intelligence, the field of financial science and technology or other related technical fields, and the method comprises the steps: obtaining an old-age care feature vector and a financial product feature vector; inputting the old-age care feature vector and the financial product feature vector into an interestingness evaluation model, and outputting an interestingness value of the target user for each old-age care financial product; screening out N candidate pension financial products from the K pension financial products based on the interestingness values; inputting the pension feature vector and financial product feature vectors of the K candidate pension financial products into a recommendation model, and outputting a matching degree value of the target user to each candidate pension financial product; and screening the target old-age care financial products based on the matching degree value, and forming an old-age care financial product recommendation strategy. According to the method and the device, the technical problem of relatively low recommendation precision of a mode of recommending the old-age care financial product to the user based on the preset rule in related technologies is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of artificial intelligence, the field of financial technology or other related technical fields, in particular, to a pension financial product recommendation method and device, electronic equipment and storage medium. BACKGROUND

[0002] With the acceleration of global population aging, pension financial products, as an important tool to help individuals and families plan for old-age funds and improve the quality of life in old age, are in increasing demand. Pension financial products cover a wide range, including but not limited to pension savings, pension insurance, pension funds, annuity products, etc., aiming to provide customized pension solutions for users of different ages and financial situations. Promoting pension financial products not only helps financial institutions expand their business scope and increase their market share, but also enhances user satisfaction and loyalty to financial institutions, achieving a win-win situation between financial institutions and users.

[0003] In related technologies, based on the basic information of the user, a preset rule or template is used for product recommendation. This method is simple to operate and easy to scale, but it ignores the user's deeper financial situation, risk preference and other personalized factors, resulting in generalized recommendation results and low product recommendation accuracy, making it difficult to meet the user's personalized needs.

[0004] To address the above problems, no effective solutions have been proposed so far. SUMMARY

[0005] The pension financial product recommendation method and device, electronic equipment and storage medium provided by the embodiments of the present application at least solve the technical problem of low recommendation accuracy in related technologies, where pension financial products are recommended for users based on a preset rule.

[0006] According to an aspect of some embodiments of the present application, a method for recommending a pension financial product is provided. The method includes: obtaining multi-dimensional pension data of a target user and pension financial product data of K pension financial products, and preprocessing the multi-dimensional pension data and the pension financial product data to obtain a pension feature vector and a financial product feature vector, wherein K is a positive integer; inputting the pension feature vector and the financial product feature vector into an interest degree evaluation model to output an interest degree value of each pension financial product for the target user, wherein the interest degree evaluation model is a model pre-constructed based on a collaborative filtering algorithm; screening K pension financial products based on the interest degree value to obtain N candidate pension financial products, wherein N is a positive integer and N is less than K; inputting the pension feature vector of the target user and the financial product feature vector of the K candidate pension financial products into a recommendation model to output a matching degree value of each candidate pension financial product for the target user, wherein the recommendation model is a model pre-constructed for evaluating the matching degree between a user and a pension financial product; screening a target pension financial product from the K candidate pension financial products based on the matching degree value, and generating a pension financial product recommendation strategy for the target user based on the target pension financial product and a strategy generation mechanism.

[0007] Further, the step of inputting the pension feature vector and the financial product feature vector into an interest degree evaluation model to output an interest degree value of each pension financial product for the target user includes: inputting the pension feature vector and the financial product feature vector into an interest degree evaluation model to construct a product interaction matrix of the target user by the interest degree evaluation model; calculating a similarity value between the target user and a known user by the interest degree evaluation model using the product interaction matrix, wherein the known user represents a historical user who has purchased a pension financial product; constructing a similar user group of the target user based on the similarity value, and calculating the interest degree value of each pension financial product for the target user based on similar users in the similar user group.

[0008] Further, the step of constructing a product interaction matrix of the target user by the interest degree evaluation model includes: extracting an interaction feature vector by the interest degree evaluation model, and calculating an interaction intensity between the target user and each pension financial product based on the interaction feature vector; constructing the product interaction matrix based on the target user, the pension financial product, and the interaction intensity.

[0009] Further, inputting the pension feature vector of the target user and the financial product feature vectors of the K pension financial products into a recommendation model outputs a matching degree value of each of the candidate pension financial products for the target user, including: inputting the pension feature vector of the target user and the financial product feature vector of each of the candidate pension financial products into an interest recommendation model, calculating the weighted cosine similarity between the target user and each of the candidate pension financial products through the interest recommendation model; the weighted cosine similarity is taken as the matching degree value of each of the candidate pension financial products for the target user, and the matching degree value of each of the candidate pension financial products for the target user is taken as the output data of the recommendation model.

[0010] Further, the step of screening K pension financial products based on the interest degree value to obtain N candidate pension financial products includes: comparing the interest degree value of each of the pension financial products for the target user with a preset interest degree threshold to obtain a comparison result; based on the comparison result, the pension financial products with an interest degree value greater than the preset interest degree threshold are screened out to obtain N candidate pension financial products.

[0011] Further, after outputting the pension financial product recommendation strategy for the target user, it further includes: collecting instant feedback information of the user for the recommended pension financial product, and inputting the instant feedback information into a reinforcement learning model, wherein the reinforcement learning model is a model constructed in advance based on a reinforcement learning algorithm; and optimizing the strategy generation mechanism through the reinforcement learning model.

[0012] Further, before obtaining the multi-dimensional pension data of the target user and the pension financial product data of the K pension financial products, it further includes: obtaining regional pension data in each region, wherein the pension data at least includes: old population data, pension security information, pension service facility information, and pension medical resource data; extracting user pension data of the user in the internal database of the financial institution; associating the regional pension data with the user, and integrating the regional pension data associated with the user with the user pension data to form the multi-dimensional pension data of the user; and constructing a knowledge base based on the multi-dimensional pension data of all users.

[0013] According to another aspect of the embodiments of the present application, a pension financial product recommendation device is also provided, comprising: an acquisition unit configured to acquire multi-dimensional pension data of a target user and pension financial product data of K pension financial products, and to pre-process the multi-dimensional pension data and the pension financial product data to obtain a pension feature vector and a financial product feature vector, wherein K is a positive integer; a first output unit configured to input the pension feature vector and the financial product feature vector into an interest degree evaluation model, and to output an interest degree value of the target user for each pension financial product, wherein the interest degree evaluation model is a model pre-constructed based on a collaborative filtering algorithm; a screening unit configured to screen the K pension financial products based on the interest degree value to obtain N candidate pension financial products, wherein N is a positive integer and N is less than K; a second output unit configured to input the pension feature vector of the target user and the financial product feature vector of the K candidate pension financial products into a recommendation model, and to output a matching degree value of the target user for each candidate pension financial product, wherein the recommendation model is a model pre-constructed for evaluating the matching degree between a user and a pension financial product; and a generation unit configured to screen a target pension financial product from the K candidate pension financial products based on the matching degree value, and to generate a pension financial product recommendation strategy for the target user based on the target pension financial product and a strategy generation mechanism.

[0014] Further, the first output unit comprises: a first construction module configured to input the pension feature vector and the financial product feature vector into an interest degree evaluation model, and to construct a product interaction matrix of the target user through the interest degree evaluation model; a first calculation module configured to calculate a similarity value of the target user and a known user through the interest degree evaluation model by using the product interaction matrix, wherein the known user represents a historical user who has purchased a pension financial product; and a second calculation module configured to construct a similar user group of the target user based on the similarity value, and to calculate the interest degree value of the target user for each pension financial product based on similar users in the similar user group.

[0015] Further, the first construction module comprises: a first calculation submodule configured to extract an interaction feature vector through the interest degree evaluation model, and to calculate an interaction intensity of the target user and each pension financial product based on the interaction feature vector; and a first construction submodule configured to construct the product interaction matrix based on the target user, the pension financial product, and the interaction intensity.

