Product recommendation method and device, equipment, medium and program product

Abstract

The invention provides a product recommendation method which can be applied to the technical field of artificial intelligence and financial science and technology. The product recommendation method comprises the steps that initial evaluation information and classification information of an object for a product are acquired, the initial evaluation information comprises first initial information and second initial information, and a second moment corresponding to the second initial information is different from a first moment corresponding to the first initial information; the classification information is weighted based on complexity information and risk level information corresponding to the product, the initial evaluation information is weighted based on time difference information, target information is obtained, and the time difference information represents a moment difference value between the first moment and the second moment; and inputting the target information into the recommendation model, and outputting a product recommendation result corresponding to the object. The invention further provides a product recommendation device and equipment, a storage medium and a program product.

Description

Technical Field

[0001] The present disclosure relates to the fields of artificial intelligence and financial technology, and specifically to a product recommendation method, device, equipment, medium and program product. Background Art

[0002] Product recommendation technology in financial services covers everything from traditional customer acquisition and market analysis to modern digital and digital-driven new technologies. Among the related technologies, the main ways to recommend products through data analysis include: building customer portraits based on information such as the customer's age, occupation, education level, annual income, etc.; analyzing the customer's preferences, habits, risk tolerance and other product and business selection preferences based on the customer's historical transaction data in the institution; or analyzing the user's recent browsing history on terminal devices (such as mobile banking) and browsing time on different pages, etc., to obtain the user's current needs and tendencies.

[0003] The inventors found that traditional marketing methods conduct customer analysis, clustering and product recommendations based on a large amount of stock data. The analysis and recommendation process relies on the customer's static and objective data (such as personal basic information, personal preference information and terminal browsing information), and lacks consideration of the customer's emotions and attitude changes towards financial products at different times. It is difficult to reflect the customer's acceptance tendency towards different recommended products, which affects the marketing accuracy and effectiveness of customers whose attitudes towards financial institutions have changed recently. Summary of the invention

[0004] In view of the above problems, the present disclosure provides a product recommendation method, apparatus, device, medium and program product.

[0005] According to a first aspect of the present disclosure, a product recommendation method is provided, comprising: obtaining initial evaluation information and classification information of an object for a product, wherein the initial evaluation information comprises first initial information and second initial information, and a second moment corresponding to the second initial information is different from a first moment corresponding to the first initial information; weighting the classification information based on complexity information and risk level information corresponding to the product, and weighting the initial evaluation information based on time difference information to obtain target information, wherein the time difference information represents the time difference between the first moment and the second moment; and inputting the target information into a recommendation model to output a product recommendation result corresponding to the object.

[0006] According to an embodiment of the present disclosure, the classification information is weighted based on the complexity information and risk level information corresponding to the product, and the initial evaluation information is weighted based on the time difference information to obtain target information, including: weighting the classification information based on the complexity information and the risk level information to obtain a weighted classification result; weighting the initial evaluation information using the time difference information to obtain weighted evaluation information; and processing the abnormal information in the weighted classification result and the weighted evaluation information to obtain the target information.

[0007] According to an embodiment of the present disclosure, weighted evaluation information includes satisfaction information and return visit information; initial evaluation information is weightedly processed using time difference information to obtain weighted evaluation information, including: determining the time difference information between a first moment and a second moment; and processing the initial evaluation information based on the time difference information and a preset weighting strategy to obtain weighted evaluation information.

[0008] According to an embodiment of the present disclosure, the method also includes: obtaining amount information, frequency information and behavior information corresponding to the product and their respective corresponding weight information; and using a weighted average algorithm to determine initial evaluation information based on the weight information and amount information, frequency information and behavior information.

[0009] According to an embodiment of the present disclosure, abnormal information in weighted classification results and weighted evaluation information is processed to obtain target information, including: determining target values ​​corresponding to the weighted classification results and weighted evaluation information respectively; replacing abnormal information with the target values ​​to obtain updated classification results and updated evaluation information; and normalizing the updated classification results and updated evaluation information to obtain target information.

[0010] According to an embodiment of the present disclosure, the target information includes target classification information and target evaluation information, and the product recommendation result represents the recommendation probability values ​​of multiple products corresponding to the object; the target information is input into the recommendation model, and the product recommendation result corresponding to the object is output, including: obtaining the recommendation probability value based on the target classification information and the target evaluation information.

[0011] According to an embodiment of the present disclosure, the recommendation model is trained by the following operations: obtaining sample initial evaluation information and sample classification information of the sample object for the sample product, wherein the sample initial evaluation information and the sample classification information include respective corresponding labels; using the sample initial evaluation information, the sample classification information and the labels to train the initial recommendation model to obtain the recommendation model.

[0012] According to an embodiment of the present disclosure, the target parameters of the recommendation model are determined by the following operations: initializing an initial population corresponding to the initial parameters, wherein the initial population includes multiple individuals, and the individual representations are initial values ​​corresponding to the initial parameters; using a fitness function to evaluate the multiple individuals to obtain an initial evaluation result; determining a parent individual based on the evaluation result, wherein the parent individual is multiple; generating a child individual using a crossover algorithm and multiple parent individuals; updating the parent individual based on the child individual to obtain an intermediate population; cyclically updating the intermediate population until a stopping strategy is met to obtain the target parameters; wherein the stopping strategy includes at least one of the following: the number of cycles is greater than or equal to a preset number threshold; the initial evaluation result is less than or equal to a preset evaluation threshold.

[0013] The second aspect of the present disclosure provides a product recommendation device, including: an information acquisition module, used to acquire the object's initial evaluation information and classification information for the product, wherein the initial evaluation information includes first initial information and second initial information, and the second moment corresponding to the second initial information is different from the first moment corresponding to the first initial information; a weighting module, used to weight the classification information based on complexity information and risk level information corresponding to the product, and to weight the initial evaluation information based on time difference information to obtain target information, wherein the time difference information represents the time difference between the first moment and the second moment; and a result output module, used to input the target information into a recommendation model and output a product recommendation result corresponding to the object.

