Generation method of rights and interests distribution prediction model and rights and interests distribution method

Abstract

The invention discloses a right distribution prediction model generation method and a right distribution method. The method comprises the steps of performing standardization processing on obtained original right distribution data to obtain target right distribution data, the target right distribution data comprising right activity information and right receiving information corresponding to at least one user; analyzing the target right distribution data, determining a preset number of right types, and generating a training sample set based on each right type and the target right distribution data; and performing federated learning training on a to-be-trained right distribution prediction model according to the training sample set in combination with training sample sets of other edge nodes to obtain a target right distribution prediction model, the target right distribution prediction model being used for predicting feedback of different users for different right types. The problems of non-uniform data formats and privacy leakage in the right distribution process are solved; the data consistency is ensured, and the accuracy of data analysis is improved; and privacy data leakage is avoided.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a method for generating a rights distribution prediction model and a rights distribution method. Background Art

[0002] In the online marketing business of banks, equity distribution is a key marketing tool, and the commonly used solution is mainly centralized data processing. This method usually involves aggregating the data of each branch to the central server of the head office for processing and analysis. It includes the following key steps: Data collection: Each branch collects data related to equity distribution and marketing. Data upload: Upload the collected data to the central server of the bank head office. However, centralized data processing usually has the following problems: The data format is not unified, and there may be differences in data recording standards and formats between branches, which makes data integration and analysis more difficult. Centralized storage and processing of data may increase the risk of leakage, involving the privacy of customer data and the security of institutional data. Summary of the invention

[0003] The present invention provides a method for generating a rights distribution prediction model and a rights distribution method to solve the problems of inconsistent data formats and privacy leakage in the rights distribution process.

[0004] According to one aspect of the present invention, a method for generating a rights distribution prediction model is provided, which is applied to an edge node and includes:

[0005] Standardize the acquired original equity distribution data to obtain target equity distribution data, where the target equity distribution data includes equity activity information and equity collection information corresponding to at least one user;

[0006] Analyze the target equity distribution data, determine a preset number of equity types, and generate a training sample set based on each of the equity types and the target equity distribution data;

[0007] The equity distribution prediction model to be trained is federated learning trained according to the training sample set combined with the training sample sets of other edge nodes to obtain a target equity distribution prediction model, and the target equity distribution prediction model is used to predict the feedback of different users on different equity types.

[0008] According to another aspect of the present invention, a method for allocating equity is provided, which is applied to an edge node and includes:

[0009] Obtaining information on rights to be distributed, rights type, users to be predicted, and rights collection information corresponding to the users to be predicted;

[0010] Inputting the equity information to be distributed, the equity type and the equity collection information corresponding to the user to be predicted into the target equity distribution prediction model to perform equity distribution prediction, wherein the target equity distribution prediction model is determined by the method for generating the equity distribution prediction model as described in any embodiment of the present invention;

[0011] Determine the equity prediction result of the to-be-predicted user according to the output of the target equity distribution prediction model;

[0012] The equity distribution strategy corresponding to the user to be predicted is determined according to the equity prediction result, so as to distribute equity to the user to be predicted according to the equity distribution strategy.

[0013] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0014] at least one processor, and a memory communicatively coupled to the at least one processor;

[0015] In which, the memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the method for generating a rights distribution prediction model or the rights distribution method described in any embodiment of the present invention.

[0016] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the method for generating a rights distribution prediction model or the rights distribution method described in any embodiment of the present invention when executed.

[0017] According to another aspect of the present invention, a computer program product is provided, which includes a computer program, and when the computer program is executed by a processor, it implements the method for generating a rights distribution prediction model or the rights distribution method described in any embodiment of the present invention.

[0018] The technical solution of the embodiment of the present invention obtains target equity distribution data by standardizing the acquired original equity distribution data, wherein the target equity distribution data includes equity activity information and equity collection information corresponding to at least one user; the target equity distribution data is analyzed to determine a preset number of equity types, and a training sample set is generated based on each of the equity types and the target equity distribution data; a to-be-trained equity distribution prediction model is federatedly trained according to the training sample set combined with the training sample sets of other edge nodes to obtain a target equity distribution prediction model, wherein the target equity distribution prediction model is used to predict the feedback of different users on different equity types, thereby solving the problems of inconsistent data formats and privacy leakage in the equity distribution process; the edge nodes In the process of data processing, the edge nodes standardize the original equity distribution data to obtain the target equity distribution data, ensure data consistency, and improve the accuracy of data analysis; the equity distribution prediction model is trained through the edge nodes, and the edge nodes combine the training samples of other edge nodes for federated learning training to generate the target equity distribution prediction model, which can conduct in-depth analysis of the data and accurately understand the user's feedback on different rights and interests, so as to determine the marketing strategy according to the user's feedback on different rights and interests and improve the accuracy of the marketing strategy; the edge nodes can process data locally in real time through the target equity distribution prediction model to improve the timeliness of the data, and the federated learning training model does not need to share the original data, thereby enhancing data privacy protection and avoiding privacy data leakage.

