Model training method and device, reminder mode prediction method and device

Abstract

The present embodiment relates to the field of artificial intelligence technology, and in particular to a model training method and device, and a reminder mode prediction method and device. The model training method includes: obtaining historical reminder data of a sample object; obtaining the effectiveness of a preset reminder mode according to the reminder mode information; screening out a target reminder mode from the preset reminder mode according to the effectiveness; performing vectorization processing on the sample information feature data to obtain a sample object feature vector, and performing vectorization processing on the sample policy feature data to obtain a sample policy feature vector; inputting the sample object feature vector and the sample policy feature vector into a preset original prediction model for training to obtain a predicted reminder mode; adjusting the parameters of the original prediction model according to the predicted reminder mode and the target reminder mode to obtain a target prediction model. The technical solution of the embodiment of the present application can predict the reminder mode of an object without the need for historical data.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence, and in particular to a model training method and device, and a reminder mode prediction method and device. Background Art

[0002] In related technologies, the current reminder method is determined by analyzing historical reminder methods. For example, in an insurance renewal scenario, the current renewal reminder method is determined by analyzing the current subject's historical renewal reminder methods.

[0003] It can be seen from this that the above method for determining the reminder mode requires obtaining historical data, which results in some objects being unable to be effectively reminded due to lack of historical data. Summary of the Invention

[0004] The main purpose of the embodiments of the present application is to propose a model training method and device, and a reminder mode prediction method and device, which can predict the reminder mode of an object without the need for historical data.

[0005] To achieve the above objectives, a first aspect of an embodiment of the present application proposes a model training method, the method comprising:

[0006] Acquire historical reminder data of the sample object; wherein the historical reminder data includes sample object information and reminder method information of the sample object, and the reminder method information is used to remind the sample object to perform the target historical behavior through a preset reminder method;

[0007] Obtaining the validity of the preset reminder method according to the reminder method information; wherein the validity includes valid;

[0008] Filtering a target reminder method from the preset reminder methods according to the validity; wherein the target reminder method is the preset reminder method with valid validity;

[0009] Acquire sample data of the sample object according to the sample object information; wherein the sample data includes sample information feature data and sample policy feature data of the sample object;

[0010] Performing vectorization processing on the sample information feature data to obtain a sample object feature vector, and performing vectorization processing on the sample policy feature data to obtain a sample policy feature vector;

[0011] Inputting the sample object feature vector and the sample policy feature vector into a preset original prediction model for training to obtain a prediction reminder method;

[0012] The parameters of the original prediction model are adjusted according to the prediction reminder mode and the target reminder mode to obtain a target prediction model.

[0013] In some embodiments, performing vectorization processing on the sample information feature data to obtain a sample object feature vector includes:

[0014] Extracting features from the sample information feature data to obtain a sample object feature group;

[0015] Encoding the sample object feature group to obtain a sample object encoding vector;

[0016] The sample object encoding vector is vectorized according to preset mapping parameters to obtain the sample object feature vector.

[0017] In some embodiments, the reminder method information includes the reminder time of the preset reminder method;

[0018] The obtaining, according to the reminder method information, of the validity of the preset reminder method includes:

[0019] Obtaining the target time according to the reminder time and preset duration;

[0020] Obtaining the time when the sample object executes the target historical behavior to obtain the behavior time;

[0021] comparing the temporal sequence of the behavior time and the target time;

[0022] If the behavior time is before the target time, the validity is valid.

[0023] In some embodiments, the sample object information includes a sample object identifier and a sample policy identifier;

[0024] The acquiring the sample data of the sample object according to the sample object information includes:

[0025] Filtering the sample information feature data from a preset first database according to the sample object identifier;

[0026] The sample policy characteristic data is obtained by screening from a preset second database according to the sample policy identifier.

[0027] To achieve the above-mentioned purpose, a second aspect of the embodiments of the present application provides a reminder mode prediction method, the method comprising:

[0028] Acquire target data of a target object; wherein the target data includes target object characteristic data and target policy characteristic data of the target object;

[0029] Performing matching processing on the target object feature data to obtain a target object feature vector, and performing matching processing on the target insurance policy feature data to obtain a target insurance policy feature vector;

[0030] The target object feature vector and the target policy feature vector are input into a target prediction model for prediction to obtain the current reminder mode; wherein, the target prediction model is obtained according to the model training method as described in any embodiment of the first aspect.

[0031] In some embodiments, a sample object vector library is constructed based on the sample object feature vectors, and a sample policy vector library is constructed based on the sample policy feature vectors;

[0032] The target object feature vector is screened out from the sample object vector library according to the target object feature data, and the target policy feature vector is screened out from the sample policy vector library according to the target policy feature data.

[0033] To achieve the above objectives, a third aspect of the embodiments of the present application provides a model training device, comprising:

[0034] A historical data acquisition module, configured to acquire historical reminder data of a sample object; wherein the historical reminder data includes sample object information and reminder method information of the sample object, wherein the reminder method information is used to remind the sample object to perform a target historical behavior through a preset reminder method;

[0035] a target reminder mode determination module, configured to obtain the validity of the preset reminder mode according to the reminder mode information; wherein the validity includes valid; and filter out a target reminder mode from the preset reminder modes according to the validity; wherein the target reminder mode is the preset reminder mode whose validity is valid;

[0036] A sample data acquisition module, configured to acquire sample data of the sample object according to the sample object information; wherein the sample data includes sample information feature data and sample policy feature data of the sample object;

[0037] A vectorization processing module, configured to perform vectorization processing on the sample information feature data to obtain a sample object feature vector, and perform vectorization processing on the sample policy feature data to obtain a sample policy feature vector;

[0038] A reminder mode prediction module, configured to input the sample object feature vector and the sample policy feature vector into a preset original prediction model for training to obtain a predicted reminder mode;

[0039] The parameter adjustment module is used to adjust the parameters of the original prediction model according to the prediction reminder mode and the target reminder mode to obtain a target prediction model.

