Insurance product recommendation method, computer device and medium

Abstract

The invention discloses an insurance product recommendation method, a computer device and a medium. The method comprises the steps of collecting basic information, preference data and user insurance policy information of a user; constructing a user portrait according to the basic information and the preference data; obtaining a guarantee gap between the user insurance policy information and a preset insurance product matching strategy; and analyzing the user portrait and the guarantee gap to generate an insurance product recommendation scheme. According to the method, through multi-dimensional user information collection, user requirements can be known more comprehensively; by constructing a multi-dimensional user portrait, user features can be described more accurately, and a basis is provided for personalized recommendation; by calculating the guarantee gap, the insurance demand of the user can be accurately positioned, and excessive insurance or insufficient guarantee is avoided; through comprehensive user portrait and guarantee gap analysis, the user does not need to consume a lot of time and energy to research insurance products on the market, and a personalized insurance product recommendation scheme meeting user requirements can be directly provided.

Description

Technical Field

[0001] The present disclosure relates to the technical field of insurance product recommendation, and in particular to an insurance product recommendation method, a computer device, and a medium. Background Art

[0002] With the continuous development of society and people's emphasis on future life security, the insurance industry has gradually become an indispensable part of people's daily life. However, faced with a wide variety of insurance products on the market, ordinary consumers often find it difficult to choose an insurance plan that suits their needs. Especially for families, the needs of each member are different, and how to screen out suitable insurance from many products has become a major problem. Traditional insurance recommendation methods mainly rely on manual experience, and the process is cumbersome and time-consuming; manual recommendations are difficult to fully analyze user needs, and are prone to protection gaps or duplicate insurance, resulting in poor accuracy; in addition, manual recommendations are limited to personal knowledge reserves, resulting in the inability to provide customized insurance plans based on the user's specific circumstances. Summary of the invention

[0003] In view of this, the embodiments of the present disclosure provide an insurance product recommendation method, a computer device, and a medium, which can solve the problems of low efficiency, poor accuracy, and insufficient personalization in existing insurance recommendation systems.

[0004] In a first aspect, the embodiments of the present disclosure provide a method for recommending insurance products, including the following technical solutions:

[0005] Collect users’ basic information, preference data, and user insurance policy information;

[0006] Building a user portrait based on the basic information and the preference data;

[0007] Obtain the protection gap between the user's insurance policy information and the preset insurance product matching strategy;

[0008] Analyze the user portrait and the protection gap to generate insurance product recommendation plans.

[0010] Optionally, the method for obtaining the preference data includes:

[0011] Collecting the user's operation behavior in a preset time period within a preset system; wherein the operation behavior includes browsing behavior, purchasing behavior and searching behavior;

[0012] Obtaining behaviors whose operation times are greater than a preset number threshold among the operation behaviors, and recording them as preferred behaviors;

[0013] Determine the weights corresponding to all the preferred behaviors;

[0014] Calculate a comprehensive score for each of the preferred behaviors according to the preferred behaviors and their corresponding weights;

[0015] The preferred behavior with the largest comprehensive score is taken as the target behavior, and preference data is obtained based on the target behavior.

[0016] Optionally, constructing a user portrait according to the basic information and the preference data includes:

[0017] Extracting age characteristics, gender characteristics, occupation characteristics, educational background characteristics, income level characteristics, marital status characteristics, number of children characteristics, home address characteristics, height characteristics, weight characteristics, living habits characteristics, and medical record characteristics from the basic information;

[0018] Extracting browsing content type features, browsing frequency features, browsing duration features, browsing preference features, or purchased commodity category features, purchase frequency features, purchase price range features, purchase preference features, or search keyword features, search frequency features, search result click characteristics, search preference features from the preference data;

[0019] Take each feature as a dimension and construct a multidimensional vector;

[0020] Get the weight of each feature and the interaction weights between different features;

[0021] Determine the eigenvalue corresponding to each eigenvalue according to the preset eigenvalue mapping database;

[0022] Performing standardization on the feature values ​​to obtain a standardization function for each feature;

[0023] Obtaining an evaluation result according to the normalization function, the weight of each feature, and the interaction weights between different features;

[0024] Constructing a user profile based on the evaluation results;

[0025] The evaluation result is Score(U):

[0026]

[0027]

[0028] Among them, ω i is the weight of each feature, v i is the eigenvalue corresponding to the i-th feature, n is the total number of features, μ i is the preset mean of the corresponding feature, σ i is the preset standard deviation of the corresponding feature, ω ij is the interaction weight between two features with an interactive relationship, σ(v i ),σ(v j) are the normalized functions of two features with interactive relationship.

[0029] Optionally, obtaining the protection gap between the user's insurance policy information and a preset insurance product matching strategy includes:

[0030] Extract key fields included in the user's existing insurance products based on the user's insurance policy information;

[0031] Set different values ​​for each key field. Each recommended insurance product includes one or more key fields with specific values. Build a mapping relationship between the key fields of insurance products and the recommended insurance products.

[0032] Configure a recommendation condition for each recommended insurance product;

[0033] Recommend recommended insurance products that meet the recommended conditions to users;

[0034] The protection gap is determined based on the user's insurance policy information and the recommended insurance products.

[0035] Optionally, configuring a recommendation condition for each recommended insurance product includes:

[0036] When the recommended insurance products include: insurance types covering outpatient expenses, hospitalization expenses, and surgical expenses, and the insured amount is set at 10%-20% of the annual income, the recommended conditions include: the user is between 18 and 65 years old, the user has no major past medical history, and the user's annual income is less than 100,000 yuan;

[0037] When the recommended insurance products include: insurance types that include a global medical network, outpatient care, hospitalization, physical examinations and emergency rescue, and the insured amount is set at 50%-100% of the annual income, the recommendation conditions include that the user is between 30 and 50 years old, the user's annual income is more than 500,000 yuan, and the user expects to receive medical treatment in a private hospital or an international hospital.

[0038] Optionally, the determining the protection gap according to the user's insurance policy information and the recommended insurance product includes:

[0039] Building a preset decision tree model based on the key fields, and not performing a pruning operation in the process of building the initial decision tree model;

[0040] Taking the recommended insurance products and the user's existing insurance products as input data;

[0041] Predicting the input data using the preset decision tree model to obtain multiple prediction results;

[0042] The category that appears most frequently in the multiple prediction results is used as the protection gap.