[0019] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 is a flowchart of a method for generating a rights distribution prediction model according to Embodiment 1 of the present invention;

[0022] Figure 2 is a flow chart of a method for generating a rights distribution prediction model according to Embodiment 2 of the present invention;

[0023] Figure 3 is a flow chart of a method for distributing rights and interests provided according to Embodiment 3 of the present invention;

[0024] Figure 4 is a structural schematic diagram of a device for generating a rights distribution prediction model provided according to a fourth embodiment of the present invention;

[0025] Figure 5 This is a schematic diagram of the structure of a rights distribution device provided according to Embodiment 5 of the present invention.

[0026] Figure 6 It is a structural schematic diagram of an electronic device provided according to Embodiment 6 of the present invention. DETAILED DESCRIPTION

[0027] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0028] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0029] Embodiment 1

[0030] Figure 1 This is a flow chart of a method for generating a rights distribution prediction model provided in the first embodiment of the present invention. This embodiment is applicable to the case of accurately generating a rights distribution prediction model. This method can be executed by an edge node. Figure 1 As shown, the method includes:

[0031] S101. Standardize the acquired original equity distribution data to obtain target equity distribution data, where the target equity distribution data includes equity activity information and equity collection information corresponding to at least one user.

[0032] In this embodiment, the original rights and interests distribution data can be understood as the original data in the rights and interests data distribution process; the target rights and interests distribution data can be understood as the rights and interests distribution data in a standard format. Both the original rights and interests distribution data and the target rights and interests distribution data can be used to represent the relevant information of the rights and interests distribution, such as the rights and interests activities distributed, the records of receipt, etc. The rights and interests activities can be payment of instant discounts, red beans, points, third-party coupons, etc. Rights and interests information can be understood as information that describes the rights and interests associated with the user. The target rights and interests distribution data includes at least one user and the rights and interests information corresponding to the user. Each user's consumption, receipt and other related information of the rights and interests are described by the rights and interests information corresponding to this user. The rights and interests information can also include other information of the user, such as the user's gender, age, consumption habits and other information, which is used to assist in the prediction of user rights and interests.

[0033] Collect and store the equity distribution data generated by the equity distribution process of different users' banks in advance. For example, the data generated by users during consumption or participation in activities will be saved in logs or corresponding storage spaces. When model training is required, obtain the equity distribution data from logs or other storage spaces and use it as the original equity distribution data. Set data processing rules or specifications in advance, standardize the original equity distribution data according to the pre-set data processing rules or specifications, process the original equity distribution data into a unified format, and obtain the target equity distribution data.

[0034] S102: Analyze the target equity distribution data, determine a preset number of equity types, and generate a training sample set based on each equity type and the target equity distribution data.

[0035] In this embodiment, the rights and interests type can be used to classify different rights and interests, and the rights and interests type can be high-frequency claiming users, points priority users, coupon priority users, etc. The preset number can be set according to the type of activity. The rights and interests activity information can be understood as information describing the published rights and interests activities, for example, the bank publishes payment instant discount, points, coupons and other activities.

[0036] Determine a preset number in advance, analyze the target equity distribution data, divide the target equity distribution data into a preset number of equity types, or divide the equity collection information in the target equity distribution data into a preset number of sets according to the degree of similarity of the data, and then determine the equity type based on the data in the set.

[0037] For each user, the rights collection information of this user is filtered out from the target rights distribution data, and the rights collection information, rights activity information and each rights type of this user are used as a training sample. Corresponding training samples are generated according to the data generated by different users in different activities, and a training sample set is formed based on all the obtained training samples.

[0038] S103. Federated learning training is performed on the equity distribution prediction model to be trained according to the training sample set combined with the training sample sets of other edge nodes to obtain a target equity distribution prediction model, which is used to predict the feedback of different users on different equity types.

[0039] In this embodiment, the equity distribution prediction model to be trained can be understood as a model with training requirements. This model is used to predict equity distribution, for example, to predict which type of equity the user is interested in. For example, the equity distribution prediction model to be trained can be a decision tree model. The target equity distribution prediction model can be understood as a trained network model, which can predict the user's feedback on different equity types based on the input data. The user's feedback on the equity type can be interested / uninterested, the degree of interest, etc.