[0040] To achieve the above-mentioned purpose, a fourth aspect of the embodiments of the present application provides a reminder mode prediction device, comprising:

[0041] A target data acquisition module is used to acquire target data of a target object; wherein the target data includes target object characteristic data and target policy characteristic data of the target object;

[0042] a matching module, configured to perform matching processing on the target object feature data to obtain a target object feature vector, and perform matching processing on the target policy feature data to obtain a target policy feature vector;

[0043] The current reminder mode prediction module is used to input the target object feature vector and the target policy feature vector into the target prediction model for prediction to obtain the current reminder mode; wherein, the target prediction model is obtained according to the model training method described in any embodiment of the first aspect.

[0044] To achieve the above-mentioned objectives, a fifth aspect of the embodiments of the present application provides an electronic device, including:

[0045] at least one memory;

[0046] at least one processor;

[0047] at least one computer program;

[0048] The computer program is stored in the memory, and the processor executes at least one computer program to implement:

[0049] The method according to the first aspect; or

[0050] The method according to the second aspect.

[0051] To achieve the above-mentioned objective, a sixth aspect of the embodiments of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to cause a computer to execute:

[0052] The method according to the first aspect; or

[0053] The method according to the second aspect.

[0054] The model prediction method and device, and the reminder mode prediction method and device provided in the embodiments of the present application obtain the sample object information and reminder mode information of the sample object, thereby obtaining the corresponding sample object feature vector and sample policy feature vector according to the sample object information association, and obtaining the target reminder mode according to the reminder mode information. Then, the preset original prediction model is trained according to the sample object feature vector and the sample policy feature vector, and the parameters of the original prediction model are adjusted according to the output data of the original prediction model (i.e., the predicted reminder mode) and the target reminder mode, thereby obtaining the target prediction model. It can be seen that the target prediction model trained by the embodiment of the present application can predict the reminder mode based on the data representing the individual attributes of the object and the attribute data of the insurance policy insured by the object, avoiding the situation in the related art where historical data (including historical reminder modes) are required to predict the reminder mode, so the embodiment of the present application improves the universality of the reminder mode prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 This is a flow chart of the model training method according to an embodiment of the present application;

[0056] Figure 2 This is another flowchart of the model training method according to an embodiment of the present application;

[0057] Figure 3 This is another flowchart of the model training method according to an embodiment of the present application;

[0058] Figure 4 This is another flowchart of the model training method according to an embodiment of the present application;

[0059] Figure 5 This is a flow chart of a reminder mode prediction method according to an embodiment of the present application;

[0060] Figure 6 This is another flowchart of the reminder mode prediction method according to an embodiment of the present application;

[0061] Figure 7 This is a module block diagram of the model training device according to an embodiment of the present application;

[0062] Figure 8 This is a module block diagram of the reminder mode prediction device according to an embodiment of the present application;

[0063] Figure 9 It is a schematic diagram of the hardware structure of the electronic device according to the embodiment of the present application. DETAILED DESCRIPTION

[0064] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0065] It should be noted that although the device schematics illustrate functional module divisions and the flowcharts illustrate logical sequences, in certain circumstances, the steps shown or described may be performed in a sequence that differs from the module divisions in the device or the sequence in the flowcharts. The terms "first," "second," and so on, in the specification, claims, and drawings, are used to distinguish similar items and are not necessarily used to describe a specific sequence or precedence.

[0066] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application pertains. The terms used herein are for the purpose of describing the embodiments of this application only and are not intended to limit this application.

[0067] First, let’s analyze some of the terms used in this application:

[0068] Artificial Intelligence (AI) is a new technical discipline that studies and develops theories, methods, technologies, and application systems for simulating, extending, and expanding human intelligence. A branch of computer science, AI seeks to understand the essence of intelligence and create new intelligent machines that can respond in a manner similar to human intelligence. Research in this field includes robotics, speech recognition, image recognition, natural language processing, and expert systems. AI can simulate the information processes of human consciousness and thinking. It also encompasses theories, methods, technologies, and application systems that use digital computers or digital computer-controlled machines to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results.

[0069] Discretization: This process converts continuous numerical attributes into discrete numerical attributes. Data discretization includes discretization of time data, discretization of multi-valued discrete data, and discretization of continuous data. Discretization of time data is primarily used for data aggregation and granularity conversion where time is the primary characteristic. Discretization converts scattered time features into higher-level time features. Common discretization operations for time data fall into two categories: The first category, discretization of the time of day, typically involves converting timestamps into seconds, minutes, hours, or morning and afternoon granularity. The second category, discretization of data at or above the daily granularity, typically involves converting dates into week numbers, days of the week, months, weekdays or weekends, quarters, years, and so on. Discretization of multi-valued discrete data involves discretizing categorical or ordinal data. Because there may be logical issues in the previous discretization process (typically caused by changes in business logic), the multi-valued discretized data needs to be re-partitioned. Discretization of continuous data is primarily used in classification or association analysis. Common methods for discretizing continuous data include the following: First, the quantile method, which uses quantiles such as quartiles and quintiles for discretization. Second, the distance interval method, which uses equidistant intervals or custom intervals for discretization, offers greater flexibility and can meet customized requirements. Furthermore, these methods can better preserve the original data distribution. Third, the frequency interval method, which sorts data according to their frequency distribution and then discretizes them according to equal or specified frequencies. This method transforms the data into a uniform distribution. While its advantage is that the observed values ​​within each interval are identical, its disadvantage is that it alters the original data distribution. Fourth, the clustering method, which uses the K-means algorithm to divide a sample set into multiple discretized clusters. Fifth, the chi-squared discretization method, which uses a chi-squared discretization method to identify the best adjacent intervals of the data and merge them to form new, larger intervals. Binarization of continuous data refers to binarizing variable features, comparing each data point to a threshold, setting data points greater than the threshold to a first fixed value (e.g., 1), and data points less than the threshold to a second fixed value (e.g., 0), thereby obtaining a binary data set. It is understood that the specific values ​​of the fixed values ​​after binarization can be adaptively adjusted depending on the application scenario. For example, for most data processing, fixed values ​​of 0 and 1 can be used, but in image processing, fixed values ​​should be set to 0 and 255.

