Model training method and device, user classification method and device, equipment and storage medium
By training the target loan demand sub-segment model, the target user demand model and the target user qualification model, and classifying the confirmed users, the problem of insufficient user behavior characteristics caused by a single data dimension in the existing technology is solved, and higher advertising delivery accuracy and conversion rate are achieved.
Patent Information
- Application Number
- CN202311333460.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-10-16
- Publication Date
- 2025-04-18
AI Technical Summary
In the prior art, loan advertising providers have a single data dimension, which leads to insufficiently portraying the behavioral characteristics of target users, which in turn leads to problems such as excessive bank marketing costs and low success conversion rate after advertising.
By obtaining user latitude data samples, including user underlying data, first data label and second data label, the target loan requirement sub-segment model, target user requirement model and target user qualification model are trained, and these models are used to classify users to determine whether they are marketing user types, and to serve advertisements to them.
By more comprehensively utilizing user underlying data, the accuracy and generalization capabilities of the target user demand model and user qualification model can be improved, and target users with loan needs and meet loan qualifications can be more accurately screened out, reducing advertisers' marketing costs and improving the successful conversion rate of advertising delivery.
Smart Images
Figure CN119850321A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of computer model training technology, and in particular to a model training method, a user classification method, an apparatus, a device and a storage medium. Background Art
[0002] At present, there are many ways to advertise bank loan business. In addition to offline marketing, banks are also actively trying to send relevant information to users through social media, the Internet, etc. When screening loan intention users, factor scoring method is often used. First, the characteristic variables that affect the customer's loan willingness and the weights of different characteristic variables are determined, and then the user qualifications are evaluated, and finally the feasibility threshold is set.
[0003] In the existing technology, the characteristic variables and weights of users' loan willingness are often set manually and are highly subjective. Some financial institutions have introduced machine learning or neural network models to predict potential loan populations in order to reduce manual intervention and improve prediction accuracy.
[0004] However, the existing technology, which predicts users' loan willingness based on existing data, has problems such as single data dimension, incomplete description of user behavior characteristics, high marketing costs for banks, and low successful conversion rate after advertising. Summary of the invention
[0005] The present application provides a model training method, a user classification method, an apparatus, a device and a storage medium, which are used to solve the problem that loan advertisement publishers in the prior art use a single data dimension, resulting in an incomplete characterization of the target user's behavioral characteristics, which in turn results in excessive marketing costs for banks and a low successful conversion rate after advertising.
[0006] In a first aspect, the present application provides a model training method, comprising:
[0007] Acquire a user latitude data sample, wherein the user latitude data sample includes user underlying data, a first data tag and a second data tag, wherein the first data tag is used to mark whether the user has a loan demand, and the second data tag is used to mark whether the user has a loan qualification;
[0008] Training the loan demand sub-model according to the user underlying data and the first data label to obtain a target loan demand sub-model;
[0009] Inputting the user behavior data and the user address data in the user underlying data into the target loan demand sub-model to obtain a loan demand output result;
[0010] Training a loan demand model according to the loan demand output result, the call data and user attribute information in the user underlying data to obtain a target user demand model;
[0011] The user qualification model is trained according to the user underlying data and the second label to obtain a target user qualification model.
[0012] In a possible design, obtaining a user dimension data sample includes:
[0013] The user underlying information is obtained, and the user underlying information includes user basic information, device information, location information, Internet log, communication information, call fee information, and social information; the user underlying information is subjected to label conversion processing to obtain a user dimension data sample.
[0014] In a second aspect, the present application provides a user classification method, comprising: obtaining dimensional data of the user to be confirmed; inputting the dimensional data of the user to be confirmed into a target loan demand model and a target user qualification model, respectively, wherein the target loan demand model and the target user qualification model are respectively the target user demand model and the target user qualification model obtained after sample training; obtaining the output result of the target loan demand model and the output result of the target user qualification model; determining the user type of the user to be confirmed based on the output result of the target loan demand model and the output result of the target user qualification model, the user type including a marketable user type and a non-marketable user type.
