Product recommendation method, device, computer equipment and storage medium

Abstract

The present application relates to a product recommendation method, apparatus, computer equipment, and storage medium, and relates to the field of big data and can be used in the field of financial technology. The method comprises: determining the similarity between a target customer and each reference customer, where a reference customer is a customer holding at least one product to be recommended; determining a target recommended product from the products to be recommended held by each reference customer based on the similarity between the target customer and each reference customer; and generating a recommendation plan for the target customer based on the target recommended product and the recommendation conditions corresponding to the target recommended product, the recommendation plan being used to recommend the target recommended product to the target customer if the target customer meets the recommendation conditions. The use of this method can improve the accuracy of product recommendations.

Description

Technical Field

[0001] The present application relates to the field of big data, and in particular to a product recommendation method, apparatus, computer equipment, and storage medium. Background Art

[0002] As banking products become more homogenized and market competition becomes increasingly fierce, the fight for high-quality customers has become a focal point among peers. At the same time, customer demands for banking products are becoming increasingly diversified and personalized. Therefore, providing customers with personalized product recommendations and improving recommendation efficiency are of great significance.

[0003] In related technologies, a recommendation system can be used to implement online product recommendations, that is, after determining the products to be recommended to customers, products are recommended to customers through online recommendation methods.

[0004] In current related technologies, the product recommendation method is single and inflexible, resulting in low product recommendation accuracy. Summary of the Invention

[0005] Based on this, it is necessary to provide a product recommendation method, device, computer equipment and storage medium to address the above technical problems.

[0006] In a first aspect, the present application provides a product recommendation method. The method comprises:

[0007] Determine the similarity between the target customer and each reference customer, wherein the reference customer is a customer who holds at least one product to be recommended;

[0008] Determining a target recommended product from the products to be recommended held by each of the reference customers based on the similarity between the target customer and each of the reference customers;

[0009] A recommendation plan for the target customer is generated based on the target recommended product and the recommendation conditions corresponding to the target recommended product. The recommendation plan is used to indicate that the target recommended product is recommended to the target customer if the target customer meets the recommendation conditions.

[0010] In one embodiment, determining a target recommended product from the products to be recommended held by each reference customer based on the similarity between the target customer and each reference customer includes:

[0011] determining a target reference customer from among the reference customers based on the similarity between the target customer and the reference customers;

[0012] For any of the to-be-recommended products held by any of the target reference customers, determining the degree of intimacy between the target customer and the to-be-recommended product, wherein the intimacy is used to represent the customer's preference for the product;

[0013] According to the intimacy, a target recommended product is determined from the products to be recommended held by each target reference customer.

[0014] In one embodiment, determining the intimacy between any of the target reference customers and the product to be recommended, for any of the target reference customers, includes:

[0015] For any of the target reference customers, determining a reference intimacy between the target reference customer and each of the to-be-recommended products held by the target reference customer;

[0016] For any of the to-be-recommended products held by the target reference customer, the intimacy between the target customer and the to-be-recommended product is determined based on the reference intimacy and the similarity between the target customer and the target reference customer.

[0017] In one embodiment, determining the reference intimacy between the target reference customer and each of the to-be-recommended products held by the target reference customer includes:

[0018] For any of the products to be recommended, obtaining product holding data corresponding to various product holding characteristics of the target reference customer;

[0019] For any of the product holding characteristics, determining a characteristic value of the target reference customer for the product holding characteristic based on the ranking of the product holding data corresponding to the product holding characteristic of the target reference customer among the reference customers, wherein the characteristic value is used to represent the importance of the product holding characteristic;

[0020] The reference intimacy between the target reference customer and the product to be recommended is determined based on the feature value of the target reference customer for each feature held by the product and the weight corresponding to the feature held by each product.

[0021] In one embodiment, determining the similarity between the target customer and each reference customer includes:

[0022] Acquiring characteristic data of the target customer and characteristic data of each reference customer based on the characteristics of each customer;

[0023] For any of the reference customers, the similarity between the target customer and the reference customer is determined based on the feature data of the target customer and the feature data of the reference customer.