[0016] Further, the second output unit comprises: a third calculation module, configured to input the pension feature vector of the target user and the financial product feature vector of each candidate pension financial product into an interest recommendation model, and calculate a weighted cosine similarity between the target user and each candidate pension financial product through the interest recommendation model; and a first taking module, configured to take the weighted cosine similarity as a matching degree value between the target user and each candidate pension financial product, and take the matching degree value of the target user to each candidate pension financial product as output data of the recommendation model.

[0017] Further, the screening unit comprises: a first comparison module, configured to compare the interest degree value of the target user to each pension financial product with a preset interest degree threshold to obtain a comparison result; and a first screening module, configured to screen out a pension financial product with an interest degree value greater than the preset interest degree threshold based on the comparison result to obtain N candidate pension financial products.

[0018] Further, the pension financial product recommendation device further comprises: a first acquisition module, configured to acquire instant feedback information of a user to a recommended pension financial product, and input the instant feedback information into a reinforcement learning model, wherein the reinforcement learning model is a model pre-constructed based on a reinforcement learning algorithm; and a first optimization module, configured to optimize the policy generation mechanism through the reinforcement learning model.

[0019] Further, the pension financial product recommendation device further comprises: a first acquisition module, configured to acquire regional pension data in each region, wherein the pension data at least comprises: old population data, pension security information, pension service facility information, and pension medical resource data; a first extraction module, configured to extract user pension data of a user in an internal database of a financial institution; a first integration module, configured to associate the regional pension data with the user, and integrate the user-associated regional pension data with the user pension data to form multi-dimensional pension data of the user; and a second construction module, configured to construct a knowledge base based on the multi-dimensional pension data of all users.

[0020] According to another aspect of the embodiment of the present application, a computer readable storage medium is also provided, which comprises a stored computer program, wherein the computer program controls a device where the computer readable storage medium is located to perform any of the above pension financial product recommendation methods when the computer program is running.

[0021] According to another aspect of the embodiments of the present application, an electronic device is also provided, including one or more processors and a memory, the memory being configured to store one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to implement any of the above-mentioned methods for recommending a pension financial product.

[0022] According to another aspect of the embodiments of the present application, a computer program product is also provided, including a computer program, wherein the computer program, when executed by a processor, implements any of the above-mentioned methods for recommending a pension financial product.

[0023] In the present application, by the following steps: obtaining multi-dimensional pension data of a target user and pension financial product data of K pension financial products, and preprocessing the multi-dimensional pension data and the pension financial product data to obtain a pension feature vector and a financial product feature vector, then inputting the pension feature vector and the financial product feature vector into an interest degree evaluation model to output an interest degree value of the target user to each pension financial product, wherein the interest degree evaluation model is a model pre-constructed based on a collaborative filtering algorithm, and based on the interest degree value, the K pension financial products are screened to obtain N candidate pension financial products, wherein N is a positive integer and N is less than K, then inputting the pension feature vector of the target user and the financial product feature vector of the K candidate pension financial products into a recommendation model to output a matching degree value of the target user to each candidate pension financial product, wherein the recommendation model is a model pre-constructed for evaluating the matching degree between the user and the pension financial product, and finally screening a target pension financial product from the K candidate pension financial products based on the matching degree value, and generating a pension financial product recommendation strategy for the target user based on the target pension financial product and a strategy generation mechanism.

[0024] In the present application, the multi-dimensional pension data of the target user is obtained, and the pension financial products are collaboratively calculated in combination with the collaborative filtering algorithm and the product recommendation algorithm to screen out the pension financial product with the highest matching degree to the target user, realizing personalized product recommendation, which can accurately evaluate the user interest degree and the product matching degree, achieving the purpose of efficiently and accurately recommending the pension financial product to the target user, achieving the technical effect of product recommendation precision, and helping to improve the conversion rate and user satisfaction of the pension financial product. Further, the technical problem of low recommendation precision in the prior art, i.e., recommending the pension financial product to the user based on the preset rules, is solved. BRIEF DESCRIPTION OF DRAWINGS

[0025] The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiments of the application and together with the description serve to explain the application. In the drawings:

[0026] Figure 1 A hardware structure block diagram of a computer terminal (or mobile device) for implementing the method for recommending the pension financial product is shown;

[0027] Figure 2 A flowchart of an optional method for recommending the pension financial product according to an embodiment of the present application is shown;

[0028] Figure 3 An architecture diagram of an optional system for recommending the pension financial product according to an embodiment of the present application is shown;

[0029] Figure 4 A schematic diagram of an optional device for recommending the pension financial product according to an embodiment of the present application is shown;

[0030] Figure 5 A hardware structure block diagram of an electronic device (or mobile device) for implementing the method for recommending the pension financial product according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0031] In order to make the personnel in the technical field better understand the present application scheme, the technical scheme in the embodiments of the present application will be described clearly and completely below in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by the person skilled in the art without creative labor should belong to the scope of protection of the present application.

[0032] It should be noted that the terms "first", "second", and the like in the specification and claims of the present application and the above-described drawings are used to distinguish similar objects, and do not necessarily have to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, 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 "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device that includes a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0033] It should be noted that the method for recommending the pension financial product and the device thereof in the present application can be used in the field of artificial intelligence to recommend the pension financial product for users based on artificial intelligence, and can also be used in any field other than the field of artificial intelligence to recommend the pension financial product for users based on artificial intelligence. The application field of the method for recommending the pension financial product and the device thereof in the present application is not limited.

[0034] It should be noted that the collected information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for display, analyzed data, etc.) involved in the present application are information and data authorized by the user or authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of related data comply with relevant laws, regulations and standards, necessary security measures are taken, do not violate public order and good customs, and provide corresponding operation portal for user to choose authorization or refusal. For example, the system and the related users or institutions are provided with an interface, which provides a corresponding operation portal for the user to choose to agree or refuse the automatic decision result; if the user chooses to refuse, the expert decision process is entered.

[0035] The following embodiments of the present application can be applied to various pension financial product recommendation systems / applications / devices. The present application realizes the automatic recommendation of pension financial products by constructing a pension data knowledge base and using collaborative filtering, content recommendation and reinforcement learning algorithm. It can quickly process a large amount of data and automatically provide personalized pension financial product recommendation scheme for users, without the need for business personnel to spend a lot of time for manual screening and analysis, greatly improving the recommendation efficiency and recommendation accuracy, and improving the service response speed and the conversion rate of pension financial products.

[0036] The present application will be described in detail below in conjunction with various embodiments.

[0037] Embodiment one

[0038] According to the embodiments of the present application, an embodiment of a pension financial product recommendation method 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 the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from here.

[0039] The method embodiment provided by the embodiment one of the present application can be executed in a mobile terminal, a computer terminal or a similar operation device. Figure 1 A hardware structure block diagram of a computer terminal (or mobile device) for implementing the pension financial product recommendation method is shown. As shown in the figure, Figure 1As shown, the computer terminal 10 (or mobile device) may include one or more (illustrated as 102a, 102b, ..., 102n in the figure) 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 will 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 than shown, or with Figure 1 Different configurations shown.

[0040] It should be noted that the one or more processors 102 and / or other data processing circuits described above may generally be referred to herein as "data processing circuitry". The data processing circuitry may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuitry may be a single independent processing module, or may be incorporated in whole or in part into any of the other components of the computer terminal 10 (or mobile device). As described in the embodiments of the present application, the data processing circuitry serves as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0041] Memory 104 can be used to store software programs and modules for application software, such as the program instructions / data storage device corresponding to the pension financial product recommendation method in the embodiments of this application. Processor 102 executes the software programs and modules stored in memory 104 to perform various functional applications and data processing, thereby implementing the pension financial product recommendation method described above. Memory 104 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, memory 104 may further include memory located remotely from processor 102, which can be connected to computer terminal 10 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0042] The transmission device 106 is configured to receive or send data via a network. The network can include, for example, a wireless network provided by a communication provider of the computer terminal 10. In one example, the transmission device 106 includes a network interface controller (NIC) that can connect to other network devices through a base station to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module that is configured to communicate with the Internet wirelessly.