[0014] A third aspect of the present disclosure provides an electronic device, comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the above method.

[0015] The fourth aspect of the present disclosure further provides a computer-readable storage medium having a computer program or instructions stored thereon, which implements the steps of the above method when the above computer program or instructions are executed by a processor.

[0016] The fifth aspect of the present disclosure further provides a computer program product, including a computer program or instructions, which implement the steps of the above method when the above computer program or instructions are executed by a processor.

[0017] According to the product recommendation method, device, equipment, medium and program product provided by the present invention, the classification information of the product is weighted by utilizing the complexity information and risk level information of the product, and the initial evaluation information is weighted by utilizing the difference between the different moments corresponding to each of the multiple initial evaluation information, so as to obtain the target information. Since the target information integrates the customer evaluation information, product complexity and risk level at different moments, the product recommendation results output by the recommendation model are more personalized, dynamic and accurate, so as to better match the actual needs and preferences of the customer, improve the overall effectiveness of the recommendation model, and further enhance the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The above contents and other objects, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:

[0019] Figure 1 A diagram schematically illustrates an application scenario of a product recommendation method, apparatus, device, medium, and program product according to an embodiment of the present disclosure;

[0020] Figure 2 A flowchart of a product recommendation method according to an embodiment of the present disclosure is schematically shown;

[0021] Figure 3 A flowchart of optimizing an initial recommendation model using a genetic algorithm according to an embodiment of the present disclosure is schematically shown;

[0022] Figure 4 An example schematic diagram schematically shows a product recommendation process according to an embodiment of the present disclosure;

[0023] Figure 5 The structural block diagram of the product recommendation device according to an embodiment of the present disclosure is schematically shown;

[0024] Figure 6 A block diagram of an electronic device suitable for implementing a product recommendation method according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0025] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0026] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0027] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0028] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0029] In the technical solution of the present disclosure, the user information (including but not limited to user personal information, user image information, user device information, such as location information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved are all information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with relevant laws, regulations and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0030] In the scenario of using personal information for automated decision-making, the methods, devices, and systems provided by the embodiments of the present disclosure provide users with corresponding operation portals for users to choose to agree or reject the automated decision-making results; if the user chooses to reject, the expert decision-making process will be entered. The expression "automated decision-making" here refers to the activity of automatically analyzing and evaluating a person's behavioral habits, interests and hobbies, or economic, health, credit status, etc. through computer programs, and making decisions. The expression "expert decision-making" here refers to the activity of making decisions by people who specialize in a certain field, have specialized experience, knowledge and skills, and have reached a certain level of professionalism.

[0031] In the process of conceiving the present disclosure, the inventors discovered that in the related art, the product marketing method relies on the static and objective data of customers (such as personal basic information, personal preference information and recent browsing information), lacks consideration of the changes in customers' emotions and attitudes towards the services and products of financial institutions at different times, and is difficult to reflect the customers' acceptance tendencies towards different recommended services and products, affecting the accuracy and effectiveness of recommendations for customers whose attitudes towards financial institutions have changed recently.

[0032] In view of this, the present invention weights the classification information of products by utilizing the complexity information and risk level information of products, and weights the initial evaluation information by utilizing the difference between the different moments corresponding to multiple initial evaluation information, so as to obtain target information. Since the target information integrates the customer evaluation information, product complexity and risk level at different moments, the product recommendation results output by the recommendation model are more personalized, dynamic and accurate, so as to better match the actual needs and preferences of customers, improve the overall effectiveness of the recommendation system, and further enhance the user experience.

[0033] Embodiments of the present disclosure provide a product recommendation method, apparatus, device, medium and program product, the recommendation method comprising: obtaining an object's initial evaluation information and classification information for a product, wherein the initial evaluation information comprises first initial information and second initial information, and a second moment corresponding to the second initial information is different from a first moment corresponding to the first initial information; weighting the classification information based on complexity information and risk level information corresponding to the product, and weighting the initial evaluation information based on time difference information to obtain target information, wherein the time difference information represents the time difference between the first moment and the second moment; and inputting the target information into a recommendation model to output a product recommendation result corresponding to the object.

[0034] Figure 1 The application scenario diagram of the product recommendation method, apparatus, device, medium and program product according to the embodiments of the present disclosure is schematically shown.

[0035] like Figure 1 As shown, the application scenario according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 is used to provide a medium for a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, the network 104, and the server 105. The network 104 may include various connection types, such as wired, wireless communication links, or optical fiber cables, etc.

[0036] Server 105 can use API integration (such as RESTful API or SOAP API), database connection (such as SQL database) or data synchronization tools and data integration platforms to obtain customers' initial evaluation information on products from a specific system or information library. The information acquisition channels include but are not limited to the counter channels of the financial institution's branches, self-service machine channels, wealth management room channels, online banking channels, mobile banking APP channels and telephone banking channels.

[0037] The user can use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications can be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only for example).

[0038] The first terminal device 101, the second terminal device 102, and the third terminal device 103 may be various electronic devices having display screens and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0039] The server 105 may be a server that provides various services, such as a background management server (only as an example) that provides support for websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103. The background management server may analyze and process the received data such as user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.

[0040] It should be noted that the product recommendation method provided in the embodiment of the present disclosure can generally be executed by the server 105. Accordingly, the product recommendation device provided in the embodiment of the present disclosure can generally be set in the server 105. The product recommendation method provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105. Accordingly, the product recommendation device provided in the embodiment of the present disclosure can also be set in a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105.

[0041] It should be understood that Figure 1The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0042] The following will be based on Figure 1 The scene described by Figure 2~Figure 4 The product recommendation method of the disclosed embodiment is described in detail.