[0015] In a possible design, determining the user type of the to-be-confirmed user according to the output result of the target loan demand model and the output result of the target user qualification model includes:
[0016] The method comprises determining a demand score of the user to be confirmed represented by an output result of the target loan demand model, and a qualification score of the user to be confirmed represented by an output result of the target user qualification model; and determining a user type of the user to be confirmed based on the demand score, the qualification score and a preset score threshold.
[0017] In a possible design, after determining the user type of the to-be-confirmed user according to the output result of the target loan demand model and the output result of the target user qualification model, the method further includes:
[0018] If the user type of the user to be confirmed is a marketable user type, an advertisement is pushed to the user to be confirmed.
[0019] In a possible design, if the user type of the to-be-confirmed user is a marketable user type, pushing an advertisement to the to-be-confirmed user includes:
[0020] If the user type of the user to be confirmed is a marketable user type, determining the tag of the user to be confirmed as a first user tag; and pushing advertisements to the user to be confirmed through an API interface according to the first user tag of the user to be confirmed.
[0021] In a possible design, after determining the user type of the to-be-confirmed user according to the output result of the target loan demand model and the output result of the target user qualification model, the method further includes:
[0022] If the user type of the user to be confirmed is a non-marketable user type, a second user tag of the user to be confirmed is determined, and advertisement pushing to the user to be confirmed with the second user tag is prohibited.
[0023] In a third aspect, the present application provides a model training method and a user classification model training device, including:
[0024] The first acquisition module is used to acquire a user latitude data sample, wherein the user latitude data sample includes user underlying data, a first data tag and a second data tag, wherein the first data tag is used to mark whether the user has a loan demand, and the second data tag is used to mark whether the user has a loan qualification;
[0025] The first training module: training the loan demand sub-model according to the user underlying data and the first data label to obtain a target loan demand sub-model;
[0026] The first processing module: inputs the user behavior data and the user address data in the user underlying data into the target loan demand sub-model to obtain a loan demand output result;
[0027] The second training module: training the loan demand model according to the loan demand output result, the call data and the user attribute information in the user underlying data to obtain the target user demand model;
[0028] The third training module: trains the user qualification model according to the user underlying data and the second label to obtain a target user qualification model.
[0029] In a fourth aspect, the present application provides a user classification device, including:
[0030] The second acquisition module: acquires the dimension data of the user to be confirmed;
[0031] The second processing module: inputs the dimension data of the user to be confirmed into the target loan demand model and the target user qualification model respectively, wherein the target loan demand model and the target user qualification model are the target user demand model and the target user qualification model respectively obtained after sample training;
[0032] The third acquisition module is used to acquire the output result of the target loan demand model and the output result of the target user qualification model;
[0033] Determination module: Determine the user type of the user to be confirmed according to the output result of the target loan demand model and the output result of the target user qualification model, wherein the user type includes a marketable user type and a non-marketable user type.
[0034] In a fifth aspect, the present application provides an electronic device, comprising: at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the model training method described in the first aspect and various possible designs of the first aspect, or the second aspect and various possible designs of the second aspect.
[0035] In a sixth aspect, the present application provides a computer storage medium, in which computer execution instructions are stored. When a processor executes the computer execution instructions, the user classification method described in the first aspect and various possible designs of the first aspect, or the second aspect and various possible designs of the second aspect is implemented.
[0036] The model training method, user classification method, device, equipment and storage medium provided by the present application first obtain user dimension data samples, use the user's underlying data and the loan demand output results output by the target loan demand sub-model as training inputs, and train the target user demand model through machine model learning and training; use the user's underlying data and the second label as training inputs, and train the target user qualification model through machine model learning and training, and use the loan demand output results output by the sub-model as the input of the training model, which can better utilize the existing user's underlying data and improve the accuracy and generalization ability of the target user demand model. In addition, the trained target user demand model and target user qualification model can screen out target users with loan needs and loan qualifications based on the characteristics of user online behavior, and deliver loan advertisements to them, solving the problems of single data dimension, insufficient comprehensive characterization of user behavior characteristics, high marketing costs for advertisers, and low conversion rate of advertising. BRIEF DESCRIPTION OF THE DRAWINGS
[0037] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.