[0040] Pre-construct the equity distribution prediction model to be trained, and perform federated learning training based on the local training sample set combined with the training sample set of other edge nodes. For example, perform model training based on some training samples in the local training sample set. When the reporting conditions are met, report the model parameters to the central server. The central server performs comprehensive processing on the received model parameters and sends new model parameters to the edge nodes. The edge nodes continue to train according to the model parameters sent by the central server, and continue to report when the reporting conditions are met. After the cutoff conditions are met, the model training is terminated, and the model parameters at this time are used as the final parameters of the model to obtain the target equity distribution prediction model. Among them, the reporting condition can be that the number of iterations exceeds the number threshold, or the accuracy exceeds the first preset value, or the accuracy improvement value exceeds the second preset value, or the training time of this iteration exceeds the first preset duration, etc.; the cutoff condition can be that the number of reports exceeds the third preset value, the total training duration exceeds the second preset duration, the accuracy of the model exceeds the fourth preset value, etc. The trained target equity distribution prediction model can be used to predict the feedback of different users on different equity types.

[0041] The method provided in the embodiment of the present application is executed by the edge node, and the edge node can be deployed in the branch to perform corresponding processing on the equity distribution data, ensuring that the node has sufficient computing and storage resources to process the data generated by the local equity distribution activities.

[0042] The embodiment of the present invention provides a method for generating a rights distribution prediction model, which solves the problems of inconsistent data formats and privacy leakage in the rights distribution process; in the process of data processing, the edge node standardizes the original rights distribution data to obtain target rights distribution data, ensures data consistency, and improves the accuracy of data analysis; the rights distribution prediction model is trained by the edge node, and the edge node combines the training samples of other edge nodes to perform federated learning training to generate a target rights distribution prediction model, which can conduct in-depth analysis of the data and accurately understand the user's feedback on different rights, so as to determine the marketing strategy according to the user's feedback on different rights and interests, and improve the accuracy of the marketing strategy; the edge node can process data locally in real time through the target rights distribution prediction model to improve the timeliness of the data, and the federated learning training model does not need to share the original data, thereby enhancing data privacy protection.

[0043] Embodiment 2

[0044] Figure 2 This is a flow chart of a method for generating a rights distribution prediction model provided in the second embodiment of the present invention. This embodiment is refined on the basis of the above embodiment. Figure 2 As shown, the method includes:

[0045] S201. Clean the acquired original rights and interests distribution data to obtain valid rights and interests distribution data.

[0046] In this embodiment, valid equity distribution data can be understood as the equity distribution data retained after invalid data is removed. Data cleaning rules are set in advance, and the original equity distribution data obtained is analyzed. According to the data cleaning rules, it is determined whether there are errors or incomplete records. Abnormal data such as errors and losses are removed through data cleaning, and the retained data is the valid equity distribution data.

[0047] S202: Format the effective equity distribution data to obtain target equity distribution data, where the target equity distribution data includes equity activity information and equity collection information corresponding to at least one user.

[0048] The standard format of the data is pre-set, and different types of data correspond to different standard formats. The effective rights distribution data is formatted according to the standard format, and the different types of data in the effective rights distribution data are processed into the standard format to obtain the target rights distribution data. During the formatting process, a data conversion algorithm can be used to convert the data. For example, all time and date formats are unified, illustratively, unified into YYYY-MM-DD; the number of collections and points are numerically standardized. The embodiment of the present application formats the data and can process the data into a format that can be processed by the machine learning model, which is convenient for subsequent model training.

[0049] Optionally, the rights and interests collection information includes at least one of the following: rights and interests collection records, user feedback, and rights and interests redemption information.

[0050] In this embodiment, the rights and interests collection record can be understood as information that records the user's collection of different rights and interests. It can be information such as the time and number of times the user collects the rights and interests. User feedback can be the user's response to different rights and interests, such as whether they are interested. Rights and interests exchange information can be information that describes the user's exchange behavior for rights and interests, such as the exchange time, exchange number, etc.

[0051] S203: Initialize a preset number of cluster centers.

[0052] The cluster centers may be initialized by randomly selecting a preset number of users as the initial cluster centers, or by selecting a specific type of users as the initial cluster centers, etc. Initializing a preset number of cluster centers represents different user groups.

[0053] S204: Clustering the rights and interests collection information corresponding to the users in the target rights and interests distribution data based on the centers of each cluster to obtain a preset number of user clusters.

[0054] In this embodiment, a user cluster can be understood as a set of users. The rights collection information corresponding to the user in the target rights distribution data is repeatedly distributed to the nearest cluster center, and the cluster center is updated until convergence, to obtain a preset number of user clusters, each of which includes different users.

