Information recommendation method and device based on classification model, electronic equipment and medium

Abstract

The embodiment of the invention provides an information recommendation method and device based on a classification model, electronic equipment and a medium, and belongs to the technical field of financial science and technology. The classification model comprises a plurality of classification sub-models; the method comprises the following steps: acquiring an insurance product set of a financial system and user behavior data of the financial system within a preset duration; generating to-be-predicted data according to the insurance product set and the user behavior data; performing feature processing on the to-be-predicted data to obtain a to-be-predicted data set; training each classification sub-model according to a to-be-predicted data set to obtain a prediction set; performing index fusion on the prediction scores in the prediction set to obtain product scores; sorting the insurance products in the insurance product set based on the product scores to obtain a recommendation list, and displaying the recommendation list to the user. According to the embodiment of the invention, the complex nonlinear relationship between the user and different insurance products can be captured, and more accurate and personalized insurance recommendation can be provided.

Description

Technical Field

[0001] The present application relates to the field of financial technology, and in particular to an information recommendation method, device, electronic device and medium based on a classification model. Background Art

[0002] In the insurance field, the performance of personalized recommendation systems is crucial. In today's digital age, the insurance industry is facing unprecedented challenges and opportunities. At the same time, with the popularization of the Internet, user expectations are also evolving. Modern insurance companies need to provide users with personalized recommendations among a variety of insurance products to meet their unique needs.

[0003] The data required for insurance recommendations are diverse, including users' personal information, historical policies, risk preferences, etc. Users' insurance needs are often affected by multiple factors, some of which have complex nonlinear relationships. Due to the high complexity of insurance products and the diversity of user needs, a single model often has difficulty capturing these complexities. Current linear models are unable to capture complex feature interactions and have limited ability to handle nonlinear relationships, which results in low accuracy in product recommendations, which cannot meet user requirements and further reduces user satisfaction. Summary of the invention

[0004] The main purpose of the embodiments of the present application is to propose an information recommendation method, device, electronic device and medium based on a classification model, which can capture the complex nonlinear relationship between users and different insurance products, provide more accurate and personalized insurance recommendations, and improve user satisfaction.

[0005] To achieve the above-mentioned purpose, a first aspect of an embodiment of the present application proposes an information recommendation method based on a classification model, wherein the classification model includes a plurality of classification sub-models; the method includes: Acquire an insurance product set of a financial system and user behavior data of the financial system within a preset time period, wherein the insurance product set includes a plurality of insurance products; Generate data to be predicted based on the insurance product set and the user behavior data; Performing feature processing on the data to be predicted to obtain a data set to be predicted; Training each of the classification sub-models according to the data set to be predicted to obtain a prediction set, wherein the prediction set includes prediction scores corresponding to the classification sub-models; Performing exponential fusion on the prediction scores in the prediction set to obtain a product score; The insurance products in the insurance product set are sorted based on the product scores to obtain a recommendation list, and the recommendation list is displayed to the user.

[0006] In some embodiments, the step of performing feature processing on the data to be predicted to obtain a data set to be predicted includes: Classify the data to be predicted to obtain numerical data and categorical data; Performing a feature bucketing operation on the numerical data to obtain a feature variable value; Performing feature crossover on the feature variable value and the categorical data to obtain a feature label; Calculate the characteristic information value of the data to be predicted based on the characteristic variable value; Performing feature screening on the data to be predicted according to a preset value threshold, screening out target data whose feature information value is greater than the value threshold; Determine a target feature label corresponding to the target data, and obtain a data set to be predicted based on the target feature label and the target data.

[0007] In some embodiments, the step of calculating the feature information value of the data to be predicted based on the feature variable value includes: Acquire the total amount of data to be predicted; Determine a first insurance product held by the user according to the user behavior data, and use the first insurance product as positive sample data; Determine a second insurance product other than the first insurance product in the insurance product set, and use the second insurance product as negative sample data; For each of the characteristic variable values, calculating an evidence weight value according to the positive sample data, the negative sample data and the total amount of data; The evidence weight values ​​are summed to obtain a feature information value.

[0008] In some embodiments, the step of training each of the classification sub-models according to the data set to be predicted to obtain a prediction set includes: For each of the classification sub-models, the data set to be predicted is input into the classification sub-model, so that the classification sub-model predicts the preference score of the insurance product in the data set to be predicted and outputs the predicted score; The prediction scores of all the classification sub-models are sorted to obtain a prediction set.

[0009] In some embodiments, performing exponential fusion on the prediction scores in the prediction set to obtain a product score includes: Classifying all the classification sub-models to obtain model categories and the number of models corresponding to the model types; determining a target prediction score corresponding to the model class in the prediction set; For each of the model categories, determining a prediction score mean according to the number of models and the target prediction score; The mean of all the predicted scores is weighted to obtain the product score.

