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

Abstract

The invention discloses a payment product recommendation method and device, electronic equipment and a storage medium, and the method comprises the steps: receiving a privacy desensitization data set which is transmitted by a branch mechanism and corresponds to a current payment product; performing user portraying on a to-be-recommended customer group corresponding to the current payment product based on the static feature data to obtain a user group portraying result; predicting a user operation behavior of the to-be-recommended customer group based on the dynamic feature data to obtain a behavior prediction result; and based on the user group portrait result and the behavior prediction result, recommending a matched target payment product to the to-be-recommended customer group. The problem that in the prior art, due to the fact that the user data collection dimension is single, the payment product recommended to the user is not accurate enough is solved, and the effect of recommending the more adaptive payment product to the user based on the multi-dimensional data is achieved by collecting the static feature data and the dynamic feature data corresponding to the to-be-recommended client base.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to a payment product recommendation method and device, electronic equipment and a storage medium. BACKGROUND

[0002] With the continuous development of payment products, the application of payment products is becoming more and more widespread, and users can apply for payment products from financial institutions according to their own needs.

[0003] At present, when recommending payment products to users, the consumption preferences and habits of users are generally determined according to the consumption behaviors of users, and payment products are recommended according to the product usage of users. However, due to the large amount of data of user-related data, user privacy information may be leaked, and only considering the basic characteristics or consumption behavior characteristics of users, the recommended payment products to users have the problem of inaccurate recommendation.

[0004] In order to solve the above problems, it is necessary to improve the recommendation method of payment products. SUMMARY

[0005] The present application provides a payment product recommendation method, device, electronic equipment and storage medium to solve the problem that the recommended payment products for users are not accurate due to the single data collection dimension of users in the prior art.

[0006] In a first aspect, the present application embodiment provides a payment product recommendation method, comprising:

[0007] receiving a privacy desensitization data set corresponding to the current payment product sent by a branch; wherein the privacy desensitization data set includes static feature data and / or dynamic feature data corresponding to user behavior;

[0008] based on the static feature data, performing user portrait on a to-be-recommended customer group corresponding to the current payment product to obtain a user group portrait result;

[0009] based on the dynamic feature data, predicting the user operation behavior of the to-be-recommended customer group to obtain a behavior prediction result;

[0010] based on the user group portrait result and the behavior prediction result, recommending a matching target payment product to the to-be-recommended customer group.

[0011] In a second aspect, the present application embodiment further provides a payment product recommendation device, comprising:

[0012] The data set receiving module is configured to receive a privacy desensitization data set corresponding to the current payment product sent by the branch; wherein the privacy desensitization data set comprises static feature data and / or dynamic feature data corresponding to user behavior;

[0013] The portrait result determining module is configured to perform user portrait on a to-be-recommended customer group corresponding to the current payment product based on the static feature data, and obtain a user group portrait result;

[0014] The prediction result determining module is configured to perform prediction on user operation behavior of the to-be-recommended customer group based on the dynamic feature data, and obtain a behavior prediction result;

[0015] The product recommendation module is configured to recommend a target payment product matched to the to-be-recommended customer group based on the user group portrait result and the behavior prediction result.

[0016] In a third aspect, an electronic device is provided, and the electronic device comprises:

[0017] at least one processor; and

[0018] a memory connected with the at least one processor in communication; wherein

[0019] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the payment product recommendation method described in any of the embodiments of the present application.

[0020] In a fourth aspect, a computer readable storage medium is provided, and the computer readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the payment product recommendation method described in any of the embodiments of the present application when executed by the processor.

[0021] In a fifth aspect, a computer program product is provided, and the computer program product comprises a computer program, and the computer program is used to implement the payment product recommendation method described in any of the embodiments of the present application when executed by a processor.

[0022] The technical scheme of the embodiment of the present application receives the privacy desensitization data set corresponding to the current payment product sent by the branch; performs user portrait on the to-be-recommended customer group corresponding to the current payment product based on static feature data to obtain a user group portrait result; predicts the user operation behavior of the to-be-recommended customer group based on dynamic feature data to obtain a behavior prediction result; and recommends a matching target payment product to the to-be-recommended customer group based on the user group portrait result and the behavior prediction result. In the technical scheme, the user portrait is performed on the static feature data of the to-be-recommended customers in the to-be-recommended customer group to obtain individual user portraits corresponding to each to-be-recommended customer, and all the individual user portraits are clustered, and the user group portrait result corresponding to the to-be-recommended customer group as a whole is determined according to the clustering result. At the same time, the time series prediction is performed on the dynamic feature data of the to-be-recommended customer group to obtain a prediction result, so as to determine the user operation behavior of the to-be-recommended customers in the to-be-recommended customer group on the payment product in the future period according to the prediction result, and analyze the preference of the user for the payment product according to the user operation behavior. On this basis, the most suitable target payment product is recommended for the to-be-recommended customer group in combination with the user group portrait result and the prediction result corresponding to the to-be-recommended customer group. The problem that the payment product recommended for the user is not accurate enough due to the single dimension of user data collection in the prior art is solved, and the effect of recommending a more suitable payment product for the user based on multi-dimensional data is achieved through the collection of static feature data and dynamic feature data corresponding to the to-be-recommended customer group.