[0055] The clustering in the embodiment of the present application can be implemented using the K-means clustering algorithm. The basic customer group labels can be pre-set and available. Clustering through the K-means clustering algorithm can be applied to different types of data and has higher versatility. There are many ways to initialize the cluster center, such as random selection; if there is some prior knowledge or specific needs, other initialization methods can also be used, such as sampling based on certain features, such as gender, province, city, asset level, etc.

[0056] S205: Determine the type of rights and interests corresponding to the user cluster according to the rights and interests collection information of the users included in the user cluster.

[0057] Determine all users included in each user cluster, determine the rights and interests collection information corresponding to the users, analyze and process the rights and interests collection information of each user, determine the rights and interests type corresponding to this user cluster, and a preset number of user clusters can obtain a preset number of rights and interests types.

[0058] Assume that a branch of a bank carries out multiple marketing activities in the first quarter, distributing instant discounts on WeChat, red bean points, Starbucks coupons, etc. The following is a specific example of how edge nodes process and analyze data from these activities:

[0059] 1. Data collection and preprocessing

[0060] Example of raw staking data:

[0061] WeChat instant discount collection: User A collected it on January 5th, and collected and redeemed it 3 times.

[0062] Red bean points collection: User B collected and redeemed 200 red beans on January 10.

[0063] Starbucks coupon collection: User C collected and used Starbucks coupons from January 5th to 15th, 3 times.

[0064] Preprocessing process:

[0065] Clean the data, remove any erroneous or incomplete records, unify all time and date formats (for example, unify to YYYY-MM-DD), and standardize the numbers of collections and points.

[0066] 2. Application of K-means clustering algorithm:

[0067] 1) Application scenarios:

[0068] Use the K-means clustering algorithm to divide users into groups based on their rights collection behavior. For example, divide users into high-frequency users, points-priority users, coupon-priority users, etc.

[0069] 2) Example of clustering results:

[0070] User A is classified as a "high-frequency recipient user".

[0071] User B is classified as a "points priority user".

[0072] User C is classified as a "coupon priority user".

[0073] The types of benefits include: high-frequency claiming users, points priority users and coupon priority users.

[0074] S206: Generate a training sample set based on each equity type and target equity distribution data.

[0075] S207: Train the equity distribution prediction model to be trained according to the training sample set to obtain candidate model parameters that meet the training cutoff requirements.

[0076] In this embodiment, the equity distribution prediction model to be trained can be understood as a pre-built model for predicting equity distribution, such as a logistic regression model or a decision tree model. The training cutoff requirement can be understood as a pre-set condition for determining whether to interrupt model training; the training cutoff requirement can be pre-set, for example, the model accuracy exceeds the accuracy threshold, the number of training samples used in training exceeds the number threshold, the training time exceeds the time threshold, etc. The candidate model parameters can be understood as the model parameters obtained by edge nodes through model training, which can be weights, biases, etc.

[0077] The training sample set includes one or more training samples. In order to meet the training accuracy requirements, the number of training samples is usually large. The embodiment of the present application can set the number of training samples in advance according to the accuracy requirements, etc., collect training samples according to the number of training samples, and form a training sample set. Pre-construct the equity distribution prediction model to be trained, and input the training samples in the training sample set into the equity distribution prediction model to be trained in turn for training to predict the reactions of different users to future equity. The model learns to identify which type of equity is more likely to attract a specific type of user. During the training process, the model parameters are continuously adjusted to detect whether the training cutoff requirements are met; if not, continue training; when it is detected that the training cutoff requirements are met, the model parameters at this time are used as candidate model parameters.

[0078] When the equity distribution prediction model to be trained is a decision tree model, the best segmentation attribute is selected based on the information gain or Gini impurity standard, and the tree is recursively constructed until the predetermined depth is reached or the amount of data contained in the node is less than the threshold. The decision tree is pruned to prevent overfitting and improve the generalization ability of the model.

[0079] The input data of the equity distribution prediction model to be trained are equity type and target equity distribution data, which are flow or log information related to users and equity, including user equity collection information and equity activity information; among which, equity collection information can be at least one of equity collection records, user feedback and equity exchange information, and can also include basic characteristics of users, such as age, gender, consumption habits and one or more information; equity activity information can be one or more information such as equity type, equity strength, activity time, activity rules, activity stage, etc., and the activity stage includes: one or more of activity participation, equity collection, equity exchange, and equity write-off. The above information can be used to identify "who has what specific actions on equity at what time, place, and activity".