[0043] Optionally, constructing a preset decision tree model based on the key fields includes:

[0044] Calculate the splitting criteria of all the key fields, determine the best key field, and record it as the target feature; wherein the splitting criteria include information gain or Gini index; the best key feature refers to the feature with the largest information gain or the smallest Gini index;

[0045] Constructing an initial decision tree model based on the target features;

[0046] Using different types of historical policy data to train the initial decision tree model respectively to obtain multiple trained initial decision tree models;

[0047] The multiple trained initial decision tree models are integrated through ensemble learning to obtain the preset decision tree model.

[0048] Optionally, analyzing the user profile and the protection gap to generate an insurance product recommendation plan includes:

[0049] Building an initial model including multiple historical user decision tree models;

[0050] Training the initial model, and using the trained initial model as a personalized recommendation model;

[0051] Based on the personalized recommendation model, respectively extract the portrait features of the user portrait and the gap features of the security gap;

[0052] Based on the image feature and the gap feature, a feature vector is obtained;

[0053] Reasoning the feature vector based on a decision tree or ensemble learning method to obtain several reference solutions and their corresponding recommendation probabilities;

[0054] The reference plan with the highest recommendation probability is used as the recommended plan for the insurance product.

[0055] In a second aspect, the embodiments of the present disclosure further provide an insurance product recommendation system, including the following solutions:

[0056] User data collection module, used to collect user basic information, preference data, and user insurance policy information;

[0057] A construction module, used to construct a user portrait based on the basic information and the preference data;

[0058] A protection gap acquisition module is used to obtain the protection gap between the user's insurance policy information and the preset insurance product matching strategy;

[0059] The personalized recommendation module is used to analyze the user portrait and the protection gap and generate insurance product recommendation plans.

[0060] In a third aspect, the embodiments of the present disclosure further provide a computer device, which adopts the following technical solution:

[0061] The computer device comprises:

[0062] at least one processor; and,

[0063] a memory communicatively connected to the at least one processor; wherein,

[0064] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute any of the above-mentioned insurance product recommendation methods.

[0065] In a fourth aspect, an embodiment of the present disclosure further provides a computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable a computer to execute any of the above-mentioned insurance product recommendation methods.

[0066] In a fifth aspect, an embodiment of the present disclosure further provides a computer program product, including a computer program / instruction, which implements the steps of any of the above methods when executed by a processor.

[0067] The insurance product recommendation method disclosed in the present application can more comprehensively understand the user's needs, risk preferences and life status through multi-dimensional user information collection; by constructing a multi-dimensional user portrait, it can more accurately describe the user's characteristics and provide a basis for personalized recommendations; by calculating the protection gap, it can accurately locate the user's insurance needs to avoid over-insurance or under-insurance; through comprehensive user portrait and protection gap analysis, the recommended products are more in line with the user's actual needs; users do not need to spend a lot of time and energy to research insurance products on the market, and can directly provide the best recommendations; according to the user's specific needs and characteristics, providing personalized insurance product recommendation plans that meet the user's needs can improve the user's acceptance and willingness to purchase the recommended products, thereby improving the conversion rate. Users can quickly find insurance products that meet their needs, improve the overall user experience, and help establish long-term customer relationships.

[0068] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the following preferred embodiments are specifically cited and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0070] Figure 1 A flowchart of the insurance product recommendation method provided in an embodiment of the present disclosure.

[0071] Figure 2 A flowchart of a method for obtaining preference data provided in an embodiment of the present disclosure.

[0072] Figure 3 A flowchart of a method for constructing a user portrait provided in an embodiment of the present disclosure.

[0073] Figure 4 A flowchart of a method for obtaining a protection gap provided in an embodiment of the present disclosure.

[0074] Figure 5 for Figure 4 A flowchart of a method for determining protection gaps based on user policy information and recommended insurance products.

[0075] Figure 6 A flowchart of a method for constructing a preset decision tree model based on key fields provided in an embodiment of the present disclosure.

[0076] Figure 7 A flowchart of a method for generating an insurance product recommendation scheme provided in an embodiment of the present disclosure.

[0077] Figure 8 A functional block diagram of the insurance product recommendation system provided in an embodiment of the present disclosure.

[0078] Fig. 9 A schematic diagram of the structure of a computer device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0079] The embodiments of the present disclosure are described in detail below with reference to the accompanying drawings.

[0080] It should be clear that the following embodiments of the present disclosure are described by specific specific examples, and those skilled in the art can easily understand other advantages and effects of the present disclosure from the contents disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present disclosure, rather than all of the embodiments. The present disclosure can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present disclosure. It should be noted that the following embodiments and features in the embodiments can be combined with each other in the absence of conflict. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present disclosure.

[0081] It should be noted that various aspects of the embodiments within the scope of the appended claims are described below. It should be apparent that the aspects described herein may be embodied in a wide variety of forms, and any specific structure and / or function described herein is merely illustrative. Based on the present disclosure, it should be understood by those skilled in the art that an aspect described herein may be implemented independently of any other aspect, and two or more of these aspects may be combined in various ways. For example, any number of aspects described herein may be used to implement the device and / or practice the method. In addition, other structures and / or functionalities other than one or more of the aspects described herein may be used to implement this device and / or practice this method.

[0082] It should also be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present disclosure. The drawings only show components related to the present disclosure rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0083] Additionally, in the following description, specific details are provided to facilitate a thorough understanding of the examples. However, it will be understood by those skilled in the art that the aspects described may be practiced without these specific details.

[0084] Reference Figure 1 , the present application discloses an insurance product recommendation method, which specifically includes:

[0085] S100, collect user's basic information, preference data, and user insurance policy information.

[0086] Among them, basic information includes age, gender, occupation, educational background, income level, marital status, number of children, home address, height, weight, living habits, and medical records.

[0087] Reference Figure 2 , the method of obtaining preference data includes:

[0088] A100, collects the user's operation behavior in a preset time period within a preset system; wherein the operation behavior includes browsing behavior, purchasing behavior and searching behavior.

[0089] By recording the user's operating behavior, we can obtain detailed information about the user's activities in the system, providing a data basis for subsequent preference analysis.