[0023] It should be understood that the content described in this part is not intended to identify key or important features of the embodiments of the present application, nor is it used to limit the scope of the present application. Other features of the present application will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creating laborious work.

[0025] Figure 1 is a flowchart of a payment product recommendation method provided by the first embodiment of the present application;

[0026] Figure 2 is a flowchart of a payment product recommendation method provided by the second embodiment of the present application;

[0027] Figure 3 is a flowchart of a payment product recommendation method provided by the second embodiment of the present application;

[0028] Figure 4 is a structural schematic diagram of a payment product recommendation device according to an embodiment of the present application;

[0029] Figure 5 is a structural schematic diagram of an electronic device for implementing a payment product recommendation method of the present application. DETAILED DESCRIPTION

[0030] In order to enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts should fall within the scope of the present application. The acquisition, transmission, storage, use, processing and the like of data in the technical solutions of the present application all comply with the relevant provisions of national laws and regulations. It should be noted that in the embodiments of the present application, some industry existing solutions such as software, components or models may be mentioned, which should be considered as exemplary, and the purpose is only to illustrate the feasibility of the implementation of the technical solutions of the present application, but it does not mean that the applicant has or will necessarily use the solutions.

[0031] 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 that illustrated or described herein.

[0032] Embodiment One

[0033] Figure 1 A flowchart of a payment product recommendation method is provided for the first embodiment of the present application. The present embodiment can be applicable to static feature data and dynamic feature data collection of a current payment product to-be-recommended customer group, and based on the static feature data, a user portrait result corresponding to the to-be-recommended customer group is determined, and based on the dynamic feature data, a user operation behavior of a to-be-recommended customer in the to-be-recommended customer group in a future period is determined, and then based on the user portrait result and the user operation behavior in the prediction result, a payment product suitable for a user in the to-be-recommended customer group is recommended. The method can be executed by a payment product recommendation device, which can be realized in the form of hardware and / or software, and the payment product recommendation device can be configured in a computing device capable of executing the payment product recommendation method.

[0034] As shown in Figure 1 , the method comprises:

[0035] S110, receiving the privacy desensitization dataset corresponding to the current payment product sent by the branch.

[0036] The branch refers to a subordinate business institution established by the business headquarters according to law, which carries out business activities in different places and has a certain independence. The payment product refers to a product issued by a financial institution, which allows users to consume first and pay later, for example, in the present technical solution, the payment product can be a credit card. The data contained in the privacy desensitization dataset is payment product associated data and user associated data after privacy processing. The payment product associated data includes application data, state data, usage information, and partner procurement data corresponding to the payment product. The user associated data can include basic identity information and user payment product usage information. It should be noted that in the present technical solution, the data in the privacy desensitization dataset is divided into static feature data and dynamic feature data, for example, the static feature data mainly contains the basic identity information, and the dynamic feature data mainly refers to the payment product associated data corresponding to the current payment product.

[0037] In actual application, taking a financial business institution as an example, in order to break through the geographical restrictions, reduce the operating costs, close to the market to realize the localization operation, at the same time, disperse the risk and optimize the resource allocation, the financial institution headquarters generally sets up branch institutions in different areas, and the branch institutions directly embed in the market to quickly respond to customer demand.

[0038] In the present technical solution, taking the payment product marketing scene as an example, in order to recommend the most suitable payment product to the customer, it is usually necessary to obtain and analyze the data associated with the customer. In the payment product marketing system, the authorized customer data contains a large amount of personal privacy information. For customers in branch institutions, if the branch institution directly sends the customer's related data to the financial institution headquarters for analysis and processing, there may be a risk of data leakage in the process of data transmission. Therefore, in order to protect the privacy of customers, the branch institution needs to perform privacy desensitization processing on the data before sending the data.

[0039] On this basis, taking the current payment product as an example, after the branch institution obtains the data associated with the current payment product and the customer data associated with the customer, the branch institution obtains the privacy desensitization dataset by performing privacy processing on the obtained data, and sends the privacy desensitization dataset to the financial institution headquarters. When detecting the data transmission prompt of the privacy desensitization dataset sent by the branch institution, the financial institution headquarters receives the privacy desensitization dataset corresponding to the current payment product sent by the branch institution.

[0040] S120, performing user portrait on the user group to be recommended corresponding to the current payment product based on the static feature data, and obtaining a user group portrait result.

[0041] The to-be-recommended customer group refers to a potential target customer group with the highest matching degree with the current payment product based on analysis and processing of static feature data. The user group portrait result refers to a classification result for describing common attributes and behavior patterns of the entire to-be-recommended customer group after feature analysis is performed on at least one to-be-recommended customer in the to-be-recommended customer group.

