Information push method, device, electronic device and medium based on deep learning

Through deep learning-based information push method, the insurance level and effective number of browsing of uninsured users are evaluated, and high-quality uninsured customers are selected, which solves the inefficiency problem caused by data loss in traditional technology, and improves marketing success rate and insurance marketing efficiency .

CN113902576BActive Publication Date: 2024-11-26CHINA PING AN PROPERTY INSURANCE CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111267482.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-10-28
Publication Date
2024-11-26
Estimated Expiration
2041-10-28

AI Technical Summary

Technical Problem

In the prior art, when screening out insurance customers based on traditional classification algorithms, the lack of data leads to inaccurate determination of high-quality out insurance customers, resulting in low marketing success rate and insurance marketing efficiency.

Method used

Using deep learning-based information push method, by obtaining insurance-related information of uninsured users, building a target network model, evaluating the user's insurance level, and filtering out potential high-quality uninsured customers based on the user's effective browsing and user information. Push to the product account manager.

Benefits of technology

It improves marketing success rate and insurance marketing efficiency, and ensures accurate screening and effective push of high-quality off-insurance customers.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN113902576B_ABST
    Figure CN113902576B_ABST
Patent Text Reader

Abstract

The embodiment of the present application discloses a method, device, electronic device and medium for information push based on deep learning, the method comprising: after obtaining the insurance-related information of the first user who lost insurance within a specified time and did not purchase an insurance product after losing insurance, obtaining the insurance characteristic information corresponding to the insurance index according to the insurance-related information, and then importing the insurance characteristic information into the trained target network model to obtain the target insurance level of the first user, if the target insurance level is greater than the preset insurance level, determining the target insurance product information whose number of effective browsing times of the first user exceeds the preset number according to the insurance-related information, obtaining the user information of the first user, and finally sending the user information and the target insurance product information to the user device of the target user. The present invention pushes the insurance product information of potential insured customers with higher insurance levels and potential insured customers who have effectively browsed more to the product account manager through screening out from the lost insurance customers, thereby improving the success rate and efficiency of insurance marketing.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a method, device, electronic device and medium for pushing information based on deep learning. Background Art

[0002] With the rapid development of the national economy, the support of relevant policies and the further upgrading of national insurance awareness, the growth trend of domestic insurance demand is very obvious. However, insurance companies have many difficulties in re-marketing to customers who have dropped out of insurance. Customers who drop out of insurance may be dissatisfied with certain services or quotations of the company, which results in more effort and cost in marketing to these customers to keep them insured.

[0003] In the traditional marketing model, marketing personnel need to visit customers multiple times and offer more favorable prices to attract customers to buy. Based on this situation, the current screening of high-quality renewal customers for such customers who have dropped out of insurance is mainly based on traditional classification algorithms, such as SVM, KNN, etc. The lack of data will lead to inaccurate identification of high-quality dropped out customers, resulting in low marketing success rate and insurance marketing efficiency. Summary of the invention

[0004] The purpose of the present invention is to solve at least one of the above-mentioned technical defects, especially the inability to eliminate low-quality customers among the uninsured customers, so as to screen out high-quality uninsured customers and push them to product account managers, thereby improving the marketing success rate and efficiency of insurance marketing.

[0005] In a first aspect, an embodiment of the present application provides an information push method based on deep learning, the method comprising:

[0006] Acquire insurance-related information of a first user, where the first user is a user who lost insurance within a specified period and did not purchase an insurance product after losing insurance, and the insurance-related information is used to indicate the operation of the first user on the product page for insurance product information after losing insurance;

[0007] Obtaining insurance characteristic information corresponding to the insurance index according to the insurance-related information;

[0008] Importing the insurance feature information into the trained target network model to obtain the target insurance level of the first user, wherein the target insurance level refers to the insurance level of the first user, and the target network model is a network model constructed and trained using the insurance feature information of the first user and the second user as samples, wherein the second user is a user who loses insurance within a specified period and purchases an insurance product after losing insurance;

[0009] If the target insurance level is greater than the preset insurance level, determining the target insurance product information whose valid browsing times of the first user exceed the preset times according to the insurance-related information, wherein the valid browsing refers to browsing that exceeds the preset time for a single browsing;

[0010] Obtaining user information of the first user;

[0011] The user information and the target insurance product information are sent to a user device of a target user.

[0012] In one possible example, the insurance characteristic information includes the first user's historical number of insurance underwriting times, the product category of each insurance underwriting, the premium of each insurance product underwriting, the insurance period of each insurance product underwriting, the purchase channel of each insurance product underwriting, the purchaser of each insurance product underwriting, the relationship with the purchaser of each insurance product underwriting, gender, age, current address and place of origin.