[0038] Figure 1 A flowchart of a model training method provided in an embodiment of the present application;
[0039] Figure 2 A flowchart of a user classification method provided in an embodiment of the present application;
[0040] Figure 3 A schematic diagram of the structure of a model training device provided in one embodiment of the present application;
[0041] Figure 4 A schematic diagram of the structure of a user classification device provided by an embodiment of the present application;
[0042] Figure 5 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application.
[0043] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION
[0044] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.
[0045] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards, and provide corresponding operation entrances for users to choose to authorize or refuse.
[0046] At present, there are many ways to advertise bank loan business. In addition to offline marketing, banks are also actively trying to send relevant information to users through social media, the Internet, etc. When screening loan intention users, factor scoring method is often used. First, the characteristic variables that affect the customer's loan willingness and the weights of different characteristic variables are determined, and then the user qualifications are evaluated, and finally the feasibility threshold is set.
[0047] In the existing technology, the characteristic variables and weights of users' loan willingness are often set manually and are highly subjective. Some financial institutions have introduced machine learning or neural network models to predict potential loan populations in order to reduce manual intervention and improve prediction accuracy.
[0048] However, the existing technology, which predicts users' loan willingness based on existing data, has problems such as single data dimension, incomplete description of user behavior characteristics, high marketing costs for banks, and low successful conversion rate after advertising.
[0049] The model training method and user classification method provided in this application first obtain user dimension data samples, use the call data and user attribute information in the user dimension data samples, and the loan demand output results of the target loan demand sub-model as training inputs, and train the target user demand model through the XGBoost analysis method; use the user's underlying data and the second label as training inputs, and train the target user qualification model through the XGBoost analysis method. According to the output results of the target user demand model, the output results of the target user qualification model and the set threshold, it is judged whether the target user belongs to the marketable user type or the non-marketable user type, and advertisements are delivered to the target users of the marketable user type through the interface. This method has a clearer positioning of the target users, and solves the problems of excessively high marketing costs for advertisers and low conversion rates of advertising.
[0050] The model training method and user classification method provided in this application are intended to solve the above technical problems of the prior art.
[0051] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.
[0052] Figure 1 Flowchart of the model training method provided in the embodiment of the present application. Figure 1 As shown, the execution subject of the model training method shown in the embodiment of the present application can be a server or a personal computer, and the model training method includes the steps of:
[0053] S101: Obtain user dimension data samples, which include user underlying data, a first data tag and a second data tag. The first data tag is used to mark whether the user has a loan demand, and the second data tag is used to mark whether the user has loan qualifications.
[0054] Among them, in the process of obtaining user latitude data samples, the user's underlying data can be multi-dimensional user feature labels such as user basic information, device information, location information, Internet log, communication information, call fee information and social information, or it can be the operation result representing the user behavior relationship obtained by substituting the above user feature labels into the label expression. For example: when the label expression Y = (A OR B) AND C, A, B, and C represent user feature labels of three dimensions: device information, location information, and Internet log, and Y is the behavior association relationship of the user represented by the label expression in a certain time series. This application does not impose any restrictions on this.
[0055] The first data label and the second data label are obtained based on the loan data, and the loan data includes: the user's loan demand and the user's loan qualification. In order to more comprehensively describe the user behavior, during the model training process, this application sets the first data label to mark whether the user has a loan demand, and sets the second data label to mark whether the user has a loan qualification, as user sample data representing the two dimensions of the user's loan demand and loan qualification.
[0056] Among them, user dimension sample data refers to data that analyzes and studies users based on their attributes, behaviors, interests, etc.