[0042] Specifically, the types of payment products can be various, and the payment product needs to be recommended in combination with the actual needs of the customer. In order to more accurately recommend the payment product for the to-be-recommended customer corresponding to the current payment product, in the technical solution, user portraits are performed on all to-be-recommended customers in the to-be-recommended customer group based on static feature data, and then a user group portrait result corresponding to the entire to-be-recommended customer group is obtained, so as to recommend the most matched target payment product for the to-be-recommended customer group according to the user group portrait result.

[0043] Optionally, the user portrait is performed on the to-be-recommended customer group corresponding to the current payment product based on the static feature data to obtain the user group portrait result, including: performing the user portrait on the static feature data of the to-be-recommended customers in the to-be-recommended customer group to obtain individual user portrait results; clustering all the individual user portrait results to obtain at least one cluster; and determining the individual user portrait result corresponding to the cluster with the highest proportion as the user group portrait result corresponding to the to-be-recommended customer group.

[0044] The individual user portrait result refers to a classification result corresponding to the to-be-recommended customer determined after analyzing the key features, behavior habits, and demand preferences of the to-be-recommended customer based on the static feature data corresponding to the to-be-recommended customer.

[0045] Exemplarily, when performing the user portrait on the to-be-recommended customer, a deep user portrait model can be constructed by using a generative adversarial network (GAN) and a variational autoencoder (VAE). The generator in the generative adversarial network is responsible for generating simulated static feature data of the authorized to-be-recommended customer, and the discriminator distinguishes between real user data and generated data. Through the adversarial training of the two, a more complex and real user behavior pattern can be learned. On this basis, the variational autoencoder is used to map the multi-dimensional data (i.e., the static feature data in multiple dimensions) of the to-be-recommended customer to a low-dimensional space while retaining the key features of the data, realizing efficient encoding and decoding of user features, so as to perform the user portrait on the to-be-recommended customer based on the deep user portrait model to obtain the individual user portrait result.

[0046] Further, in order to determine the feature data corresponding to the entire recommended customer group, after obtaining the individual user portrait result, the individual user portrait results of all users are clustered based on a clustering analysis algorithm to obtain at least one cluster. It can be understood that the recommended customers in each cluster have similar feature attributes. On this basis, the individual user portrait result corresponding to the cluster with the highest proportion is determined as the user group portrait result corresponding to the recommended customer group.

[0047] In S130, the user operation behavior of the recommended customer group is predicted based on the dynamic feature data to obtain a behavior prediction result.

[0048] In the technical solution, the dynamic feature data refers to the active behavior data of the user for the current payment product. For example, the dynamic feature data can be payment product marketing response data, payment product usage data associated with the current payment product, and cooperation partner procurement data corresponding to the current payment product. The prediction result refers to the prediction output of the behavior characteristics in the future period based on the historical dynamic feature data of the recommended customer group.

[0049] For example, the payment product marketing response data can be the advertisement click data of the user for the payment product advertisement published on the advertisement delivery platform; the payment product usage data can be the function click data based on the payment product application software or the trigger data of the payment product preferential activity; or the procurement data of the cooperation partner to the financial institution or branch for the current payment product.

[0050] Optionally, the user operation behavior of the recommended customer group is predicted based on the dynamic feature data to obtain a behavior prediction result, including: generating an input time sequence feature based on the dynamic feature data, and performing vector conversion on the input time sequence feature to obtain a to-be-used vector; inputting the to-be-used vector into a pre-constructed reasoning model to obtain a behavior prediction result corresponding to the recommended customer group.

[0051] The input time sequence feature refers to the dynamic feature sequence data organized by the time dimension, for example, the advertisement click frequency of the customer for the current payment product on the advertisement delivery platform within 12 consecutive months. The to-be-used vector refers to a multi-dimensional numerical matrix formed after mathematical conversion of the input time sequence feature, for example, a 10-dimensional feature vector generated after conversion of the input time sequence feature.

[0052] In practical applications, in order to break the limitations of single feature extraction, multi-modal learning technology can be used, such as a method based on the combination of convolutional neural network and recurrent neural network to perform behavior prediction on the recommended customer group. For example, for unstructured data such as pictures or videos in advertising data, convolutional neural network can effectively extract visual features, and recurrent neural network can model the time series data of payment product click volume to capture the periodic characteristics of click volume. In terms of function click data, graph neural network is used to regard the function modules in the payment product application software as nodes and the click path as edges to construct a function click graph, thereby mining the potential association features between functions. Specifically, the dynamic feature data is input into the pre-constructed multi-modal learning model to generate the input time series feature, and the input time series feature is vector converted to obtain the to-be-used vector.