[0013] In a possible example, the target network model is obtained by performing the following operations:

[0014] Extracting insurance feature information from the insurance-related information of the first user and the second user;

[0015] Extracting insurance feature information corresponding to part of the first user and part of the second user respectively, and constructing a training set;

[0016] Extract the remaining second users and the insurance feature information corresponding to the remaining second users respectively, and construct a test set;

[0017] Adopting multiple classification methods and constructing a classification model of the insurance grade to be tested according to the training set;

[0018] The test set is input into each insurance grade classification model to be tested respectively, the verification results of each insurance grade classification model to be tested are sorted, and several insurance grade classification models to be tested are selected as the insurance grade classification models according to the sorting.

[0019] In a possible example, sending the user information and the target insurance product information to a user device of a target user includes:

[0020] Determine n third users whose office addresses are closest to the current residence and / or office address of the first user, where n is a positive integer;

[0021] If n is 1, the third user is determined to be the target user.

[0022] In a possible example, after determining the n third users whose office addresses are closest to the current residence and / or office address of the first user, the method further includes:

[0023] If n is greater than or equal to 2, obtaining the points of the n third users, where the points are used to indicate the user service level of the third users;

[0024] A user with the highest score among the n third users is determined as the target user.

[0025] In a possible example, after determining the n third users whose office addresses are closest to the current residence and / or office address of the first user, the method further includes:

[0026] If n is greater than or equal to 2, obtaining the points of the n third users, where the points are used to indicate the user service level of the third users;

[0027] Determine m fourth users with the highest scores among the n third users, where m is an integer less than or equal to n;

[0028] If m is 1, the fourth user is determined to be the target user;

[0029] If m is greater than or equal to 2, obtaining the to-be-processed task amounts of the m fourth users;

[0030] Determine p fifth users with the least amount of tasks to be processed among the m fourth users as the target users, where p is an integer less than or equal to m.

[0031] In a possible example, after importing the insurance feature information into a trained target network model to obtain the target insurance level of the first user, the method further includes: if the target insurance level is less than or equal to the preset insurance level, obtaining the insurance-related information of the first user.

[0032] In a second aspect, an embodiment of the present application provides an information push device based on deep learning, the device comprising:

[0033] An insurance information acquisition unit, used to acquire insurance information of a first user, where the first user is a user who loses insurance within a specified period and does not purchase an insurance product after losing insurance, and the insurance information is used to indicate the operation of the first user on the product page for insurance product information after losing insurance;

[0034] An insurance characteristic information acquisition unit, used to obtain insurance characteristic information corresponding to the insurance index according to the insurance-related information;

[0035] an insurance grade acquisition unit, configured to import the insurance feature information into a trained target network model to obtain a target insurance grade of the first user, wherein the target insurance grade refers to the insurance grade of the first user, and the target network model is a network model constructed and trained using the insurance feature information of the first user and the second user as samples, wherein the second user is a user who loses insurance within a specified period and purchases an insurance product after losing insurance;

[0036] a determination unit, configured to determine, if the target insurance level is greater than a preset insurance level, target insurance product information whose valid browsing times by the first user exceed a preset number of times according to the insurance-related information, wherein the valid browsing refers to browsing that exceeds a preset time for a single browsing;

[0037] A user information acquisition unit, configured to acquire user information of the first user;

[0038] A communication unit is used to send the user information and the target insurance product information to a user device of a target user.

[0039] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a processor, a memory, a communication interface, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the processor, and the one or more programs include instructions for executing the steps in the method shown in the first aspect.

[0040] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium storing a computer program for electronic data exchange, wherein the computer program enables a computer to execute the steps in the method shown in the first aspect.

[0041] It can be seen that in the embodiment of the present application, after obtaining the insurance-related information of the first user who lost insurance within a specified period and did not purchase insurance products after losing insurance, the insurance feature information corresponding to the insurance index is obtained according to the insurance-related information, and then the insurance feature information is imported into the network model constructed and trained using the insurance feature information of the first user and the second user as samples to obtain the target insurance level of the first user. The second user is a user who lost insurance within a specified period and purchased insurance products after losing insurance. If the target insurance level is greater than the preset insurance level, the user information of the first user is obtained, and the target insurance product information whose effective browsing times of the first user exceed the preset times is determined according to the insurance-related information. Effective browsing refers to browsing that exceeds the preset time for a single browsing, and the user information of the first user is obtained. Finally, the user information and the target insurance product information are sent to the user device of the target user. It can be seen that by adopting the present application, potential insured customers with higher insurance levels are screened out from the customers who lost insurance, and the insurance product information that the potential insured customers have effectively browsed more is obtained, so as to push to the product account manager, thereby improving the marketing success rate and the efficiency of insurance marketing. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0043] Figure 1 An application environment diagram provided for an embodiment of the present application;

[0044] Figure 2 It is a flowchart of a method for pushing information based on deep learning provided in an embodiment of the present application;

[0045] Figure 3 It is a structural schematic diagram of an information push device based on deep learning provided in an embodiment of the present application;

[0046] Figure 4 It is a structural schematic diagram of an electronic device provided in another embodiment of the present application. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0048] The terms "first", "second", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, device, storage medium, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices.