[0090] Among them, browsing behavior includes the number of times and length of time users browse different pages in the system and view the insurance product details page; purchasing behavior includes the type, amount, and time of purchase of insurance products purchased by users; search behavior includes the keywords users search for in the system, search frequency, and clicks on search results.

[0091] By collecting a variety of operation behaviors, we can fully understand the user's behavior trajectory in the system, capture the user's behavior patterns and preferences, and provide a rich data foundation for subsequent behavior data analysis and user preference mining.

[0092] A200, obtain the behavior whose operation times are greater than the preset times threshold in the operation behavior, and record it as the preferred behavior.

[0093] By setting thresholds, users' frequently performed operations can be screened out, and these behaviors can better reflect the users' true preferences.

[0094] A300, determine the weights corresponding to all preferred behaviors.

[0095] Specifically, a weight can be assigned to each preferred behavior according to the type and importance of the behavior. For example, the weight of purchase behavior can be higher than that of browsing behavior and search behavior. By assigning weights to different types of behaviors, the contribution of various behaviors to user preferences can be quantified.

[0096] Make the analysis more scientific.

[0097] A400, based on the preferred behaviors and their corresponding weights, calculates the comprehensive score of each preferred behavior.

[0098] Specifically, the weighted average method or other suitable algorithms can be used to calculate the comprehensive score of each preference behavior, for example, comprehensive score = number of operations × weight. By calculating the comprehensive score, the user's preference behavior can be comprehensively evaluated to avoid the deviation caused by single data.

[0099] A500,takes the preferred behavior with the largest comprehensive score as the target behavior, and obtains the preference data based on the target behavior.

[0100] By selecting the preference behavior with the largest comprehensive score, we can accurately identify the user's main preferences and provide key data for subsequent personalized recommendations. Based on the preference data of the target behavior, we can generate personalized recommendation plans that are more in line with user preferences, thereby improving recommendation accuracy and user satisfaction.

[0101] Furthermore, the preference data of browsing behavior may include analyzing which pages or content the user is interested in based on the user's browsing behavior, thereby forming browsing preference data; the preference data of purchasing behavior may include analyzing the product category, frequency, amount, etc. purchased by the user based on the user's purchasing behavior, thereby forming purchasing preference data; the preference data of search behavior may include analyzing the user's frequently used keywords and search result clicks based on the user's search behavior, thereby forming search preference data.

[0102] S200, construct a user profile based on basic information and preference data.

[0103] specifically refer to Figure 3 , the construction methods of user portraits include:

[0104] S210, extracting age characteristics, gender characteristics, occupation characteristics, educational background characteristics, income level characteristics, marital status characteristics, number of children characteristics, home address characteristics, height characteristics, weight characteristics, living habits characteristics, and medical record characteristics from the basic information.

[0105] S220, extracting browsing content type features, browsing frequency features, browsing duration features, browsing preference features, or purchased product category features, purchase frequency features, purchase price range features, purchase preference features, or search keyword features, search frequency features, search result click characteristics, and search preference features from the preference data.

[0106] S230, taking each feature as a dimension and constructing a multi-dimensional vector.

[0107] By extracting detailed basic information features and preference data features to construct a multidimensional vector, we can comprehensively analyze user characteristics from multiple angles and improve the completeness and accuracy of user portraits.

[0108] S240, obtaining the weight of each feature and the interaction weights between different features.

[0109] Specifically, historical data or expert evaluation can be used to determine the initial weight of each feature. For example, the weight of income level may be higher than that of lifestyle characteristics. The relationship between different features can be analyzed to determine the interaction weights between features. For example, there is an interaction between age and health status. By assigning weights to each feature and the interaction between features, the contribution of various features to the user profile can be quantified.

[0110] S250, determining a feature value corresponding to each feature according to a preset feature value mapping database.

[0111] Specifically, a preset feature value mapping database can be established to map each feature to a specific value or level. For example, age features can be mapped to specific age values, and occupational features can be mapped to occupational levels. Through feature value mapping, different types of features can be unified into standardized values ​​to facilitate subsequent calculations and analysis.

[0112] S260, normalizing the feature values ​​to obtain a normalization function for each feature.

[0113] Specifically, the i-th feature normalization function is σ(v i ):

[0114] Among them, v i is the eigenvalue corresponding to the i-th feature, μ i is the preset mean of the corresponding feature, σ i is the preset standard deviation of the corresponding feature.

[0115] S270, obtaining an evaluation result according to the normalization function, the weight of each feature, and the interaction weights between different features.

[0116] Specifically, the evaluation result is Score(U):

[0117]

[0118] Among them, ω i is the weight of each feature, n is the total number of features, ω ij is the interaction weight between two features with an interactive relationship, σ(v i ),σ(v j ) are two YY+242207P with interactive relationship

[0119] The normalization function of the features.

[0120] By considering the weights of features and interaction weights, the overall characteristics of users can be comprehensively evaluated to generate more accurate evaluation results.

[0121] S280, construct a user portrait based on the evaluation results.

[0122] Specifically, the evaluation results can be classified using machine learning algorithms (such as K-means clustering, decision tree, random forest, etc.), and the classification results can be mapped to predefined user labels, such as "high-risk user", "medium-risk user", "low-risk user", "high-value user", "potential user", etc.

[0123] Furthermore, the features in the basic information and preference data can be mapped to specific user portrait dimensions; for example, age features can be mapped to the "age group" dimension, and purchase preference features can be mapped to the "product preference" dimension; according to business needs, more user portrait dimensions can be expanded, such as "risk tolerance", "security needs", "investment preferences", etc. Through feature mapping and dimension expansion, a multi-dimensional user portrait can be constructed to fully reflect the characteristics and needs of users.

[0124] S300, obtaining the protection gap between the user's insurance policy information and the preset insurance product matching strategy.

[0125] Specific reference Figure 4 , the methods for obtaining the protection gap include:

[0126] S310, extracting key fields included in the user's existing insurance products based on the user's insurance policy information.

[0127] Specifically, key fields may include insurance type (such as life insurance, health insurance, car insurance, etc.), coverage (such as death protection, disease protection, medical protection, etc., insured amount (i.e. the protection amount of the insurance product), premium (i.e. the premium that the user needs to pay annually or monthly), policyholder information (i.e. the policyholder's name, age, gender, etc.), insured person information (i.e. the insured's name, age, gender, etc.), insurance period (i.e. the validity period of the insurance product), and payment method (such as annual payment, monthly payment, etc.).