[0053] Further, the to-be-used vector is input into the pre-constructed time series prediction model based on the Transformer architecture (i.e., the inference model) to capture the dynamic behavior characteristics of the recommended customers in the recommended customer group in real time, and through the self-attention mechanism, the association between behaviors at different time points is focused on, the next possible behavior of the user is predicted in advance, and the prediction result of the recommended customer group in the future period is obtained. For example, within a month in the future, the advertising click data of the recommended customer group on the advertising platform for the current payment product corresponding to the advertisement.

[0054] The advantage of such a setting is that the inference model based on the self-attention mechanism can automatically learn the importance weight of different source features. For example, when recommending payment products, if there is a large amount of travel-related payment product mall product purchase data in the recent period, the travel feature weight in the product purchase data will increase, and when the travel-related features are fused with the click features of the travel-related payment product advertisements in the advertising click data and the click features of the travel-related functions in the function click data, these travel-related features will dominate in the final feature representation, providing strong support for subsequent precise recommendation of travel theme payment products and preparing for the next promotion process of the payment product.

[0055] S140, based on the user group portrait result and the behavior prediction result, recommending a matching target payment product to the recommended customer group.

[0056] Among them, the target payment product refers to the payment product that is most suitable for the recommended customers in the recommended customer group, etc.

[0057] In practical applications, referring to Figure 3Taking payment product marketing as an example, the payment product marketing process is regarded as a sequential decision problem, and a deep reinforcement learning algorithm such as the proximal policy optimization algorithm (PPO) is used. The user portrait results of the customer group to be recommended and the user operation behavior data of the customer to be recommended are taken as inputs, and a series of marketing actions (such as recommending payment products or providing payment product related activities) are output. Through continuous interaction with the user, the model adjusts its strategy according to the user's feedback to the marketing action (whether to respond, the degree of response, etc.), so as to maximize the long-term cumulative reward (such as the number of users applying for payment products).

[0058] The technical scheme of the embodiment of the application receives the privacy desensitization data set corresponding to the current payment product sent by the branch; performs user portrait on the customer group to be recommended corresponding to the current payment product based on the static feature data, to obtain the user portrait result; predicts the user operation behavior of the customer group to be recommended based on the dynamic feature data, to obtain the behavior prediction result; and recommends the matching target payment product to the customer group to be recommended based on the user portrait result and the behavior prediction result. In the technical scheme, the static feature data of the customer to be recommended in the customer group to be recommended is used to perform user portrait to obtain individual user portraits corresponding to each customer to be recommended, and all individual user portraits are clustered, and the user portrait result corresponding to the whole customer group to be recommended is determined according to the clustering result. At the same time, the time series prediction is performed on the dynamic feature data of the customer group to be recommended to obtain the prediction result, so as to determine the user operation behavior of the customer to be recommended in the customer group to be recommended to the payment product in the future period according to the prediction result, and analyze the preference of the user to the payment product according to the user operation behavior. On this basis, the most suitable target payment product is recommended to the customer group to be recommended in combination with the user portrait result and the prediction result corresponding to the customer group to be recommended. The problem that the payment product recommended for the user is not accurate enough due to the single dimension of user data collection in the prior art is solved, and the effect of recommending a more suitable payment product for the user based on multi-dimensional data is achieved by collecting the static feature data and the dynamic feature data corresponding to the customer group to be recommended.

[0059] Embodiment two

[0060] Figure 2An optional flowchart of a payment product recommendation method provided by Embodiment Two of the present application further includes, before the receiving of the privacy desensitization dataset corresponding to the current payment product sent by the branch, obtaining a to-be-used dataset associated with the current payment product, and performing privacy desensitization processing on the to-be-used data in the to-be-used dataset to obtain a privacy desensitization dataset; the to-be-used dataset includes static feature data and / or dynamic feature data corresponding to the current payment product; and sending the privacy desensitization dataset to the headquarters to enable the headquarters to recommend a matching target payment product to the to-be-recommended customer based on the privacy desensitization dataset.

[0061] As shown in Figure 2 , the method includes:

[0062] S210, obtaining a to-be-used dataset associated with the current payment product, and performing privacy desensitization processing on the to-be-used data in the to-be-used dataset to obtain a privacy desensitization dataset.

[0063] The to-be-used dataset includes original static feature data and dynamic feature data associated with the current payment product and not subjected to privacy processing. The privacy desensitization dataset refers to the to-be-used dataset subjected to privacy processing.

[0064] In one specific example, as shown in Figure 3 , the to-be-used dataset associated with the current payment product is obtained based on the data integration layer, and the to-be-used dataset includes static feature data and dynamic feature data. The static feature data includes user portrait data corresponding to each customer in the to-be-recommended customer group, and the dynamic feature data can include advertisement click data, payment product application software function click data, and external data (i.e., partner procurement data).

[0065] It can be understood that federated learning allows model training without sharing original data. For example, branch offices associated with the headquarters of a business organization can use respective to-be-used datasets on local devices to perform model training, and then upload trained model parameters to a central server of the headquarters of the business organization for aggregation. The advantage of this setup is that the original data of each branch office is always saved locally and does not leak the privacy information of users. Through federated learning, more data can be used for model training to jointly train a more powerful user behavior prediction model (i.e., the inference model mentioned in the present technical solution), without the need to share the original data of users.