[0049] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0050] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, 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 obtain the best results.

[0051] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0052] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0053] At present, the screening of uninsured customers who access insurance product information on the product page but do not purchase the insurance product is mainly based on traditional classification algorithms, such as SVM and KNN. Due to the lack of data, the high-quality uninsured customers finally pushed to the product account manager are not accurate enough, resulting in low marketing success rate and insurance marketing efficiency.

[0054] Based on the above problems, an embodiment of the present invention provides an information push method based on deep learning, which is described in detail as follows.

[0055] See also Figure 1 , Figure 1 FIG. 1 is an application environment diagram of an embodiment of the present invention; in this embodiment, the technical solution of the present invention can be implemented on a mobile electronic device, such as Figure 1 In the embodiment, the first electronic device 120, the second electronic device 130 and the third electronic device 140 can access the server 110 through the internet network, and the corresponding clients can be installed on the first electronic device 120, the second electronic device 130 and the third electronic device 140 respectively in the form of APP, etc. The user interacts with the server 110 by logging in to the client, and implements relevant functions on the user side on the client. In the embodiment of the present invention, the first user can log in through the first electronic device 120, the second user can log in through the second electronic device 130, and the third user can log in through the third electronic device 140; the server 110 referred to here refers to a server device that implements various background functions. Considering the actual situation of server processing at present, the device mentioned here may also refer to a cluster of one device or multiple devices. Figure 1 In the figure, three users and corresponding mobile electronic devices are used as an example to illustrate, and it can be applied to scenarios with other numbers of users in other usage scenarios.

[0056] See also Figure 2 , Figure 2 A flowchart of a method for pushing information based on deep learning is provided in an embodiment of the present application, such as Figure 2 As shown, the information push method based on deep learning specifically includes the following steps S101 to S105.

[0057] S101. Obtain insurance-related information of the first user.

[0058] Among them, the first user, also known as the uninsured user, refers to a user who loses insurance within a specified period of time and does not purchase an insurance product after losing insurance. The insurance-related information is used to indicate the first user's operations on the insurance product information on the product page after losing insurance.

[0059] The insurance-related information includes but is not limited to the first user's historical number of insurance underwritings, product categories underwritten each time, product premiums underwritten each time, insurance periods of products underwritten each time, product purchase channels underwritten each time, product purchasers underwritten each time, relationship with the product purchasers underwritten each time, gender, age, current address and at least one of native place.

[0060] S102. Obtain insurance characteristic information corresponding to the insurance index according to the insurance-related information.

[0061] Among them, the insurance characteristic information includes but is not limited to at least one of the first user's historical number of insurance underwritings, product category of each insurance underwriting, product premium of each insurance underwriting, insurance period of each insurance underwriting, product purchase channel of each insurance underwriting, product purchaser of each insurance underwriting, relationship with the product purchaser of each insurance underwriting, gender, age, current address and native place.

[0062] S103: Import the insurance feature information into the trained target network model to obtain the target insurance level of the first user.

[0063] Among them, the target insurance level refers to the insurance level of the first user, and the target network model is a network model constructed and trained using the insurance feature information of the first user and the second user as samples. The second user is a user who loses insurance within a specified time and purchases an insurance product after losing insurance.

[0064] It should be noted that the higher the insurance level, the higher the possibility that the uninsured user will be re-insured.

[0065] There is no specific limitation on the total number of insurance levels. For example, the total number of insurance levels can be 4, 6, 7, and so on.

[0066] S104: If the target insurance level is greater than a preset insurance level, determine target insurance product information whose number of valid browsing times by the first user exceeds a preset number according to the insurance-related information.

[0067] The effective browsing refers to a single browsing that exceeds a preset time.

[0068] The preset number of times may be 20 times, 30 times, 100 times, etc., and there is no specific limitation on the preset number of times.

[0069] S105: Obtain user information of the first user.

[0070] The user information includes but is not limited to the name, gender, age, current address and contact number of the first user.

[0071] S106: Send the user information and the target insurance product information to a user device of the target user.

[0072] The target user may be a product manager of insurance, that is, a promoter of insurance products.