[0057] The method for obtaining user dimension data samples may include:
[0058] Sa: Obtain user underlying information, including basic user information, device information, location information, Internet logs, communication information, call fee information, and social information;
[0059] Sb: Perform label conversion on the user's underlying information to obtain user dimension data samples.
[0060] Among them, label conversion processing can refer to converting non-numeric data into numerical data so that the machine learning algorithm can better process it.
[0061] S102: Training the loan demand sub-model according to the user underlying data and the first data label to obtain a target loan demand sub-model.
[0062] Among them, the user sample data with loan demand in the first data label is used as the positive sample data for model training, and the user sample data without loan demand in the first loan label is used as the negative sample data for model training.
[0063] Specifically, a neural network is used as a feature extractor to extract features of the user's underlying data and the first data label, and the features of the user's underlying data and the first data label extracted by the neural network are input into a machine learning training model for XGBoost analysis method training. During the training process, the recursive feature data is continuously screened, and important feature data is re-input into the model for further training. The overfitting of the model is reduced through cross-validation and other methods to achieve the training effect and obtain the target loan demand sub-segmentation model.
[0064] S103: Input the user behavior data and user address data in the user underlying data into the target loan demand sub-model to obtain a loan demand output result.
[0065] Among them, user behavior data refers to the list of apps visited in the past 6 months (a collection of all apps installed on the mobile phone), and user address data refers to the 5-digit geohash (a method of encoding geographic location information) address list.
[0066] Among them, the loan demand output result is a score of 0-100. The larger the score, the stronger the loan demand willingness.
[0067] S104: Train the loan demand model according to the loan demand output result, the call data in the user underlying data, and the user attribute information to obtain a target user demand model.
[0068] Among them, the call data in the user's underlying data refers to the call records generated when the user uses the communication service, including call time, call location, call duration, call object, etc. User attribute information refers to the user's basic information and status, including natural characteristics and characteristics of behavioral reminders.
[0069] Specifically, the target user demand model is obtained by training through the XGBoost analysis method. The loan demand output results, call data in the user's underlying data, and user attribute information are input into the loan demand model for XGBoost analysis method training. During the training process, the recursive feature data is continuously screened, and the important feature data is re-input into the model for re-training. The overfitting degree of the model is reduced through cross-validation and other methods, so that it can achieve the training effect and obtain the target user demand model.
[0070] S105: Training the user qualification model according to the user underlying data and the second label to obtain a target user qualification model.
[0071] Among them, the user sample data of users who meet the loan qualification requirements in the second loan label is used as the positive sample data for model training, and the user sample data of users who do not meet the loan qualification requirements in the second loan label is used as the negative sample data for model training.
[0072] Specifically, the target user qualification model is obtained by training through the XGBoost analysis method. The target user qualification model is trained using the XGBoost analysis method and framework, and the user's underlying data and the second label are used as training parameters, and input into the target user qualification model using the XGBoost model as the framework for training to obtain the target user qualification model, and the target user qualification model is used to predict the target user's qualification.
[0073] In summary, in this embodiment, firstly, a user dimension data sample is obtained, and the call data and user attribute information in the user dimension data sample, as well as the loan demand output result output by the target loan demand sub-model are used as training inputs, and the target user demand model is trained through the XGBoost analysis method; the user underlying data and the second label are used as training inputs, and the target user qualification model is trained through the XGBoost analysis method. The trained target user demand model and target user qualification model can screen out target users with loan needs and loan qualifications based on the characteristics of user online behavior, and deliver loan advertisements to them, which can solve the problems of single data dimension, incomplete characterization of user behavior characteristics, high marketing costs for advertisers, and low conversion rate of advertising delivery.
[0074] Figure 2 is a flow chart of the user classification method provided in the embodiment of the present application. This embodiment describes a method for predicting loan demand and loan qualifications based on a trained model and delivering advertisements to the target population. Figure 2 As shown, the user classification method shown in this embodiment includes:
[0075] S201: Obtain dimensional data of the user to be confirmed.