[0128] In this step, by extracting detailed key fields, we can fully understand the user's existing insurance products and their coverage.

[0129] S320, setting different values ​​for each key field, each recommended insurance product includes one or more key fields with specific values, and constructing a mapping relationship between the key fields of the insurance product and the recommended insurance product.

[0130] By setting standardized values ​​for each key field, subsequent matching and analysis are facilitated; by building a clear mapping relationship, user needs and recommended insurance products can be quickly matched.

[0131] S330, configuring a recommendation condition for each recommended insurance product.

[0132] Specifically, when the recommended insurance products include: insurance types that cover outpatient expenses, hospitalization expenses, and surgical expenses, and the insured amount is set at 10%-20% of the annual income, the recommended conditions include: the user is between 18 and 65 years old, the user has no major past medical history, and the user's annual income is less than 100,000 yuan;

[0133] When the recommended insurance products include: insurance types that cover global medical networks, outpatient clinics, hospitalization, physical examinations and emergency rescue, and the insured amount is set at 50%-100% of annual income, the recommended conditions include that the user is between 30 and 50 years old, the user's annual income is more than 500,000 yuan, and the user expects to receive medical treatment in private hospitals or international hospitals.

[0134] By configuring precise recommendation conditions, we can ensure that the recommended insurance products meet the actual needs of users.

[0135] S340, recommending recommended insurance products that meet the recommendation conditions to the user.

[0136] Specifically, recommendation algorithms (such as collaborative filtering, content recommendation, rule-based recommendation, etc.) can be used to recommend insurance products that meet the recommendation conditions to users, and the recommendation results can be displayed to users, including product name, coverage, insured amount, premium, insurance period and other information.

[0137] S350, determines the protection gap based on the user’s policy information and recommended insurance products.

[0138] Reference Figure 5 , based on the user's policy information and recommended insurance products, the method of determining the protection gap specifically includes:

[0139] B100, builds a preset decision tree model based on key fields, and does not perform pruning operations during the construction of the initial decision tree model.

[0140] Specifically, these key fields are used to construct an initial decision tree model, using the CART (Classification and Regression Tree) algorithm or other suitable decision tree algorithms; in the process of constructing the initial decision tree model, no pruning operation is performed to retain the complexity of the model and ensure that more details and features can be captured.

[0141] B200,takes the recommended insurance products and the user’s existing insurance products as input data,,predicts the input data through the preset decision tree model, and obtains multiple,prediction results.

[0142] For example, recommended insurance product A: Insurance type = health insurance, coverage = disease coverage, insured amount = 500,000 yuan, premium = 5,000 yuan / year, insurance period = 10 years.

[0143] The user already has insurance product B: Insurance type = life insurance, Coverage = death protection, Sum insured = RMB 300,000, Premium = RMB 2,000 / year, Insurance period = lifetime.

[0144] By clarifying the input data, we can ensure that the decision tree model can accurately identify and process the differences between the recommended insurance products and the user's existing insurance products; through the predictions of multiple initial decision tree models, we can obtain diverse and comprehensive prediction results, increase the comprehensiveness and reliability of the analysis, and the prediction results of multiple models can complement and verify each other, thereby improving the accuracy and stability of the prediction results.

[0145] B300, takes the category that appears most frequently in multiple prediction results as the protection gap.

[0146] Count multiple prediction results and calculate the number of occurrences of each prediction category; select the prediction category with the most occurrences as the user's protection gap. If multiple categories have the same number of occurrences, you can select one of them or conduct further analysis.

[0147] By counting the number of occurrences of prediction results, we can determine the user's main protection gaps and ensure the targetedness and effectiveness of the recommended plan. The categories with the highest number of occurrences usually represent the consensus and majority opinion of the model, which can improve the reliability and credibility of decision-making. The determination of protection gaps can help optimize the recommendation strategy of insurance products, fill the user's protection gaps, and improve protection coverage.

[0148] Among them, the protection gap includes one or more of the health protection gap, property protection gap, life protection gap, and accident protection gap.

[0149] Further, refer to Figure 6 ,The method of building a preset decision tree model based on key fields includes:

[0151] B111, calculate the splitting criteria of all key fields, determine the best key field, and record it as the target feature; wherein the splitting criteria include information gain or Gini index; the best key feature refers to the feature with the largest information gain or the smallest Gini index.

[0152] Specifically, for information gain, the information gain (Information Gain) of each key feature is calculated. The larger the information gain, the better the feature splitting effect; for Gini index, the Gini index (GiniIndex) of each key feature is calculated. The smaller the Gini index, the better the feature splitting effect.

[0153] By calculating the splitting criteria, the optimal features can be accurately selected to ensure the accuracy and effectiveness of the decision tree model.

[0154] B112, build an initial decision tree model based on the target features.

[0155] By clarifying the target features, a clearly structured initial decision tree model can be constructed, which facilitates subsequent training and optimization.

[0156] Specifically, the target features are used as nodes, and data segmentation is performed recursively to build an initial decision tree model. During the data segmentation process, the target features are selected each time for node splitting until the stopping condition is met (such as reaching the preset tree depth, the sample size is too small, etc.).

[0157] In the process of building the initial decision tree model, no pruning operation is performed, all nodes and branches are retained to ensure the maximum depth of each tree, thereby capturing more sample features.

[0158] Without pruning operations, all the information and complex structure of the decision tree model can be retained, ensuring that the model can capture the complex relationship of the data. The complexity of the initial decision tree model can improve the model's expressiveness and predictive capabilities, providing a basis for subsequent integrated learning.

[0159] Furthermore, for classification problems, majority voting is used to determine the final prediction category; for regression problems, the average of all decision tree prediction values ​​is used as the final prediction result.

[0160] B113, using different types of historical policy data to train the initial decision tree model respectively, to obtain multiple trained initial decision tree models.

[0161] By using different types of historical policy data for training, the diversity and generalization ability of the model can be improved and overfitting can be avoided.

[0162] Specifically, historical policy data can be divided into multiple types (such as by policy type, user type, time period, etc.), and each type corresponds to a set of training data; for each type of training data, an initial decision tree model is used for training to generate multiple trained initial decision tree models; different types of training data can train initial decision tree models with diversity, which can better adapt to different types of user data and protection needs, and improve the adaptability and flexibility of the model.