[0066] Optionally, the to-be-used data in the to-be-used data set is subjected to privacy desensitization processing to obtain a privacy desensitization data set, including: performing privacy processing on each to-be-used data in the to-be-used data set based on a differential privacy algorithm to obtain privacy desensitization data corresponding to each to-be-used data; and constructing the privacy desensitization data set based on all the privacy desensitization data.

[0067] Specifically, in the data collection and processing stage, after obtaining the to-be-used data set, the differential privacy algorithm can be used to perturb the to-be-used data in the to-be-used data set to obtain privacy desensitization data, and the privacy desensitization data set can be obtained based on all the privacy desensitization data to protect the privacy of the user. For example, when the to-be-used data is authorized user data, a certain amount of noise can be added to the user data, so that others cannot accurately infer the specific information of a certain user from the data, thereby ensuring the security of the data.

[0068] Optionally, in the process of obtaining the to-be-used data set associated with the current payment product, if it is detected that the data flow is greater than the preset data flow, a message queue is generated based on the to-be-used data; at least one to-be-used data temporarily stored in the message queue is read, and each to-be-used data is stored in a corresponding target storage area.

[0069] The target storage area is a dynamic data storage area or a static data storage area, for example, the dynamic data storage area can be a Redis module, and the static data storage area can be a Hive module.

[0070] Based on the above example, continue to refer to Figure 3 In the payment product marketing system, the to-be-used data in the to-be-used data set will continue to be generated, and Kafka can be used as a message queue to collect data from different data sources. For example, an advertising platform sends user advertising click data to a specific topic of Kafka; a financial business system pushes authorized customer data to the corresponding Kafka topic. Since the generation speed and processing speed of the to-be-used data may not be consistent, Kafka can be used as a data buffer layer. When the data generation speed is too fast, Kafka can temporarily store these data to avoid data loss. For example, during a promotion campaign, payment product advertising click data will increase significantly, and Kafka can store these data and wait for Flink to process them.

[0071] On this basis, Flink reads data from each topic of Kafka and performs real-time processing. For example, for advertising click data, Flink can calculate the click rate, conversion rate and other indicators of different advertisements. Flink can perform real-time analysis and decision-making based on the processed data.

[0072] Further, Redis has high-performance read-write capabilities and is suitable for storing real-time feature data. In the payment product marketing system, dynamic feature data obtained after Flink processing, such as advertisement click counts and function use frequencies, can be stored in Redis. When real-time recommendation or decision-making is needed, these feature data can be quickly obtained from Redis. Redis can serve as a cache layer to accelerate data access speed. For example, when performing real-time marketing recommendation, the system can first search for the relevant feature data of a user in Redis, and if found, directly use it, avoiding the delay of obtaining data from other data sources.

[0073] Hive is a Hadoop-based data warehouse tool suitable for storing large-scale static feature data. In the payment product marketing system, static feature data can be stored in the Hive module for long-term data analysis, model training, and trend prediction. In addition, Hive supports SQL queries, which can facilitate data analysis and mining. For example, Hive queries can be used to analyze payment product advertisement click behavior, function usage, and consumption trends in different time periods to provide data support for strategy formulation of the payment product marketing system.

[0074] S220, sending the privacy desensitization dataset to the headquarters institution, so that the headquarters institution recommends a matching target payment product to the to-be-recommended customer based on the privacy desensitization dataset.

[0075] S230, receiving the privacy desensitization dataset corresponding to the current payment product sent by the branch institution.

[0076] S240, performing user profiling on the to-be-recommended customer group corresponding to the current payment product based on the static feature data to obtain a user group profiling result.

[0077] S250, predicting user operation behavior of the to-be-recommended customer group based on the dynamic feature data to obtain a behavior prediction result.

[0078] S260, recommending a matching target payment product to the to-be-recommended customer group based on the user group profiling result and the behavior prediction result.

[0079] Optionally, after recommending a matching target payment product to the to-be-recommended customer group based on the user group profiling result and the behavior prediction result, the method further includes: real-time obtaining response information of the to-be-recommended customer in the to-be-recommended customer group to the target payment product, and updating the to-be-used dataset based on the response information; repeating the steps of performing privacy desensitization processing on the to-be-used data in the to-be-used dataset to obtain a privacy desensitization dataset, and sending the privacy desensitization dataset to the headquarters institution, so that the headquarters institution recommends a matching target payment product to the to-be-recommended customer based on the privacy desensitization dataset.

[0080] Specifically, after recommending the target payment product to the to-be-recommended customer group, in order to timely recommend a more matched payment product to the to-be-recommended customer according to the demand of the to-be-recommended customer, the response information of the to-be-recommended customer to the target payment product can be acquired in real time. For example, the application data of the customer in the to-be-recommended customer group to the target payment product, the advertisement click data, the click data of the related activities on the payment product application software to the target payment product, and the like are acquired, so as to update the to-be-used data set based on the response information of all the to-be-recommended customers, and in the next payment product recommendation process, the to-be-recommended customer group is recommended a more matched payment product based on the updated to-be-used data set.