[0043] Figure 2 The flowchart of the product recommendation method according to the embodiment of the present disclosure is schematically shown.

[0044] like Figure 2 As shown, the product recommendation method of this embodiment includes operations S210 to S230.

[0045] In operation S210, initial evaluation information and classification information of a product by an object is obtained, wherein the initial evaluation information includes first initial information and second initial information, and a second time corresponding to the second initial information is different from a first time corresponding to the first initial information.

[0046] In the embodiments of the present disclosure, the object may be a customer who purchases a product or handles related business. Products may represent different types of financial products, and the classification information of the products includes, but is not limited to, deposit, debt, stock, fund, foreign exchange and insurance information for individual customers. Considering that the attitude and emotion of the object towards the same product change at different times, the initial evaluation information may be divided into the first initial information and the second initial information based on the different evaluation times. The second initial information may be the second evaluation information made by the same customer for the same product after the first initial information is made, after a preset time period.

[0047] In an embodiment of the present disclosure, before obtaining the user's evaluation information, the user's consent or authorization may be obtained. For example, before operation S210, a request to obtain the user's evaluation information may be issued to the user. When the user agrees or authorizes that the user's evaluation information may be obtained, operation S210 is performed.

[0048] In operation S220, the classification information is weighted based on the complexity information and risk level information corresponding to the product, and the initial evaluation information is weighted based on the time difference information to obtain target information, wherein the time difference information represents the time difference between the first moment and the second moment.

[0049] In the embodiments of the present disclosure, for different types of products (standardized products and non-standardized products), the complexity information of the products may include structural complexity and process complexity. Considering the differences in returns, volatility and term, different types of products correspond to different risk level information, and the risk level information after integrating the return, volatility and term information may include high risk level, medium risk level and low risk level. It can be understood that the higher the complexity and risk level, the higher the corresponding product weight value.

[0050] In the embodiments of the present disclosure, the time information corresponding to when the customer purchases the recommended product (target product) can be used as the first moment; after the customer completes the purchase of the target product and a preset time period (for example, three days, seven days, or ten days) is passed, the time information corresponding to the second evaluation information obtained by returning the customer is used as the second moment. The difference information is obtained by calculating the time difference between the first moment and the second moment, and the initial evaluation information is weighted according to the difference information, and the time difference between the first moment and the second moment is inversely proportional to the weight value. After weighting the initial evaluation information and the classification information, the target information can be obtained, and the target information can include target classification information and target evaluation information.

[0051] In operation S230, the target information is input into the recommendation model, and a product recommendation result corresponding to the object is output.

[0052] In an embodiment of the present disclosure, the recommendation model may be a neural network model optimized based on a genetic algorithm. By inputting target classification information and target evaluation information into the neural network model, a recommendation probability value of a product may be output. The larger the recommendation probability value, the higher the recommendation priority of the product.

[0053] According to the embodiments of the present disclosure, the classification information of the product is weighted by utilizing the complexity information and risk level information of the product, and the initial evaluation information is weighted by utilizing the difference between the different moments corresponding to multiple initial evaluation information, so as to obtain the target information. Since the target information integrates the customer evaluation information, product complexity and risk level at different moments, the product recommendation results output by the recommendation model are more personalized, dynamic and accurate, so as to better match the actual needs and preferences of customers, improve the overall effectiveness of the recommendation system, and further enhance the user experience.

[0054] It is understandable that the above has explained how to obtain the recommendation results of the product, and the following will explain how to determine the target information.

[0055] According to an embodiment of the present disclosure, the classification information is weighted based on the complexity information and risk level information corresponding to the product, and the initial evaluation information is weighted based on the time difference information to obtain target information, including: weighting the classification information based on the complexity information and the risk level information to obtain a weighted classification result; weighting the initial evaluation information using the time difference information to obtain weighted evaluation information; and processing the abnormal information in the weighted classification result and the weighted evaluation information to obtain the target information.

[0056] In the embodiments of the present disclosure, complexity information can measure the complexity of product structure and transaction methods. For example, whether it is a structured product, whether derivatives are involved, whether there is leverage, etc. Risk level information can be evaluated through the risk characteristics of the product (such as market risk, credit risk, liquidity risk and other dimensions). Risk level information can be characterized as high risk level, medium risk level and low risk level, or level A, level B and level C. The form of expression of risk level information can be determined according to actual conditions and is not limited here. By weighting the classification information of the product (including deposit, debt, stock, fund, foreign exchange and insurance dimensions) using structural complexity, process complexity and different risk level information, corresponding weights are assigned to the importance of classification information of different dimensions, and weighted classification results can be obtained.

[0057] For example, for products with higher complexity, higher weights can be set, because complex products are usually accompanied by higher uncertainty and risk. Weights can also be set based on the importance of the product risk level to the customer's adaptability and risk tolerance, and products with higher risks can be set with greater weight impact.

[0058] For example, after determining the classification dimensions and weights, you can define evaluation indicators for each classification dimension and ensure that these indicators can be quantified. For the quantification of complexity information, it can be measured by the structural complexity index (for example, whether the product involves derivatives such as options and futures, the hedging mechanism of the product, etc.); the score of complex products can be expressed as an integer or a fraction, and the higher the complexity, the higher the score. For the quantification of risk level, it can be quantified according to the risk characteristics of financial products. The risk level of the product can be divided according to the rating model (such as Moody's rating) or according to indicators such as volatility and historical drawdown. The risk level is converted into a numerical value (for example, low risk = 1, medium risk = 2, high risk = 3) for processing; then the weighted algorithm is used to comprehensively weight the classification information using the quantified complexity information and risk level information to obtain a comprehensive weighted result (weighted classification result).

[0059] Furthermore, based on the comprehensive weighted results, further classification or sorting can be performed, and products can be divided into multiple categories according to the weighted score range. For example, products with low scores can be classified into low-complexity, low-risk product categories; products with high scores may be high-complexity, high-risk products.