[0043] The display can be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with the user interface of the computer terminal 10 (or mobile device).

[0044] In the above operating environment, the present application provides a method for recommending a pension financial product, as shown in Figure 2 The method is implemented by a pension financial product recommendation system.

[0045] Figure 2 An optional method for recommending a pension financial product according to an embodiment of the present application is shown in the flowchart of Figure 2 The method includes the following steps:

[0046] A pension financial product is a financial tool designed specifically to help individuals or families plan, manage, and increase their retirement income. This type of product includes various forms such as savings plans, pension insurance, pension funds, annuities, etc., and aims to provide economic security for future retirement life through long-term accumulation and appreciation of funds. Compared with other financial products, pension financial products focus more on long-term and stability, usually have a longer investment period or guarantee period, and a relatively conservative risk management strategy to ensure the safety of funds and the sustainability of future retirement life. In the context of increasing population aging, it is increasingly important to prepare for retirement funds in advance. Pension financial products can provide professional and systematic fund management and growth plans to help users resist inflation, achieve asset preservation and appreciation, and reduce the pressure of life after retirement. In addition, many pension financial products also have tax incentives, which can further increase their appeal.

[0047] Personalized pension financial product recommendation aims to meet the unique economic conditions, risk tolerance and pension planning goals of different users. Everyone's financial situation, lifestyle, life expectancy and health status are different, which means their needs and ways of using pension funds are also very different. Through personalized recommendation, the most suitable pension financial product portfolio can be provided according to the specific circumstances of each individual, helping users more effectively plan and manage their pension funds, thereby improving their quality of life in retirement. Personalized recommendations can also enhance users' trust and satisfaction with financial institutions, promote long-term user relationship building, and are an important way for financial institutions to improve competitiveness and expand business.

[0048] Further, before acquiring the multi-dimensional pension data of the target user and the pension financial product data of the K pension financial products, it also includes: acquiring regional pension data in each region, wherein the pension data at least includes: old population data, pension security information, pension service facility information, and pension medical resource data; extracting user pension data of the user in the internal database of the financial institution; associating the regional pension data with the user, and integrating the user-associated regional pension data with the user pension data to form the multi-dimensional pension data of the user; and constructing a knowledge base based on the multi-dimensional pension data of all users.

[0049] Pension financial products have regional characteristics, and pension service configurations in different regions are not the same, and users' demand for pension financial products is also different. In the embodiment of the present application, the pension data knowledge base is constructed based on two aspects of regional pension data and user pension data, which provides a data basis for personalized recommendation of pension financial products. Specifically, first, an interface mechanism with regional pension data sources is established, including data cooperation with civil affairs departments, social security agencies, health and health departments, etc., to collect comprehensive regional pension data covering old population data (such as age distribution, gender ratio), pension security information (such as pension payment standards, subsidy policies), pension service facility information (the number and distribution of nursing homes and community nursing centers), and pension medical resource data (such as the number of hospitals, pension department configuration, medical insurance reimbursement range). These data will serve as the basis for building the knowledge base, ensuring that the recommendation system can fully consider the differences in pension environment and policy in different regions.

[0050] Secondly, the user information associated with pension is extracted from the internal database of the financial institution, including but not limited to age, income level, asset status, risk preference, purchased pension financial product records, and pension financial product consultation history, etc. These data are the key to forming user portraits, which help to accurately grasp the specific needs and preferences of each user.

[0051] Next, through data matching techniques, the regional information of each user is linked to the corresponding regional pension data, forming regional pension feature vectors. Then, these regional feature vectors are deeply integrated with the user's user pension data, forming user comprehensive pension data containing multi-dimensional information such as personal attributes, financial status, and regional background. This integration process ensures that the recommendation strategy can reflect both the user's personalized needs and the pension reality of the user's location, improving the accuracy of the recommendation.

[0052] Finally, based on the integration of all user multi-dimensional pension data, a comprehensive pension data knowledge base is constructed. This knowledge base not only contains user data, but also integrates regional pension data, reflecting the pension demand characteristics of different regions, different age groups, and different economic conditions. Through continuous updating and maintenance, the knowledge base will become the core of the dynamic adjustment of the recommendation system, ensuring that the recommendation results always keep pace with the current market situation, policy direction, and user changes.

[0053] Step S201, obtaining multi-dimensional pension data of the target user and pension financial product data of K pension financial products, and preprocessing the multi-dimensional pension data and the pension financial product data to obtain pension feature vectors and financial product feature vectors.

[0054] In the above step S201, the multi-dimensional pension data of the target user is extracted based on the pre-constructed pension data knowledge base, which can specifically include user attribute information (such as age, gender, income level), financial behavior data (such as savings habits, investment preferences), business data related to pension (such as whether to have a pension account, insurance purchase record), risk preference assessment results, and historical consultation records and purchase behavior. These data comprehensively reflect the user's personal attributes, economic status, and preferences and needs for pension financial products. At the same time, detailed information of K (K is a positive integer, which can be adjusted according to actual situation) pension financial products is obtained, including but not limited to product type, risk level, expected return, investment period, additional services (such as health management, legal consultation), etc. These data constitute the basis of financial product features and are the key basis for the recommendation algorithm to judge the matching degree of products and users.

[0055] After obtaining the original data, necessary data preprocessing is performed. Data preprocessing includes data cleaning to remove duplicate, invalid or missing data, data conversion to convert non-numeric data into numeric data, such as encoding product types into numbers, data standardization to scale numeric data to the same range and eliminate dimension effects. Finally, the multi-dimensional pension data of the target user is converted into pension feature vectors for model calculation, while the data of K pension financial products is converted into corresponding financial product feature vectors.

[0056] Step S202, input the pension feature vector and the financial product feature vector into the interest degree evaluation model, and output the interest degree value of the target user to each pension financial product.

[0057] In the above step S202, the potential interest of the target user to each pension financial product is quantified based on the interest degree evaluation model, which provides a decision basis for subsequent personalized recommendation. The interest degree evaluation model is a model pre-constructed based on a collaborative filtering algorithm. The pension feature vector and the financial product feature vector are input into the interest degree evaluation model. The pension feature vector contains personalized information such as user age, gender, income level, and risk preference, while the financial product feature vector covers product attributes such as product type, risk level, and yield range. The interest degree evaluation model calculates the interest degree value of the target user to each pension financial product based on the collaborative filtering algorithm, thereby screening the interested pension financial products for the target user.

[0058] Further, the step of inputting the pension feature vector and the financial product feature vector into the interest degree evaluation model to output the interest degree value of the target user to each pension financial product includes: inputting the pension feature vector and the financial product feature vector into the interest degree evaluation model, constructing a product interaction matrix of the target user through the interest degree evaluation model; calculating the similarity value between the target user and the known user through the interest degree evaluation model by using the product interaction matrix, wherein the known user represents a historical user who has purchased a pension financial product; constructing a similar user group of the target user based on the similarity value, and calculating the interest degree value of the target user to each pension financial product based on similar users in the similar user group.

[0059] Specifically, in calculating the interest degree value, first, the obtained target user pension feature vector and the financial product feature vector of K pension financial products are input into the pre-trained interest degree evaluation model. Based on the collaborative filtering algorithm, the model first constructs a product interaction matrix of the target user. The construction of the product interaction matrix is completed by analyzing the interaction data between the target user and the known user, i.e. the historical user who has purchased the pension financial product and the pension financial product. Each element represents the interaction intensity of the user to the specific product. The interaction intensity is quantified based on the purchase frequency, the number of consultations and other information. The similarity value between the target user and the known user is calculated using the product interaction matrix, aiming to find a user group with similar behavior patterns and preference tendencies as the target user.