[0081] On the basis of the above examples, continuing to refer to Figure 3 , a real-time model evaluation system is established, a dynamic time warping algorithm is used to measure the similarity between the prediction result of the to-be-recommended customer and the actual behavior, and the response information of the to-be-recommended customer to the target payment product is calculated in real time. At the same time, an online A / B test framework is introduced, which can compare the effects of different model versions or different marketing strategies in real time without interrupting the business. On this basis, the meta-learning technology can also be used to optimize the inference model, so that the inference model can quickly adapt to new payment product customer groups and payment product business scenarios, so that when facing new market trends or user behavior patterns change greatly, the meta-learning model can guide the model layer of the payment product marketing system to quickly adjust the model parameters and structure, and realize rapid optimization. For example, when a new consumer hotspot appears in the market or a competitor launches a new payment product, the system can adaptively optimize the target payment product recommended to the to-be-recommended customer group in a short time, and maintain a competitive advantage.

[0082] The technical scheme of the embodiment of the application acquires the to-be-used data set associated with the current payment product

[0083] The to-be-used data set includes static feature data and / or dynamic feature data corresponding to the current payment product. The privacy desensitization data set is sent to the headquarters institution, so that the headquarters institution recommends a matched target payment product to the to-be-recommended customer based on the privacy desensitization data set. In this technical scheme, after each branch institution collects the to-be-used data associated with the user, in order to protect the user's privacy data from being leaked, the differential privacy algorithm is used to perform privacy processing on the to-be-used data to obtain a privacy desensitization data set, and then the privacy desensitization data set is sent to the headquarters institution, so as to ensure the data security in the data transmission process. The problem of data leakage caused by the fact that the security measures are not in place when the user data is processed in the prior art is solved, and the effect of protecting the user data security is achieved.

[0084] Embodiment three

[0085] Figure 4 A structural schematic diagram of a payment product recommendation device provided for embodiment three of the present application is shown in FIG. 3. As shown in the figure, the device comprises a data set receiving module 310, a portrait result determining module 320, a prediction result determining module 330, and a product recommendation module 340. Figure 4

[0086] The data set receiving module 310 is configured to receive a privacy de-sensitization data set corresponding to a current payment product sent by a branch office, wherein the privacy de-sensitization data set comprises static feature data and / or dynamic feature data corresponding to user behavior.

[0087] The portrait result determining module 320 is configured to perform user portrait on a to-be-recommended customer group corresponding to the current payment product based on the static feature data, to obtain a user group portrait result.

[0088] The prediction result determining module 330 is configured to perform prediction on user operation behavior of the to-be-recommended customer group based on the dynamic feature data, to obtain a behavior prediction result.

[0089] The product recommendation module 340 is configured to recommend a matching target payment product to the to-be-recommended customer group based on the user group portrait result and the behavior prediction result.

[0090] ​The technical scheme of the embodiment of the present application receives the privacy desensitization data set corresponding to the current payment product sent by the branch; performs user portrait on the to-be-recommended customer group corresponding to the current payment product based on the static feature data, to obtain a user group portrait result; predicts the user operation behavior of the to-be-recommended customer group based on the dynamic feature data, to obtain a behavior prediction result; and recommends a matching target payment product to the to-be-recommended customer group based on the user group portrait result and the behavior prediction result. In the technical scheme, the user portrait is performed on the static feature data of the to-be-recommended customers in the to-be-recommended customer group, to obtain individual user portraits corresponding to each to-be-recommended customer, and all the individual user portraits are clustered, and the user group portrait result corresponding to the to-be-recommended customer group as a whole is determined according to the clustering result. At the same time, the time series prediction is performed on the dynamic feature data of the to-be-recommended customer group, to obtain a prediction result, so as to determine the user operation behavior of the to-be-recommended customers in the to-be-recommended customer group on the payment product in the future period according to the prediction result, and analyze the preference of the users for the payment product according to the user operation behavior. On this basis, the most suitable target payment product is recommended to the to-be-recommended customer group in combination with the user group portrait result and the prediction result corresponding to the to-be-recommended customer group. The problem that the payment product recommended for the user is not accurate enough due to the single dimension of user data collection in the prior art is solved, and the effect of recommending a more suitable payment product for the user based on multi-dimensional data is achieved by collecting the static feature data and the dynamic feature data corresponding to the to-be-recommended customer group.

[0091] Optionally, the payment product recommendation apparatus further comprises a desensitization module configured to acquire a to-be-used data set associated with the current payment product, and perform privacy desensitization processing on to-be-used data in the to-be-used data set to obtain a privacy desensitization data set; the to-be-used data set comprises static feature data and / or dynamic feature data corresponding to the current payment product;

[0092] The data set sending module is configured to send the privacy desensitization data set to the headquarters, so that the headquarters recommends a matching target payment product to the to-be-recommended customer based on the privacy desensitization data set.