[0060] For example, there are three financial products (products a, b, and c), with a complexity score ranging from 0 to 10, and a risk level score representing: 1 = low risk, 2 = medium risk, and 3 = high risk. Their complexity and risk levels are shown in Table 1 below.

[0061] Table 1

[0062]

[0063] Through weighted calculation, we can get the comprehensive score of each product (weighted classification result). Among them, product a: 4x0.6+2x0.4=3.2; product b: 7x0.6+3x0.4=5.4; product c: 3x0.6+1x0.4=2.2. It can be seen that product b has a higher comprehensive risk and complexity, and is suitable for investors with higher risk tolerance; product c has the lowest score and is suitable for investors with lower risk appetite.

[0064] According to the embodiments of the present disclosure, by utilizing complexity information and risk level information to perform weighted processing on the classification information of products, the classification of products can be adjusted according to their risk level and complexity, so that the classification results can better reflect the actual risk status and complexity level, which helps financial institutions make more informed decisions. The weighted classification can help financial institutions identify products with high risk and complexity, thereby more reasonably recommending products to customers.

[0065] In the embodiments of the present disclosure, after obtaining the weighted evaluation information, an outlier detection method (such as the Z-score method) may be used to determine abnormal information, including but not limited to information that deviates from the average value or information that is missing. The abnormal information may be processed using a substitution method, a conversion method, etc.

[0066] It can be understood that an example of how to obtain target information has been described above, and how to determine weighted evaluation information will be further described below.

[0067] According to an embodiment of the present disclosure, weighted evaluation information includes satisfaction information and return visit information; initial evaluation information is weightedly processed using time difference information to obtain weighted evaluation information, including: determining the time difference information between a first moment and a second moment; and processing the initial evaluation information based on the time difference information and a preset weighting strategy to obtain weighted evaluation information.

[0068] In an embodiment of the present disclosure, the preset weighting strategy can characterize a weighting function, such as exponential decay, linear decay, etc. The method for determining the weighted evaluation information may include: determining the time difference information, the time difference information can characterize the time difference between the first moment when the first initial information is generated and the second moment when the second initial information is generated, which can be expressed in days, months or years; the time difference affects the timeliness of the product evaluation, so the longer the time interval, the lower the correlation between the first initial information and the second initial information. It can be understood that the time difference and the weighted value are inversely proportional. After determining the time difference information, the initial evaluation information can be weighted by a weighting function (such as an exponential decay function, a linear decay function) and the time difference to obtain the weighted evaluation information.

[0069] According to an embodiment of the present disclosure, by using the time difference to weight the initial evaluation information, the evaluation can be made more timely and reflect the attitude or emotional changes of the subject towards the product. The influence of the initial evaluation information can be adjusted by using a weighting function to ensure that the final evaluation information can better reflect the subject's true attitude or evaluation towards the product.

[0070] It can be understood that an example of how to determine weighted evaluation information has been described above, and how to determine initial evaluation information will be further described below.

[0071] According to an embodiment of the present disclosure, the method also includes: obtaining amount information, frequency information and behavior information corresponding to the product and their respective corresponding weight information; and using a weighted average algorithm to determine initial evaluation information based on the weight information and amount information, frequency information and behavior information.

[0072] In an embodiment of the present disclosure, the first initial information in the initial evaluation information may represent the evaluation information of the object when purchasing the product, and the initial evaluation information may also be referred to as a satisfaction evaluation. The satisfaction evaluation may be a satisfaction evaluation obtained by integrating the relevant information of the object (customer) for the business recently handled by the object (customer) or the product purchased at a financial institution, and the relevant information includes but is not limited to the amount information, frequency information, behavior information, and the corresponding weight information. For example, the type and name of the business that the customer has recently handled at a financial institution through offline or online channels, the amount of the business, the frequency of the business, the attitude score of the business personnel, the professional level score of the business personnel, the behavior information of the business personnel during the handling process (whether there are non-business related behaviors, such as asking for personal non-business information), whether the business after handling is consistent with the introduction before handling, etc.

[0073] For example, corresponding weights are assigned to relevant information items, with the weight of the credit information being W1, the weight of the frequency information being W2, and the weight of the behavior information being W3, and satisfying W1+W2+W3=1; in order to avoid the impact of differences in information scales (for example, the possible numerical range of credit information is much larger than that of frequency information), a normalization algorithm (for example, minimum-maximum normalization or Z-score normalization) can be used to normalize each piece of information; the standardized credit information, frequency information, and behavior information are multiplied by their corresponding weights, and the sum is calculated to obtain the initial evaluation information.

[0074] According to the embodiments of the present disclosure, by introducing information of different dimensions (such as amount, frequency and behavior information), the recommendation system can adjust the initial evaluation for each user in a more personalized manner; by combining weight information, amount information, frequency information and behavior information, the information quality of the initial evaluation information can be improved through comprehensive consideration of multiple dimensions, and different weights can be assigned according to the importance of users or products, so that subsequent recommendation models can pay more attention to factors with greater influence, thereby improving the recommendation accuracy.

[0075] It can be understood that an example of how to determine the initial evaluation information has been described above, and how to determine the target information will be further described below.

[0076] According to an embodiment of the present disclosure, abnormal information in weighted classification results and weighted evaluation information is processed to obtain target information, including: determining target values ​​corresponding to the weighted classification results and weighted evaluation information respectively; replacing abnormal information with the target values ​​to obtain updated classification results and updated evaluation information; and normalizing the updated classification results and updated evaluation information to obtain target information.

[0077] In the embodiments of the present disclosure, an outlier identification method can be used to identify outliers in weighted classification information and weighted evaluation information. For example, a box plot can be used to identify outliers in data. By displaying the quartiles, upper and lower boundaries, and outliers of the data, extreme values ​​can be found. If a data point is outside the upper quartile (Q3) + 1.5 times the interquartile range (IQR) or the lower quartile (Q1) - 1.5 times the IQR, it can be considered an outlier.