[0060] In the embodiment of the present application, the cosine similarity can be used to calculate the similarity value between the target user and the known user:

[0061]

[0062] Wherein, Ru and Rv are the product interaction matrices of users u and v respectively.

[0063] Then, the known users similar to the target user are screened according to a preset similarity threshold, and a similar user group N(u) of the target user is constructed.

[0064] Finally, the interest degree value of the target user u for the pension financial product is predicted based on the similar user group:

[0065]

[0066] wherein, R vi represents the interest degree value of the user v in the similar user group for the pension financial product i.

[0067] Further, the step of constructing the product interaction matrix of the target user through the interest degree evaluation model includes: extracting an interaction feature vector through the interest degree evaluation model, and calculating the interaction intensity of the target user with each pension financial product based on the interaction feature vector; and constructing the product interaction matrix based on the target user, the pension financial product, and the interaction intensity.

[0068] Specifically, in the construction of the product interaction matrix, the interest degree evaluation model performs in-depth analysis on the pension feature vector of the target user and the financial product feature vector of each pension financial product, extracts an interaction feature vector that can reflect the potential interaction mode between the user and the product, the interaction feature vector can include: the attention degree of the target user to the pension financial product, the consultation frequency of the target user to the pension financial product, the financial behavior mode of the target user, etc., and then calculates the interaction intensity of the target user with each pension financial product according to the interaction feature vector, so as to quantify the close degree of the association between the user and the product. Finally, the target user, each pension financial product, and the calculated interaction intensity are integrated into a product interaction matrix. In the matrix, each row represents a target user, each column represents a pension financial product, and the element value in the matrix corresponds to the specific interaction intensity of the target user and the product. The purpose of constructing this matrix is to provide a structured data representation for the subsequent recommendation algorithm, so as to facilitate the algorithm to understand and process the complex relationship between the user and the product.

[0069] In step S203, the K pension financial products are screened based on the interest degree value, and N candidate pension financial products are obtained.

[0070] In the above step S203, the interest degree value of the user for each pension financial product is sorted, and the K pension financial products are screened based on a preset interest degree threshold, and N candidate pension financial products are obtained. Through the screening based on the interest degree value, N candidate products that are most likely to attract the attention of the target user can be refined from the K pension financial products. This strategy significantly improves the personalization and pertinence of the recommendation, reduces the interference to the user, and improves the success rate of the recommendation and the user satisfaction.

[0071] Further, the step of screening the K pension financial products based on the interest degree values to obtain N candidate pension financial products includes: comparing the interest degree value of each pension financial product of the target user with a preset interest degree threshold to obtain a comparison result; screening the pension financial products with the interest degree value greater than the preset interest degree threshold based on the comparison result to obtain the N candidate pension financial products.

[0072] Specifically, for each pension financial product, the interest degree value of the target user for the pension financial product is compared with the preset interest degree threshold, and if the interest degree value is greater than the preset interest degree threshold, the pension financial product is taken as a candidate pension financial product of the target user, and finally the N candidate pension financial products screened from the K initial pension financial products are obtained. The interest degree threshold is set according to historical data, business strategy or user feedback, and is used for preliminary screening of the initial pension financial products.

[0073] In step S204, the pension feature vector of the target user and the financial product feature vectors of the K candidate pension financial products are input into the recommendation model, and the matching degree values of the target user for each candidate pension financial product are output.

[0074] In the above step S204, the matching degree values of the target user for each candidate pension financial product are calculated by the recommendation model, and the candidate pension financial products are further screened based on the matching degree values, so as to realize accurate screening of the candidate pension financial products and ensure the personalization of the recommendation. The recommendation model is a model pre-constructed for evaluating the matching degree between the user and the pension financial product. The pension feature vector of the target user and the financial product feature vectors of the N candidate pension financial products screened are input into the pre-constructed recommendation model. The recommendation model internally operates through weighting, comparison, prediction and other algorithms to quantitatively evaluate the matching degree between the target user and each candidate pension financial product, and the output result is the matching degree value. These values reflect how the product meets the specific needs and preferences of the user, and are an important basis for the recommendation decision. The calculated matching degree value of each candidate pension financial product is output for further screening of the pension financial products.

[0075] Further, the step of inputting the pension feature vector of the target user and the financial product feature vectors of the K candidate pension financial products into the recommendation model and outputting the matching degree values of the target user to each candidate pension financial product includes: inputting the pension feature vector of the target user and the financial product feature vectors of each candidate pension financial product into an interest recommendation model, calculating the weighted cosine similarity between the target user and each candidate pension financial product through the interest recommendation model; taking the weighted cosine similarity as the matching degree value of the target user to each candidate pension financial product, and taking the matching degree value of the target user to each candidate pension financial product as the output data of the recommendation model.

[0076] Specifically, first, the pension feature vector of the target user and the financial product feature vectors of the K candidate pension financial products are input into the recommendation model. Then, the recommendation model uses the weighted cosine similarity algorithm to calculate the matching degree value between the target user and each candidate pension financial product. The weighted cosine similarity considers the importance of each dimension of the vector, can provide more reasonable similarity evaluation results, quantitatively evaluate the closeness between user features and product features, and reflect the possible interest and matching of the user to the candidate product. Finally, the calculated weighted cosine similarity is taken as the matching degree value of the target user to each candidate pension financial product, and is taken as the output data of the recommendation model.

[0077] An optional embodiment, the financial product feature vector of the candidate pension financial product can be represented as: fi=[fi1,fi2,…,fik], the user pension feature vector can be represented as: cu=[cu1,cu2,…,cul], and the matching degree value of the two is calculated by using the recommendation model:

[0078]

[0079] wherein, w f represents the weight of the feature f, which can be automatically configured by the recommendation model, and the weight values of each feature can be updated based on the feedback of the user to the recommendation strategy.

[0080] Step S205, based on the matching degree value, the target pension financial product is selected from the K candidate pension financial products, and a pension financial product recommendation strategy is generated for the target user based on the target pension financial product and the strategy generation mechanism.

[0081] In step S205, the matching degree values of the N candidate products output in step S204 are sorted, usually in descending order of matching degree values, to identify the product that best matches the user's needs. Based on the sorted matching degree values, the product or product combination with the highest matching degree value is selected as the target pension financial product. This decision-making process ensures that the recommendation system can provide financial products that best meet the user's individual pension planning needs. Using a strategy generation mechanism, a specific recommendation strategy is designed based on the target pension financial product and user characteristics. The strategy may include product presentation, recommended time, recommended channels (such as email, SMS, financial APP push, etc.), and interaction with the product (such as providing detailed product consultation links, setting up a direct purchase button, etc.). Finally, the generated recommendation strategy is output to the target user, ensuring that the user receives the most suitable pension financial product recommendation without disturbing the user, thereby improving the conversion rate of pension financial products.

[0082] Further, after outputting the pension financial product recommendation strategy for the target user, it also includes collecting the user's immediate feedback information on the recommended pension financial product, and inputting the immediate feedback information into a reinforcement learning model, wherein the reinforcement learning model is a model pre-constructed based on a reinforcement learning algorithm; and optimizing the strategy generation mechanism through the reinforcement learning model.