[0010] In some embodiments, sorting the insurance products in the insurance product set based on the product scores to obtain a recommendation list includes: Sorting the insurance products in the insurance product set according to the product scores to obtain a sorted sequence; Filtering target insurance products with preset rankings in the sorting sequence to obtain a target insurance product set; A recommendation list is generated according to the target insurance product set.

[0011] In some embodiments, performing a feature bucketing operation on the numerical data to obtain a feature variable value includes: Obtain the data quantity of the numerical data, and set the number of buckets according to the data quantity; Determine the quantile according to the amount of data and the amount of buckets; Divide the numerical data into intervals according to the quantiles to obtain multiple bucket intervals; The bucket intervals are numbered to obtain characteristic variable values.

[0012] To achieve the above-mentioned purpose, a second aspect of an embodiment of the present application proposes an information recommendation device based on a classification model, wherein the classification model includes a plurality of classification sub-models; the device includes: A data acquisition module, used to acquire an insurance product set of a financial system and user behavior data of the financial system within a preset time period, wherein the insurance product set includes a plurality of insurance products; A data generation module, used to generate data to be predicted based on the insurance product set and the user behavior data; A feature processing module, used for performing feature processing on the data to be predicted to obtain a data set to be predicted; A model training module, used for training each of the classification sub-models according to the data set to be predicted to obtain a prediction set, wherein the prediction set includes prediction scores corresponding to the classification sub-models; An index fusion module, used to perform index fusion on the prediction scores in the prediction set to obtain a product score; The product recommendation module is used to sort the insurance products in the insurance product set based on the product scores, obtain a recommendation list, and display the recommendation list to the user.

[0013] To achieve the above-mentioned purpose, the third aspect of an embodiment of the present application proposes an electronic device, an electronic device comprising a memory and a processor, the memory storing a computer program, and the processor implementing the information recommendation method based on the classification model as described in the first aspect when executing the computer program.

[0014] To achieve the above-mentioned purpose, the fourth aspect of an embodiment of the present application proposes a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the information recommendation method based on the classification model as described in the first aspect.

[0015] The information recommendation method, device, electronic device and storage medium based on the classification model proposed in the present application first obtain the insurance product set of the financial system and the user behavior data of the financial system within a preset time period, then generate the data to be predicted based on the insurance product set and the user behavior data, perform feature processing on the data to be predicted, thereby reducing the influence of outliers in the data to be predicted, and being able to remove a part of the features with relatively low value for the prediction target and redundant features, then train each classification sub-model according to the data set to be predicted, train multiple classification sub-models through the same data set to be predicted, obtain the prediction score corresponding to each classification sub-model, and further obtain the prediction set, which can predict the user's preference score for different insurance products through multiple classification sub-models, thereby more accurately capturing the user's preference for various types of insurance products, and then exponentially fuse the prediction scores in the prediction set, use nonlinear fusion of each model result to obtain the product score, realize the capture of the complex nonlinear relationship between the user and different insurance products, and improve the robustness of the model in actual scenarios, and finally, sort the insurance products in the insurance product set based on the product score to obtain a recommendation list, and display the recommendation list to the user, provide more accurate and personalized insurance recommendations, and improve user satisfaction. The embodiment of the present application first trains multiple classification sub-models through the data set to be predicted, and then uses nonlinear fusion to fuse the prediction results of each classification sub-model, which can capture the complex nonlinear relationship between users and different insurance products, improve recommendation accuracy and user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is a flow chart of an information recommendation method based on a classification model provided in an embodiment of the present application; Figure 2 yes Figure 1 Flow chart of step S102 in FIG. Figure 3 A flow chart of a method for setting multiple cutoff values ​​according to a sample sequence and an interaction frequency provided in an embodiment of the present application; Figure 4A flow chart of a method for setting a suppression function according to a demarcation value provided in an embodiment of the present application; Figure 5 yes Figure 3 Flow chart of step S301 in FIG. Figure 6 yes Figure 1 Flow chart of step S106 in FIG. Figure 7 yes Figure 6 Flowchart of step S602 in FIG. Figure 8 is a structural diagram of an information recommendation device based on a classification model provided in an embodiment of the present application; Fig. 9 It is a schematic diagram of the hardware structure of the electronic device provided in the embodiment of the present application. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0018] It should be noted that, although the functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first", "second", etc. in the specification, claims and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0020] First, some nouns involved in this application are analyzed: Natural Language Processing (NLP): NLP uses computers to process, understand, and apply human languages ​​(such as Chinese, English, etc.). NLP is a branch of artificial intelligence and an interdisciplinary subject between computer science and linguistics. It is often referred to as computational linguistics. Natural language processing includes grammatical analysis, semantic analysis, and text understanding. Natural language processing is commonly used in technical fields such as machine translation, handwritten and printed character recognition, speech recognition and text-to-speech conversion, information intent recognition, information extraction and filtering, text classification and clustering, public opinion analysis, and opinion mining. It involves data mining related to language processing, machine learning, knowledge acquisition, knowledge engineering, artificial intelligence research, and linguistic research related to language computing.