[0070] One-hot encoding (also known as single-bit encoding) uses an N-bit register to encode N states. Each state has its own register bit, and at any given time, only one of the bits is active. That is, only one bit is 1, and the rest are zero. One-hot encoding uses 0s and 1s to represent some parameters, using an N-bit state register to encode N states. Before the advent of one-hot encoding, machine learning classifiers could not process data with unordered, discrete categorical features. This is because classifiers typically process continuous and ordered data. However, mapping tables can be created for these discrete features to make them ordered and continuous. For example, individual attribute information can be mapped as follows: gender [male, female] can be mapped to two-dimensional data (0, 1); residential location [A, B, C] can be mapped to three-dimensional data (0, 1, 2); and occupation [actor, teacher, civil servant, engineer, firefighter] can be mapped to five-dimensional data (0, 1, 2, 3, 4). Therefore, for sample A (female, place A, civil servant), a feature mapping of (1, 0, 2) can be performed. When using one-hot encoding to transform the above attribute information, according to the principle of encoding N states corresponding to N-bit registers, the above attribute information can be transformed into the following ordered and continuous form: for the gender feature [male, female], "male" can be encoded as 10, "female" can be encoded as 01, and N = 2; for the residence feature [place A, place B, place C], "place A" can be encoded as 100, "place B" can be encoded as 010, and "place C" can be encoded as 001, and N = 3; for the occupation feature [actor, teacher, civil servant, engineer, firefighter], "actor" can be encoded as 10000, "teacher" can be encoded as 01000, "civil servant" can be encoded as 00100, "engineer" can be encoded as 00010, and "firefighter" can be encoded as 00001, and N = 5. Correspondingly, the attribute information of sample A is encoded as (0, 1, 1, 0, 0, 0, 1, 0, 0).

[0071] Embedding layer: The embedding layer is a layer in the neural network structure. The embedding layer is a word embedding learned jointly with a neural network model for specific natural language processing. This embedding method encodes the words in the cleaned text into one-hot encoding, and the size or dimension of the vector space is specified as part of the model, such as 50, 100, or 300 dimensions. The vector is initialized with a small random number. The embedding layer is used at the front end of the neural network and is supervised by the backpropagation algorithm. The encoded words are mapped into word vectors, and if a multi-layer perceptron model MLP is used, the word vectors are cascaded before being input into the model. If a recurrent neural network RNN ​​is used, each word can be used as an input in the sequence. This method of learning the embedding layer requires a large amount of training data and can be slow, but it can learn and train embedding models that are both specific to specific text data and for NLP. Embedding is a type of vector representation, meaning it uses a low-dimensional vector to represent an object. This object can be a word, a product, a movie, or anything else. The property of this embedding vector is that it allows objects corresponding to closely spaced vectors to have similar meanings. For example, the distance between embedding (Avengers) and embedding (Iron Man) will be very close, while the distance between embedding (Avengers) and embedding (Gone with the Wind) will be farther. Embedding is essentially a mapping from semantic space to vector space, while simultaneously preserving the relationship between the original samples in the semantic space as much as possible. For example, two semantically similar words will be positioned relatively close in the vector space. Embedding can encode objects using low-dimensional vectors while preserving their meaning. It is often used in machine learning. When building machine learning models, objects are encoded as low-dimensional dense vectors and then passed to the DNN to improve efficiency.

[0072] In related technologies, the current reminder method is determined by analyzing historical reminder methods. For example, in an insurance renewal scenario, the current renewal reminder method is determined by analyzing the current subject's historical renewal reminder methods.

[0073] It can be seen from this that the above method for determining the reminder mode requires obtaining historical data, which results in some objects being unable to be effectively reminded due to lack of historical data.

[0074] Based on this, the embodiments of the present application provide a model training method and device, and a reminder mode prediction method and device, which can predict the reminder mode for objects without historical data, thereby improving the universality of reminder mode prediction.

[0075] The model training method and device, and reminder mode prediction method and device provided in the embodiments of the present application are specifically illustrated through the following embodiments. First, the model training method in the embodiments of the present application is described.

[0076] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to achieve optimal results.

[0077] Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interaction systems, and mechatronics. AI software technologies primarily encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0078] The model training method and device, and the reminder mode prediction method and device provided in the embodiments of the present application relate to the field of artificial intelligence technology, and in particular to the field of prediction model technology. The model training method and device, and the reminder mode prediction method and device provided in the embodiments of the present application can be applied to a terminal, can be applied to a server side, or can be software running in a terminal or a server side. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, or a smart watch, etc.; the server can be an independent server, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms; the software can be an application that implements a model training method or a reminder mode prediction method, etc., but is not limited to the above forms.

[0079] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, and the like. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, and the like that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments in which tasks are performed by remote processing devices connected via a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media, including storage devices.

[0080] It should be noted that in each specific embodiment of the present application, when it comes to the need to perform relevant processing based on data related to the user's identity or characteristics, such as user information, user behavior data, user historical data, and user location information, the user's permission or consent will be obtained first, and the collection, use, and processing of such data will comply with the relevant laws, regulations, and standards of the relevant countries and regions. In addition, when the embodiment of the present application needs to obtain the user's sensitive personal information, the user's separate permission or consent will be obtained through a pop-up window or by jumping to a confirmation page. After clearly obtaining the user's separate permission or consent, the necessary user-related data for the normal operation of the embodiment of the present application will be obtained.

[0081] Reference Figure 1 , an embodiment of the present application provides a model prediction method, which includes but is not limited to steps S110 to S170.