[0078] In an embodiment of the present disclosure, outliers can be further processed. The target value can represent the median or mean value corresponding to the weighted classification result and the weighted evaluation information. For example, outliers can be processed by substitution, replacing the outliers with the mean value of the feature, if the data distribution is more skewed, it is more appropriate to use the median to replace the outliers, or based on the trend of the data, use an interpolation method (such as linear interpolation) to replace the outliers. It is also possible to reduce the impact of outliers by transforming the abnormal information through a conversion method. For example, logarithmic transformation, square root transformation and other operations are performed on the data. For missing values ​​(missing items), the mean or median can be used to fill in, so as to obtain an updated classification result or updated evaluation information. Then, the updated classification result or updated evaluation information is normalized using a normalization method (such as minimum-maximum normalization or Z-score standardization) to obtain the target information.

[0079] It can be understood that an example of how to determine target information has been described above, and how to output product recommendation results will be further described below.

[0080] According to an embodiment of the present disclosure, the target information includes target classification information and target evaluation information, and the product recommendation result represents the recommendation probability values ​​of multiple products corresponding to the object; the target information is input into the recommendation model, and the product recommendation result corresponding to the object is output, including: obtaining the recommendation probability value based on the target classification information and the target evaluation information.

[0081] In the embodiment of the present disclosure, after obtaining the target classification information and the target evaluation information, they can be input into a feedforward neural network. The feedforward neural network can be composed of an input layer, several hidden layers and an output layer. The input layer can accept the target classification information and the target evaluation information; the hidden layer can be used to perform nonlinear transformation on the input information, and can have multiple layers and neurons. The output layer can output a value representing the recommendation probability according to business needs (it can be a value between 0 and 1, representing the recommendation probability)

[0082] It is understandable that the above has explained how to output product recommendation results, and the training method of the recommendation model will be further explained below.

[0083] According to an embodiment of the present disclosure, the recommendation model is trained by the following operations: obtaining sample initial evaluation information and sample classification information of the sample object for the sample product, wherein the sample initial evaluation information and the sample classification information include respective corresponding labels; using the sample initial evaluation information, the sample classification information and the labels to train the initial recommendation model to obtain the recommendation model.

[0084] In an embodiment of the present disclosure, the initial recommendation model may be an untrained BP neural network model. Before the initial recommendation model is trained using the sample initial evaluation information and the sample classification information, the information may be preprocessed, including: normalizing the initial evaluation information so that the range of the information is within a suitable interval (e.g., between 0 and 1); performing unique hot encoding on the classification information (e.g., product category) and converting it into a binary vector; and dividing the data set into a training set and a test set. The BP neural network model is trained using the sample information in the training set, comprising an input layer, a hidden layer, and an output layer. The output layer may be a single neuron, which uses a Sigmoid activation function to output the probability of recommendation; thereby using the test set to verify the accuracy of the recommendation results output by the model, and obtaining a recommendation model when the accuracy of the recommendation results meets a preset accuracy threshold.

[0085] For example, the sample initial evaluation information, sample classification information and sample object information can be used to construct an information set. The information can represent the pre-processed sample satisfaction information and sample follow-up result information, and then each piece of information can be labeled. The label can be the business recommended for the customer and the successfully marketed product after statistical satisfaction and follow-up surveys. The sample information and labels can be used to train the initial recommendation model (BP neural network model). By taking the sample classification information, sample satisfaction information and sample follow-up result information as input, the hidden layer network is used to calculate the distribution weights, and customers with different sample information are analyzed to obtain a recommendation model that meets the preset recommendation accuracy.

[0086] It can be understood that the training method of the recommendation model has been described above, and how to determine the target parameters of the recommendation model will be further described below.

[0087] According to an embodiment of the present disclosure, the target parameters of the recommendation model are determined by the following operations: initializing an initial population corresponding to the initial parameters, wherein the initial population includes multiple individuals, and the individual representations are initial values ​​corresponding to the initial parameters; using a fitness function to evaluate the multiple individuals to obtain an initial evaluation result; determining a parent individual based on the evaluation result, wherein the parent individual is multiple; generating a child individual using a crossover algorithm and multiple parent individuals; updating the parent individual based on the child individual to obtain an intermediate population; cyclically updating the intermediate population until a stopping strategy is met to obtain the target parameters; wherein the stopping strategy includes at least one of the following: the number of cycles is greater than or equal to a preset number threshold; the initial evaluation result is less than or equal to a preset evaluation threshold.

[0088] In the embodiments of the present disclosure, the target parameters may be the weights and thresholds of the model. The initial evaluation results may characterize the fitness values ​​corresponding to each individual. The parent individuals may characterize the individuals with high fitness values ​​among the offspring individuals. The intermediate population may characterize the population in the iterative update cycle without stopping the iteration.

[0089] Figure 3 The flowchart of optimizing the initial recommendation model by using a genetic algorithm according to an embodiment of the present disclosure is schematically shown.

[0090] like Figure 3 As shown, the weights and thresholds of the initial recommendation model may be updated and optimized using a genetic algorithm, and the updating method may include operations S301 to S307.

[0091] Operation S301, initializing the population. Each individual (i.e., a solution) represents a set of initial values ​​of weights and thresholds of the neural network. The weights and thresholds can be randomly generated so as to start optimization from multiple initial points. It is understandable that appropriate initialization weights and thresholds have an important impact on the initial state of the model and the convergence speed during training.

[0092] Operation S302, calculate the fitness value. Each individual can be evaluated according to the fitness function. The fitness value can reflect the performance of the neural network, and the training error (such as mean square error MSE) or classification accuracy of the network can be used as the fitness function. Individuals with high fitness values ​​represent that the initial settings of weights and thresholds are better, and the individual is more likely to be selected for the next genetic operation.