[0021] Categorical Boosting (CatBoost): CatBoost was developed by the Russian search giant Yandex and is particularly good at processing data sets containing categorical features. It is based on the gradient boosting decision tree algorithm, but has made many improvements to efficiently handle categorical features and achieve better results in many practical problems.

[0022] Light Gradient Boosting Machine (LightGBM): LightGBM was launched by Microsoft to solve the problem of gradient boosting decision trees in massive data, so that gradient boosting decision trees can be used in industrial scenarios better and faster. It uses a variety of optimization methods such as histogram algorithm, gradient-based single-side sampling algorithm, and mutually exclusive feature bundling algorithm to increase training speed and reduce memory usage.

[0023] The classification model-based information recommendation method and device, electronic device and storage medium provided in the embodiments of the present application can capture the complex nonlinear relationship between users and different insurance products, provide more accurate and personalized insurance recommendations, and improve user satisfaction.

[0024] The following examples are used to illustrate the method. First, the information recommendation method based on the classification model in the example of the present application is described.

[0025] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0026] The basic technologies of artificial intelligence generally include sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. Artificial intelligence software technologies mainly include computer vision technology, robotics technology, biometrics technology, module management technology of online meeting systems, natural language processing technology, and machine learning / deep learning.

[0027] The information recommendation method based on the classification model provided in the embodiment of the present application relates to the field of financial technology technology. The information recommendation method based on the classification model provided in the embodiment of the present application can be applied to the terminal, can also be applied to the server side, and can also be software running in the terminal or the server side. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc.; the server side can be configured as an independent physical server, or it can be configured as a server cluster or distributed system composed of multiple physical servers, and can also be configured as a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the information recommendation method based on the classification model, etc., but is not limited to the above forms.

[0028] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data information types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0029] It should be noted that in each specific implementation of the present application, when it comes to the need to perform relevant processing based on data related to user identity or characteristics such as user information, user behavior data, user historical data, and user location information, the user's permission or consent will be obtained first, and the collection, use, and processing of these data will comply with the relevant laws, regulations, and standards of the relevant countries and regions. In addition, when the embodiment of the present application needs to obtain the user's sensitive personal information, the user's separate permission or consent will be obtained through a pop-up window or by jumping to a confirmation page. After clearly obtaining the user's separate permission or consent, the necessary user-related data for the normal operation of the embodiment of the present application will be obtained.

[0030] In the insurance field, the performance of personalized recommendation systems is crucial. In today's digital age, the insurance industry is facing unprecedented challenges and opportunities. At the same time, with the popularization of the Internet, user expectations are also evolving. Modern insurance companies need to provide users with personalized recommendations among a variety of insurance products to meet their unique needs.

[0031] The data required for insurance recommendations are diverse, including users' personal information, historical policies, risk preferences, etc. Users' insurance needs are often affected by multiple factors, some of which have complex nonlinear relationships. Due to the high complexity of insurance products and the diversity of user needs, a single model often has difficulty capturing these complexities. Current linear models are unable to capture complex feature interactions and have limited ability to handle nonlinear relationships, which results in low accuracy in product recommendations, which cannot meet user requirements and further reduces user satisfaction.

[0032] In order to solve the above problems, the present embodiment provides an information recommendation method, device, electronic device and storage medium based on a classification model. First, a set of insurance products of a financial system and user behavior data of the financial system within a preset time period are obtained. Then, data to be predicted is generated based on the insurance product set and the user behavior data. Feature processing is performed on the data to be predicted, thereby reducing the influence of outliers in the data to be predicted, and removing a part of features with relatively low value for the prediction target and redundant features. Then, each classification sub-model is trained according to the data set to be predicted, and multiple classification sub-models are trained by the same data set to be predicted to obtain a prediction score corresponding to each classification sub-model, and further obtain a prediction set. The user's preference scores for different insurance products can be predicted by multiple classification sub-models, thereby more accurately capturing the user's preference for various types of insurance products. Then, the prediction scores in the prediction set are exponentially fused, and each model result is nonlinearly fused to obtain a product score, thereby capturing the complex nonlinear relationship between the user and different insurance products, and improving the robustness of the model in actual scenarios. Finally, the insurance products in the insurance product set are sorted based on the product score to obtain a recommendation list, and the recommendation list is displayed to the user, providing more accurate and personalized insurance recommendations, and improving user satisfaction. The embodiment of the present application first trains multiple classification sub-models through the data set to be predicted, and then uses nonlinear fusion to fuse the prediction results of each classification sub-model, which can capture the complex nonlinear relationship between users and different insurance products, improve recommendation accuracy and user experience.

[0033] The following is a detailed description with reference to the accompanying drawings.

[0034] Figure 1 is an optional flow chart of the information recommendation method based on the classification model provided in the embodiment of the present application. Figure 1 The method may include but is not limited to steps S101 to S106.

[0035] It should be noted that the classification model includes multiple classification sub-models.