[0082] S110: Acquire historical reminder data of the sample object; wherein the historical reminder data includes sample object information and reminder method information of the sample object, and the reminder method information is used to remind the sample object to perform the target historical behavior through a preset reminder method;

[0083] It is understandable that the sample object is a user with known historical reminder data, and the historical reminder data is obtained through an API (Application Programming Interface) interface or the like. Among them, the historical reminder data includes sample object information used to search and match sample information feature data and sample policy feature data, as well as reminder method information. Specifically, the sample information feature data is used to characterize the individual attribute data of the sample object, and the sample policy feature data is used to characterize the attribute data of the insurance policy insured by the sample object. The reminder method information includes a plurality of identical or different preset reminder methods, and the preset reminder methods are used to characterize the channels for reminding the sample object to perform the target historical behavior, such as SMS, robot outbound calls, manual telephone calls, face-to-face interviews and other channels. The target historical behavior is used to characterize the operations expected to be performed by the sample object. For example, when the embodiment of the present application is applied to an insurance renewal reminder scenario, a renewal reminder is given to the sample user through any of the above channels, in order to expect the sample user to perform a renewal operation (i.e., the target historical behavior is a renewal operation).

[0084] S120, obtaining the validity of the preset reminder method according to the reminder method information; wherein the validity includes valid;

[0085] It is understandable that the reminder method information includes multiple identical or different preset reminder methods. When the sample user obtains a reminder to perform the target historical behavior according to the preset reminder method, the sample user will perform the target historical behavior, or not perform the target historical behavior, or perform the target historical behavior after a period of time. It can be seen from this that there is a certain correlation between the preset reminder method and the target historical behavior. The correlation is used to characterize the effectiveness of the preset reminder method information, that is, when the numerical value corresponding to the correlation is larger, it indicates that after the sample user obtains the reminder to perform the target historical behavior through the corresponding preset reminder method, the probability of performing the target historical behavior is higher, and at this time, the effectiveness of the preset reminder method is considered to be valid. When the numerical value corresponding to the correlation is smaller, it indicates that after the sample user obtains the reminder to perform the target historical behavior through the corresponding preset reminder method, the probability of performing the target historical behavior is lower, and at this time, the effectiveness of the preset reminder method is considered to be invalid.

[0086] It is understandable that for the same type of preset reminder method, when the trigger time and trigger frequency are different, the corresponding correlation degree may be different. For example, when the preset reminder method is text message, the effectiveness of reminding the sample user to perform the renewal operation through text message during working hours may be lower than the effectiveness of reminding the sample user to perform the renewal operation through text message during rest time. Or, the effectiveness of reminding the sample user to perform the renewal operation by sending one text message during rest time may be lower than the effectiveness of reminding the sample user to perform the renewal operation by sending two text messages during rest time. Therefore, according to the accuracy requirements for judging the effectiveness, the reminder method information may also include the trigger time, trigger frequency, etc. of the preset reminder method, which is not specifically limited in the embodiments of the present application.

[0087] It is understandable that the effectiveness of the preset reminder method is calculated based on the preset correlation evaluation strategy and reminder method information. For example, the effectiveness is predicted by a preset correlation evaluation model, wherein the reminder method information serves as the input data of the correlation evaluation model. Alternatively, by setting different weights for the reminder method information, the response time of the sample user, etc., the effectiveness is calculated according to the weighted summation method, which is not specifically limited in the embodiments of the present application. Among them, the response time of the sample user is used to characterize the time of reminding the sample user through the corresponding preset reminder method, and the interval time of the sample user performing the target historical behavior.

[0088] S130, filtering out a target reminder method from the preset reminder methods according to validity; wherein the target reminder method is a preset reminder method with valid validity;

[0089] It is understood that the correlation degree is calculated for multiple preset reminder methods to determine the effectiveness of the corresponding preset reminder method. The preset reminder method with effective effectiveness is used as the target reminder method. For example, when calculating the correlation degree by weighted summation, the preset reminder method with a calculated correlation degree value greater than the preset value is used as the target reminder method.

[0090] S140. Acquire sample data of the sample object according to the sample object information; wherein the sample data includes sample information feature data and sample policy feature data of the sample object;

[0091] It is understood that historical reminder data includes sample object information used to search and match sample information feature data and sample policy feature data. This sample object information is used to associate reminder method information with sample data. Specifically, sample information feature data is used to represent the individual attribute data of the sample object, such as name, age, gender, occupation, income, etc. Sample policy feature data is used to represent the attribute data of the insurance policy purchased by the sample object, such as premium, insurance type, number of payments, and whether it is annual payment.

[0092] S150. Perform vectorization processing on the sample information feature data to obtain a sample object feature vector, and perform vectorization processing on the sample policy feature data to obtain a sample policy feature vector;

[0093] It can be understood that in order to facilitate the subsequent training of the original prediction model, the original prediction model can learn the valid data in the sample information feature data and the sample policy feature data, and to improve the training speed of the original prediction model, the sample information feature data and the sample policy feature data are vectorized respectively to obtain the corresponding sample object feature vector and sample policy feature vector.

[0094] S160: Input the sample object feature vector and the sample policy feature vector into a preset original prediction model for training to obtain a prediction reminder method;

[0095] It can be understood that the sample object feature vector and the sample policy feature vector are used as input data for the original prediction model to obtain the reminder method output by the original prediction model, and the reminder method with the highest probability is used as the predicted reminder method. The predicted reminder method is the reminder method with the highest correlation predicted by the original prediction model based on analysis of the sample object feature vector and the sample policy feature vector. That is, when the corresponding sample user is reminded according to the predicted reminder method, the probability of the sample user performing the target historical behavior is the highest.

[0096] S170: Adjust parameters of the original prediction model according to the prediction reminder mode and the target reminder mode to obtain a target prediction model.

[0097] It can be understood that the target reminder method obtained according to the above steps is used as label data, and the loss value is calculated based on the label data and the predicted reminder method output by the original prediction model. The parameters of the original prediction model are adjusted according to the loss value, so as to obtain a target prediction model whose output data is close to the label data. It can be seen that the target prediction model is a model that predicts the reminder method based on the individual attribute data of the object and the attribute data of the insurance policy insured by the object. Therefore, when it is necessary to remind a new object, that is, a user for whom there is no historical data, of the target behavior (such as renewal operation), it is only necessary to obtain the individual attribute data of the new object and the attribute data of the insurance policy insured by the new object, thereby improving the universality of the reminder method prediction.