[0093] Operation S303, selection operation. According to the fitness value, select the individuals with better performance as the parent generation, which will be used to generate the next generation of individuals. The selection method can adopt strategies such as roulette selection and tournament selection.

[0094] Operation S304 is a crossover operation. The genes (i.e., weights and thresholds) of two parent generations can be combined into a new individual (offspring). New individuals are generated by exchanging certain parts of the parent generations (e.g., a portion of the weights and thresholds), so that different weight and threshold combinations can be explored.

[0095] Operation S305, mutation operation. After the crossover operation, the newly generated individuals may be randomly adjusted in a small range to increase the diversity of the search space. It is understandable that the mutation operation may make small random changes to certain weights or thresholds in the individuals to prevent the algorithm from falling into a local optimum too early.

[0096] Operation S306, replacement operation. In each generation, new individuals will replace part or all of the parent individuals to form a new population. The population of each generation may tend to have better performing individuals, that is, a more optimal combination of weights and thresholds.

[0097] Operation S307, stop iteration condition. The iteration process continues until a certain stop condition is reached, such as reaching a preset number of generations (upper limit of the number of generations) or when the fitness reaches a certain threshold (such as the error is less than a certain value), the optimization process is stopped.

[0098] According to the embodiments of the present disclosure, considering that in the traditional gradient descent method, the optimization process may fall into the local optimal solution, especially in high-dimensional and complex neural networks. The present disclosure utilizes a genetic algorithm through global search (through crossover and mutation operations), which can more effectively avoid falling into the local optimal solution and has a strong global search capability. The genetic algorithm does not rely on gradient information, so it can find the appropriate initial weights and thresholds faster than the gradient descent method in some cases.

[0099] Figure 4 An example schematic diagram of a product recommendation process according to an embodiment of the present disclosure is schematically shown.

[0100] like Figure 4 As shown, the initial evaluation information 410 and the classification information 420 of the product P1 of the financial institution by the object (customer) can be obtained. The initial evaluation information 410 may include satisfaction information (first initial information) 411 and return visit result information (second initial information) 412. The return visit result information 412 may be the telephone return visit information obtained five days after obtaining the satisfaction information 411. It is understandable that the initial evaluation information 410 may be obtained by integrating information such as the type and name of the product P1 purchased by the customer, the amount of the product, the frequency of purchasing the product P1, and the frequency of successful recommendation. It is understandable that the way to obtain the initial evaluation information includes but is not limited to the counter channel, self-service machine channel, financial management room channel, online banking channel, mobile banking APP channel and telephone banking channel of the financial institution's outlets.

[0101] After obtaining the initial evaluation information 410 and the classification information 420 of the product, the initial evaluation information 410 and the classification information 420 may be preprocessed. For example, the classification information 420 may be weighted according to the complexity and risk level of the business and the product, the initial evaluation information 410 may be weighted according to the time difference information between the satisfaction information 411 and the follow-up result information 412, the missing information may be filled with zeros, and each item may be standardized to obtain the target classification information 430 and the target evaluation information 440.

[0102] After determining the target classification information 430 and the target evaluation information 440, the target classification information 430 and the target evaluation information 440 can be input into the recommendation model 450 to obtain the product recommendation result 460 corresponding to the object. The recommendation result 460 can provide the business personnel with the customer's subsequent business / product tendency information.

[0103] It should be noted that the product recommendation method disclosed in the present invention can also be combined with static objective information of customers, such as portrait information (age, occupation, education level, annual income, etc.), historical transaction information (user preferences, habits and risk tolerance) and historical browsing information of customer terminals to comprehensively determine the product recommendation results for users.

[0104] Based on the above product recommendation method, the present disclosure also provides a product recommendation device. Figure 5 The device is described in detail.

[0105] Figure 5 The structural block diagram of the product recommendation device according to an embodiment of the present disclosure is schematically shown.

[0106] like Figure 5 As shown, the product recommendation device of this embodiment includes an information acquisition module 510 , a weighting module 520 and a result output module 530 .

[0107] The information acquisition module 510 is used to obtain the object's initial evaluation information and classification information for the product, wherein the initial evaluation information includes first initial information and second initial information, and the second time corresponding to the second initial information is different from the first time corresponding to the first initial information. In one embodiment, the information acquisition module 510 can be used to perform the operation S210 described above, which will not be repeated here.

[0108] The weighting module 520 is used to perform weighted processing on the classification information based on the complexity information and risk level information corresponding to the product, and to perform weighted processing on the initial evaluation information based on the time difference information, so as to obtain the target information, wherein the time difference information represents the time difference between the first moment and the second moment. In one embodiment, the weighting module 520 can be used to perform the operation S220 described above, which will not be described in detail here.

[0109] The result output module 530 is used to input the target information into the recommendation model and output the product recommendation result corresponding to the target. In one embodiment, the result output module 530 can be used to perform the operation S230 described above, which will not be described in detail here.

[0110] According to the embodiments of the present disclosure, through the information acquisition module 510, the weighting module 520 and the result output module 530 in the product recommendation device, the classification information of the product is weighted by using the complexity information and the risk level information of the product, and the initial evaluation information is weighted by using the difference between the different moments corresponding to the multiple initial evaluation information, so as to obtain the target information. Since the target information integrates the customer evaluation information, product complexity and risk level at different moments, the product recommendation results output by the recommendation model are more personalized, dynamic and accurate, so as to better match the actual needs and preferences of customers, improve the overall effectiveness of the recommendation model, and further enhance the user experience.