[0163] Furthermore, you can use cross-validation or holdout method to evaluate the performance of the model. The main indicators include accuracy, precision, recall, F1 score, etc.; adjust hyperparameters, such as the number of decision trees (n_estimators), the maximum depth of each tree (max_depth), the minimum number of sample splits (min_samples_split), etc., to optimize model performance.

[0164] B114, integrating multiple trained initial decision tree models through ensemble learning to obtain a preset decision tree model.

[0165] Specifically, ensemble learning methods (such as bagging, boosting, random forest, etc.) can be used to integrate multiple trained initial decision tree models; through ensemble learning, the prediction results of multiple models are combined to generate the final preset decision tree model; in the integration process, voting, weighted averaging, etc. can be used to combine the prediction results of multiple models to improve the accuracy and stability of the model.

[0166] Through ensemble learning, the prediction results of multiple models can be combined to significantly improve the accuracy and prediction ability of the model. Ensemble learning can reduce the errors and deviations of a single model and enhance the stability and robustness of the model. The integration of preset decision tree models can optimize the decision-making process and generate more accurate and reliable prediction results, providing strong support for the determination of security gaps.

[0167] S400 analyzes user profiles and protection gaps to generate insurance product recommendations.

[0168] Specific reference Figure 7 , a method for generating an insurance product recommendation scheme, comprising:

[0169] S410, constructing an initial model including multiple historical user decision tree models.

[0170] Specifically, based on historical user data, multiple historical user decision tree models are constructed. Each decision tree model is generated based on the data of specific historical users and can capture the needs and behavior patterns of different user groups. By constructing multiple historical user decision tree models, the diversity and comprehensiveness of the initial model are ensured, providing a rich data foundation for subsequent personalized recommendations.

[0171] S420: Train the initial model and use the trained initial model as a personalized recommendation model.

[0172] By training the initial model, a personalized recommendation model is generated, which means that the model can understand user needs and preferences more deeply, thereby providing more accurate recommendations.

[0173] Train the initial model and use the initial model that meets the preset indicators as the personalized recommendation model;

[0174] Among them, the preset indicators include any one of the prediction accuracy of the model on the training set or the validation set, the proportion of user groups or scenarios covered by the model, and the complexity of the model.

[0175] By screening preset indicators (such as prediction accuracy), the initial model with the best performance can be selected to improve the accuracy and prediction ability of the personalized recommendation model; specifically, consideration of coverage ratio can ensure that the personalized recommendation model is applicable to a wider range of user groups and scenarios, thereby improving the applicability and popularity of the model; limiting the complexity of the model can reduce the computing resource requirements and operating costs of the model, ensuring efficient operation and rapid response of the model; personalized recommendation models that meet preset indicators can better meet the personalized needs of users, provide accurate and effective recommendation solutions, and significantly improve user experience and satisfaction.

[0176] S430, extracting the portrait features of the user portrait and the gap features of the security gap based on the personalized recommendation model.

[0177] By extracting portrait features, we can fully understand the user's personal attributes and behavioral habits, providing a comprehensive data basis for personalized recommendations; by extracting gap features, we can accurately identify the user's protection gaps and provide a basis for recommending suitable insurance products.

[0178] S440, obtaining a feature vector based on the image feature and the gap feature.

[0179] Specifically, the user portrait features and gap features are combined into a feature vector as input data for subsequent reasoning and analysis, ensuring that the dimension of the feature vector is moderate to avoid overfitting or underfitting. By constructing the feature vector, the user portrait and guarantee gaps can be comprehensively analyzed to improve the comprehensiveness and accuracy of the recommendation plan. The feature vector form enables the model to efficiently process and analyze data, simplify the input process, and improve the operating efficiency of the model.

[0180] Steps S430 and S440 combine user portraits and guarantee gap analysis to extract key features and convert them into feature vectors. This process ensures that the recommended solutions can be closely centered around the user's actual situation and needs.

[0181] S450, inferring the feature vector based on a decision tree or an integrated learning method to obtain several reference solutions and their corresponding recommendation probabilities.

[0182] Use decision trees or ensemble learning methods (such as random forests) to infer feature vectors and generate several reference plans and their corresponding recommendation probabilities. Each reference plan corresponds to a recommendation probability, indicating the possibility of the plan being recommended. Through model reasoning, multiple different reference plans can be generated to increase the diversity and selectivity of recommended plans. The recommendation probability corresponding to each reference plan can quantify the recommendation priority of the plan and help users and recommendation systems make decisions. The introduction of recommendation probability can optimize the recommendation strategy and ensure the accuracy and effectiveness of the recommended plan. Ensemble learning methods (such as random forests) can improve the robustness and generalization ability of the model, reduce the risk of overfitting, and improve the reliability of recommendation results.

[0183] S460, the reference plan with the highest recommendation probability is used as the recommended plan for the insurance product.

[0184] The solution with the highest recommendation probability usually represents the optimal recommendation of the model, which can accurately meet the needs of users, improve the accuracy of the recommended solution and user satisfaction. By recommending the optimal solution, it can significantly improve user acceptance and purchase conversion rate, improve business benefits, and accurately recommend solutions to optimize user experience, reduce user selection time and decision-making costs, and improve user satisfaction and loyalty.

[0185] The entire recommendation process is user-centric, from user profiling to protection gap analysis to final product recommendations, all of which embody the concept of personalized service and help improve user experience. Through accurate recommendations, this method can better meet users' insurance needs, thereby improving user satisfaction and loyalty.

[0186] Furthermore, in the process of generating insurance product recommendation plans, adjustments are made according to the actual situation of the protection gap to ensure that the recommended plans can effectively fill the protection gap. The specific steps are as follows: First, analyze the coverage, insured amount, and term of the existing insurance to identify which risks or needs are not covered or met; adjust the type of insurance products in the recommended plan for the identified protection gap, for example, if there is a lack of major disease protection, it is recommended to add major disease insurance; adjust the insured amount and insurance period in the recommended plan according to the specific situation of the protection gap, for example, if the existing insurance amount is insufficient, increase the insurance amount; if the insurance period is insufficient, extend the insurance period; for specific risks, such as occupational risks and regional risks, adjust the recommended plan and add corresponding protection products; if the existing insurance fails to cover all family members, adjust the recommended plan to increase the protection of uncovered members; comprehensively consider the user portrait and protection gap, optimize the recommended plan, and ensure the comprehensiveness and rationality of the plan. Through the above steps, the generated insurance product recommendation plan can accurately fill the user's protection gap and provide personalized insurance protection suggestions.