[0098] The model prediction method provided by the embodiment of the present application obtains the sample object information and reminder method information of the sample object, thereby obtaining the corresponding sample object feature vector and sample policy feature vector according to the sample object information association, and obtaining the target reminder method according to the reminder method information. Then, the preset original prediction model is trained according to the sample object feature vector and the sample policy feature vector, and the parameters of the original prediction model are adjusted according to the output data of the original prediction model (i.e., the predicted reminder method) and the target reminder method, so as to obtain the target prediction model. It can be seen that the target prediction model trained by the embodiment of the present application can predict the reminder method based on the data representing the individual attributes of the object and the attribute data of the insurance policy insured by the object, avoiding the situation in the related art where historical data (including historical reminder methods) are required to predict the reminder method, so the embodiment of the present application improves the universality of the reminder method prediction.

[0099] Reference Figure 2 In some embodiments, step S150 includes but is not limited to sub-steps S210 to S230.

[0100] S210, extracting features from the sample information feature data to obtain a sample object feature group;

[0101] It is understood that in order to obtain valid information from the sample information feature data and reduce the data processing workload of the original prediction model, feature extraction is performed on the sample information feature data to obtain a sample object feature group. For example, continuous sample information feature data is discretized. For example, if the sample information feature data includes continuous feature data such as income and age, using the distance interval method as an example, age is divided into interval feature groups such as [0, 10], [11, 21], [22, 32], and [33, 43]. Based on the sample information feature data, a feature group containing the sample feature data is searched and matched, and this feature group is used as the sample object feature group.

[0102] It is understandable that the above feature group settings can be adaptively adjusted based on the actual business scope. For example, when the embodiment of the present application is applied to the insurance scenario, the age distribution of the insured is relatively wide, but the age distribution of the policyholders performing the renewal operation may be relatively concentrated. Therefore, based on the actual application scenario, the age feature can be divided into feature groups with multiple custom intervals such as [0,25], [26,30], [31,35], [36,40], [41,50], and [50,70].

[0103] It is understandable that, for discrete sample information feature data, each type of discrete sample information feature data can be set as a sample object feature group. For example, for gender, two sample object feature groups, male and female, can be set.

[0104] S220, encoding according to the sample object feature group to obtain a sample object encoding vector;

[0105] It is understandable that since machine learning algorithms cannot process disordered discrete categorical features, it is necessary to encode the sample object feature groups obtained according to the above steps to convert multiple sample object feature groups into an ordered and continuous form. For example, when using one-hot encoding to encode the sample object feature groups, the gender feature male can be encoded as 10, the gender feature female can be encoded as 01, the age feature [0, 25] can be encoded as 100000, the age feature [26, 30] can be encoded as 010000, the age feature [31, 35] can be encoded as 001000, the age feature [36, 40] can be encoded as 000100, the age feature [41, 50] can be encoded as 000010, and the age feature [50, 70] can be encoded as 000001. From this, we can see that when the sample information feature data of the sample object is (female, 27 years old), according to the "female" sample object feature group corresponding to the sample object and the [26,30] sample object feature group, the sample object encoding vector of (0, 1, 0, 1, 0, 0, 0, 0) will be obtained.

[0106] It is understandable that the above encoding method is only exemplary. According to actual needs, other encoding methods can also be used to encode the sample object feature group, and this embodiment of the present application does not specifically limit this.

[0107] S230: Perform vectorization processing on the sample object encoding vector according to preset mapping parameters to obtain the sample object feature vector.

[0108] It is understandable that, as the encoding results of the above steps show, the more individual attribute data included in the sample information feature data, the larger the dimension of the resulting sample object encoding vector. Therefore, to reduce the computational complexity of the original prediction model, the sample object encoding vector can be subjected to dimensionality reduction. Specifically, the sample object encoding vector is input into a preset embedding layer, which then performs a matrix multiplication operation on the sample object encoding vector according to preset mapping parameters to obtain the sample object feature vector.

[0109] It is understandable that in practical applications, multiple sample object encoding vectors corresponding to each individual attribute can be input into the embedding layer, and then feature concatenation is performed to form a 1024-dimensional intermediate vector. Then, through the 768-dimensional fully connected layer, the sample object feature vector is obtained.

[0110] It is understandable that, in some embodiments, the sample policy feature data will be processed in the same manner as steps S210 to S230 to obtain a sample policy feature vector.

[0111] The embodiment of the present application vectorizes the sample object feature group corresponding to the sample information feature data and the sample policy feature group corresponding to the sample policy feature data, thereby extracting valid data from the sample information feature data and the sample policy feature data, and reducing the amount of data subsequently input into the original prediction model, thereby improving the training accuracy and training speed of the original prediction model.

[0112] Reference Figure 3 In some embodiments, step S120 includes but is not limited to sub-steps S310 to S340.

[0113] S310, obtaining the target time according to the reminder time and preset duration;

[0114] It is understandable that in the embodiment of the present application, the preset duration is used to judge the correlation between the preset reminder method and the sample user's execution of the target historical behavior. Specifically, the reminder method information includes the reminder time of the preset reminder method, and the preset duration is used to characterize the duration of judging the effectiveness of the preset reminder method, that is, the preset duration is used to limit the response time of the sample object to execute the target historical behavior after receiving the reminder through the target reminder method. For example, if the preset reminder method is SMS, the reminder time is 10:00:00 on July 25, 2022, and the preset duration is 72 hours, then the target time is 10:00:00 on July 28, 2022.

[0115] It is understandable that different preset durations can be set for different preset reminder methods based on actual needs and the characteristics of different preset reminder methods. For example, compared with machine reminder methods such as text messages and robot calls, manual reminder methods such as face-to-face interviews and manual phone calls are more effective for sample subjects. Therefore, the preset duration of manual reminder methods should be shorter than the preset duration of machine reminder methods.

[0116] Understandably, in actual applications, it's possible that a sample subject might be reminded multiple times over a period of time using a variety of similar or different preset reminder methods. Therefore, to clearly identify which preset reminder method is associated with the sample subject's historical target behavior, the preset duration should not be set too long; typically, 72 hours is acceptable.