[0073] Among them, the preset insurance level can be set according to actual needs without any specific restrictions.

[0074] It can be seen that in the embodiment of the present application, after obtaining the insurance-related information of the first user who lost insurance within a specified period and did not purchase insurance products after losing insurance, the insurance feature information corresponding to the insurance index is obtained according to the insurance-related information, and then the insurance feature information is imported into the network model constructed and trained using the insurance feature information of the first user and the second user as samples to obtain the target insurance level of the first user. The second user is a user who lost insurance within a specified period and purchased insurance products after losing insurance. If the target insurance level is greater than the preset insurance level, the user information of the first user is obtained, and the target insurance product information whose effective browsing times of the first user exceed the preset times is determined according to the insurance-related information. Effective browsing refers to browsing that exceeds the preset time for a single browsing, and the user information of the first user is obtained. Finally, the user information and the target insurance product information are sent to the user device of the target user. It can be seen that by adopting the present application, potential insured customers with higher insurance levels are screened out from the customers who lost insurance, and the insurance product information that the potential insured customers have effectively browsed more is obtained, so as to push to the product account manager, thereby improving the marketing success rate and the efficiency of insurance marketing.

[0075] In a possible example, the target network model is obtained by performing the following operations: extracting insurance feature information from insurance-related information of the first user and the second user; extracting insurance feature information corresponding to part of the first user and part of the second user respectively to construct a training set; extracting insurance feature information corresponding to the remaining second user and the remaining second user respectively to construct a test set; using multiple classification methods and constructing an insurance grade classification model to be tested based on the training set; inputting the test set into each insurance grade classification model to be tested respectively, sorting the verification results of each insurance grade classification model to be tested, and selecting several insurance grade classification models to be tested as the insurance grade classification model according to the sorting.

[0076] For example, the training process of the target network model includes: the feature input layer is connected to a fully connected layer with 256 neurons, and then the Relu function is used as the activation function; batch normalization layer and Dropout layer are used to avoid model overfitting, and the dropout parameter is set to 0.4; then two fully connected layers are connected in succession, containing 128 and 64 neurons respectively, and the Relu function is used as the activation function to complete the further abstract representation of the data features; finally, a decision layer with 6 neurons is used to determine the insurance level of each customer. The softmax function will output the probability of the customer corresponding to the 6 insurance levels. We take the highest one among the six insurance levels as the insurance level of this customer.

[0077] In one possible example, sending the user information and the target insurance product information to the user device of the target user includes: determining n third users whose office addresses are closest to the current residence and / or office address of the first user, where n is a positive integer; if n is 1, determining that the third user is the target user.

[0078] Among them, the higher the points, the higher the qualifications and stronger the ability of the insurance product account manager. The points can be determined by factors such as the insurance product account manager's signing volume, customer visits, and signing rate.

[0079] In a possible example, after determining the n third users whose office addresses are closest to the current residence and / or office address of the first user, the method further includes: if n is greater than or equal to 2, obtaining points of the n third users, the points being used to indicate the user service levels of the third users; and determining the user with the highest points among the n third users as the target user.

[0080] It can be understood that, when there are multiple users with the highest scores among the n third users, the multiple users are the target users.

[0081] In the specific implementation, if the current address of user A who accesses the insurance product information on the product page but does not purchase the insurance product is A, and the target insurance level of user A is greater than the preset insurance level, and the office addresses of insurance product account manager B, insurance product account manager C, and insurance product account manager D are all closest to A, then the points of insurance product account manager B, insurance product account manager C, and insurance product account manager D will be obtained. If the points of insurance product account manager B, insurance product account manager C, and insurance product account manager D are 100 points, 300 points, and 300 points respectively, insurance product account manager C and insurance product account manager D will be determined as target users.

[0082] It can be seen that in this example, the user information of the uninsured users with high insurance levels can be sent to the user device of the nearest most capable insurance product account manager, so that the nearest most capable insurance product account manager can serve the users with high insurance levels.

[0083] In a possible example, after determining the n third users whose office addresses are closest to the current residence and / or office address of the first user, the method further includes: if n is greater than or equal to 2, obtaining points of the n third users, the points being used to indicate the user service level of the third users; determining m fourth users with the highest points among the n third users, where m is an integer less than or equal to n; if m is 1, determining the fourth users as the target users; if m is greater than or equal to 2, obtaining the amount of pending tasks of the m fourth users; determining p fifth users with the least amount of pending tasks among the m fourth users as the target users, where p is an integer less than or equal to m.