[0093] Optionally, the payment product recommendation apparatus further comprises a message queue generation module configured to, in the process of acquiring the to-be-used data set associated with the current payment product, generate a message queue based on the to-be-used data if it is detected that the data flow is greater than a preset data flow.

[0094] The data storage module is configured to read at least one piece of to-be-used data temporarily stored in the message queue, and store each piece of to-be-used data in a corresponding target storage area; the target storage area is a dynamic data storage area or a static data storage area.

[0095] Optionally, the payment product recommendation device further comprises a privacy processing module configured to perform privacy processing on each of the to-be-used data in the to-be-used data set based on a differential privacy algorithm to obtain privacy desensitization data corresponding to each of the to-be-used data.

[0096] The data set generation module is configured to generate a privacy desensitization data set based on all the privacy desensitization data.

[0097] Optionally, the portrait result determination module comprises an individual user portrait result determination unit configured to perform user portrait on the static feature data of the to-be-recommended customers in the to-be-recommended customer group to obtain individual user portrait results.

[0098] The clustering unit is configured to cluster all the individual user portrait results to obtain at least one cluster.

[0099] The user group portrait result determination unit is configured to determine the individual user portrait result corresponding to the cluster with the highest proportion as the user group portrait result corresponding to the to-be-recommended customer group.

[0100] Optionally, the prediction result determination module comprises a vector conversion unit configured to generate to-be-input time series features based on the dynamic feature data and perform vector conversion on the to-be-input time series features to obtain to-be-used vectors.

[0101] The prediction result determination unit is configured to input the to-be-used vectors into a pre-constructed inference model to obtain behavior prediction results corresponding to the to-be-recommended customer group.

[0102] Optionally, the payment product recommendation device further comprises a response information acquisition module configured to acquire response information of the to-be-recommended customers in the to-be-recommended customer group to the target payment product in real time after recommending the target payment product matching the user group portrait result and the behavior prediction result to the to-be-recommended customer group, and update the to-be-used data set based on the response information.

[0103] The repeating module is configured to repeatedly perform the steps of performing privacy desensitization processing on the to-be-used data in the to-be-used data set to obtain a privacy desensitization data set, and sending the privacy desensitization data set to the headquarters institution, so that the headquarters institution recommends the target payment product matching the privacy desensitization data set to the to-be-recommended customers.

[0104] The payment product recommendation device provided in the embodiments of the present application can perform the payment product recommendation method provided in any of the embodiments of the present application, and has the corresponding function modules and beneficial effects of the execution method.

[0105] Embodiment Four

[0106] Figure 5A schematic diagram of the structure of an electronic device 10 according to an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0107] like Figure 5 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 may also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0108] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0109] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as payment product recommendation methods.

[0110] In some embodiments, the payment product recommendation method can be implemented as a computer program tangibly embodied in a computer readable storage medium, e.g., storage unit 18. In some embodiments, parts or all of the computer program can be loaded and / or installed onto electronic device 10 via, e.g., ROM 12 and / or communication unit 19. When the computer program is loaded onto RAM 13 and executed by processor 11, one or more steps of the payment product recommendation method described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to perform the payment product recommendation method by other means, e.g., with the aid of firmware.

[0111] Various implementations of the systems and techniques described above can be realized in digital electronic circuitry, integrated circuitry, a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), a system on a chip (SOC), a complex programmable logic device (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0112] Computer programs used to implement the payment product recommendation method of the present application can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the computer program running on the processor implements the functions / operations specified in the flowcharts and / or the block diagrams. The computer program can be executed entirely on a machine, partially on a machine, partially on a machine as part of a standalone software package, or entirely on a remote machine or server.

[0113] In the context of the present application, a computer-readable storage medium can be a tangible medium that can contain or store computer programs for use by or in connection with an instruction execution system, apparatus, or device. Computer-readable storage media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. More specific examples of a machine-readable storage medium will include one or more lines of a program of instructions in a transitory signal, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0114] To provide for interaction with a user, the systems and techniques described here can be implemented on an electronic device having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the electronic device. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0115] The systems and techniques described here can be implemented in a computing system that includes a back end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front end component (e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0116] The computing system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. Servers can be cloud servers, also known as cloud computing servers or cloud hosts, which are a host product in the cloud computing service system to solve the defects of great management difficulty and weak business scalability in traditional physical hosts and VPS services.

[0117] Embodiment five

[0118] The embodiment of the application further provides a computer program product comprising a computer program which, when executed by a processor, implements the payment product recommendation method provided in any embodiment of the application.

[0119] The computer program product, in implementation, can be written in one or more programming languages or combinations thereof to implement computer program codes for performing operations of the application, including object-oriented programming languages such as Java, Smalltalk, C++, and conventional procedural programming languages such as C language or similar programming languages. The program codes can be executed entirely on a user computer, partially on a user computer, as an independent software package, partially on a user computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user computer through any kind of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, through the Internet using an Internet service provider).