[0117] S320, obtaining the time when the sample object executes the target historical behavior to obtain the behavior time;

[0118] It is understood that, by obtaining the feedback instruction when the sample object executes the target historical behavior through the relevant API interface or other means, the time when the feedback instruction is obtained is used as the behavior time when the sample object executes the target historical behavior. Alternatively, the behavior time can be directly obtained through the API interface or other means, and the behavior time is recorded by the corresponding terminal (or application) that responds to the sample object executing the target historical behavior.

[0119] S330, comparing the time sequence of the behavior time and the target time;

[0120] It is understood that the behavior time and the target time are compared according to the order of date, time, etc. to obtain a comparison result. Among them, the comparison result includes a first result for representing that the behavior time is before the target time, or a second result for representing that the behavior time is after the target time. Specifically, when the behavior time is the same as the target time, according to actual needs, the first result can be used as the comparison result at this time, or the second result can be used as the comparison result at this time. This embodiment of the application does not specifically limit this.

[0121] S340: If the behavior time is before the target time, the validity is valid.

[0122] It is understandable that when the comparison result is the first result, that is, after the sample subject is reminded by a preset reminder method, the response time of the sample subject is less than the preset time, then the validity of the preset reminder method is determined to be valid.

[0123] The embodiment of the present application obtains the target time through the preset duration and the reminder time, and judges the effectiveness of the preset reminder method through the target time, thereby realizing the screening of the preset reminder method, and avoiding the subsequent impact on the prediction accuracy of the target preset model when the parameters of the original prediction model are adjusted through invalid preset reminder methods.

[0124] Reference Figure 4 In some embodiments, step S140 also includes but is not limited to sub-steps S410 to S420.

[0125] S410, filtering sample information feature data from a preset first database according to the sample object identifier;

[0126] It is understood that the sample object information includes a sample object identifier used to query and match the sample information feature data. The sample object identifier can be in any form, such as a numeric code, a symbolic code, or an identifier. For example, the sample object identifier is the corresponding sample user ID number. The corresponding sample information feature data is searched and matched in a preset first database using this ID number and a related API interface, thereby associating the reminder method information with the sample information feature data.

[0127] S420: Filter and obtain sample policy characteristic data from a preset second database according to the sample policy identifier.

[0128] It is understood that the sample object information includes a sample policy identifier used to query and match the sample policy characteristic data. The sample policy identifier can be in any form, such as a digital code, a symbolic code, or an identifier. For example, the sample policy identifier is the policy ID number of the policy insured by the sample object. Using this ID number and the relevant API interface, the corresponding sample policy characteristic data is searched and matched in the preset second database, thereby associating the reminder method information with the sample policy characteristic data.

[0129] The embodiment of the present application obtains sample information characteristic data from a preset first database through a sample object identifier, and obtains sample policy characteristic data from a preset second database through a sample policy identifier, thereby avoiding the problem of difficulty in finding valid data due to a large amount of data when the sample information characteristic data, sample policy characteristic data, and reminder method information are integrated and stored.

[0130] The model training method provided in the embodiment of the present application obtains the sample object information and reminder method information of the sample object, obtains the corresponding sample object feature vector and sample policy feature vector through the sample object information association, and filters the preset reminder method through the reminder time and the preset duration to obtain the target reminder method. Then, the preset original prediction model is trained according to the sample object feature vector and the sample policy feature vector, and the parameters of the original prediction model are adjusted according to the output data of the original prediction model (i.e., the predicted reminder method) and the target reminder method, so as to obtain the target prediction model. It can be seen that the target prediction model obtained by training in the embodiment of the present application can predict the reminder method based on the data representing the individual attributes of the object and the attribute data of the insurance policy insured by the object, thereby avoiding the situation in the related art where historical data (including historical reminder methods) are required to predict the reminder method, thereby improving the universality of the reminder method prediction.

[0131] Reference Figure 5 , an embodiment of the present application further provides a reminder mode prediction method, which includes but is not limited to steps S510 to S530.

[0132] S510: Acquire target data of a target object; wherein the target data includes target object characteristic data and target policy characteristic data of the target object;

[0133] It is understandable that the target object is the user who is currently to be reminded. The target object can be a new user for whom no historical data exists, or an old user for whom historical data exists. The historical data includes the reminder method used to remind the target object to perform the target behavior, the reminder time corresponding to the reminder method, and the response time of the target object. The target data of the target object is obtained through the relevant API interface and other methods. The target data includes target object feature data used to characterize the individual attributes of the target object, such as name, age, gender, occupation, income, etc.; and target policy feature data used to characterize the attributes of the insurance policy purchased by the target object, such as premium, insurance type, number of payments, whether it is paid annually, etc.

[0134] S520: Match the target object feature data to obtain a target object feature vector, and match the target policy feature data to obtain a target policy feature vector;

[0135] It is understood that the target object feature data is matched to obtain a target object feature vector corresponding to the target object feature data from the training data of the target prediction model. Similarly, the target policy feature data is matched to obtain a target policy feature vector corresponding to the target policy feature data from the training data of the target prediction model.

[0136] S530. Input the target object feature vector and the target policy feature vector into the target prediction model for prediction to obtain the current reminder mode; wherein, the target prediction model is obtained according to the model training method of the above embodiment.

[0137] It is understood that the target object feature vector and the target policy feature vector are used as input data for the target prediction model to obtain the current reminder method. When the target object is reminded to perform the target behavior (such as renewal) using the current reminder method, the probability of the target object performing the target behavior is maximized, thereby ensuring the effectiveness of the target behavior reminder.

[0138] The reminder mode prediction method provided in the embodiment of the present application predicts the current reminder mode of the target object through the target reminder model obtained by training in the above embodiment, thereby avoiding the phenomenon in the related art that it is necessary to obtain the historical data of the target object to predict the current reminder mode. It can be seen that the reminder mode prediction method provided in the embodiment of the present application can predict the reminder mode of the target object without historical data, thereby improving the universality of the reminder mode prediction. In addition, when the target data of the target object changes, such as when the age, income, etc. change, the reminder mode prediction method provided in the embodiment of the present application will predict the new current reminder mode based on the changed target data, thereby improving the accuracy of the current reminder mode prediction.