[0084] In the specific implementation, if the current address of user A who accesses the insurance product information on the product page but does not purchase the insurance product is A, and the target insurance level of user A is greater than the preset insurance level, and the office addresses of insurance product account manager B, insurance product account manager C, and insurance product account manager D are all closest to A, then the points of insurance product account manager B, insurance product account manager C, and insurance product account manager D will be obtained. If the points of insurance product account manager B, insurance product account manager C, and insurance product account manager D are 100 points, 300 points, and 300 points respectively, then the pending task volume of insurance product account manager C and insurance product account manager D will be obtained. If the pending task volume of insurance product account manager C and insurance product account manager D are 100 and 105 respectively, then insurance product account manager D is determined to be the target user.

[0085] It can be seen that in this example, the user information of the users who have lost their insurance with a high insurance level can be sent to the user device of the nearest insurance product account manager with the strongest capabilities and relatively the least amount of tasks to be processed, so that the nearest insurance product account manager with the strongest capabilities and relatively the least amount of tasks to be processed can serve the users who have lost their insurance with a high insurance level.

[0086] In a possible example, after importing the insurance feature information into a trained target network model to obtain the target insurance level of the first user, the method further includes: if the target insurance level is less than or equal to the preset insurance level, obtaining the insurance-related information of the first user.

[0087] It can be seen that in this example, when the target insurance level is less than or equal to the preset insurance level, the insurance-related information of the first user will be continuously obtained. That is, when the uninsured user is a non-high-quality uninsured user, the insurance-related information of the uninsured user will be continuously tracked, and when he becomes a high-quality uninsured user, he can be provided with timely services.

[0088] In the case of integrated units, see Figure 3 , Figure 3A block diagram of the functional units of an information push device based on deep learning is provided. The information push device based on deep learning 200 includes:

[0089] An insurance information acquisition unit 210 is used to acquire insurance information of a first user, where the first user is a user who loses insurance within a specified period and does not purchase an insurance product after losing insurance, and the insurance information is used to indicate the operation of the first user on the product page for insurance product information after losing insurance;

[0090] The insurance characteristic information acquisition unit 220 is used to obtain the insurance characteristic information corresponding to the insurance index according to the insurance-related information;

[0091] The insurance grade acquisition unit 230 is used to import the insurance feature information into the trained target network model to obtain the target insurance grade of the first user, wherein the target insurance grade refers to the insurance grade of the first user, and the target network model is a network model constructed and trained using the insurance feature information of the first user and the second user as samples, wherein the second user is a user who loses insurance within a specified period and purchases an insurance product after losing insurance;

[0092] A determination unit 240 is configured to determine, if the target insurance level is greater than a preset insurance level, target insurance product information for which the first user has browsed effectively more than a preset number of times according to the insurance-related information, wherein the effective browsing refers to browsing that exceeds a preset time for a single browsing;

[0093] A user information acquisition unit 250, configured to acquire user information of the first user;

[0094] The communication unit 260 is used to send the user information and the target insurance product information to a user device of a target user.

[0095] The information push device 200 based on deep learning may further include a storage unit 270 for storing program code and data of the terminal. The storage unit 270 may be a memory. When the storage unit 270 is a memory, the information push device 200 based on deep learning involved in the embodiment of the present application may be Figure 4 Electronic equipment shown.

[0096] It should be noted that the specific implementation of each operation can be found in the above Figure 1 The description in the method embodiment shown will not be repeated in detail here.

[0097] It can be seen that in the embodiment of the present application, after obtaining the insurance-related information of the first user who lost insurance within a specified period and did not purchase insurance products after losing insurance, the insurance feature information corresponding to the insurance index is obtained according to the insurance-related information, and then the insurance feature information is imported into the network model constructed and trained using the insurance feature information of the first user and the second user as samples to obtain the target insurance level of the first user. The second user is a user who lost insurance within a specified period and purchased insurance products after losing insurance. If the target insurance level is greater than the preset insurance level, the user information of the first user is obtained, and the target insurance product information whose effective browsing times of the first user exceed the preset times is determined according to the insurance-related information. Effective browsing refers to browsing that exceeds the preset time for a single browsing, and the user information of the first user is obtained. Finally, the user information and the target insurance product information are sent to the user device of the target user. It can be seen that by adopting the present application, potential insured customers with higher insurance levels are screened out from the customers who lost insurance, and the insurance product information that the potential insured customers have effectively browsed more is obtained, so as to push to the product account manager, thereby improving the marketing success rate and the efficiency of insurance marketing.

[0098] In one possible example, the insurance characteristic information includes the first user's historical number of insurance underwriting times, the product category of each insurance underwriting, the premium of each insurance product underwriting, the insurance period of each insurance product underwriting, the purchase channel of each insurance product underwriting, the purchaser of each insurance product underwriting, the relationship with the purchaser of each insurance product underwriting, gender, age, current address and place of origin.