[0024] In one embodiment, obtaining the characteristic data of the target customer and the characteristic data of each reference customer based on the characteristics of each customer includes:

[0025] Determining a target product field, where the target product field corresponds to at least one of the products to be recommended;

[0026] Determining a reference customer based on the target product field, wherein at least one of the products to be recommended held by the reference customer belongs to the target product field;

[0027] According to the customer characteristics corresponding to the target product field, characteristic data of the target customer and characteristic data of each of the reference customers are obtained.

[0028] In one embodiment, for any of the reference customers, determining the similarity between the target customer and the reference customer based on the characteristic data of the target customer and the characteristic data of the reference customer includes:

[0029] Determining the information value of each of the customer characteristics, and using the information value of each of the customer characteristics as a weighted weight of each of the customer characteristics;

[0030] For any of the reference customers, determining the distance between the target customer and the reference customer in terms of each of the customer characteristics;

[0031] For any of the reference customers, the distances between the target customers and the reference customers in terms of the customer characteristics are fused according to the weighted weights of the customer characteristics to obtain the similarity between the target customer and the reference customer.

[0032] In one embodiment, determining the information value of each of the customer characteristics includes:

[0033] Performing binning processing on the customer features to obtain multiple customer feature bins;

[0034] For any of the customer feature bins, determine the weight of evidence for the customer feature bin based on the number of customers who purchased the product corresponding to the customer feature bin, the number of customers who did not purchase the product corresponding to the customer feature bin, the number of customers who purchased the product corresponding to the customer feature, and the number of customers who did not purchase the product corresponding to the customer feature;

[0035] The information value of the customer feature is determined based on the evidence weight of each of the customer feature bins.

[0036] In one embodiment, generating a recommendation plan for the target customer based on the target recommended product and the recommendation conditions corresponding to the target recommended product includes:

[0037] Obtaining a customer tag corresponding to the target customer;

[0038] Obtaining the recommendation conditions corresponding to each of the target recommended products;

[0039] For any of the target recommended products, if the customer tag corresponding to the target customer meets the recommendation condition corresponding to the target recommended product, the target recommended product is used as the recommended product;

[0040] A recommendation plan for the target customer is generated based on each of the recommended products and the recommendation conditions corresponding to each of the recommended products.

[0041] In a second aspect, the present application further provides a product recommendation device. The device comprises:

[0042] A first determination module is configured to determine the similarity between the target customer and each reference customer, wherein the reference customer is a customer who holds at least one product to be recommended;

[0043] A second determining module is configured to determine a target recommended product from the products to be recommended held by each reference customer based on the similarity between the target customer and each reference customer;

[0044] A generation module is used to generate a recommendation plan for the target customer based on the target recommended product and the recommendation conditions corresponding to the target recommended product. The recommendation plan is used to indicate that the target recommended product is recommended to the target customer when the target customer meets the recommendation conditions.

[0045] In one embodiment, the second determining module is further configured to:

[0046] determining a target reference customer from among the reference customers based on the similarity between the target customer and the reference customers;

[0047] For any of the to-be-recommended products held by any of the target reference customers, determining the degree of intimacy between the target customer and the to-be-recommended product, wherein the intimacy is used to represent the customer's preference for the product;

[0048] According to the intimacy, a target recommended product is determined from the products to be recommended held by each target reference customer.

[0049] In one embodiment, the second determining module is further configured to:

[0050] For any of the target reference customers, determining a reference intimacy between the target reference customer and each of the to-be-recommended products held by the target reference customer;

[0051] For any of the to-be-recommended products held by the target reference customer, the intimacy between the target customer and the to-be-recommended product is determined based on the reference intimacy and the similarity between the target customer and the target reference customer.

[0052] In one embodiment, the second determining module is further configured to:

[0053] For any of the products to be recommended, obtaining product holding data corresponding to various product holding characteristics of the target reference customer;

[0054] For any of the product holding characteristics, determining a characteristic value of the target reference customer for the product holding characteristic based on the ranking of the product holding data corresponding to the product holding characteristic of the target reference customer among the reference customers, wherein the characteristic value is used to represent the importance of the product holding characteristic;

[0055] The reference intimacy between the target reference customer and the product to be recommended is determined based on the feature value of the target reference customer for each feature held by the product and the weight corresponding to the feature held by each product.