[0080] The output of the training benefit distribution prediction model is to predict the reaction of different users to different benefit types and predict which type of benefit is more likely to attract a specific type of user. The output is a prediction result about the benefit type. For example, for a certain user, the model may output that the user is more inclined to a certain type of benefit (such as WeChat instant discount, payment instant discount, etc.). It is "user-benefit type" rather than "user-activity type". Benefits are part of an activity, so it is more detailed to guide business personnel on how to distribute specific benefits to certain users to improve the effectiveness of the activity.

[0081] Application example: For an upcoming marketing campaign, the target benefit distribution prediction model predicts that user D (who belongs to the "points priority user" group) may be more interested in activities that increase points. Based on this, the branch provides user D with more points-related benefits. In this way, edge computing nodes can analyze benefit distribution data in real time, help understand and meet customer needs more effectively, and improve the relevance and effectiveness of marketing activities.

[0082] S208. Send the candidate model parameters to the central server so that the central server can determine the global model parameters based on the received edge nodes and the candidate model parameters sent by other edge nodes. The candidate model parameters sent by other edge nodes are obtained by model training based on their corresponding training sample sets.

[0083] In this embodiment, the global model parameters can be understood as model parameters applicable to all edge nodes. The edge node sends the candidate model parameters to the central server; other edge nodes also perform model training based on the training sample set formed by local data, and send the candidate model parameters to the central server. The central server determines the global model parameters based on the candidate model parameters received from this edge node and the candidate model parameters sent by other edge nodes, that is, the central server analyzes and determines the global model parameters based on all the received candidate model parameters, for example, the average, maximum value, minimum value of all candidate model parameters, or the optimization processing using complex algorithms, etc., to obtain the global model parameters. The updated global model parameters reflect the data characteristics of all branches and provide a global perspective. The central server sends the global model parameters to each edge node to guide the local marketing decisions of the edge nodes.

[0084] Optionally, the global model parameters are obtained by performing a weighted operation on the parameters of each candidate model.

[0085] S209: Receive the global model parameters fed back by the central server, and generate a target equity distribution prediction model based on the global model parameters.

[0086] The edge node receives the global model parameters fed back by the central server, uses the global model parameters as the model parameters, and generates a target equity distribution prediction model. The target equity distribution prediction model is used to predict the feedback of different users on different equity types.

[0087] Through federated learning and edge computing, the data of the head office and branches are efficiently integrated, and the equity distribution data of each branch is pre-processed and analyzed locally, which reduces data transmission and processing time and promotes cross-branch collaboration. At the same time, the federated learning framework allows model collaboration between different branches without sharing original data, ensuring data privacy while achieving data insights across the entire bank.

[0088] The method for generating the equity distribution prediction model provided in the embodiment of the present application effectively protects privacy data through federated learning, and the original data does not leave the local area, reducing the risk of data leakage; it is resource-efficient, disperses the computing load, and reduces the dependence on the central server; it has global insights and provides comprehensive marketing strategy suggestions by aggregating the learning results of each branch; it is updated in real time, and can quickly respond to market changes and update marketing strategies in real time, thereby taking into account customer privacy protection and effective equity distribution decisions, and providing a marketing optimization solution that is both safe and efficient. The data flow in the embodiment of the present application is from the branch to the edge node for local processing and analysis; the model flow is the flow of model parameters in the federated network to achieve global model optimization and update. Since federated learning only shares model parameters rather than original data, the privacy of customer data is better protected. At the same time, edge computing processes data locally, reducing the risk of data being intercepted or tampered with during transmission; federated learning ensures that data is local and not transmitted in the network, protecting customer privacy; edge computing reduces the centralized storage of data and reduces security risks.

[0089] Embodiment 3

[0090] Figure 3 This is a flow chart of a method for distributing rights and interests provided in the third embodiment of the present invention. This embodiment is applicable to accurately distributing rights and interests to different users. The method can be executed by an edge node. Figure 3 As shown, the method includes:

[0091] S301. Obtain information on rights to be distributed, rights type, users to be predicted, and rights collection information corresponding to the users to be predicted.

[0092] In this embodiment, the rights and interests information to be distributed can be understood as the rights and interests information of activities with distribution needs. The activities can refer to activities that are about to be launched or have been launched. For example, the rights and interests information to be distributed can be one or more information such as the strength of rights and interests, activity time, activity rules, and activity stages. The activity stages include: one or more of activity participation, rights and interests collection, rights and interests exchange, and rights and interests write-off. The users to be predicted can be understood as users who need to predict the rights and interests they are interested in during the rights and interests distribution process. The rights and interests collection information corresponding to the users to be predicted can be one or more of rights and interests collection records, user feedback, rights and interests exchange information, user age, gender, consumption habits, etc.