[0036] Step S101, obtaining an insurance product set of a financial system and user behavior data of the financial system within a preset time period, wherein the insurance product set includes a plurality of insurance products.

[0037] In step S101 of some embodiments, an insurance product set of the financial system is obtained, wherein the insurance product set includes multiple insurance products, and the insurance products include information such as product name, product premium, product category, etc., and user behavior data of the financial system within a preset time period is obtained, such as the user's historical click data, historical purchase data, user age data, user income level data, etc., to facilitate subsequent analysis of the user behavior data and further recommend corresponding products to the user in combination with the user behavior data.

[0038] It should be noted that user behavior data includes but is not limited to the user's own characteristic information and historical behavior information, wherein the user's own characteristic information includes but is not limited to the user's gender, age, income level, etc., and the historical behavior information includes but is not limited to historical click information, historical browsing information, historical purchase information, etc. Insurance products include but are not limited to institutional dimension characteristics and insurance product characteristics, wherein the institutional dimension characteristics include but are not limited to the institutional region information where the insurance product is located, the effective region information, etc., and the insurance product characteristics include but are not limited to product premium information, product category information, etc., and the embodiments of this application do not make specific restrictions.

[0039] Step S102, generating data to be predicted based on the insurance product set and user behavior data.

[0040] In step S102 of some embodiments, the insurance product set and user behavior data are sorted to generate data to be predicted, so that the data to be predicted includes both relevant information of the insurance products and relevant information of the user behavior data, which facilitates subsequent prediction of the insurance products.

[0041] Step S103: perform feature processing on the data to be predicted to obtain a data set to be predicted.

[0042] In step S103 of some embodiments, feature processing is performed on the data to be predicted to reduce noise and incompleteness in the data set, improve the overall quality of the data, and reduce the complexity of the model, so as to obtain the data set to be predicted, thereby facilitating the subsequent improvement of the generalization ability of the model, so that the model can more accurately capture the needs and preferences of users, thereby providing more personalized insurance product recommendations.

[0043] Step S104, training each classification sub-model according to the data set to be predicted to obtain a prediction set, wherein the prediction set includes prediction scores corresponding to the classification sub-models.

[0044] In step S104 of some embodiments, each classification sub-model is trained according to the data set to be predicted, and different classification sub-models are used to predict the user's preference scores for different insurance products. This enables the model to learn the user's behavior and preferences from different angles, further improving the generalization ability of the model. The prediction scores of all classification sub-models are then sorted to obtain a prediction set, which facilitates the subsequent provision of personalized insurance product recommendations to each user, thereby improving user experience and satisfaction.

[0045] It should be noted that the classification sub-models in the embodiments of the present application include but are not limited to models such as CatBoost and LightGBM, and the embodiments of the present application do not impose specific restrictions.

[0046] Step S105, performing exponential fusion on the prediction scores in the prediction set to obtain a product score.

[0047] In step S105 of some embodiments, the prediction scores in the prediction set are exponentially fused to obtain the prediction results after multi-model fusion and the product scores, which can further better capture the nonlinear relationships in complex personalized insurance recommendations and provide more accurate and personalized insurance products.

[0048] Step S106, sorting the insurance products in the insurance product set based on the product scores, obtaining a recommendation list, and displaying the recommendation list to the user.

[0049] In step S106 of some embodiments, the insurance products in the insurance product set are sorted based on the product scores, so that insurance products with higher scores are ranked higher, a recommendation list is obtained, and the recommendation list is displayed to the user, providing more accurate and personalized insurance products and improving user satisfaction.

[0050] See also Figure 2 In some embodiments, step S103 may also include but is not limited to steps S201 to S206.

[0051] Step S201, classify the data to be predicted to obtain numerical data and categorical data.

[0052] In step S201 of some embodiments, since the data to be predicted includes insurance product data and user behavior data, the categories of these data may not be unified, the embodiments of the present application first classify the data to be predicted, determine the different categories of data, and obtain numerical data and categorical data, so as to facilitate the subsequent feature analysis of different types of data and improve the accuracy of data processing.

[0053] It should be noted that in the process of classifying the predicted data, the embodiments of the present application will also perform missing value processing on the predicted data. For numerical data, missing value processing is performed by the mean filling method. Specifically, the mean of the non-missing values ​​of all numerical data is first calculated, and then this mean is used to fill all missing values ​​in the feature, thereby ensuring the integrity of the data, and mean filling will not change the size of the data set, avoiding information loss caused by deleting missing values; for categorical data, missing value processing is performed by the mode filling method. Specifically, the value that appears most frequently in the categorical data, that is, the mode, is found, and then this mode is used to fill the missing values ​​in the feature to ensure the quality of the data and the accuracy of the model.

[0054] Step S202, performing a feature bucketing operation on the numerical data to obtain feature variable values.