[0099] In a possible example, the target network model is obtained by performing the following operations: extracting insurance feature information from insurance-related information of the first user and the second user; extracting insurance feature information corresponding to part of the first user and part of the second user respectively to construct a training set; extracting insurance feature information corresponding to the remaining second user and the remaining second user respectively to construct a test set; using multiple classification methods and constructing an insurance grade classification model to be tested based on the training set; inputting the test set into each insurance grade classification model to be tested respectively, sorting the verification results of each insurance grade classification model to be tested, and selecting several insurance grade classification models to be tested as the insurance grade classification model according to the sorting.

[0100] In a possible example, in terms of sending the user information and the target insurance product information to a user device of a target user, the communication unit 260 is specifically configured to:

[0101] Determine n third users whose office addresses are closest to the current residence and / or office address of the first user, where n is a positive integer; if n is 1, determine the third user as the target user.

[0102] In a possible example, the communication unit 260 is also used to: after determining the n third users whose office addresses are closest to the current residence and / or office address of the first user, if n is greater than or equal to 2, obtain the points of the n third users, and the points are used to indicate the user service level of the third users; determine the user with the highest points among the n third users as the target user.

[0103] In a possible example, the communication unit 260 is also used to: after determining the n third users whose office addresses are closest to the current residence and / or office address of the first user, if n is greater than or equal to 2, obtain the points of the n third users, and the points are used to indicate the user service level of the third users; determine the m fourth users with the highest points among the n third users, where m is an integer less than or equal to n; if m is 1, determine the fourth users as the target users; if m is greater than or equal to 2, obtain the amount of pending tasks of the m fourth users; determine the p fifth users with the least amount of pending tasks among the m fourth users as the target users, where p is an integer less than or equal to m.

[0104] In a possible example, the insurance-related information acquisition unit 210 is further used to: if the target insurance level is less than or equal to the preset insurance level, acquire the insurance-related information of the first user.

[0105] See also Figure 4 , is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application, such as Figure 4 The electronic device 300 in the embodiment shown may include: a processing module 310, a memory 320, a communication interface 330, and one or more programs 321, wherein the one or more programs 321 are stored in the memory 320 and are configured to be executed by the processing module 310, and the one or more programs 321 include instructions for executing the steps in the method described below:

[0106] Acquire insurance-related information of a first user, where the first user is a user who lost insurance within a specified period and did not purchase an insurance product after losing insurance, and the insurance-related information is used to indicate the operation of the first user on the product page for insurance product information after losing insurance;

[0107] Obtaining insurance characteristic information corresponding to the insurance index according to the insurance-related information;

[0108] Importing the insurance feature information into the trained target network model to obtain the target insurance level of the first user, wherein the target insurance level refers to the insurance level of the first user, and the target network model is a network model constructed and trained using the insurance feature information of the first user and the second user as samples, wherein the second user is a user who loses insurance within a specified period and purchases an insurance product after losing insurance;

[0109] If the target insurance level is greater than the preset insurance level, determining the target insurance product information whose valid browsing times of the first user exceed the preset times according to the insurance-related information, wherein the valid browsing refers to browsing that exceeds the preset time for a single browsing;

[0110] Obtaining user information of the first user;

[0111] The user information and the target insurance product information are sent to a user device of a target user.

[0112] The processing module 310 may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0113] It should be noted that the specific implementation of each operation can be found in the above Figure 2 The description in the method embodiment shown will not be repeated in detail here.

[0114] It can be seen that in the embodiment of the present application, after obtaining the insurance-related information of the first user who lost insurance within a specified period and did not purchase insurance products after losing insurance, the insurance feature information corresponding to the insurance index is obtained according to the insurance-related information, and then the insurance feature information is imported into the network model constructed and trained using the insurance feature information of the first user and the second user as samples to obtain the target insurance level of the first user. The second user is a user who lost insurance within a specified period and purchased insurance products after losing insurance. If the target insurance level is greater than the preset insurance level, the user information of the first user is obtained, and the target insurance product information whose effective browsing times of the first user exceed the preset times is determined according to the insurance-related information. Effective browsing refers to browsing that exceeds the preset time for a single browsing, and the user information of the first user is obtained. Finally, the user information and the target insurance product information are sent to the user device of the target user. It can be seen that by adopting the present application, potential insured customers with higher insurance levels are screened out from the customers who lost insurance, and the insurance product information that the potential insured customers have effectively browsed more is obtained, so as to push to the product account manager, thereby improving the marketing success rate and the efficiency of insurance marketing.