[0093] Banks can publish activities regularly or on holidays, and different activities have different rights and interests information. The edge node obtains the rights and interests information to be distributed, the rights and interests type, the users to be predicted, and the rights and interests collection information corresponding to the users to be predicted. Among them, the number of users to be predicted can be one or more, and the users to be predicted can be obtained from the pre-generated user list. After determining the user to be predicted, the historical collection record of the rights and interests information of the user to be predicted is read accordingly to obtain the rights and interests collection information.

[0094] S302. Input the rights information to be distributed, the rights type and the rights collection information corresponding to the predicted user into the target rights distribution prediction model to perform rights distribution prediction. The target rights distribution prediction model is determined by the generation method of the rights distribution prediction model of any embodiment of the present invention.

[0095] The information on the rights to be distributed, the types of rights and the rights collection information corresponding to the users to be predicted are input into the target rights distribution prediction model. The target rights distribution prediction model analyzes the information on the rights to be distributed, the types of rights and the rights collection information, and analyzes the rights type of the users to be predicted based on the rights collection information and the rights type. Then, the rights that the users to be predicted may be interested in are analyzed in combination with the rights information to be distributed, so as to realize the rights distribution prediction.

[0096] S303: Determine the equity prediction result of the user to be predicted according to the output of the target equity distribution prediction model.

[0097] In this embodiment, the equity prediction result can be understood as the equity that the user is interested in. The output result of the target equity distribution prediction model is obtained, and the output result is used as the equity prediction result of the user to be predicted. For example, if the equity prediction result is a third-party coupon, the user to be predicted is more interested in the equity of the third-party coupon.

[0098] S304: Determine the equity distribution strategy corresponding to the user to be predicted according to the equity prediction result, so as to distribute equity to the user to be predicted according to the equity distribution strategy.

[0099] In this embodiment, the equity distribution strategy can be understood as a description of the distribution of equity, for example, which activity is recommended to which user. According to the equity prediction results, the equity that the predicted user is interested in can be determined, and then according to the different equity corresponding to different activities, the corresponding equity distribution strategy is generated, and the equity is distributed to the predicted user according to the equity distribution strategy, providing the user with more equity that he is interested in.

[0100] The embodiment of the present application provides a method for equity distribution. The edge node predicts the equity distribution through the target equity distribution prediction model, and can analyze the equity distribution data in real time, helping to more effectively understand and meet customer needs, and improve the pertinence and effectiveness of marketing activities. In addition, the target equity distribution prediction model in the embodiment of the present application uses the data processed locally by the edge node to train the target equity distribution prediction model. Through the federated learning model, it can more accurately characterize the customer's equity preferences and consumption behaviors, and provide a marketing strategy that is more in line with customer needs. Because the model can quickly respond to the update of local data, it can adjust the marketing strategy in time to adapt to changes in market and customer needs. The prediction of equity distribution through edge nodes makes data processing more decentralized, reduces the computing burden of the central server, and reduces operation and maintenance costs.

[0101] Embodiment 4

[0102] Figure 4 This is a schematic diagram of a device for generating a prediction model for equity distribution provided in the fourth embodiment of the present invention, which is applied to edge nodes. Figure 4 As shown, the device includes: a standardization processing module 41, a data analysis module 42 and a federated training module 43.

[0103] A standardization processing module 41 is used to perform standardization processing on the acquired original equity distribution data to obtain target equity distribution data, wherein the target equity distribution data includes equity activity information and equity collection information corresponding to at least one user;

[0104] A data analysis module 42, configured to analyze the target equity distribution data, determine a preset number of equity types, and generate a training sample set based on each of the equity types and the target equity distribution data;

[0105] The federated training module 43 is used to perform federated learning training on the equity distribution prediction model to be trained according to the training sample set combined with the training sample sets of other edge nodes to obtain a target equity distribution prediction model, and the target equity distribution prediction model is used to predict the feedback of different users on different equity types.

[0106] The embodiment of the present invention provides a method for generating a rights distribution prediction model, which solves the problems of inconsistent data formats and privacy leakage in the rights distribution process; in the process of data processing, the edge node standardizes the original rights distribution data to obtain target rights distribution data, ensures data consistency, and improves the accuracy of data analysis; the rights distribution prediction model is trained by the edge node, and the edge node combines the training samples of other edge nodes to perform federated learning training to generate a target rights distribution prediction model, which can conduct in-depth analysis of the data and accurately understand the user's feedback on different rights, so as to determine the marketing strategy according to the user's feedback on different rights and interests, and improve the accuracy of the marketing strategy; the edge node can process data locally in real time through the target rights distribution prediction model to improve the timeliness of the data, and the federated learning training model does not need to share the original data, thereby enhancing data privacy protection.