[0187] Furthermore, the insurance product recommendation method disclosed in the present application also includes: using statistical analysis and simulation techniques (such as Monte Carlo simulation) to evaluate the user's future risk exposure and possible claims situation; collecting user feedback information on the recommended plan, obtaining an improved model, and obtaining a recommendation strategy based on the improved model.

[0188] Among them, using statistical analysis and simulation techniques (such as Monte Carlo simulation) to evaluate users' future risk exposure and possible claims specifically includes: 1) defining risk events: determining the types of risk events that need to be evaluated, such as accidents, diseases, and unemployment; 2) determining the probability distribution of risk events: based on historical data and industry experience, determining the probability distribution of each risk event; assuming that the probability of accident events follows a uniform distribution, the probability of disease events follows a normal distribution, and the probability of unemployment events follows a Poisson distribution. 3) setting simulation parameters: defining the number of simulations (such as 10,000 times) to ensure the reliability of the results.

[0189] Furthermore, the method for obtaining the improved model specifically includes: 1) collecting detailed information of users (such as age, family status, health status, existing policy type and insured amount, etc.) and constructing a feature matrix X and corresponding protection gap labels y; wherein each row represents a user, each column represents a feature, and the matrix elements represent the values ​​of the user features.

[0190] 2) Initialization parameters: Determine the initial parameters of the decision tree and random forest models, such as the number of trees (n_estimators), maximum depth (max_depth), minimum number of sample splits (min_samples_split), etc.

[0191] 3) Define the model optimization strategy: Use cross-validation to evaluate model performance, and balance model complexity and bias by adjusting the regularization parameter a of the decision tree and the number of trees n_estimators of the random forest. For each set of parameters, divide the dataset into K subsets, use one of them as the test set in turn, and the rest as the training set, and calculate the average performance of the model.

[0192] Among them, calculating the model performance score each time and finding the average value includes:

[0193] Among them, score i is the validation performance score of the model on the i-th subset.

[0194] 4) Iteratively reduce the error: Optimize the model by traversing the parameter space through grid search. Grid search will try different parameter combinations to find the optimal parameter setting to minimize the average error E. cv .

[0195]

[0196] Finally, the norm error between the predicted rating matrix R1 and the actual rating matrix R0 is calculated to optimize the model.

[0197] Based on the optimal parameter model obtained through training, the user's protection gap can be predicted and a personalized insurance recommendation plan can be generated. Based on the prediction results, a personalized recommendation list is generated for each user, and the most suitable insurance product can be selected as the recommendation result by sorting according to the prediction score.

[0198] Furthermore, the application also includes: using symmetric encryption algorithms such as AES (Advanced Encryption Standard) to encrypt sensitive data stored in the database; using appropriate encryption strategies for different types of data to ensure the confidentiality and integrity of the data; desensitizing sensitive information during data display and analysis, such as masking some character replacements to ensure that user privacy information is not exposed during data sharing and analysis; formulating a clear privacy policy to inform users of how their data is collected, used and protected, providing users with control over data access and management, and allowing users to view, modify and delete personal data.

[0199] In the prior art, faced with a wide variety of insurance products on the market, ordinary consumers often find it difficult to choose an insurance plan that suits their needs. The insurance product recommendation method disclosed in this application can fully understand the user's actual needs and preferences by collecting user behavior data and actively provided data, and build a multi-dimensional user portrait based on the pre-processed data, which can accurately describe the user's characteristic information. Through a personalized recommendation model, according to the user portrait and protection gap, an accurate insurance product recommendation plan is generated, which can help users overcome selection difficulties. Through data-driven personalized recommendations, users can quickly find insurance products that meet their needs and solve the problem of selection difficulties.

[0200] In the prior art, for a family, the needs of each member are different, and how to screen out suitable insurance from numerous products has become a major problem; the insurance product recommendation method disclosed in the present application can fully understand the needs of each family member by collecting personal information, family information, health information, etc. of family members, and can build an independent user portrait for each family member, describe their characteristic information respectively, analyze their protection gaps for each family member, and ensure that the recommended plan meets the needs of all family members. According to the portrait and protection gap of each family member, a personalized insurance product recommendation plan can be generated to meet the needs of all family members. By generating a personalized recommendation plan for each family member, the problem of different needs of family members is solved to ensure that the insurance needs of each member are met.

[0201] In the prior art, the traditional insurance recommendation method mainly relies on manual experience, and the process is cumbersome and time-consuming. The insurance product recommendation method disclosed in the present application collects user data in an automated manner to reduce manual operations, reduces manual intervention and time costs through automated data preprocessing and user portrait construction, automatically analyzes protection gaps through computational models to reduce the time and complexity of manual analysis, automatically generates recommendation plans through machine learning algorithms to improve recommendation efficiency, and greatly improves the efficiency of insurance product recommendations and reduces the time and cost of manual operations through automated data processing and recommendation models.

[0202] In the prior art, manual recommendations are difficult to fully analyze user needs, and protection gaps or duplicate insurance may easily occur, resulting in poor accuracy. The insurance product recommendation method disclosed in the present application, through multi-dimensional data collection, comprehensively understands the user's actual needs and risk preferences, constructs a multi-dimensional user portrait, and comprehensively analyzes the user's characteristic information. Through a preset insurance product matching strategy, it comprehensively analyzes the user's protection gap to avoid insufficient protection or duplicate insurance. Through a personalized recommendation model, an accurate recommendation plan is generated based on the user's comprehensive needs. Through multi-dimensional data analysis and a personalized recommendation model, the user's actual needs are comprehensively analyzed to avoid protection gaps or duplicate insurance, thereby improving the accuracy of recommendations.