[0055] In step S202 of some embodiments, in order to fully discretize the numerical data into multiple variable values, the embodiments of the present application perform a feature bucketing operation on the numerical data, improve the stability and robustness of the model through equal-frequency bucketing, and make the number of data points in each bucket roughly equal, so that the data distribution is more uniform, obtain characteristic variable values, and realize the processing of abnormal data. In addition, by bucketing continuous features, nonlinearity can be introduced into the model, thereby improving the feature expression ability of the model.

[0056] Step S203, performing feature crossover on the feature variable values ​​and the categorical data to obtain feature labels.

[0057] In step S203 of some embodiments, feature crossover is performed on feature variable values ​​and categorical data. In the embodiment of the present application, feature crossover is performed on feature variable values ​​and categorical data through second-order crossover to construct new labels and obtain feature labels. Feature crossover can help the model capture the complex and nonlinear relationship between features and improve the model's fitting ability. By combining features, the model can learn richer information, thereby improving the accuracy of prediction.

[0058] Specifically, the embodiment of the present application first selects the features that need to be crossed in the numerical data according to the feature variable value, and selects the features that need to be crossed in the categorical data, for example, the insurance type and the institutional region code, and then combines the selected features to form a new cross feature, for example, combining the insurance type with the institutional region, and then encodes the combined features to ensure that the model can correctly understand the new label and obtain the feature label. Among them, the combined features can be encoded using unique hot encoding or label encoding, and the embodiment of the present application does not make specific restrictions.

[0059] Step S204, calculating the characteristic information value of the data to be predicted based on the characteristic variable value.

[0060] In step S204 of some embodiments, the characteristic information value of the data to be predicted is calculated based on the characteristic variable value, so that the contribution of each data in the data to be predicted can be determined, which facilitates the subsequent screening of the data to be predicted by the characteristic information value.

[0061] Step S205, performing feature screening on the data to be predicted according to a preset value threshold, and screening out target data whose feature information value is greater than the value threshold.

[0062] In step S205 of some embodiments, feature screening is performed on the data to be predicted according to a preset value threshold, and target data whose feature information value is greater than the value threshold is screened out, thereby screening out information with a greater contribution, removing some features with relatively low value to the predicted target and redundant features, and improving the accuracy of information prediction.

[0063] It should be noted that the value threshold of the embodiment of the present application can be set according to the needs of the user, for example, set to 0.1, 0.2, 0.3, etc., and the embodiment of the present application does not make any specific restrictions.

[0064] It is understandable that the embodiment of the present application will eliminate data that is less than or equal to the value threshold in the data to be predicted, thereby ensuring a high degree of contribution of the information and facilitating the prediction of the data by the subsequent model.

[0065] Step S206, determining the target feature label corresponding to the target data, and obtaining the data set to be predicted according to the target feature label and the target data.

[0066] In step S206 of some embodiments, the target feature label corresponding to the target data is determined, and the data set to be predicted is obtained based on the target feature label and the target data, thereby reducing noise and incompleteness in the data set, improving the overall quality of the data, and reducing the complexity of the model, which facilitates the subsequent improvement of the generalization ability of the model, so that the model can more accurately capture the needs and preferences of users, thereby providing more personalized insurance product recommendations.

[0067] See also Figure 3 In some embodiments, step S204 may also include but is not limited to steps S301 to S305.

[0068] Step S301, obtaining the total amount of data to be predicted.

[0069] Step S302: determine the first insurance product held by the user based on the user behavior data, and use the first insurance product as positive sample data.

[0070] Step S303: determine a second insurance product other than the first insurance product in the insurance product set, and use the second insurance product as negative sample data.

[0071] Step S304: for each characteristic variable value, the evidence weight value is calculated according to the positive sample data, the negative sample data and the total amount of data.

[0072] Step S305, summing up the evidence weight values ​​to obtain the feature information value.

[0073] In steps S301 to S305 of some embodiments, in the process of calculating the characteristic information value of the data to be predicted based on the characteristic variable value, the embodiment of the present application first obtains the total amount of data of the data to be predicted, and then determines the first insurance product that the user is holding based on the user behavior data, and uses the first insurance product as the positive sample data. After that, a second insurance product other than the first insurance product is determined in the insurance product set, wherein the second insurance product is a product that the user does not hold, and uses the second insurance product as the negative sample data. For each characteristic variable value, the evidence weight value is calculated based on the positive sample data, the negative sample data, and the total amount of data. Specifically, Specifically, the evidence weight value of the embodiment of the present application can be calculated by WOE=ln[(number of positive examples / total number of examples) / (number of negative examples / total number of examples)], wherein the number of positive examples is the number of positive sample data, the number of negative examples is the number of negative sample data, and the total number of examples is the total amount of data, thereby being able to determine the ratio of insurance products already held by the user to insurance products not held, being able to measure the ratio of insurance products held by the user, and further being able to obtain the nonlinear relationship between the feature and the target variable through the evidence weight value, then performing a sum operation on the evidence weight value to obtain the feature information value value, so as to facilitate the subsequent determination of the information contribution degree of the data through the feature information value value.