[0083] Specifically, after recommending the pension financial product to the target user, it is also necessary to determine the user's feedback information on the product recommendation. After the user receives the recommendation strategy, the system will immediately start a monitoring mechanism to collect the user's immediate feedback information on the recommended product. This information can quickly reflect whether the recommendation has aroused the user's interest and whether the user has taken further action, such as clicking on the link to view detailed information, seeking advice or making a purchase, etc. The collected immediate feedback information is then input into the reinforcement learning model. The purpose of this process is to enable the model to "learn" the user's real reactions and evaluate the effectiveness and optimization points of the recommendation strategy. By linking feedback information with previous recommendation strategies, the reinforcement learning model can determine which recommendation strategies have received positive feedback and which strategies may need to be adjusted. Based on the reinforcement learning model's analysis of immediate feedback, the system automatically adjusts the parameters and rules in the strategy generation mechanism, enabling future recommendation strategies to better adapt to user needs and market changes. This optimization process is dynamic and continuous, ensuring that the recommendation strategy evolves over time and improves the individualization and effectiveness of the recommendation.

[0084] Through the above steps, the multi-dimensional pension data of the target user and the pension financial product data of the K pension financial products are obtained, and the multi-dimensional pension data and the pension financial product data are preprocessed to obtain a pension feature vector and a financial product feature vector. The pension feature vector and the financial product feature vector are input into an interest degree evaluation model, and an interest degree value of the target user for each pension financial product is output. The interest degree evaluation model is a model pre-constructed based on a collaborative filtering algorithm, and the K pension financial products are screened based on the interest degree value to obtain N candidate pension financial products, wherein N is a positive integer and N is less than K. Then, the pension feature vector of the target user and the financial product feature vector of the K candidate pension financial products are input into a recommendation model, and a matching degree value of the target user for each candidate pension financial product is output. The recommendation model is a model pre-constructed for evaluating the matching degree between the user and the pension financial product. Finally, the target pension financial product is screened from the K candidate pension financial products based on the matching degree value, and a pension financial product recommendation strategy is generated for the target user based on the target pension financial product and the strategy generation mechanism.

[0085] In the embodiment, the multi-dimensional pension data of the target user is obtained, and the pension financial products are collaboratively calculated in combination with the collaborative filtering algorithm and the product recommendation algorithm to screen the pension financial product with the highest matching degree for the target user, realize personalized product recommendation, accurately evaluate the user interest degree and the product matching degree, achieve the purpose of efficiently and accurately recommending the pension financial product for the target user, and achieve the technical effect of product recommendation precision, which helps to improve the conversion rate and user satisfaction of the pension financial product. Further, the technical problem of low recommendation precision in the related art that the pension financial product is recommended for the user based on the preset rules is solved.

[0086] Figure 3 is an optional pension financial product recommendation system architecture diagram according to an embodiment of the application, as Figure 3 shown, the pension financial product recommendation system includes a data collection module, a data preprocessing module, a pension financial product knowledge base, and a pension financial product recommendation module, specifically,

[0087] The data collection module is used for regional pension data collection and financial institution user data collection. Various data pipelines are established to cooperate with regional civil affairs departments, social security agencies, health and health departments, etc. to collect multi-dimensional pension-related data including local elderly population statistics (age distribution, gender ratio, etc.), pension security policies (pension payment standards, subsidy policies, etc.), pension service facilities (number of nursing homes, distribution, service quality, etc.), elderly medical resources (number of hospitals, specialty characteristics, medical insurance reimbursement range, etc.), and local residents' average life expectancy, incidence of common old-age diseases, etc. to comprehensively collect and integrate regional pension data. These data provide rich regional background information for the recommendation system, enabling the recommendation results to fully consider regional differences and pension resource distribution, tailor pension financial product solutions more suitable for the actual situation of the user's region, and improve the scientificity and rationality of the recommendation.

[0088] Meanwhile, user basic information (name, age, gender, occupation, income level, asset status, family status, etc.), financial behavior data (savings habits, investment preferences, insurance purchase records, credit history, etc.), pension-related business data (whether to open a pension account, enterprise annuity payment, etc.), user risk preference assessment results, and past pension financial product consultation and purchase records are extracted from the financial institution's internal database. Through the construction of user pension data (corresponding to the above-mentioned user pension data), the accurate portrait of user pension demand is realized, providing strong support for personalized recommendation.

[0089] The data preprocessing module includes a data cleaning module, a data standardization module, and a data fusion module. The data cleaning module is used to clean the collected regional pension data and user pension data, removing duplicate values, missing values, and incorrect data. For missing data items, interpolation algorithms or statistical model-based methods are used for reasonable estimation and supplementation. For obviously abnormal data points, they are identified and removed or corrected to ensure the accuracy and integrity of the data. Differential privacy technology is used to desensitize sensitive fields, and entity alignment algorithms are used to establish user-pension resource association relationships.

[0090] The data standardization module is used to standardize different sources and formats of data. For example, for elderly population age distribution data, it is converted to a standardized proportion format with five-year intervals. For user income level and asset status data, Z-score standardization or range standardization methods are used to map them to the same numerical range for subsequent analysis and comparison.

[0091] The data fusion module is used to fuse regional pension data and user pension data, taking the region where the user is located and the user characteristics as the association key, to integrate the information of regional pension resources and individual user data. For example, the distribution data of pension service facilities in the city where a user is located is combined with the user's living location, economic status and other information to form a comprehensive data record of the user containing regional pension environment characteristics.

[0092] After the regional pension data and the user pension data are fused, deep feature analysis is performed on the fused data to mine key features related to pension planning. For regional pension data, the pension burden coefficient (a comprehensive index of the proportion of the elderly population and the local economic level), the pension resource accessibility index (a weighted combination of indicators such as the density of pension service facilities and the convenience of public transportation), the medical and pension integration degree (the cooperation between medical institutions and pension institutions, the coverage rate of medical and pension combination services, etc.), and other macro features are extracted; for user pension data, the pension wealth accumulation progress (based on user age, income, assets, etc. to predict the wealth gap between the wealth size at retirement and the wealth required for the ideal pension life level), the pension risk exposure degree (based on user health status, occupational risk, etc. to assess the main pension risk types and degree), and other micro features are extracted.

[0093] A hierarchical framework structure of the pension product knowledge base is designed, including three levels of regional level, user group level and individual user level. At the regional level, the pension macro feature data of each region is stored according to the administrative division levels of province, city, district, etc.; at the user group level, users are clustered and divided according to user age interval, income level range, occupation category, risk preference type, etc., forming a plurality of user group portraits with similar pension demand characteristics, and associating the corresponding regional pension resources and policy adaptation knowledge; at the individual user level, the comprehensive feature data of each user and the pension planning suggestion template and historical recommendation record customized for each user are recorded in detail.

[0094] Based on the results of feature analysis extraction, the corresponding knowledge content is filled into the knowledge base framework. At the same time, a dynamic updating mechanism of the knowledge base is established, and the latest data is regularly collected from the data source, and after preprocessing and feature analysis, the regional pension data, customer group characteristics and individual customer information in the knowledge base are updated to ensure the timeliness and accuracy of the knowledge base content to reflect the changes of the pension environment and customer demand.

[0095] The pension financial product recommendation module selects suitable pension financial products for the target user based on collaborative filtering algorithm and recommendation algorithm and generates a recommendation strategy to send to the target user, and finally optimizes the recommendation strategy based on the real-time feedback information of the target user using reinforcement learning algorithm.

[0096] Based on the collaborative filtering algorithm, multiple pension financial products can be preliminarily screened. Specifically, the cosine similarity can be used to calculate the similarity between the target user and the known users:

[0097]

[0098] Among them, Ru and Rv are the product interaction matrices of users u and v respectively.

[0099] Then, known users similar to the target user are screened out according to the preset similarity threshold, and a similar user group N(u) of the target user is constructed.

[0100] Finally, based on similar user groups, the target user u’s interest in pension financial products is predicted:

[0101]

[0102] Among them, R vi Represents the interest value of user v in the similar user group towards pension financial product i.

[0103] Based on the interest value, N pension financial products that the target user is interested in are selected.