[0056] In one embodiment, the first determining module is further configured to:

[0057] Acquiring characteristic data of the target customer and characteristic data of each reference customer based on the characteristics of each customer;

[0058] For any of the reference customers, the similarity between the target customer and the reference customer is determined based on the feature data of the target customer and the feature data of the reference customer.

[0059] In one embodiment, the first determining module is further configured to:

[0060] Determining a target product field, where the target product field corresponds to at least one of the products to be recommended;

[0061] Determining a reference customer based on the target product field, wherein at least one of the products to be recommended held by the reference customer belongs to the target product field;

[0062] According to the customer characteristics corresponding to the target product field, characteristic data of the target customer and characteristic data of each of the reference customers are obtained.

[0063] In one embodiment, the first determining module is further configured to:

[0064] Determining the information value of each of the customer characteristics, and using the information value of each of the customer characteristics as a weighted weight of each of the customer characteristics;

[0065] For any of the reference customers, determining the distance between the target customer and the reference customer in terms of each of the customer characteristics;

[0066] For any of the reference customers, the distances between the target customers and the reference customers in terms of the customer characteristics are fused according to the weighted weights of the customer characteristics to obtain the similarity between the target customer and the reference customer.

[0067] In one embodiment, the first determining module is further configured to:

[0068] Performing binning processing on the customer features to obtain multiple customer feature bins;

[0069] For any of the customer feature bins, determine the weight of evidence for the customer feature bin based on the number of customers who purchased the product corresponding to the customer feature bin, the number of customers who did not purchase the product corresponding to the customer feature bin, the number of customers who purchased the product corresponding to the customer feature, and the number of customers who did not purchase the product corresponding to the customer feature;

[0070] The information value of the customer feature is determined based on the evidence weight of each of the customer feature bins.

[0071] In one embodiment, the generating module is further configured to:

[0072] Obtaining a customer tag corresponding to the target customer;

[0073] Obtaining the recommendation conditions corresponding to each of the target recommended products;

[0074] For any of the target recommended products, if the customer tag corresponding to the target customer meets the recommendation condition corresponding to the target recommended product, the target recommended product is used as the recommended product;

[0075] A recommendation plan for the target customer is generated based on each of the recommended products and the recommendation conditions corresponding to each of the recommended products.

[0076] In a third aspect, the present application further provides a computer device, wherein the computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements any of the above methods when executing the computer program.

[0077] In a fourth aspect, the present application further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, any of the above methods is implemented.

[0078] In a fifth aspect, the present application further provides a computer program product, wherein the computer program product comprises a computer program, and the computer program implements any of the above methods when executed by a processor.

[0079] The above-mentioned product recommendation method, device, computer equipment and storage medium can determine the target recommended product for the target customer from the recommended products held by the reference customer by determining the similarity between the target customer and the reference customer, and generate a recommendation plan for the target customer through the recommendation conditions corresponding to the target recommended product, so as to assist offline marketing personnel in recommending the target recommended product to the target customer. The embodiment of the present application generates a recommendation plan for the target customer to assist the marketing personnel in offline product marketing through the recommendation plan, that is, the recommendation plan instructs the marketing personnel to recommend the target recommended product to the target customer when the target customer meets the recommendation conditions. The present application integrates online marketing and offline marketing through the recommendation plan, which can improve the flexibility of product recommendation, enrich the product recommendation method, and improve the accuracy of product recommendation. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] Figure 1 Schematic diagram of a product recommendation method in one embodiment;

[0081] Figure 2 104 is a flow chart of step 104 in one embodiment;

[0082] Figure 3 204 is a flow chart of step 204 in one embodiment;

[0083] Figure 4 302 is a flowchart of an embodiment;

[0084] Figure 5 102 is a flow chart of step 102 in one embodiment;

[0085] Figure 6 5 is a flow chart of step 502 in one embodiment;

[0086] Figure 7 5 is a flow chart of step 504 in one embodiment;

[0087] Figure 8 FIG. 7 is a flow chart of step 702 in one embodiment;