[0076] Among them, the users to be confirmed are users who need to be predicted for loan demand and loan qualification and to whom advertisements need to be placed. The dimensional data of the users to be confirmed include: basic user information, device information, location information, Internet logs, communication information, call fee information, and social information. Among them, in some embodiments, the storage period requirements for dimensional data can be: 6 months for daily data and 12 months for monthly data.
[0077] S202: Inputting the dimension data of the user to be confirmed into the target loan demand model and the target user qualification model respectively, wherein the target loan demand model and the target user qualification model are the target user demand model and the target user qualification model respectively obtained through sample training.
[0078] The call data and user attribute information in the user dimension data sample and the loan demand output result output by the target loan demand sub-model are input into the target user demand model to obtain the user loan demand score output by the target user demand model. The user underlying data and the second label are input into the target user qualification model to obtain the user loan qualification score output by the target user qualification model.
[0079] Among them, the target user demand model and the target user qualification model are obtained after evaluation of indicators such as classification threshold analysis, learning curve, cross validation and loss function value to meet the training effect.
[0080] S203: Obtaining the output result of the target loan demand model and the output result of the target user qualification model.
[0081] Among them, the target user demand model and the target user qualification model will output the target user's loan demand score and loan qualification score respectively. The higher the target user's loan demand, the higher the loan demand score output by the target user demand model; the higher the target user's loan qualification, the higher the loan qualification score output by the target user qualification model.
[0082] S204: Determine the user type of the user to be confirmed according to the output result of the target loan demand model and the output result of the target user qualification model, where the user type includes a marketable user type and a non-marketable user type.
[0083] Among them, the output result of the target loan demand model is a score of 0-100. The larger the score, the stronger the target user's loan demand willingness; the output result of the target loan qualification model is a score of 0-100. The larger the score, the higher the target user's loan qualification.
[0084] In summary, this embodiment obtains the dimensional data of the user to be confirmed and inputs them into the target loan demand model and the target user qualification model respectively, to obtain the output results of the target user demand model and the output results of the target user qualification model, and then judges whether the target user belongs to the marketable user type or the non-marketable user type based on the output results and the set threshold, thereby having a clearer positioning of the target user and greatly increasing the success rate of advertising after delivery.
[0085] In one embodiment of the present application, based on the above embodiment, the present application embodiment focuses on how to determine the user type of the user to be confirmed, and the method includes:
[0086] S301: Obtaining dimension data of the user to be confirmed;
[0087] S302: Inputting the dimension data of the user to be confirmed into the target loan demand model and the target user qualification model respectively, wherein the target loan demand model and the target user qualification model are the target user demand model and the target user qualification model respectively obtained through sample training;
[0088] S303: Obtaining the output result of the target loan demand model and the output result of the target user qualification model;
[0089] In this embodiment, the contents of steps S301 to S303 are consistent with the contents of steps S201 to S203. Please refer to steps S201 to S203 for detailed description, which will not be repeated here.
[0090] S304: Determine the demand score of the user to be confirmed represented by the output result of the target user demand model, and the qualification score of the user to be confirmed represented by the output result of the target user qualification model.
[0091] Among them, the output result of the target user demand model is the demand degree of the target user's loan demand, and the output result of the target user qualification model is the qualification degree of the target user's loan qualification.
[0092] S305: Determine the user type of the user to be confirmed based on the demand score, the qualification score and a preset score threshold.
[0093] Among them, user types include marketable user types and non-marketable user types.
[0094] The output result score of the target loan demand model ranges from 0 to 100, and the output result score of the target user qualification model ranges from 0 to 100. Set the threshold of the target user's loan demand to 60 points, and the threshold of the target user's loan qualification to 60 points. If the output result score of the target loan demand model exceeds 60 points, the target user belongs to the marketable user type; if the output result score of the target loan qualification model exceeds 60 points, the target user belongs to the marketable user type.