[0111] According to an embodiment of the present disclosure, the weighting module 520 includes: a first weighting submodule, a second weighting submodule and an abnormal information processing submodule. The first weighting submodule is used to perform weighted processing on the classification information based on the complexity information and the risk level information to obtain a weighted classification result; the second weighting submodule is used to perform weighted processing on the initial evaluation information using the time difference information to obtain weighted evaluation information; and the abnormal information processing submodule is used to process the abnormal information in the weighted classification result and the weighted evaluation information to obtain the target information.

[0112] According to an embodiment of the present disclosure, the weighted evaluation information includes satisfaction information and return visit information; the second weighted submodule includes: a time difference information determination unit and an evaluation information processing unit. The time difference information determination unit is used to determine the time difference information between the first moment and the second moment; and the evaluation information processing unit is used to process the initial evaluation information based on the time difference information and a preset weighting strategy to obtain weighted evaluation information.

[0113] According to an embodiment of the present disclosure, the device further includes: an information acquisition module and an initial evaluation information determination module. The information acquisition module is used to acquire the amount information, frequency information and behavior information corresponding to the product and the weight information corresponding to each; and the initial evaluation information determination module is used to determine the initial evaluation information based on the weight information and the amount information, frequency information and behavior information using a weighted average algorithm.

[0114] According to an embodiment of the present disclosure, the abnormal information processing submodule includes: a target value determination unit, an abnormal information replacement unit and a result processing unit. The target value determination unit is used to determine the target value corresponding to the weighted classification result and the weighted evaluation information; the abnormal information replacement unit is used to replace the abnormal information with the target value to obtain the updated classification result and the updated evaluation information; and the result processing unit is used to normalize the updated classification result and the updated evaluation information to obtain the target information.

[0115] According to an embodiment of the present disclosure, the target information includes target classification information and target evaluation information, and the product recommendation result represents the recommendation probability values ​​of multiple products corresponding to the object; the result output module 530 includes: a probability value acquisition submodule, which is used to obtain the recommendation probability value based on the target classification information and the target evaluation information.

[0116] According to an embodiment of the present disclosure, the recommendation model is trained by the following operations: obtaining sample initial evaluation information and sample classification information of the sample object for the sample product, wherein the sample initial evaluation information and the sample classification information include respective corresponding labels; using the sample initial evaluation information, the sample classification information and the labels to train the initial recommendation model to obtain the recommendation model.

[0117] According to an embodiment of the present disclosure, the target parameters of the recommendation model are determined by the following operations: initializing an initial population corresponding to the initial parameters, wherein the initial population includes multiple individuals, and the individual representations are initial values ​​corresponding to the initial parameters; using a fitness function to evaluate the multiple individuals to obtain an initial evaluation result; determining a parent individual based on the evaluation result, wherein the parent individual is multiple; generating a child individual using a crossover algorithm and multiple parent individuals; updating the parent individual based on the child individual to obtain an intermediate population; cyclically updating the intermediate population until a stopping strategy is met to obtain the target parameters; wherein the stopping strategy includes at least one of the following: the number of cycles is greater than or equal to a preset number threshold; the initial evaluation result is less than or equal to a preset evaluation threshold.

[0118] According to an embodiment of the present disclosure, any multiple modules of the information acquisition module 510, the weighting module 520 and the result output module 530 can be combined into one module for implementation, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules and implemented in one module. According to an embodiment of the present disclosure, at least one of the information acquisition module 510, the weighting module 520 and the result output module 530 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application-specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation methods of software, hardware and firmware or in any appropriate combination of any of them. Alternatively, at least one of the information acquisition module 510, the weighting module 520 and the result output module 530 can be at least partially implemented as a computer program module, and when the computer program module is run, the corresponding function can be performed.

[0119] Figure 6A block diagram of an electronic device suitable for implementing a product recommendation method according to an embodiment of the present disclosure is schematically shown.

[0120] like Figure 6 As shown, the electronic device according to the embodiment of the present disclosure includes a processor 601, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 602 or the program loaded from the storage part 608 to the random access memory (RAM) 603. The processor 601 may include, for example, a general-purpose microprocessor (such as a CPU), an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (for example, an application-specific integrated circuit (ASIC)), etc. The processor 601 may also include an on-board memory for caching purposes. The processor 601 may include a single processing unit or multiple processing units for performing different actions of the method flow according to the embodiment of the present disclosure.

[0121] In RAM 603, various programs and data required for the operation of the electronic device are stored. The processor 601, ROM 602 and RAM 603 are connected to each other through a bus 604. The processor 601 performs various operations of the method flow according to the embodiment of the present disclosure by executing the program in ROM 602 and / or RAM 603. It should be noted that the program can also be stored in one or more memories other than ROM 602 and RAM 603. The processor 601 can also perform various operations of the method flow according to the embodiment of the present disclosure by executing the program stored in the one or more memories.

[0122] According to an embodiment of the present disclosure, the electronic device may further include an input / output (I / O) interface 605, which is also connected to the bus 604. The electronic device may further include one or more of the following components connected to the input / output (I / O) interface 605: an input portion 606 including a keyboard, a mouse, etc.; an output portion 607 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 608 including a hard disk, etc.; and a communication portion 609 including a network interface card such as a LAN card, a modem, etc. The communication portion 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the input / output (I / O) interface 605 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 610 as needed, so that a computer program read therefrom is installed into the storage portion 608 as needed.

[0123] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the method according to the embodiment of the present disclosure is implemented.

[0124] According to an embodiment of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, it may include but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, an apparatus or a device. For example, according to an embodiment of the present disclosure, the computer-readable storage medium may include the ROM 602 and / or RAM 603 described above and / or one or more memories other than ROM 602 and RAM 603.

[0125] The embodiment of the present disclosure also includes a computer program product, which includes a computer program, and the computer program contains program code for executing the method shown in the flowchart. When the computer program product is run in a computer system, the program code is used to enable the computer system to implement the product recommendation method provided by the embodiment of the present disclosure.