[0107] Optionally, the standardization processing module 41 includes:

[0108] A data cleaning unit, used to clean the acquired original rights distribution data to obtain valid rights distribution data;

[0109] The formatting processing unit is used to format the effective rights distribution data to obtain target rights distribution data.

[0110] Optionally, the data analysis module 42 includes:

[0111] A cluster center initialization unit, used to initialize a preset number of cluster centers;

[0112] A user cluster generating unit, configured to cluster the rights collection information corresponding to the users in the target rights distribution data based on the cluster centers, to obtain a preset number of user clusters;

[0113] The equity type determination module is used to determine the equity type corresponding to the user cluster according to the equity collection information of the users included in the user cluster.

[0114] Optionally, the federated training module 43 includes:

[0115] A candidate parameter determination unit, configured to train the equity distribution prediction model to be trained according to the training sample set to obtain candidate model parameters that meet the training cutoff requirements;

[0116] A candidate parameter sending unit, used to send the candidate model parameters to a central server, so that the central server determines the global model parameters according to the received candidate model parameters sent by the edge node and other edge nodes, and the candidate model parameters sent by other edge nodes are obtained by model training according to their corresponding training sample sets;

[0117] The global parameter receiving unit is used to receive the global model parameters fed back by the central server and generate a target equity distribution prediction model according to the global model parameters.

[0118] Optionally, the global model parameters are obtained by performing a weighted operation on the candidate model parameters.

[0119] Optionally, the rights and interests collection information includes at least one of the following: rights and interests collection records, user feedback, and rights and interests redemption information.

[0120] The device for generating the equity distribution prediction model provided in the embodiment of the present invention can execute the method for generating the equity distribution prediction model provided in any embodiment of the present invention, and has the functional modules and beneficial effects corresponding to the execution method.

[0121] Embodiment 5

[0122] Figure 5 This is a schematic diagram of the structure of a rights distribution device provided in Embodiment 5 of the present invention, which is applied to edge nodes. Figure 5 As shown, the device includes: an information acquisition module 51, a prediction module 52, a prediction result determination module 53 and a strategy determination module 54.

[0123] The information acquisition module 51 is used to acquire information on the rights and interests to be distributed, the type of rights and interests, the users to be predicted, and the rights and interests collection information corresponding to the users to be predicted;

[0124] A prediction module 52, used for inputting the information of the equity to be distributed, the type of equity and the equity collection information corresponding to the predicted user into the target equity distribution prediction model to perform equity distribution prediction, wherein the target equity distribution prediction model is determined by the generation method of the equity distribution prediction model according to any one of claims 1 to 6;

[0125] A prediction result determination module 53, used to determine the equity prediction result of the to-be-predicted user according to the output of the target equity distribution prediction model;

[0126] The strategy determination module 54 is used to determine the equity distribution strategy corresponding to the user to be predicted according to the equity prediction result, so as to distribute equity to the user to be predicted according to the equity distribution strategy.

[0127] The embodiment of the present application provides a method for equity distribution. The edge node predicts the equity distribution through the target equity distribution prediction model, and can analyze the equity distribution data in real time, helping to more effectively understand and meet customer needs, and improve the pertinence and effectiveness of marketing activities. In addition, the target equity distribution prediction model in the embodiment of the present application uses the data processed locally by the edge node to train the target equity distribution prediction model. Through the federated learning model, it can more accurately characterize the customer's equity preferences and consumption behaviors, and provide a marketing strategy that is more in line with customer needs. Because the model can quickly respond to the update of local data, it can adjust the marketing strategy in time to adapt to changes in market and customer needs. The prediction of equity distribution through edge nodes makes data processing more decentralized, reduces the computing burden of the central server, and reduces operation and maintenance costs.

[0128] The equity distribution method device provided in the embodiment of the present invention can execute the equity distribution method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0129] Embodiment 6

[0130] Figure 6 A schematic diagram of the structure of an electronic device 60 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0131] like Figure 6 As shown, the electronic device 60 includes at least one processor 61, and a memory connected to the at least one processor 61, such as a read-only memory (ROM) 62, a random access memory (RAM) 63, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 61 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 62 or the computer program loaded from the storage unit 68 to the random access memory (RAM) 63. In the RAM 63, various programs and data required for the operation of the electronic device 60 can also be stored. The processor 61, the ROM 62, and the RAM 63 are connected to each other via a bus 64. An input / output (I / O) interface 65 is also connected to the bus 64.