[0074] It should be noted that the characteristic information value of the embodiment of the present application can be obtained by the following formula: ; Among them, positive examples are positive sample data, and negative examples are negative sample data. is the weight of evidence.

[0075] See also Figure 4 In some embodiments, step S104 may also include but is not limited to steps S401 to S402.

[0076] Step S401: for each classification sub-model, the data set to be predicted is input into the classification sub-model, so that the classification sub-model predicts the preference score of the insurance product in the data set to be predicted and outputs the predicted score.

[0077] Step S402, sorting the prediction scores of all classification sub-models to obtain a prediction set.

[0078] In steps S401 to S402 of some embodiments, in the process of training each of the classification sub-models according to the data set to be predicted, for each classification sub-model, the data set to be predicted is input into the classification sub-model so that the classification sub-model predicts the preference score of the insurance product in the data set to be predicted, and outputs the prediction score. By predicting the user's preference score for different insurance products through different classification sub-models, the model can learn the user's behavior and preferences from different angles, further improving the generalization ability of the model, and then the prediction scores of all classification sub-models are sorted to obtain a prediction set, so as to facilitate the subsequent provision of personalized insurance product recommendations to each user and improve user experience and satisfaction.

[0079] It should be noted that the classification sub-models in the embodiments of the present application include but are not limited to models such as CatBoost and LightGBM, and the embodiments of the present application do not impose specific restrictions.

[0080] See also Figure 5 In some embodiments, step S105 may also include but is not limited to steps S501 to S504.

[0081] Step S501, classify all classification sub-models to obtain model categories and the number of models corresponding to the model types.

[0082] Step S502: determine the target prediction score corresponding to the model category in the prediction set.

[0083] Step S503: for each model category, determine the mean prediction score according to the number of models and the target prediction score.

[0084] Step S504: perform weighted calculation on the mean of all prediction scores to obtain the product score.

[0085] In steps S501 to S504 of some embodiments, in the process of exponentially fusing the prediction scores in the prediction set, the embodiments of the present application first classify all classification sub-models to obtain model categories and the number of models of each model type. For example, there are three categories of classification sub-models, the number of models in the first model category is two, the number of models in the second model category is three, the number of models in the third model category is five, and so on. Then, the target prediction score corresponding to the model category is searched in the prediction set. For each model category, the mean prediction score under the model category is calculated according to the number of models and the target prediction score. Finally, a weighted calculation is performed on the mean prediction scores to obtain the product score, so as to obtain the prediction result after multi-model fusion, which can further better capture the nonlinear relationship in complex personalized insurance recommendations and provide more accurate and personalized insurance products.

[0086] Specifically, taking the classification sub-models including the CatBoost model and the LightGBM model as an example, the formula for calculating the product score in the embodiment of the present application is as follows: ; in, represents the mean prediction score of the five CatBoost models, Represents the mean prediction score of 5 LightGBM models, Indicates the product score.

[0087] See also Figure 6 , Figure 6 A flow chart of a method for sorting insurance products in an insurance product set based on product scores provided in an embodiment of the present application, the method including but not limited to steps S601 to S603.

[0088] Step S601, sorting the insurance products in the insurance product set according to the product scores to obtain a sorting sequence.

[0089] Step S602: Filter out target insurance products with preset rankings in the sorting sequence to obtain a target insurance product set.

[0090] Step S603: Generate a recommendation list based on the target insurance product set.

[0091] In steps S601 to S603 of some embodiments, in the process of sorting the insurance products in the insurance product set based on the product scores, the embodiments of the present application first sort the insurance products in the insurance product set according to the product scores, wherein the sorting can be performed from high to low according to the product scores, or from low to high according to the product scores, to obtain a sorting sequence, thereby avoiding missing insurance products, and then filtering out target insurance products with preset rankings in the sorting sequence to obtain a target insurance product set. Specifically, when the product scores of the sorting sequence decrease successively, the embodiments of the present application filter out the first 50 target insurance products in the sorting sequence to obtain a target insurance product set; when the product scores of the sorting sequence increase successively, the embodiments of the present application filter out the last 50 target insurance products in the sorting sequence, thereby realizing the screening of insurance products with higher scores, and then generating a recommendation list based on the target insurance product set to realize accurate recommendation of products and improve the accuracy of recommendations.

[0092] It should be noted that the preset ranking in the embodiment of the present application can be set according to the user's needs and the sorting rules of the sorting sequence, and the embodiment of the present application does not make any specific restrictions.

[0093] See also Figure 7 In some embodiments, step S202 may also include but is not limited to steps S701 to S704.

[0094] Step S701, obtaining the data quantity of numerical data, and setting the number of buckets according to the data quantity.

[0095] Step S702, determining the quantile according to the amount of data and the number of buckets.

[0096] Step S703, dividing the numerical data into intervals according to the quantiles to obtain a plurality of bucket intervals.

[0097] Step S704: number the bucket intervals to obtain characteristic variable values.