[0095] In one type of marketing, specifically, step S305 includes:
[0096] Sa: If the user type of the user to be confirmed is a marketable user type, determine that the tag of the user to be confirmed is the first user tag, and push advertisements to the user to be confirmed.
[0097] The user's tag may be basic information, behavior information or business preference.
[0098] Sb: Push advertisements to the user to be confirmed through an API interface according to the first user tag of the user to be confirmed.
[0099] Among them, API allows different software applications to communicate and interact with each other to achieve functions such as data exchange and function sharing.
[0100] In another marketing type, specifically, step S305 includes:
[0101] If the user type of the user to be confirmed is a non-marketable user type, a second user tag of the user to be confirmed is determined, and pushing advertisements to the user to be confirmed with the second user tag is prohibited.
[0102] In summary, this embodiment obtains the output results of the target user demand model and the target user qualification model by inputting the acquired dimensional data of the user to be confirmed into the target loan demand model and the target user qualification model, and then determines whether the target user belongs to the marketable user type or the non-marketable user type according to the output results and the set threshold, so that the positioning of the target user is clearer. In addition, advertising is placed on the target user of the marketable user type through the API interface, which increases the accuracy of targeted advertising and reduces the advertising cost of advertisers.
[0103] Figure 3 A schematic diagram of the structure of a model training device provided in one embodiment of the present application. Figure 3 As shown, the model training device includes: a first acquisition module 301, a first training module 302, a first processing module 303, a second training module 304 and a third training module 305.
[0104] The first acquisition module 301: acquires a user latitude data sample, which includes user underlying data, a first data tag and a second data tag, wherein the first data tag is used to mark whether the user has a loan demand, and the second data tag is used to mark whether the user has a loan qualification;
[0105] First training module 302: training the loan demand sub-model according to the user underlying data and the first data label to obtain a target loan demand sub-model;
[0106] The first processing module 303: inputting the user behavior data and the user address data in the user underlying data into the target loan demand sub-model to obtain the loan demand output result;
[0107] The second training module 304: trains the loan demand model according to the loan demand output result, the call data in the user underlying data and the user attribute information to obtain the target user demand model;
[0108] The third training module 305: trains the user qualification model according to the user underlying data and the second label to obtain a target user qualification model.
[0109] The device provided in this embodiment can be used to execute the technical solution of the above method embodiment. Its implementation principle and technical effect are similar, and this embodiment will not be repeated here.
[0110] In a possible design, the first acquisition module 401 is specifically used to: acquire user underlying information, including user basic information, device information, location information, Internet log, communication information, call fee information, and social information. Perform label conversion processing on the user underlying information to obtain user dimension data samples.
[0111] The device provided in this embodiment can be used to execute the technical solution of the above method embodiment. Its implementation principle and technical effect are similar, and this embodiment will not be repeated here.
[0112] Figure 4 This is a schematic diagram of the structure of a user classification device provided by an embodiment of the present application. Figure 4 As shown, the user classification device includes: a second acquisition module 401, a second processing module 402, a third acquisition module 403, a determination module 404, a first determination module 405 and a second determination module 406.
[0113] The second acquisition module 401: acquiring dimension data of the user to be confirmed;
[0114] The second processing module 402: inputs the dimension data of the user to be confirmed into the target loan demand model and the target user qualification model respectively, wherein the target loan demand model and the target user qualification model are the target user demand model and the target user qualification model respectively obtained through sample training;
[0115] The third acquisition module 403: acquires the output result of the target loan demand model and the output result of the target user qualification model;
[0116] Determination module 404: Determine the user type of the user to be confirmed according to the output result of the target loan demand model and the output result of the target user qualification model, where the user type includes a marketable user type and a non-marketable user type.
[0117] The device provided in this embodiment can be used to execute the technical solution of the above method embodiment. Its implementation principle and technical effect are similar, and this embodiment will not be repeated here.