[0115] In one possible example, the insurance characteristic information includes the first user's historical number of insurance underwriting times, the product category of each insurance underwriting, the premium of each insurance product underwriting, the insurance period of each insurance product underwriting, the purchase channel of each insurance product underwriting, the purchaser of each insurance product underwriting, the relationship with the purchaser of each insurance product underwriting, gender, age, current address and place of origin.

[0116] In a possible example, the target network model is obtained by performing the following operations:

[0117] Extracting insurance feature information from the insurance-related information of the first user and the second user;

[0118] Extracting insurance feature information corresponding to part of the first user and part of the second user respectively, and constructing a training set;

[0119] Extract the remaining second users and the insurance feature information corresponding to the remaining second users respectively, and construct a test set;

[0120] Adopting multiple classification methods and constructing a classification model of the insurance grade to be tested according to the training set;

[0121] The test set is input into each insurance grade classification model to be tested respectively, the verification results of each insurance grade classification model to be tested are sorted, and several insurance grade classification models to be tested are selected as the insurance grade classification models according to the sorting.

[0122] In one possible example, in the aspect of sending the user information and the target insurance product information to the user device of the target user, the instructions in the one or more programs 321 are specifically used to: determine n third users whose office addresses are closest to the current residence address and / or office address of the first user, where n is a positive integer; if n is 1, then determine that the third user is the target user

[0123] In one possible example, the instructions in the one or more programs 321 are also used to: after determining the n third users whose office addresses are closest to the current residence and / or office address of the first user, if n is greater than or equal to 2, obtain the points of the n third users, and the points are used to indicate the user service level of the third users; determine the user with the highest points among the n third users as the target user.

[0124] In one possible example, the instructions in the one or more programs 321 are also used to: after determining the n third users whose office addresses are closest to the current residence and / or office address of the first user, if n is greater than or equal to 2, obtain the points of the n third users, and the points are used to indicate the user service level of the third users; determine the m fourth users with the highest points among the n third users, where m is an integer less than or equal to n; if m is 1, determine the fourth users as the target users; if m is greater than or equal to 2, obtain the amount of pending tasks of the m fourth users; determine the p fifth users with the least amount of pending tasks among the m fourth users as the target users, where p is an integer less than or equal to m.

[0125] In a possible example, the insurance-related information acquisition unit is further used to: if the target insurance level is less than or equal to the preset insurance level, acquire the insurance-related information of the first user.

[0126] The present application also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the following is implemented: Figure 2 The information push method based on deep learning shown in the embodiment.

[0127] The computer-readable storage medium may be an internal storage unit of the electronic device described in any of the foregoing embodiments, such as a hard disk or memory of a control device. The computer-readable storage medium may also be an external storage device of the control device, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the control device. Furthermore, the computer-readable storage medium may also include both an internal storage unit of the control device and an external storage device. The computer-readable storage medium is used to store the computer program and other programs and data required by the control device. The computer-readable storage medium may also be used to temporarily store data that has been output or is to be output.

[0128] As an example, the computer-readable storage medium may be deployed on a computer device for execution, or deployed on multiple computer devices located at one location for execution, or executed on multiple computer devices distributed at multiple locations and interconnected through a communication network. Multiple computer devices distributed at multiple locations and interconnected through a communication network may constitute a blockchain network.

[0129] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A method for pushing information based on deep learning, characterized in that: include: Acquire insurance-related information of a first user, where the first user is a user who lost insurance within a specified period and did not purchase an insurance product after losing insurance, and the insurance-related information is used to indicate the operation of the first user on the product page for insurance product information after losing insurance; Obtaining insurance characteristic information corresponding to the insurance index according to the insurance-related information; Importing the insurance feature information into the trained target network model to obtain the target insurance level of the first user, wherein the target insurance level refers to the insurance level of the first user, and the target network model is a network model constructed and trained using the insurance feature information of the first user and the second user as samples, wherein the second user is a user who loses insurance within a specified period and purchases an insurance product after losing insurance; If the target insurance level is greater than the preset insurance level, determining the target insurance product information whose valid browsing times of the first user exceed the preset times according to the insurance-related information, wherein the valid browsing refers to browsing that exceeds the preset time for a single browsing; Obtaining user information of the first user; Sending the user information and the target insurance product information to a user device of a target user; The step of determining the target user includes: determining n third users whose office addresses are closest to the current residence address and / or office address of the first user, where n is a positive integer; if n is 1, determining the third user as the target user; The training process of the target network model includes: connecting the feature input layer with a fully connected layer having 256 neurons, using the relu function as an activation function; using a batch normalization layer and a dropout layer to avoid model overfitting, and the dropout parameter is set to 0.4; continuously connecting two fully connected layers, including 128 and 64 neurons respectively, both using the relu function as an activation function to complete further abstract representation of data features; finally using a decision layer with 6 neurons to determine the insurance grade of each customer, the softmax function will output the probability of the customer corresponding to the 6 insurance grades, and the highest one among the 6 insurance grades will be taken as the insurance grade of this customer; The target network model is obtained by performing the following operations: extracting insurance feature information from the insurance-related information of the first user and the second user; extracting the insurance feature information corresponding to part of the first user and part of the second user respectively to construct a training set; extracting the insurance feature information corresponding to the remaining second user and the remaining second user respectively to construct a test set; using multiple classification methods and constructing an insurance grade classification model to be tested based on the training set; inputting the test set into each insurance grade classification model to be tested respectively, sorting the verification results of each insurance grade classification model to be tested, and selecting several insurance grade classification models to be tested as the insurance grade classification model according to the sorting.