[0104] Then, based on the recommendation algorithm, N pension financial products are accurately screened. The financial product feature vector of the candidate pension financial product can be expressed as: fi = [fi1, fi2, …, fik], and the user pension feature vector can be expressed as: cu = [cu1, cu2, …, cul]. The recommendation model is used to calculate the matching value between the two:

[0105]

[0106] Among them, w f The weight of feature f can be automatically configured by the recommendation model, and the weight value of each feature can be updated based on the user's feedback on the recommendation strategy.

[0107] Based on the matching value, the target pension financial products with the highest adaptability to the target users are screened out, and a strategy generation mechanism is used to design a specific pension financial product recommendation strategy based on the target pension financial products and user characteristics.

[0108] Finally, the recommendation strategy is updated based on the instant feedback from the user. The state space S, action space A, and reward function R are defined. The state s includes customer characteristics, product features, and environmental characteristics. The action a is the recommended pension financial product or product portfolio. The reward r is composed of multi-dimensional indicators, including the user's instant feedback (browsing time, clicks, purchases, etc.), long-term returns, and the degree of fit with the pension planning goal. The deep Q-network algorithm is used to initialize the neural network parameters θ. The Q-value function Q(s, a; θ) represents the expected cumulative reward of choosing action a in state s. The ε-greedy strategy is used to select actions:

[0109]

[0110] The recommended product portfolio is displayed to the customer, and the customer's behavior feedback is recorded and the reward is calculated. The experience replay mechanism is used to randomly sample (s, a, r, s') from the experience buffer, and the network parameters θ are updated:

[0111]

[0112] where α is the learning rate and γ is the discount factor.

[0113] By linking feedback information with previous recommendation strategies, the reinforcement learning model can determine which recommendation strategies have received positive feedback and which strategies may need adjustment. Based on the analysis of instant feedback by the reinforcement learning model, the system automatically adjusts the parameters and rules in the strategy generation mechanism, enabling future recommendation strategies to better adapt to user needs and market changes. This optimization process is dynamic and continuous, ensuring that the recommendation strategy evolves over time, improving the personalization and effectiveness of recommendations.

[0114] The recommendation system can also include a recommendation result presentation and interaction module to visually display recommended products, providing product details, advantages, and target audience information. It allows customers to sort and filter products, supports viewing details, consulting, or purchasing operations. Behavior data such as customer browsing time and click path are recorded, and an evaluation portal is set up to collect feedback for the reinforcement learning algorithm to optimize the recommendation strategy.

[0115] The embodiment of the present application realizes the automatic recommendation of pension financial products by constructing a pension data knowledge base and using collaborative filtering, content recommendation, and reinforcement learning algorithms. It can quickly process large amounts of data and automatically provide personalized pension financial product recommendation solutions for users, without the need for business personnel to spend a lot of time on manual screening and analysis, greatly improving the recommendation efficiency and accuracy, and improving the service response speed and the conversion rate of pension financial products.

[0116] The following will be described in detail in conjunction with another embodiment.

[0117] Embodiment Two

[0118] The pension financial product recommendation device provided in the embodiment comprises a plurality of implementation units, each of which corresponds to each implementation step in Embodiment I, and the specific implementation and beneficial effects thereof can refer to the foregoing method embodiment, which will not be described here.

[0119] Figure 4 is a schematic diagram of an optional pension financial product recommendation device according to an embodiment of the present application, as shown in Figure 4 The pension financial product recommendation device can comprise an acquisition unit 41, a first output unit 42, a screening unit 43, a second output unit 44, and a generation unit 45, wherein

[0120] The acquisition unit 41 is configured to acquire multi-dimensional pension data of a target user and pension financial product data of K pension financial products, and to preprocess the multi-dimensional pension data and the pension financial product data to obtain a pension feature vector and a financial product feature vector, wherein K is a positive integer.

[0121] The first output unit 42 is configured to input the pension feature vector and the financial product feature vector into an interest degree evaluation model, and to output an interest degree value of the target user for each pension financial product, wherein the interest degree evaluation model is a model pre-constructed based on a collaborative filtering algorithm.

[0122] The screening unit 43 is configured to screen the K pension financial products based on the interest degree value to obtain N candidate pension financial products, wherein N is a positive integer and N is less than K.

[0123] The second output unit 44 is configured to input the pension feature vector of the target user and the financial product feature vector of the K candidate pension financial products into a recommendation model, and to output a matching degree value of the target user for each candidate pension financial product, wherein the recommendation model is a model pre-constructed for evaluating the matching degree between a user and a pension financial product.

[0124] The generation unit 45 is configured to screen a target pension financial product from the K candidate pension financial products based on the matching degree value, and to generate a pension financial product recommendation strategy for the target user based on the target pension financial product and a strategy generation mechanism.

[0125] The recommendation device of the pension financial product can acquire multi-dimensional pension data of a target user and pension financial product data of K pension financial products through an acquisition unit 41, and pre-process the multi-dimensional pension data and the pension financial product data to obtain a pension feature vector and a financial product feature vector, where K is a positive integer; input the pension feature vector and the financial product feature vector into an interest degree evaluation model through a first output unit 42, and output an interest degree value of the target user for each pension financial product, where the interest degree evaluation model is a model pre-constructed based on a collaborative filtering algorithm; filter the K pension financial products based on the interest degree value through a filtering unit 43 to obtain N candidate pension financial products, where N is a positive integer and N is less than K; input the pension feature vector of the target user and the financial product feature vector of the K candidate pension financial products into a recommendation model through a second output unit 44, and output a matching degree value of the target user for each candidate pension financial product, where the recommendation model is a model pre-constructed for evaluating the matching degree between a user and a pension financial product; and filter a target pension financial product from the K candidate pension financial products based on the matching degree value through a generation unit 45, and generate a pension financial product recommendation strategy for the target user based on the target pension financial product and a strategy generation mechanism.

[0126] In this embodiment, multi-dimensional pension data of a target user is acquired, and a pension financial product is collaboratively calculated in combination with a collaborative filtering algorithm and a product recommendation algorithm to filter out a pension financial product with the highest matching degree for the target user, realize personalized product recommendation, accurately evaluate user interest degree and product matching degree, achieve the goal of efficiently and accurately recommending a pension financial product for a target user, and achieve the technical effect of product recommendation precision, which helps to improve the conversion rate and user satisfaction of the pension financial product. Furthermore, the technical problem of low recommendation precision in the related art, i.e., recommending a pension financial product for a user based on a pre-set rule, is solved.

[0127] Further, the first output unit includes: a first construction module configured to input the pension feature vector and the financial product feature vector into the interest degree evaluation model, and construct a product interaction matrix of the target user through the interest degree evaluation model; a first calculation module configured to calculate a similarity value of the target user and a known user through the interest degree evaluation model by using the product interaction matrix, where the known user represents a historical user who has purchased a pension financial product; and a second calculation module configured to construct a similar user group of the target user based on the similarity value, and calculate the interest degree value of the target user for each pension financial product based on similar users in the similar user group.

[0128] Further, the first construction module comprises: a first calculation submodule configured to extract an interaction feature vector by using the interest degree evaluation model, and calculate the interaction intensity between the target user and each pension financial product based on the interaction feature vector; and a first construction submodule configured to construct a product interaction matrix based on the target user, the pension financial product, and the interaction intensity.

[0129] Further, the second output unit comprises: a third calculation module configured to input the pension feature vector of the target user and the financial product feature vector of each candidate pension financial product into the interest recommendation model, and calculate the weighted cosine similarity between the target user and each candidate pension financial product by using the interest recommendation model; and a first taking module configured to take the weighted cosine similarity as the matching degree value between the target user and each candidate pension financial product, and take the matching degree value of the target user to each candidate pension financial product as the output data of the recommendation model.