[0118] In one possible design, the determination module 404 is specifically used to: determine the demand score of the user to be confirmed represented by the output result of the target loan demand model, and the qualification score of the user to be confirmed represented by the output result of the target user qualification model. According to the demand score, the qualification score and a preset score threshold, the user type of the user to be confirmed is determined.
[0119] In a possible design, the above device further includes:
[0120] The first determination module 405: if the user type of the user to be confirmed is a marketable user type, push an advertisement to the user to be confirmed.
[0121] In a possible design, the sending module 405 is specifically used to: if the user type of the user to be confirmed is a marketable user type, determine that the tag of the user to be confirmed is the first user tag, and push advertisements to the user to be confirmed through the API interface according to the first user tag of the user to be confirmed.
[0122] In a possible design, the above device further includes:
[0123] Second determination module 406: If the user type of the user to be confirmed is a non-marketable user type, a second user tag of the user to be confirmed is determined, and advertisements are prohibited from being pushed to the user to be confirmed with the second user tag.
[0124] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 5 As shown, the electronic device includes:
[0125] The electronic device may include one or more processors 501 of processing cores, one or more computer-readable storage media memories 502 , communication components 503 and other components. The processor 501 , the memory 502 and the communication component 503 are connected via a bus 504 .
[0126] In the specific implementation process, at least one processor 501 executes the computer execution instructions stored in the memory 502, so that at least one processor 501 executes the above model training method and user classification method.
[0127] The specific implementation process of the processor 501 can be found in the above method embodiment, and its implementation principle and technical effect are similar, so this embodiment will not be repeated here.
[0128] In the above Figure 5In the illustrated embodiment, it should be understood that the processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), etc. A general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the invention may be directly implemented as being executed by a hardware processor, or may be executed by a combination of hardware and software modules in the processor.
[0129] The memory may include a high-speed memory (Random Access Memory, RAM), and may also include a non-volatile memory (Non-volatile Memory, NVM), such as at least one disk memory.
[0130] The bus may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, the bus in the drawings of the present application is not limited to only one bus or one type of bus.
[0131] In some embodiments, a computer program product is also proposed, including a computer program or instructions, which, when executed by a processor, implements the steps of any of the above-mentioned model training methods and user classification methods.
[0132] The specific implementation of the above operations can be found in the previous embodiments, which will not be described in detail here.
[0133] A person of ordinary skill in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be completed by instructions, or by controlling related hardware through instructions. The instructions may be stored in a computer-readable storage medium and loaded and executed by a processor.
[0134] To this end, an embodiment of the present application provides a computer-readable storage medium, which stores multiple instructions, and the instructions can be loaded by a processor to execute the steps of any model training method or user classification method provided in the embodiment of the present application.
[0135] The storage medium may include: a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.
[0136] According to one aspect of the present application, a computer program product or a computer program is provided. The computer program product or the computer program comprises computer instructions stored in a computer-readable storage medium.
[0137] Since the instructions stored in the storage medium can execute the steps in any model training method and user classification method provided in the embodiments of the present application, the beneficial effects that can be achieved by any model training method and user classification method provided in the embodiments of the present application can be achieved. Please refer to the previous embodiments for details and will not be repeated here.
[0138] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. The present application is intended to cover any modification, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary techniques in the art that are not disclosed in the present application. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present application are indicated by the following claims.
[0139] It should be understood that the present application is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.
Claims
1. A model training method, characterized in that: The method comprises: Acquire a user latitude data sample, wherein the user latitude data sample includes user underlying data, a first data tag and a second data tag, wherein the first data tag is used to mark whether the user has a loan demand, and the second data tag is used to mark whether the user has a loan qualification; Training the loan demand sub-model according to the user underlying data and the first data label to obtain a target loan demand sub-model; Inputting the user behavior data and the user address data in the user underlying data into the target loan demand sub-model to obtain a loan demand output result; Training a loan demand model according to the loan demand output result, the call data and user attribute information in the user underlying data to obtain a target user demand model; The user qualification model is trained according to the user underlying data and the second label to obtain a target user qualification model.