[0098] In steps S701 to S704 of some embodiments, in the process of performing feature bucketing operations on numerical data, the embodiments of the present application first obtain the data quantity of the numerical data, and set the number of buckets according to the data quantity, that is, determine whether the numerical data is to be divided into multiple buckets, and then determine the quantiles according to the data quantity and the number of buckets. For example, if it is divided into 4 buckets, it is necessary to calculate the three quantiles of 25%, 50%, and 75%. Thereafter, the numerical data is interval-divided according to the quantiles, and the data is divided into different buckets according to the calculated quantiles to obtain multiple bucket intervals, wherein each bucket corresponds to an interval, and the data points fall into the corresponding intervals according to their values. The bucket intervals are then numbered, and each bucket is assigned a unique identifier to obtain the feature variable value, thereby realizing the processing of abnormal data. The stability and robustness of the model are improved by equal-frequency bucketing, and the number of data points in each bucket is made approximately equal, so that the data distribution is more uniform. In addition, by bucketing continuous features, nonlinearity can be introduced into the model, thereby improving the feature expression ability of the model.

[0099] See also Figure 8 The embodiment of the present application further provides an information recommendation device based on a classification model, wherein the classification model includes a plurality of classification sub-models, and the device includes: The data acquisition module 801 is used to acquire the insurance product set of the financial system and the user behavior data of the financial system within a preset time period, wherein the insurance product set includes multiple insurance products; The data generation module 802 is used to generate data to be predicted based on the insurance product set and user behavior data; The feature processing module 803 is used to perform feature processing on the data to be predicted to obtain a data set to be predicted; A model training module 804 is used to train each classification sub-model according to the data set to be predicted to obtain a prediction set, wherein the prediction set includes a prediction score corresponding to the classification sub-model; An index fusion module 805 is used to perform index fusion on the prediction scores in the prediction set to obtain a product score; The product recommendation module 806 is used to sort the insurance products in the insurance product set based on the product scores, obtain a recommendation list, and display the recommendation list to the user.

[0100] The specific implementation of the information recommendation device based on the classification model is basically the same as the specific implementation of the information recommendation method based on the classification model, and will not be repeated here.

[0101] The embodiment of the present application also provides an electronic device, the electronic device comprising: a memory, a processor, a program stored in the memory and executable on the processor, and a data bus for realizing connection and communication between the processor and the memory, and the program is executed by the processor to realize the above-mentioned information recommendation method based on the classification model. The electronic device can be any intelligent terminal including a tablet computer, a car computer, etc.

[0102] See also Fig. 9 , Fig. 9 The hardware structure of an electronic device of another embodiment is illustrated, and the electronic device includes: The processor 901 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application; The memory 902 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 902 can store an operating system and other application programs. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 902, and the processor 901 calls and executes the information recommendation method based on the classification model in the embodiment of this application; Input / output interface 903, used to implement information input and output; Communication interface 904, used to realize communication interaction between the device and other devices, which can be realized through wired mode (such as USB, network cable, etc.) or wireless mode (such as mobile network, WIFI, Bluetooth, etc.); A bus 905 that transmits information between various components of the device (e.g., the processor 901, the memory 902, the input / output interface 903, and the communication interface 904); The processor 901 , the memory 902 , the input / output interface 903 and the communication interface 904 are connected to each other in communication within the device via a bus 905 .

[0103] An embodiment of the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned information recommendation method based on the classification model is implemented.

[0104] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0105] The information recommendation method, device, electronic device and storage medium based on the classification model provided by the embodiment of the present application first obtain the insurance product set of the financial system and the user behavior data of the financial system within a preset time period, then generate the data to be predicted based on the insurance product set and the user behavior data, perform feature processing on the data to be predicted, thereby reducing the influence of outliers in the data to be predicted, and being able to remove a part of the features with relatively low value for the prediction target and redundant features, then train each classification sub-model according to the data set to be predicted, train multiple classification sub-models through the same data set to be predicted, obtain the prediction score corresponding to each classification sub-model, and further obtain the prediction set, and can predict the user's preference score for different insurance products through multiple classification sub-models, so as to more accurately capture the user's preference for various types of insurance products, and then perform exponential fusion on the prediction scores in the prediction set, use nonlinear fusion of each model result to obtain the product score, realize the capture of the complex nonlinear relationship between the user and different insurance products, and improve the robustness of the model in actual scenarios, and finally, sort the insurance products in the insurance product set based on the product score to obtain a recommendation list, and display the recommendation list to the user, so as to provide more accurate and personalized insurance recommendations and improve user satisfaction. The embodiment of the present application first trains multiple classification sub-models through the data set to be predicted, and then uses nonlinear fusion to fuse the prediction results of each classification sub-model, which can capture the complex nonlinear relationship between users and different insurance products, improve recommendation accuracy and user experience.

[0106] The embodiments described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0107] It can be understood by those skilled in the art that Figure 1-9 The technical solutions shown in the figure do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than those shown in the figure, or a combination of certain steps, or different steps.