[0203] In the prior art, manual recommendations are limited to an individual's knowledge reserves, resulting in an inability to provide customized insurance plans based on the user's specific circumstances. The insurance product recommendation method disclosed in this application, through multi-dimensional data collection, comprehensively understands the user's specific needs and circumstances, builds a user portrait based on comprehensive data, accurately describes the user's specific circumstances, and comprehensively analyzes the user's protection gap through preset matching strategies and calculation models. Through a personalized recommendation model, a customized recommendation plan is generated based on the user's specific circumstances. Through a data-driven personalized recommendation model, a customized insurance product recommendation plan can be generated based on the user's specific circumstances, thereby solving the problem of knowledge limitations of manual recommendations.

[0204] The insurance product recommendation method disclosed in the present application can more comprehensively understand the user's needs, risk preferences and life status through multi-dimensional user information collection; by constructing a multi-dimensional user portrait, it can more accurately describe the user's characteristics and provide a basis for personalized recommendations; by calculating the protection gap, it can accurately locate the user's insurance needs to avoid over-insurance or under-insurance; through comprehensive user portrait and protection gap analysis, the recommended products are more in line with the user's actual needs; users do not need to spend a lot of time and energy to research insurance products on the market, and can directly provide the best recommendations; according to the user's specific needs and characteristics, providing personalized insurance product recommendation plans that meet the user's needs can improve the user's acceptance and willingness to purchase the recommended products, thereby improving the conversion rate. Users can quickly find insurance products that meet their needs, improve the overall user experience, and help establish long-term customer relationships.

[0205] Second, refer to Figure 8 The present disclosure also provides an insurance product recommendation system, including the following solutions:

[0206] User data collection module 11, used to collect user basic information, preference data, and user insurance policy information;

[0207] A construction module 12 is used to construct a user portrait based on basic information and preference data;

[0208] The protection gap acquisition module 13 is used to obtain the protection gap between the user's insurance policy information and the preset insurance product matching strategy;

[0209] The personalized recommendation module 14 is used to analyze user profiles and protection gaps and generate insurance product recommendation plans.

[0210] It should be noted that the solutions in the insurance product recommendation method disclosed in the first aspect of this application are all applicable to the acquisition of corresponding solutions in the insurance product recommendation system disclosed in the second aspect of this application, so they will not be repeated here.

[0211] The computer device according to the embodiment of the present disclosure includes a memory and a processor. The memory is used to store non-temporary computer-readable instructions. Specifically, the memory may include one or more computer program products, and the computer program product may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may, for example, include a random access memory (RAM) and / or a cache memory (cache), etc. The non-volatile memory may, for example, include a read-only memory (ROM), a hard disk, a flash memory, etc.

[0212] The processor may be a central processing unit (CPU) or other forms of processing units with data processing capabilities and / or instruction execution capabilities, and may control other components in the computer device to perform desired functions. In one embodiment of the present disclosure, the processor is used to run the computer-readable instructions stored in the memory, so that the computer device executes all or part of the steps of the insurance product recommendation method of each embodiment of the present disclosure.

[0213] Those skilled in the art should be able to understand that in order to solve the technical problem of how to obtain a good user experience, the present embodiment may also include well-known structures such as a communication bus and an interface, and these well-known structures should also be included in the protection scope of the present disclosure.

[0214] like Fig. 9 A schematic diagram of the structure of a computer device provided in an embodiment of the present disclosure is shown, which is a schematic diagram of the structure of a computer device suitable for implementing the embodiment of the present disclosure. Fig. 9 The computer device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0215] like Fig. 9 As shown, the computer device may include a processor (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) or a program loaded from a storage device into a random access memory (RAM). In the RAM, various programs and data required for the operation of the computer device are also stored. The processor, ROM, and RAM are connected to each other via a bus. An input / output (I / O) interface is also connected to the bus.

[0216] Typically, the following devices can be connected to the I / O interface: input devices such as sensors or visual information acquisition devices; output devices such as display screens; storage devices such as tapes, hard disks, etc.; and communication devices. The communication device can allow the computer device to communicate with other devices (such as edge computing devices) wirelessly or by wire to exchange data. Fig. 9 A computer device having various devices is shown, but it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed instead.

[0217] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device, or installed from a ROM. When the computer program is executed by a processor, all or part of the steps of the insurance product recommendation method of the embodiment of the present disclosure are executed.

[0218] For detailed description of this embodiment, reference may be made to the corresponding descriptions in the aforementioned embodiments, which will not be repeated here.

[0219] According to the computer-readable storage medium of the embodiment of the present disclosure, non-transitory computer-readable instructions are stored thereon. When the non-transitory computer-readable instructions are executed by a processor, all or part of the steps of the insurance product recommendation method of each embodiment of the present disclosure are executed.

[0220] The above-mentioned computer-readable storage media include, but are not limited to: optical storage media (e.g., CD-ROM and DVD), magneto-optical storage media (e.g., MO), magnetic storage media (e.g., magnetic tape or mobile hard disk), media with built-in rewritable non-volatile memory (e.g., memory card) and media with built-in ROM (e.g., ROM box).

[0221] For detailed description of this embodiment, reference may be made to the corresponding descriptions in the aforementioned embodiments, which will not be repeated here.

[0222] The basic principles of the present disclosure are described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, effects, etc. mentioned in the present disclosure are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. are required by each embodiment of the present disclosure. In addition, the specific details disclosed above are only for the purpose of illustration and ease of understanding, and are not limiting. The above details do not limit the present disclosure to the necessity of adopting the above specific details to be implemented.

[0223] In the present disclosure, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is any such actual relationship or order between these entities or operations. The block diagrams of the devices, devices, equipment, and systems involved in the present disclosure are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagram. As will be appreciated by those skilled in the art, these devices, devices, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including", "comprising", "having", etc. are open words, referring to "including but not limited to", and can be used interchangeably with them. The words "or" and "and" used here refer to the words "and / or" and can be used interchangeably with them, unless the context clearly indicates otherwise. The words "such as" used here refer to the phrase "such as but not limited to", and can be used interchangeably with them.

[0224] Additionally, as used herein, "or" used in a list of items beginning with "at least one" indicates a separate list, so that, for example, a list of "at least one of A, B, or C" means A or B or C, or AB or AC or BC, or ABC (i.e., A and B and C). Furthermore, the word "exemplary" does not mean that the example described is preferred or better than other examples.