[0120] It should be understood that the steps shown above can be reordered, added or deleted. For example, the steps described in the application can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions of the application can be achieved, which are not limited herein.

[0121] The above detailed description does not constitute a limitation on the scope of protection of the application. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the application shall be included in the scope of protection of the application.

Claims

1. A payment product recommendation method characterized by comprising: The method comprises the following steps: receiving a privacy de-identification data set corresponding to a current payment product sent by a branch; wherein the privacy de-identification data set comprises static feature data and / or dynamic feature data corresponding to user behavior; conducting user profiling on a to-be-recommended customer group corresponding to the current payment product based on the static feature data, to obtain a user group profiling result; predicting user operation behavior of the to-be-recommended customer group based on the dynamic feature data, to obtain a behavior prediction result; based on the user group profiling result and the behavior prediction result, recommending a matching target payment product to the to-be-recommended customer group.

2. The method of claim 1, wherein, Before the step of receiving a privacy de-identification data set corresponding to a current payment product sent by a branch, the method further comprises the following steps: obtaining a to-be-used data set associated with the current payment product, and conducting privacy de-identification processing on to-be-used data in the to-be-used data set, to obtain a privacy de-identification data set; the to-be-used data set comprises static feature data and / or dynamic feature data corresponding to the current payment product; sending the privacy de-identification data set to a headquarters, so that the headquarters recommends a matching target payment product to a to-be-recommended customer based on the privacy de-identification data set.

3. The method of claim 2, wherein, The method further comprises the following steps: during the process of obtaining a to-be-used data set associated with the current payment product, if it is detected that data flow is greater than a preset data flow, generating a message queue based on to-be-used data; reading at least one piece of to-be-used data temporarily stored in the message queue, and storing each piece of to-be-used data to a corresponding target storage area; wherein the target storage area is a dynamic data storage area or a static data storage area.

4. The method of claim 2, wherein, The step of conducting privacy de-identification processing on to-be-used data in the to-be-used data set, to obtain a privacy de-identification data set, comprises the following steps: conducting privacy processing on each piece of to-be-used data in the to-be-used data set based on a differential privacy algorithm, to obtain privacy de-identification data corresponding to each piece of to-be-used data; based on all privacy de-identification data, forming a privacy de-identification data set.

5. The method of claim 1, wherein, The step of conducting user profiling on a to-be-recommended customer group corresponding to the current payment product based on the static feature data, to obtain a user group profiling result, comprises the following steps: conducting user profiling on static feature data of a to-be-recommended customer in the to-be-recommended customer group, to obtain an individual user profiling result; clustering all individual user profiling results, to obtain at least one cluster; determining an individual user profiling result corresponding to a cluster with the highest proportion as a user group profiling result corresponding to the to-be-recommended customer group.

6. The method of claim 1, wherein, The step of predicting user operation behavior of the to-be-recommended customer group based on the dynamic feature data, to obtain a behavior prediction result, comprises the following steps: generating a to-be-input time series feature based on the dynamic feature data, and conducting vector conversion on the to-be-input time series feature, to obtain a to-be-used vector; inputting the to-be-used vector into a pre-constructed inference model, to obtain a behavior prediction result corresponding to the to-be-recommended customer group.

7. The method of claim 2, wherein, After the step of recommending a matching target payment product to the to-be-recommended customer group based on the user group profiling result and the behavior prediction result, the method further comprises the following steps: obtaining response information of the to-be-recommended customers in the to-be-recommended customer group to the target payment product in real time, and updating the to-be-used data set based on the response information; repeating the steps of performing privacy desensitization processing on the to-be-used data in the to-be-used data set to obtain a privacy desensitized data set, and sending the privacy desensitized data set to the headquarters institution, so that the headquarters institution recommends a matching target payment product to the to-be-recommended customers based on the privacy desensitized data set.

8. A payment product recommendation apparatus characterized by comprising: Comprise: a data set receiving module configured to receive a privacy desensitized data set corresponding to a current payment product sent by a branch institution; wherein the privacy desensitized data set comprises static feature data and / or dynamic feature data corresponding to user behavior; a portrait result determining module configured to perform user portrait on a to-be-recommended customer group corresponding to the current payment product based on the static feature data, and obtain a user group portrait result; a prediction result determining module configured to predict user operation behavior of the to-be-recommended customer group based on the dynamic feature data, and obtain a behavior prediction result; a product recommendation module configured to recommend a matching target payment product to the to-be-recommended customer group based on the user group portrait result and the behavior prediction result.

9. An electronic device, comprising: The electronic device comprises: at least one processor; and a memory connected in communication with the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to perform the payment product recommendation method of any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer instructions, and the computer instructions are used to enable the processor to perform the payment product recommendation method of any one of claims 1-7 when executed.

Images