[0126] The above functions defined in the system / device of the embodiment of the present disclosure are performed when the computer program is executed by the processor 601. According to the embodiment of the present disclosure, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0127] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices, magnetic storage devices, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of signals on a network medium, and downloaded and installed through the communication part 609, and / or installed from a removable medium 611. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0128] In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 609, and / or installed from the removable medium 611. When the computer program is executed by the processor 601, the above functions defined in the system of the embodiment of the present disclosure are performed. According to the embodiment of the present disclosure, the system, device, apparatus, module, unit, etc. described above can be implemented by a computer program module.

[0129] According to an embodiment of the present disclosure, the program code for executing the computer program provided by the embodiment of the present disclosure can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, using an Internet service provider to connect through the Internet).

[0130] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0131] It will be appreciated by those skilled in the art that the features described in the various embodiments of the present disclosure may be combined and / or combined in a variety of ways, even if such combinations or combinations are not explicitly described in the present disclosure. In particular, without departing from the spirit and teachings of the present disclosure, the features described in the various embodiments of the present disclosure may be combined and / or combined in a variety of ways. All of these combinations and / or combinations fall within the scope of the present disclosure.

[0132] The embodiments of the present disclosure are described above. However, these embodiments are only for illustrative purposes and are not intended to limit the scope of the present disclosure. Although the embodiments are described above, this does not mean that the measures in the various embodiments cannot be used in combination to advantage. Without departing from the scope of the present disclosure, those skilled in the art may make a variety of substitutions and modifications, which should all fall within the scope of the present disclosure.

Claims

1. A product recommendation method, characterized in that: The method comprises: Acquire the object's initial evaluation information and classification information of the product, wherein the initial evaluation information includes first initial information and second initial information, and the second time corresponding to the second initial information is different from the first time corresponding to the first initial information; Performing weighted processing on the classification information based on complexity information and risk level information corresponding to the product, and performing weighted processing on the initial evaluation information based on time difference information, to obtain target information, wherein the time difference information represents a time difference between the first moment and the second moment; and The target information is input into a recommendation model, and a product recommendation result corresponding to the object is output.

2. The recommendation method according to claim 1, characterized in that: The classification information is weighted based on the complexity information and risk level information corresponding to the product, and the initial evaluation information is weighted based on the time difference information to obtain target information, including: Performing weighted processing on the classification information based on the complexity information and the risk level information to obtain a weighted classification result; Using the time difference information to perform weighted processing on the initial evaluation information to obtain weighted evaluation information; and The weighted classification result and abnormal information in the weighted evaluation information are processed to obtain the target information.

3. The recommendation method according to claim 2, characterized in that: The weighted evaluation information includes satisfaction information and return visit information; The initial evaluation information is weighted by using the time difference information to obtain weighted evaluation information, including: Determine the time difference between the first moment and the second moment; as well as The initial evaluation information is processed based on the time difference information and a preset weighting strategy to obtain the weighted evaluation information.

4. The recommendation method according to claim 1, characterized in that: The method further comprises: Obtaining the amount information, frequency information and behavior information corresponding to the product and their corresponding weight information; and The initial evaluation information is determined by using a weighted average algorithm based on the weight information, the amount information, the frequency information and the behavior information.

5. The recommendation method according to claim 2, characterized in that: Processing the weighted classification result and abnormal information in the weighted evaluation information to obtain the target information includes: Determining target values ​​corresponding to the weighted classification result and the weighted evaluation information respectively; Replacing the abnormal information with the target value to obtain updated classification results and updated evaluation information; and The update classification result and the update evaluation information are normalized to obtain the target information.

6. The recommendation method according to claim 1, characterized in that: The target information includes target classification information and target evaluation information, and the product recommendation result represents the recommendation probability values ​​of multiple products corresponding to the object; Inputting the target information into a recommendation model and outputting a product recommendation result corresponding to the object includes: The recommendation probability value is obtained based on the target classification information and the target evaluation information.

7. The recommendation method according to claim 1, characterized in that: The recommendation model is trained by the following operations: Obtaining sample initial evaluation information and sample classification information of the sample object for the sample product, wherein the sample initial evaluation information and the sample classification information include respective corresponding labels; An initial recommendation model is trained using the sample initial evaluation information, the sample classification information and the label to obtain the recommendation model.

8. The recommendation method according to claim 7, characterized in that: The target parameters of the recommendation model are determined by the following operations: Initializing an initial population corresponding to an initial parameter, wherein the initial population includes a plurality of individuals, and the individuals represent initial values ​​corresponding to the initial parameter; Using a fitness function to evaluate the plurality of individuals to obtain an initial evaluation result; Determine a parent individual based on the evaluation result, wherein the parent individual is multiple; Generate offspring individuals using a crossover algorithm and a plurality of the parent individuals; Update the parent individuals based on the offspring individuals to obtain an intermediate population; The intermediate population is updated cyclically until a stopping strategy is satisfied to obtain the target parameter; wherein the stopping strategy includes at least one of the following: the number of cycles is greater than or equal to a preset number threshold; the initial evaluation result is less than or equal to a preset evaluation threshold.

9. A product recommendation device, characterized in that: The device comprises: An information acquisition module, used to acquire the object's initial evaluation information and classification information on the product, wherein the initial evaluation information includes first initial information and second initial information, and the second time corresponding to the second initial information is different from the first time corresponding to the first initial information; a weighting module, configured to perform weighted processing on the classification information based on complexity information and risk level information corresponding to the product, and to perform weighted processing on the initial evaluation information based on time difference information, so as to obtain target information, wherein the time difference information represents a time difference between the first moment and the second moment; and The result output module is used to input the target information into the recommendation model and output the product recommendation result corresponding to the object.

10. An electronic device comprising: one or more processors; a memory for storing one or more computer programs, It is characterized in that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 8.

11. A computer-readable storage medium having a computer program or instruction stored thereon, characterized in that: When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

12. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images