[0130] Further, the screening unit comprises: a first comparison module configured to compare the interest degree value of the target user to each pension financial product with a preset interest degree threshold to obtain a comparison result; and a first screening module configured to screen out the pension financial product with an interest degree value greater than the preset interest degree threshold based on the comparison result to obtain N candidate pension financial products.

[0131] Further, the pension financial product recommendation device further comprises: a first acquisition module configured to acquire instant feedback information of the user to the recommended pension financial product, and input the instant feedback information into a reinforcement learning model, wherein the reinforcement learning model is a model pre-constructed based on a reinforcement learning algorithm; and a first optimization module configured to optimize the strategy generation mechanism by using the reinforcement learning model.

[0132] Further, the pension financial product recommendation device further comprises: a first acquisition module configured to acquire regional pension data in each region, wherein the pension data at least comprises: old population data, pension security information, pension service facility information, and pension medical resource data; a first extraction module configured to extract user pension data of the user in the internal database of the financial institution; a first integration module configured to associate the regional pension data with the user, and integrate the regional pension data associated with the user with the user pension data to form multi-dimensional pension data of the user; and a second construction module configured to construct a knowledge base based on the multi-dimensional pension data of all users.

[0133] It should be noted that the acquisition unit 41, the first output unit 42, the screening unit 43, the second output unit 44, and the generation unit 45 correspond to steps S201 to S204 in the first embodiment. The examples and application scenarios implemented by the above units and the corresponding steps are the same, but are not limited to the contents disclosed in the first embodiment. It should be noted that the above modules or units can be hardware components or software components stored in a memory (e.g., memory 104) and processed by one or more processors (e.g., processors 102a, 102b, ..., 102n). The above modules or units can also be part of a device and can be run in the computer terminal 10 provided in the first embodiment.

[0134] The present invention is described below in conjunction with another optional embodiment.

[0135] Example 3

[0136] An embodiment of the present invention may further provide an electronic device, Figure 5 is a hardware structure block diagram of an electronic device (or mobile device) for executing an optional method for recommending pension financial products according to an embodiment of the present invention, such as Figure 5 As shown, the electronic device may include: one or more ( Figure 5 Only one is shown) processor 502, memory 504, storage controller, and peripheral interface, wherein the peripheral interface is connected to the radio frequency module, audio module and display.

[0137] 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. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, implementing 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 a combination thereof.

[0138] The processor can call information and application programs stored in the memory through the transmission device to perform the following steps: obtaining multi-dimensional pension data of a target user and pension financial product data of K pension financial products, and preprocessing the multi-dimensional pension data and the pension financial product data to obtain a pension feature vector and a financial product feature vector, wherein K is a positive integer; inputting the pension feature vector and the financial product feature vector into an interest degree evaluation model to output an interest degree value of the target user for each pension financial product, wherein the interest degree evaluation model is a model pre-constructed based on a collaborative filtering algorithm; screening K pension financial products based on the interest degree value to obtain N candidate pension financial products, wherein N is a positive integer and N is less than K; inputting the pension feature vector of the target user and the financial product feature vector of the K candidate pension financial products into a recommendation model to output a matching degree value of the target user for each candidate pension financial product, wherein the recommendation model is a model pre-constructed for evaluating the matching degree between the user and the pension financial product; screening a target pension financial product from the K candidate pension financial products based on the matching degree value, and generating a pension financial product recommendation strategy for the target user based on the target pension financial product and a strategy generation mechanism.

[0139] The processor can also call information and application programs stored in the memory through the transmission device to perform the following steps: inputting the pension feature vector and the financial product feature vector into the interest degree evaluation model to construct a product interaction matrix of the target user through the interest degree evaluation model; calculating a similarity value of the target user and a known user through the interest degree evaluation model by using the product interaction matrix, wherein the known user represents a historical user who has purchased a pension financial product; constructing a similar user group of the target user based on the similarity value, and calculating the interest degree value of the target user for each pension financial product based on similar users in the similar user group.

[0140] The processor can also call information and application programs stored in the memory through the transmission device to perform the following steps: extracting an interaction feature vector through the interest degree evaluation model, and calculating an interaction intensity of the target user and each pension financial product based on the interaction feature vector; constructing a product interaction matrix based on the target user, the pension financial product, and the interaction intensity.

[0141] The processor can also call information and application programs stored in the memory through the transmission device to perform the following steps: inputting the pension feature vector of the target user and the financial product feature vector of each candidate pension financial product into the interest recommendation model to calculate a weighted cosine similarity of the target user and each candidate pension financial product through the interest recommendation model; taking the weighted cosine similarity as a matching degree value of the target user and each candidate pension financial product, and taking the matching degree value of the target user for each candidate pension financial product as output data of the recommendation model.

[0142] The processor can also call information and application programs stored in the memory through the transmission device to perform the following steps: comparing the interest degree value of the target user to each pension financial product with a preset interest degree threshold to obtain a comparison result; and screening the pension financial products with an interest degree value greater than the preset interest degree threshold based on the comparison result to obtain N candidate pension financial products.

[0143] The processor can also call information and application programs stored in the memory through the transmission device to perform the following steps: collecting instant feedback information of the user on the recommended pension financial product, and inputting the instant feedback information into a reinforcement learning model, wherein the reinforcement learning model is a model pre-constructed based on a reinforcement learning algorithm; and optimizing a strategy generation mechanism through the reinforcement learning model.

[0144] The processor can also call information and application programs stored in the memory through the transmission device to perform the following steps: obtaining regional pension data in each region, wherein the pension data at least includes: old population data, pension security information, pension service facility information, and pension medical resource data; extracting user pension data of the user in the internal database of the financial institution; associating the regional pension data with the user, and integrating the regional pension data associated with the user with the user pension data to form multi-dimensional pension data of the user; and constructing a knowledge base based on the multi-dimensional pension data of all users.

[0145] By adopting the embodiment of the present application, a pension financial product recommendation method is provided. Multi-dimensional pension data of a target user is obtained, and a pension financial product is calculated in combination with a collaborative filtering algorithm and a product recommendation algorithm to screen a pension financial product with the highest matching degree to the target user, so as to realize personalized product recommendation, accurately evaluate user interest degree and product matching degree, achieve the purpose of efficiently and accurately recommending a pension financial product to a target user, and achieve the technical effect of product recommendation precision, which helps to improve the conversion rate and user satisfaction of the pension financial product. Furthermore, the technical problem of low recommendation precision in the related art that a pension financial product is recommended to a user based on a preset rule is solved.

[0146] Those skilled in the art can understand that Figure 5 The structure shown is only schematic, and the electronic device can also be a smart phone, a tablet computer, a palm computer, a Mobile Internet Device (MID), a PAD, or the like. Figure 5 It does not limit the structure of the above-mentioned electronic device. For example, the electronic device can include more or less components (such as a network interface, a display device, etc.) than those shown in the figure, or have a different configuration from that shown in the figure. Figure 5 It does not limit the structure of the above-mentioned electronic device. For example, the electronic device can include more or less components (such as a network interface, a display device, etc.) than those shown in the figure, or have a different configuration from that shown in the figure. Figure 5 It does not limit the structure of the above-mentioned electronic device. For example, the electronic device can include more or less components (such as a network interface, a display device, etc.) than those shown in the figure, or have a different configuration from that shown in the figure.

[0147] Those skilled in the art can understand that all or part of the steps of various methods in the above embodiments can be completed by instructing the terminal device related hardware through a program, and the program can be stored in a computer readable storage medium, which can include a flash disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0148] The application will be described in detail below in combination with another alternative embodiment.

[0149] Embodiment four

[0150] The embodiment of the application further provides a computer readable storage medium. Optionally, in the embodiment of the application, the computer readable storage medium can be used to save the program code executed by the recommendation method of the pension financial product provided in the embodiment one.

[0151] Optionally, in the embodiment of the application, the storage medium can be located in any one of the computer terminals in the computer terminal group in the computer network, or in any one of the mobile terminals in the mobile terminal group.