2. The method according to claim 1, characterized in that: The obtaining of user dimension data samples includes: Obtaining user underlying information, including user basic information, device information, location information, Internet logs, communication information, call fee information, and social information; The user's underlying information is subjected to label conversion processing to obtain a user dimension data sample.
3. A user classification method, characterized in that: The method comprises: Get the dimension data of the user to be confirmed; Inputting the dimension data of the user to be confirmed into a target loan demand model and a target user qualification model respectively, wherein the target loan demand model and the target user qualification model are the target user demand model and the target user qualification model described in claims 1 to 2 respectively; Obtaining the output result of the target loan demand model and the output result of the target user qualification model; The user type of the user to be confirmed is determined according to the output result of the target loan demand model and the output result of the target user qualification model, and the user type includes a marketable user type and a non-marketable user type.
4. The method according to claim 3, characterized in that: The step of determining the user type of the user to be confirmed according to the output result of the target loan demand model and the output result of the target user qualification model includes: Determine the demand score of the user to be confirmed represented by the output result of the target loan demand model, and the qualification score of the user to be confirmed represented by the output result of the target user qualification model; The user type of the user to be confirmed is determined according to the demand score, the qualification score and a preset score threshold.
5. The method according to claim 3, characterized in that: After determining the user type of the to-be-confirmed user according to the output result of the target loan demand model and the output result of the target user qualification model, the method further includes: If the user type of the user to be confirmed is a marketable user type, an advertisement is pushed to the user to be confirmed.
6. The method according to claim 5, characterized in that If the user type of the to-be-confirmed user is a marketable user type, pushing an advertisement to the to-be-confirmed user includes: If the user type of the user to be confirmed is a marketable user type, determining that the mark of the user to be confirmed is a first user mark; According to the first user tag of the user to be confirmed, an advertisement is pushed to the user to be confirmed through an API interface.
7. The method according to claim 3, characterized in that After determining the user type of the to-be-confirmed user according to the output result of the target loan demand model and the output result of the target user qualification model, the method further includes: If the user type of the user to be confirmed is a non-marketable user type, a second user tag of the user to be confirmed is determined, and pushing advertisements to the user to be confirmed with the second user tag is prohibited.
8. A model training device, characterized in that: include: The first acquisition module is used to acquire a user latitude data sample, wherein the user latitude data sample includes user underlying data, a first data tag and a second data tag, wherein the first data tag is used to mark whether the user has a loan demand, and the second data tag is used to mark whether the user has a loan qualification; The first training module: training the loan demand sub-model according to the user underlying data and the first data label to obtain a target loan demand sub-model; The first processing module: inputs the user behavior data and the user address data in the user underlying data into the target loan demand sub-model to obtain a loan demand output result; The second training module: training the loan demand model according to the loan demand output result, the call data and the user attribute information in the user underlying data to obtain the target user demand model; The third training module: trains the user qualification model according to the user underlying data and the second label to obtain a target user qualification model.
9. A user classification device, characterized in that: include: The second acquisition module: acquires the dimension data of the user to be confirmed; The second processing module: inputs the dimension data of the user to be confirmed into the target loan demand model and the target user qualification model respectively, wherein the target loan demand model and the target user qualification model are the target user demand model and the target user qualification model described in claims 1 to 2 respectively; The third acquisition module is used to acquire the output result of the target loan demand model and the output result of the target user qualification model; Determination module: Determine the user type of the user to be confirmed according to the output result of the target loan demand model and the output result of the target user qualification model, wherein the user type includes a marketable user type and a non-marketable user type.
10. An electronic device, characterized in that: include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the method according to any one of claims 1 to 2, or any one of claims 3 to 7.
11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 2, or any one of claims 3 to 7 when executed by a processor.