[0139] Reference Figure 6In some embodiments, step S520 includes but is not limited to sub-steps S610 to S620.

[0140] S610: Construct a sample object vector library based on the sample object feature vectors, and construct a sample policy vector library based on the sample policy feature vectors;

[0141] It is understood that a sample object vector library is constructed based on the sample object feature vectors obtained in any of the above-mentioned model training method embodiments, and the sample object vector library includes multiple sample object feature vectors and a sample object feature group corresponding to each sample object feature vector. Similarly, a sample policy vector library is constructed based on the sample policy feature vectors obtained in any of the above-mentioned model training method embodiments, and the sample policy vector library includes multiple sample policy feature vectors and a sample policy feature group corresponding to each sample policy feature vector.

[0142] S620: Filter out target object feature vectors from the sample object vector library according to the target object feature data, and filter out target policy feature vectors from the sample policy vector library according to the target policy feature data.

[0143] It is understood that the sample object vector library is filtered based on the target object feature data, and the sample object feature vector corresponding to the sample object feature group containing the target object feature data is used as the target object feature vector. Similarly, the sample policy vector library is filtered based on the target policy feature data, and the sample policy feature vector corresponding to the sample policy feature group containing the target policy feature data is used as the target policy feature vector. For example, the sample object vector library includes multiple sample object feature groups representing age, such as [0,25], [26,30], [31,35], [36,40], [41,50], and [50,70], as well as the sample object feature vectors corresponding to these sample object feature groups. If the target object is 29 years old, the sample object feature vector corresponding to the sample object feature group [26,30] can be used as the target object feature vector.

[0144] The reminder mode prediction method provided in the embodiment of the present application obtains the current reminder mode through the target data and target prediction model of the target object, avoiding the phenomenon in related technologies that the current reminder mode needs to be predicted based on historical data, thereby enriching the types of target objects, that is, the target object can be either a new user (i.e., a user for whom no historical data exists) or an old user (i.e., a user for whom historical data exists).

[0145] Reference Figure 7 , the embodiment of the present application further provides a model training device, the model training device comprising:

[0146] The historical data acquisition module 710 is used to acquire historical reminder data of the sample object; wherein the historical reminder data includes sample object information and reminder method information of the sample object, and the reminder method information is used to remind the sample object to perform the target historical behavior through a preset reminder method;

[0147] The target reminder mode determination module 720 is configured to obtain the validity of the preset reminder mode according to the reminder mode information; wherein the validity includes valid; and to filter the target reminder mode from the preset reminder modes according to the validity; wherein the target reminder mode is the preset reminder mode whose validity is valid;

[0148] The sample data acquisition module 730 is used to acquire sample data of the sample object according to the sample object information; wherein the sample data includes sample information feature data and sample policy feature data of the sample object;

[0149] Vectorization processing module 740, configured to perform vectorization processing on the sample information feature data to obtain a sample object feature vector, and perform vectorization processing on the sample policy feature data to obtain a sample policy feature vector;

[0150] The reminder mode prediction module 750 is used to input the sample object feature vector and the sample policy feature vector into a preset original prediction model for training to obtain a predicted reminder mode;

[0151] The parameter adjustment module 760 is used to adjust the parameters of the original prediction model according to the prediction reminder mode and the target reminder mode to obtain the target prediction model.

[0152] It can be seen that the contents of the above-mentioned model prediction method embodiments are all applicable to the embodiments of the present model prediction device. The functions specifically implemented by the present model prediction device embodiments are the same as those of the above-mentioned model prediction method embodiments, and the beneficial effects achieved are also the same as those achieved by the above-mentioned model prediction method embodiments.

[0153] Reference Figure 8 The present application further provides a reminder mode prediction device, which includes:

[0154] The target data acquisition module 810 is used to acquire target data of a target object; wherein the target data includes target object characteristic data and target policy characteristic data of the target object;

[0155] Matching module 820, configured to perform matching processing on target object feature data to obtain a target object feature vector, and perform matching processing on target policy feature data to obtain a target policy feature vector;

[0156] The current reminder mode prediction module 830 is used to input the target object feature vector and the target policy feature vector into the target prediction model for prediction to obtain the current reminder mode; wherein the target prediction model is obtained according to the model training method of any of the above embodiments.

[0157] It can be seen that the contents of the above-mentioned reminder mode prediction method embodiment are all applicable to the embodiment of the present reminder mode prediction device. The functions specifically implemented by the present reminder mode prediction device embodiment are the same as those in the above-mentioned reminder mode prediction method embodiment, and the beneficial effects achieved are also the same as those achieved by the above-mentioned reminder mode prediction method embodiment.

[0158] An embodiment of the present application further provides an electronic device, including:

[0159] at least one memory;

[0160] at least one processor;

[0161] at least one program;

[0162] The program is stored in the memory, and the processor executes at least one program to implement the model training method or reminder mode prediction method described above. The electronic device can be any smart terminal including a mobile phone, a tablet computer, a personal digital assistant (PDA), an in-vehicle computer, etc.

[0163] The following combination Figure 9 The electronic device according to the embodiment of the present application is introduced in detail.

[0164] like Figure 9 , Figure 9 The hardware structure of an electronic device according to another embodiment is shown. The electronic device includes:

[0165] The processor 910 may be implemented as a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is configured to execute relevant programs to implement the technical solutions provided in the embodiments of the present application.

[0166] The memory 920 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 920 can store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 920 and is called by the processor 910 to execute the model training method or reminder mode prediction method of the embodiments of this application;

[0167] Input / output interface 930, used to implement information input and output;

[0168] Communication interface 940, used to implement communication interaction between this device and other devices, which can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WiFi, Bluetooth, etc.);

[0169] bus 950 , which transmits information between various components of the device (e.g., processor 910 , memory 920 , input / output interface 930 , and communication interface 940 );

[0170] The processor 910 , the memory 920 , the input / output interface 930 , and the communication interface 940 are connected to each other in communication within the device via a bus 950 .