[0132] A number of components in the electronic device 60 are connected to the I / O interface 65, including: an input unit 66, such as a keyboard, a mouse, etc.; an output unit 67, such as various types of displays, speakers, etc.; a storage unit 68, such as a disk, an optical disk, etc.; and a communication unit 69, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 69 allows the electronic device 60 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0133] The processor 61 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 61 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The processor 61 executes the various methods and processes described above, such as a method for generating a rights distribution prediction model or a rights distribution method.

[0134] In some embodiments, the method for generating a rights distribution prediction model or the rights distribution method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 68. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 60 via the ROM 62 and / or the communication unit 69. When the computer program is loaded into the RAM 63 and executed by the processor 61, one or more steps of the method for generating a rights distribution prediction model or the rights distribution method described above may be executed. Alternatively, in other embodiments, the processor 61 may be configured to execute the method for generating a rights distribution prediction model or the rights distribution method by any other appropriate means (e.g., by means of firmware).

[0135] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0136] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0137] An embodiment of the present invention provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the method for generating a rights distribution prediction model or the rights distribution method described in any embodiment of the present invention.

[0138] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, 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 disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0139] To provide interaction with a user, the systems and techniques described herein may 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 trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0140] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may 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.

[0141] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0142] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0143] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for generating a prediction model for equity distribution, characterized in that: Applied to edge nodes, including: Standardize the acquired original equity distribution data to obtain target equity distribution data, where the target equity distribution data includes equity activity information and equity collection information corresponding to at least one user; Analyze the target equity distribution data, determine a preset number of equity types, and generate a training sample set based on each of the equity types and the target equity distribution data; The equity distribution prediction model to be trained is federated learning trained according to the training sample set combined with the training sample sets of other edge nodes to obtain a target equity distribution prediction model, and the target equity distribution prediction model is used to predict the feedback of different users on different equity types.

2. The method according to claim 1, characterized in that The processing of the obtained original equity distribution data to obtain target equity distribution data includes: Clean the original equity distribution data to obtain effective equity distribution data; The effective rights distribution data is formatted to obtain target rights distribution data.

3. The method according to claim 1, characterized in that The analyzing the target equity distribution data to determine the equity type corresponding to each user includes: Initialize a preset number of cluster centers; Clustering the rights collection information corresponding to the users in the target rights distribution data based on the cluster centers to obtain a preset number of user clusters; The type of rights and interests corresponding to the user cluster is determined according to the rights and interests collection information of the users included in the user cluster.

4. The method according to claim 1, characterized in that: The method of performing federated learning training on the equity distribution prediction model to be trained according to the training sample set combined with the training sample sets of other edge nodes to obtain a target equity distribution prediction model includes: Training the equity distribution prediction model to be trained according to the training sample set to obtain candidate model parameters that meet the training cutoff requirements; Sending the candidate model parameters to a central server so that the central server determines global model parameters based on the received edge nodes and the candidate model parameters sent by other edge nodes, and the candidate model parameters sent by other edge nodes are obtained by model training based on their corresponding training sample sets; Receive the global model parameters fed back by the central server, and generate a target equity distribution prediction model according to the global model parameters.

5. The method according to claim 4, characterized in that The global model parameters are obtained by performing a weighted operation on the candidate model parameters.

6. The method according to any one of claims 1 to 5, characterized in that: The rights and interests collection information includes at least one of the following: rights and interests collection records, user feedback, and rights and interests redemption information.

7. A method for distributing rights and interests, characterized in that: Applied to edge nodes, including: Obtaining information on rights to be distributed, rights type, users to be predicted, and rights collection information corresponding to the users to be predicted; Inputting the equity information to be distributed, the equity type and the equity collection information corresponding to the user to be predicted into the target equity distribution prediction model to perform equity distribution prediction, wherein the target equity distribution prediction model is determined by the generation method of the equity distribution prediction model according to any one of claims 1 to 6; Determine the equity prediction result of the to-be-predicted user according to the output of the target equity distribution prediction model; The equity distribution strategy corresponding to the user to be predicted is determined according to the equity prediction result, so as to distribute equity to the user to be predicted according to the equity distribution strategy.

8. An electronic device, characterized in that: The electronic device comprises: at least one processor, and a memory communicatively coupled to the at least one processor; Wherein, the memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the method for generating the equity distribution prediction model described in any one of claims 1 to 6 or the equity distribution method described in claim 7.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the method for generating a rights distribution prediction model described in any one of claims 1 to 6 or the rights distribution method described in claim 7 when executed.

10. A computer program product, characterized in that The computer program product comprises a computer program, which, when executed by a processor, implements the method for generating a rights distribution prediction model according to any one of claims 1 to 6 or the rights distribution method according to claim 7.

Images