[0225] It should also be noted that in the system and method of the present disclosure, each component or each step can be decomposed and / or recombined. Such decomposition and / or recombination should be regarded as equivalent solutions of the present disclosure.

[0226] Various changes, substitutions, and modifications of the techniques described herein may be made without departing from the teachings defined by the appended claims. Furthermore, the scope of the claims of the present disclosure is not limited to the specific aspects of the processes, machines, manufactures, compositions of events, means, methods, and actions described above. Currently existing or later to be developed processes, machines, manufactures, compositions of events, means, methods, or actions that perform substantially the same functions or achieve substantially the same results as the corresponding aspects described herein may be utilized. Thus, the appended claims include such processes, machines, manufactures, compositions of events, means, methods, or actions within their scope.

[0227] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present disclosure. Therefore, the present disclosure is not intended to be limited to the aspects shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

[0228] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present disclosure to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.

Claims

1. A method for recommending insurance products, characterized in that: include: Collect users’ basic information, preference data, and user insurance policy information; Building a user portrait based on the basic information and the preference data; Obtain the protection gap between the user's insurance policy information and the preset insurance product matching strategy; Analyze the user portrait and the protection gap to generate insurance product recommendation plans.

2. The insurance product recommendation method according to claim 1, characterized in that: The method for obtaining the preference data includes: Collecting the user's operation behavior in a preset time period within a preset system; wherein the operation behavior includes browsing behavior, purchasing behavior and searching behavior; Obtaining behaviors whose operation times are greater than a preset number threshold among the operation behaviors, and recording them as preferred behaviors; Determine the weights corresponding to all the preferred behaviors; Calculate a comprehensive score for each of the preferred behaviors according to the preferred behaviors and their corresponding weights; The preferred behavior with the largest comprehensive score is taken as the target behavior, and preference data is obtained based on the target behavior.

3. The insurance product recommendation method according to claim 1, characterized in that: The constructing a user portrait according to the basic information and the preference data includes: Extracting age characteristics, gender characteristics, occupation characteristics, educational background characteristics, income level characteristics, marital status characteristics, number of children characteristics, home address characteristics, height characteristics, weight characteristics, living habits characteristics, and medical record characteristics from the basic information; Extracting browsing content type features, browsing frequency features, browsing duration features, browsing preference features, or purchased commodity category features, purchase frequency features, purchase price range features, purchase preference features, or search keyword features, search frequency features, search result click characteristics, search preference features from the preference data; Take each feature as a dimension and construct a multidimensional vector; Get the weight of each feature and the interaction weights between different features; Determine the eigenvalue corresponding to each eigenvalue according to the preset eigenvalue mapping database; Performing standardization on the feature values ​​to obtain a standardization function for each feature; Obtaining an evaluation result according to the normalization function, the weight of each feature, and the interaction weights between different features; Constructing a user profile based on the evaluation results; The evaluation result is Score(U): Among them, ω i is the weight of each feature, v i is the eigenvalue corresponding to the i-th feature, n is the total number of features, μ i is the preset mean of the corresponding feature, σ i is the preset standard deviation of the corresponding feature, ω ij is the interaction weight between two features with an interactive relationship, σ(v i ),σ(v j ) are the normalized functions of two features with interactive relationship.

4. The insurance product recommendation method according to claim 1, characterized in that: The obtaining of the protection gap between the user's insurance policy information and the preset insurance product matching strategy includes: Extract key fields included in the user's existing insurance products based on the user's insurance policy information; Set different values ​​for each key field. Each recommended insurance product includes one or more key fields with specific values. Build a mapping relationship between the key fields of insurance products and the recommended insurance products. Configure a recommendation condition for each recommended insurance product; Recommend recommended insurance products that meet the recommended conditions to users; The protection gap is determined based on the user's insurance policy information and the recommended insurance products.

5. The insurance product recommendation method according to claim 4, characterized in that: The configuration of a recommendation condition for each recommended insurance product includes: When the recommended insurance products include: insurance types covering outpatient expenses, hospitalization expenses, and surgical expenses, and the insured amount is set at 10%-20% of the annual income, the recommended conditions include: the user is between 18 and 65 years old, the user has no major past medical history, and the user's annual income is less than 100,000 yuan; When the recommended insurance products include: insurance types that include a global medical network, outpatient care, hospitalization, physical examinations and emergency rescue, and the insured amount is set at 50%-100% of the annual income, the recommendation conditions include that the user is between 30 and 50 years old, the user's annual income is more than 500,000 yuan, and the user expects to receive medical treatment in a private hospital or an international hospital.

6. The insurance product recommendation method according to claim 4, characterized in that: The determining of the protection gap according to the user's insurance policy information and the recommended insurance product includes: Building a preset decision tree model based on the key fields, and not performing a pruning operation in the process of building the initial decision tree model; Taking the recommended insurance products and the user's existing insurance products as input data; The input data is predicted by the preset decision tree model to obtain multiple predicted Test results; The category that appears most frequently in the multiple prediction results is used as the protection gap.

7. The insurance product recommendation method according to claim 6, characterized in that: The step of constructing a preset decision tree model based on the key fields includes: Calculate the splitting criteria of all the key fields, determine the best key field, and record it as the target feature; wherein the splitting criteria include information gain or Gini index; the best key feature refers to the feature with the largest information gain or the smallest Gini index; Constructing an initial decision tree model based on the target features; Using different types of historical policy data to train the initial decision tree model respectively to obtain multiple trained initial decision tree models; The multiple trained initial decision tree models are integrated through ensemble learning to obtain the preset decision tree model.

8. The insurance product recommendation method according to claim 1, characterized in that: The analyzing the user profile and the protection gap to generate an insurance product recommendation plan includes: Building an initial model including multiple historical user decision tree models; Training the initial model, and using the trained initial model as a personalized recommendation model; Based on the personalized recommendation model, respectively extract the portrait features of the user portrait and the gap features of the security gap; Based on the image feature and the gap feature, a feature vector is obtained; The feature vector is inferred based on a decision tree or ensemble learning method to obtain a number of YY+242207P Reference plans and their corresponding recommendation probabilities; The reference plan with the highest recommendation probability is used as the recommended plan for the insurance product.

9. A computer device, characterized in that: The computer device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the insurance product recommendation method described in any one of claims 1-8.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the insurance product recommendation method described in any one of claims 1-8.

Images