[0152] The embodiment of the application further provides a computer program product, when executed on a data processing device, is suitable for executing the steps of the recommendation method of the pension financial product: obtaining multi-dimensional pension data of a target user and pension financial product data of K pension financial products, and pre-processing the multi-dimensional pension data and the pension financial product data to obtain a pension feature vector and a financial product feature vector, wherein K is a positive integer; inputting the pension feature vector and the financial product feature vector into an interest degree evaluation model to output an interest degree value of the target user to each pension financial product, wherein the interest degree evaluation model is a model pre-constructed based on a collaborative filtering algorithm; screening the K pension financial products based on the interest degree value to obtain N candidate pension financial products, wherein N is a positive integer, and N is less than K; inputting the pension feature vector of the target user and the financial product feature vector of the K candidate pension financial products into a recommendation model to output a matching degree value of the target user to each candidate pension financial product, wherein the recommendation model is a model pre-constructed for evaluating the matching degree between the user and the pension financial product; screening a target pension financial product from the K candidate pension financial products based on the matching degree value, and generating a pension financial product recommendation strategy for the target user based on the target pension financial product and a strategy generation mechanism.

[0153] The above embodiment numbers of the application are only for description, and do not represent the advantages and disadvantages of the embodiments.

[0154] In the above-mentioned embodiments of the present application, the description of each embodiment is focused on, and the part not described in detail in a certain embodiment can be referred to the relevant description of other embodiments.

[0155] In several embodiments provided in the present application, it should be understood that the disclosed technical contents can be implemented by other ways. Among them, the above-mentioned device embodiments are only schematic, for example, the division of the units can be a logical function division, and in actual implementation, there can be another division way, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units or modules shown or discussed can be indirect coupling or communication connection through some interfaces, and can be electrical or other forms.

[0156] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed to multiple units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0157] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of software functional unit.

[0158] The integrated unit, if realized in the form of software functional unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part of the prior art that makes contributions or the whole or part of the technical solutions can be embodied in the form of software product, which is stored in a storage medium and includes a plurality of instructions to make a computer device (which can be a personal computer, a server or a network device, etc.) execute all or part of the steps of the method described in each embodiment of the present application. The foregoing storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and various program code storage media.

[0159] The above-mentioned is only the preferred embodiment of the present application, and it should be pointed out that for ordinary skilled in the art, without departing from the principles of the present application, a number of improvements and refinements can be made, and these improvements and refinements should be regarded as the protection scope of the present application.

Claims

1. A method for recommending pension financial products, characterized in that: include: Obtain multidimensional pension data of a target user and pension financial product data of K pension financial products, and preprocess the multidimensional pension data and the pension financial product data to obtain a pension feature vector and a financial product feature vector, where K is a positive integer; Inputting the pension feature vector and the financial product feature vector into an interest evaluation model, and outputting the target user's interest value for each pension financial product, wherein the interest evaluation model is a model pre-built based on a collaborative filtering algorithm; Screening the K pension financial products based on the interest values ​​to obtain N candidate pension financial products, where N is a positive integer and is less than K; Inputting the target user's pension feature vector and the K financial product feature vectors of the candidate pension financial products into a recommendation model, and outputting the target user's matching value for each of the candidate pension financial products, wherein the recommendation model is a pre-built model for evaluating the matching degree between a user and a pension financial product; A target pension financial product is screened out from the K candidate pension financial products based on the matching value, and a pension financial product recommendation strategy is generated for the target user based on the target pension financial product and a strategy generation mechanism.

2. The method according to claim 1, characterized in that The steps of inputting the pension feature vector and the financial product feature vector into an interest evaluation model and outputting the target user's interest value for each pension financial product include: Inputting the pension feature vector and the financial product feature vector into an interest evaluation model, and constructing a product interaction matrix for the target user through the interest evaluation model; Utilizing the product interaction matrix, and using the interest evaluation model, calculating similarity values ​​between the target user and known users, wherein the known users represent historical users who have purchased pension financial products; A similar user group of the target user is constructed based on the similarity value, and the interest value of the target user in each of the pension financial products is calculated based on similar users in the similar user group.

3. The method according to claim 2, characterized in that The step of constructing the product interaction matrix of the target user by using the interest evaluation model includes: Extracting interaction feature vectors through the interest evaluation model, and calculating the interaction intensity between the target user and each of the pension financial products based on the interaction feature vectors; The product interaction matrix is ​​constructed based on the target user, the pension financial product, and the interaction intensity.

4. The method according to claim 1, wherein The steps of inputting the pension feature vector of the target user and the financial product feature vectors of the K candidate pension financial products into the recommendation model and outputting the matching value of the target user to each of the candidate pension financial products include: Inputting the pension feature vector of the target user and the financial product feature vectors of each candidate pension financial product into an interest recommendation model, and calculating the weighted cosine similarity between the target user and each candidate pension financial product through the interest recommendation model; The weighted cosine similarity is used as a matching value between the target user and each of the candidate pension financial products, and the matching value between the target user and each of the candidate pension financial products is used as output data of the recommendation model.

5. The method according to claim 1, characterized in that The steps of screening the K pension financial products based on the interest values ​​to obtain N candidate pension financial products include: Comparing the target user's interest value in each of the pension financial products with a preset interest threshold to obtain a comparison result; Based on the comparison result, pension financial products whose interest value is greater than a preset interest threshold are screened out to obtain N candidate pension financial products.

6. The method according to claim 1, characterized in that After outputting the pension financial product recommendation strategy for the target user, the method further includes: Collecting real-time feedback information from users on recommended pension financial products and inputting the real-time feedback information into a reinforcement learning model, wherein the reinforcement learning model is a model pre-built based on a reinforcement learning algorithm; The strategy generation mechanism is optimized through the reinforcement learning model.

7. The method according to claim 1, characterized in that Before obtaining the multi-dimensional pension data of target users and the pension financial product data of K pension financial products, it also includes: Acquire regional elderly care data in each region, wherein the elderly care data at least includes: elderly population data, elderly care security information, elderly care service facility information, and elderly care medical resource data; Extracting user pension data from the internal database of financial institutions; Associating the regional pension data with the user, and integrating the user-associated regional pension data with the user pension data to form multi-dimensional pension data for the user; Build a knowledge base based on the multi-dimensional pension data of all users.

8. A device for recommending pension financial products, characterized in that: include: an acquisition unit, configured to acquire multidimensional pension data of a target user and pension financial product data of K pension financial products, and preprocess the multidimensional pension data and the pension financial product data to obtain a pension feature vector and a financial product feature vector, wherein K is a positive integer; a first output unit, configured to input the pension feature vector and the financial product feature vector into an interest evaluation model, and output the target user's interest value in each pension financial product, wherein the interest evaluation model is a model pre-built based on a collaborative filtering algorithm; a screening unit, configured to screen the K pension financial products based on the interest values ​​to obtain N candidate pension financial products, where N is a positive integer and is less than K; a second output unit, configured to input the pension feature vector of the target user and the financial product feature vectors of the K candidate pension financial products into a recommendation model, and output a matching value of the target user to each of the candidate pension financial products, wherein the recommendation model is a pre-built model for evaluating the matching degree between a user and a pension financial product; A generating unit is configured to screen out a target pension financial product from the K candidate pension financial products based on the matching value, and generate a pension financial product recommendation strategy for the target user based on the target pension financial product and a strategy generating mechanism.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the pension financial product recommendation method according to any one of claims 1 to 7.

10. An electronic device, characterized in that: The system comprises one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method for recommending pension financial products as described in any one of claims 1 to 7.

11. A computer program product, characterized in that The computer program product includes a computer program, wherein when the computer program is executed by a processor, the method for recommending pension financial products according to any one of claims 1 to 7 is implemented.

Images