2. The method according to claim 1, characterized in that: After determining the n third users whose office addresses are closest to the current residence and / or office address of the first user, the method further includes: If n is greater than or equal to 2, obtaining the points of the n third users, where the points are used to indicate the user service level of the third users; A user with the highest score among the n third users is determined as the target user.

3. The method according to claim 1, characterized in that After determining the n third users whose office addresses are closest to the current residence and / or office address of the first user, the method further includes: If n is greater than or equal to 2, obtaining the points of the n third users, where the points are used to indicate the user service level of the third users; Determine m fourth users with the highest scores among the n third users, where m is an integer less than or equal to n; If m is 1, the fourth user is determined to be the target user; If m is greater than or equal to 2, obtaining the to-be-processed task amounts of the m fourth users; Determine p fifth users with the least amount of tasks to be processed among the m fourth users as the target users, where p is an integer less than or equal to m.

4. The method according to claim 1, characterized in that: After importing the insurance feature information into the trained target network model to obtain the target insurance level of the first user, the method further includes: If the target insurance level is less than or equal to the preset insurance level, the insurance-related information of the first user is obtained.

5. A device for information push based on deep learning, characterized in that: include: an insurance information acquisition unit, used to acquire insurance information of a first user, where the first user is a user who loses insurance within a specified period and does not purchase an insurance product after losing insurance, and the insurance information is used to indicate the operation of the first user on the product page for insurance product information after losing insurance; An insurance characteristic information acquisition unit, used to obtain insurance characteristic information corresponding to the insurance index according to the insurance-related information; an insurance grade acquisition unit, configured to import the insurance feature information into a trained target network model to obtain a target insurance grade of the first user, wherein the target insurance grade refers to the insurance grade of the first user, and the target network model is a network model constructed and trained using the insurance feature information of the first user and the second user as samples, wherein the second user is a user who loses insurance within a specified period and purchases an insurance product after losing insurance; a determination unit, configured to determine, if the target insurance level is greater than a preset insurance level, target insurance product information whose valid browsing times by the first user exceed a preset number of times according to the insurance-related information, wherein the valid browsing refers to browsing that exceeds a preset time for a single browsing; A user information acquisition unit, configured to acquire user information of the first user; A communication unit, configured to send the user information and the target insurance product information to a user device of a target user; The step of determining the target user includes: determining n third users whose office addresses are closest to the current residence address and / or office address of the first user, where n is a positive integer; if n is 1, determining the third user as the target user; The training process of the target network model includes: connecting the feature input layer with a fully connected layer having 256 neurons, using the relu function as an activation function; using a batch normalization layer and a dropout layer to avoid model overfitting, and the dropout parameter is set to 0.4; continuously connecting two fully connected layers, including 128 and 64 neurons respectively, both using the relu function as an activation function to complete further abstract representation of data features; finally using a decision layer with 6 neurons to determine the insurance grade of each customer, the softmax function will output the probability of the customer corresponding to the 6 insurance grades, and the highest one among the 6 insurance grades will be taken as the insurance grade of this customer; The target network model is obtained by performing the following operations: extracting insurance feature information from the insurance-related information of the first user and the second user; extracting the insurance feature information corresponding to part of the first user and part of the second user respectively to construct a training set; extracting the insurance feature information corresponding to the remaining second user and the remaining second user respectively to construct a test set; using multiple classification methods and constructing an insurance grade classification model to be tested based on the training set; inputting the test set into each insurance grade classification model to be tested respectively, sorting the verification results of each insurance grade classification model to be tested, and selecting several insurance grade classification models to be tested as the insurance grade classification model according to the sorting.

6. An electronic device, characterized in that: The method comprises a processor, a memory, a communication interface, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the processor, and the one or more programs include instructions for executing the steps in the method according to any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that: A computer program for electronic data exchange is stored, wherein the computer program enables a computer to execute the steps in the method according to any one of claims 1 to 4.