[0108] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0109] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.

[0110] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0111] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0112] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the above units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0113] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0114] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0115] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including multiple instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store programs.

[0116] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but the scope of the rights of the present invention is not limited thereto. Any modification, equivalent substitution and improvement made by a person skilled in the art without departing from the scope and essence of the present invention should be within the scope of the rights of the present invention.

Claims

1. An information recommendation method based on a classification model, characterized in that: The classification model includes a plurality of classification sub-models; the method includes: Acquire an insurance product set of a financial system and user behavior data of the financial system within a preset time period, wherein the insurance product set includes a plurality of insurance products; Generate data to be predicted based on the insurance product set and the user behavior data; Performing feature processing on the data to be predicted to obtain a data set to be predicted; Training each of the classification sub-models according to the data set to be predicted to obtain a prediction set, wherein the prediction set includes prediction scores corresponding to the classification sub-models; Performing exponential fusion on the prediction scores in the prediction set to obtain a product score; The insurance products in the insurance product set are sorted based on the product scores to obtain a recommendation list, and the recommendation list is displayed to the user.

2. The information recommendation method based on the classification model according to claim 1, characterized in that: The step of performing feature processing on the data to be predicted to obtain a data set to be predicted includes: Classify the data to be predicted to obtain numerical data and categorical data; Performing a feature bucketing operation on the numerical data to obtain a feature variable value; Performing feature crossover on the feature variable value and the categorical data to obtain a feature label; Calculate the characteristic information value of the data to be predicted based on the characteristic variable value; Performing feature screening on the data to be predicted according to a preset value threshold, screening out target data whose feature information value is greater than the value threshold; Determine a target feature label corresponding to the target data, and obtain a data set to be predicted based on the target feature label and the target data.

3. The information recommendation method based on the classification model according to claim 2, characterized in that: The calculating the characteristic information value of the data to be predicted based on the characteristic variable value includes: Acquire the total amount of data to be predicted; Determine a first insurance product held by the user according to the user behavior data, and use the first insurance product as positive sample data; Determine a second insurance product other than the first insurance product in the insurance product set, and use the second insurance product as negative sample data; For each of the characteristic variable values, calculating an evidence weight value according to the positive sample data, the negative sample data and the total amount of data; The evidence weight values ​​are summed to obtain a feature information value.

4. The information recommendation method based on the classification model according to claim 1, characterized in that: The step of training each of the classification sub-models according to the data set to be predicted to obtain a prediction set includes: For each of the classification sub-models, the data set to be predicted is input into the classification sub-model, so that the classification sub-model predicts the preference score of the insurance product in the data set to be predicted and outputs the predicted score; The prediction scores of all the classification sub-models are sorted to obtain a prediction set.

5. The information recommendation method based on the classification model according to claim 1, characterized in that: The exponential fusion of the prediction scores in the prediction set to obtain a product score includes: Classifying all the classification sub-models to obtain model categories and the number of models corresponding to the model types; determining a target prediction score corresponding to the model class in the prediction set; For each of the model categories, determining a prediction score mean according to the number of models and the target prediction score; The mean of all the predicted scores is weighted to obtain the product score.

6. The information recommendation method based on the classification model according to claim 1, characterized in that: The step of sorting the insurance products in the insurance product set based on the product scores to obtain a recommendation list includes: Sorting the insurance products in the insurance product set according to the product scores to obtain a sorted sequence; Filtering target insurance products with preset rankings in the sorting sequence to obtain a target insurance product set; A recommendation list is generated according to the target insurance product set.

7. The information recommendation method based on the classification model according to claim 2, characterized in that: The performing a feature bucketing operation on the numerical data to obtain a feature variable value includes: Obtain the data quantity of the numerical data, and set the number of buckets according to the data quantity; Determine the quantile according to the amount of data and the amount of buckets; Divide the numerical data into intervals according to the quantiles to obtain multiple bucket intervals; The bucket intervals are numbered to obtain characteristic variable values.

8. An information recommendation device based on a classification model, characterized in that: The classification model includes a plurality of classification sub-models; the device includes: A data acquisition module, used to acquire an insurance product set of a financial system and user behavior data of the financial system within a preset time period, wherein the insurance product set includes a plurality of insurance products; A data generation module, used to generate data to be predicted based on the insurance product set and the user behavior data; A feature processing module, used for performing feature processing on the data to be predicted to obtain a data set to be predicted; A model training module, used for training each of the classification sub-models according to the data set to be predicted to obtain a prediction set, wherein the prediction set includes prediction scores corresponding to the classification sub-models; An index fusion module, used to perform index fusion on the prediction scores in the prediction set to obtain a product score; The product recommendation module is used to sort the insurance products in the insurance product set based on the product scores, obtain a recommendation list, and display the recommendation list to the user.

9. An electronic device, characterized in that: The electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the information recommendation method based on the classification model as described in any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the information recommendation method based on a classification model according to any one of claims 1 to 7 is implemented.

Images