[0171] An embodiment of the present application also provides a storage medium, which is a computer-readable storage medium. The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the above-mentioned model training method or reminder mode prediction method.

[0172] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0173] The embodiments described in the embodiments of this application are intended to more clearly illustrate the technical solutions of the embodiments of this application and do not constitute a limitation on the technical solutions provided by the embodiments of this application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0174] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0175] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, i.e., they may be located in one place or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of this embodiment.

[0176] Those skilled in the art will appreciate that all or some of the steps in the methods, systems, and functional modules / units in the devices disclosed above may be implemented as software, firmware, hardware, or appropriate combinations thereof.

[0177] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any 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 clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0178] It should be understood that in this application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0179] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0180] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0181] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0182] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes multiple instructions for enabling an electronic device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes various media that can store programs, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0183] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but are not intended to limit the scope of the present invention. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and essence of the present invention should be within the scope of the present invention.

Claims

1. A model training method, characterized in that: The method comprises: Acquire historical reminder data of the sample object; wherein the historical reminder data includes sample object information and reminder method information of the sample object, and the reminder method information is used to remind the sample object to perform the target historical behavior through a preset reminder method; Obtaining the validity of the preset reminder method according to the reminder method information; wherein the validity includes valid; Filtering a target reminder method from the preset reminder methods according to the validity; wherein the target reminder method is the preset reminder method with valid validity; Acquire sample data of the sample object according to the sample object information; wherein the sample data includes sample information feature data and sample policy feature data of the sample object; Performing vectorization processing on the sample information feature data to obtain a sample object feature vector, and performing vectorization processing on the sample policy feature data to obtain a sample policy feature vector; Inputting the sample object feature vector and the sample policy feature vector into a preset original prediction model for training to obtain a prediction reminder method; The parameters of the original prediction model are adjusted according to the prediction reminder mode and the target reminder mode to obtain a target prediction model.

2. The model training method according to claim 1, characterized in that The step of performing vectorization processing on the sample information feature data to obtain a sample object feature vector includes: Extracting features from the sample information feature data to obtain a sample object feature group; Encoding the sample object feature group to obtain a sample object encoding vector; The sample object encoding vector is vectorized according to preset mapping parameters to obtain the sample object feature vector.

3. The model training method according to claim 1, characterized in that The reminder method information includes the reminder time of the preset reminder method; The obtaining, according to the reminder method information, of the validity of the preset reminder method includes: Obtaining the target time according to the reminder time and preset duration; Obtaining the time when the sample object executes the target historical behavior to obtain the behavior time; comparing the temporal sequence of the behavior time and the target time; If the behavior time is before the target time, the validity is valid.

4. The model training method according to any one of claims 1 to 3, characterized in that The sample object information includes a sample object identifier and a sample policy identifier; The acquiring the sample data of the sample object according to the sample object information includes: Filtering the sample information feature data from a preset first database according to the sample object identifier; The sample policy characteristic data is obtained by screening from a preset second database according to the sample policy identifier.

5. A reminder mode prediction method, characterized in that: The method comprises: Acquire target data of a target object; wherein the target data includes target object characteristic data and target policy characteristic data of the target object; Performing matching processing on the target object feature data to obtain a target object feature vector, and performing matching processing on the target insurance policy feature data to obtain a target insurance policy feature vector; The target object feature vector and the target policy feature vector are input into a target prediction model for prediction to obtain a current reminder mode; wherein, the target prediction model is obtained according to the model training method according to any one of claims 1 to 4.

6. The reminder mode prediction method according to claim 5, characterized in that: The matching process of the target object feature data to obtain a target object feature vector and the matching process of the target policy feature data to obtain a target policy feature vector include: Construct a sample object vector library based on the sample object feature vector, and construct a sample policy vector library based on the sample policy feature vector; The target object feature vector is screened out from the sample object vector library according to the target object feature data, and the target policy feature vector is screened out from the sample policy vector library according to the target policy feature data.

7. A model training device, characterized in that: The device comprises: A historical data acquisition module, configured to acquire historical reminder data of a sample object; wherein the historical reminder data includes sample object information and reminder method information of the sample object, wherein the reminder method information is used to remind the sample object to perform a target historical behavior through a preset reminder method; a target reminder mode determination module, configured to obtain the validity of the preset reminder mode according to the reminder mode information; wherein the validity includes valid; and filter out a target reminder mode from the preset reminder modes according to the validity; wherein the target reminder mode is the preset reminder mode whose validity is valid; A sample data acquisition module, configured to acquire sample data of the sample object according to the sample object information; wherein the sample data includes sample information feature data and sample policy feature data of the sample object; A vectorization processing module, configured to perform vectorization processing on the sample information feature data to obtain a sample object feature vector, and perform vectorization processing on the sample policy feature data to obtain a sample policy feature vector; A reminder mode prediction module, configured to input the sample object feature vector and the sample policy feature vector into a preset original prediction model for training to obtain a predicted reminder mode; The parameter adjustment module is used to adjust the parameters of the original prediction model according to the prediction reminder mode and the target reminder mode to obtain a target prediction model.

8. A reminder mode prediction device, characterized in that: include: A target data acquisition module is used to acquire target data of a target object; wherein the target data includes target object characteristic data and target policy characteristic data of the target object; a matching module, configured to perform matching processing on the target object feature data to obtain a target object feature vector, and perform matching processing on the target policy feature data to obtain a target policy feature vector; The current reminder mode prediction module is used to input the target object feature vector and the target policy feature vector into the target prediction model for prediction to obtain the current reminder mode; wherein, the target prediction model is obtained according to the model training method according to any one of claims 1 to 4.

9. An electronic device, characterized in that: include: at least one memory; at least one processor; at least one computer program; The computer program is stored in the memory, and the processor executes at least one computer program to implement: The method according to any one of claims 1 to 4; or The method according to any one of claims 5 to 6.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to cause a computer to execute: The method according to any one of claims 1 to 4; or The method according to any one of claims 5 to 6.

Images