[0088] Figure 9 106 is a flow chart of step 106 in one embodiment;

[0089] Figure 10 A schematic diagram of a product recommendation method in one embodiment;

[0090] Figure 11 is a structural block diagram of a product recommendation device in one embodiment;

[0091] Figure 12 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0092] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0093] In one embodiment, Figure 1 As shown, a product recommendation method is provided. This embodiment uses the method applied to a terminal as an example for illustration. It is understandable that the method can also be applied to a server, or to a system including a terminal and a server, and implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0094] Step 102 : Determine the similarity between the target customer and each reference customer, where the reference customer is a customer who holds at least one product to be recommended.

[0095] In this embodiment of the present application, a target customer is a customer who does not hold the product to be recommended and to whom a product recommendation is to be made, and a reference customer is a customer who already holds at least one product to be recommended. By determining the similarity between the target customer and each reference customer, this embodiment of the present application can determine a target recommended product to be recommended to the target customer from among the products to be recommended held by the reference customers.

[0096] It should be noted that the embodiment of the present application does not specifically limit the method for determining the similarity between the target customer and each reference customer. Any method that can determine the similarity between the target customer and each reference customer is applicable to the embodiment of the present application.

[0097] Step 104 : Determine a target recommended product from the products to be recommended held by each reference customer based on the similarity between the target customer and each reference customer.

[0098] In this embodiment of the present application, the target recommended product is a product that is determined to be more suitable for the target customer and has a higher predicted recommendation success rate, among all the products to be recommended held by each reference customer. The target recommended product can be part of the products to be recommended, or it can be all the products to be recommended, and this embodiment of the present application does not specifically limit this.

[0099] For example, embodiments of the present application can determine, based on the similarity between the target customer and multiple reference customers, a target reference customer whose similarity to the target customer meets a preset condition, thereby determining a target recommended product for the target customer from among the recommended products held by the target reference customer. For example, the preset condition can include customers with a high similarity ranking, or customers with a similarity greater than a threshold, etc., and embodiments of the present application do not specifically limit the preset condition.

[0100] Step 106 , generating a recommendation plan for the target customer based on the target recommended product and the recommendation conditions corresponding to the target recommended product. The recommendation plan is used to indicate that the target recommended product should be recommended to the target customer if the target customer meets the recommendation conditions.

[0101] In an embodiment of the present application, the recommendation conditions corresponding to the target recommended product are preset conditions, and the recommendation conditions are logically associated with the target recommended product, that is, customers who meet the recommendation conditions will have a higher demand or acceptance level for the recommended product.

[0102] For example, if historical data shows that customers with low risk tolerance have a high demand for net asset value (NAV) financial products, then when the product to be recommended is a NAV financial product, one of the recommendation conditions for the NAV financial product may be the customer's low risk tolerance. When a target customer meets any of the recommendation conditions for a target recommended product, it can be inferred that the target customer may need the target recommended product, and the target recommended product can be recommended to the target customer. However, when a target customer does not meet any of the recommendation conditions for a target recommended product, it can be estimated that even if the target recommended product is recommended to the target customer, the target customer is unlikely to purchase the target recommended product due to the lack of reasonable reasons for recommending the target recommended product to the target customer. Therefore, the target recommended product can be not recommended to the target customer.

[0103] For example, if the target recommended products corresponding to a target customer are: XXX net value wealth management product, XXXX growth insurance, among which the recommendation conditions corresponding to the XXX net value wealth management product are: low risk tolerance, recent transfer of funds from an account with the same name outside the bank, and monthly income exceeding a fixed threshold, and the target customer meets the recommendation condition of recent transfer of funds from an account with the same name outside the bank, then you can indicate in the recommendation plan to recommend the target recommended product to the target customer, for example: "Recently, there has been a transfer from an account with the same name outside the bank to the bank, and the XXXX net value wealth management product can be recommended."

[0104] Similarly, if the target customer also meets the recommendation condition of "recent consumption of baby products" in XXXX Growth Insurance, the recommendation plan can also indicate that XXXX Growth Insurance should be recommended to the target customer.

[0105] In this embodiment, the recommendation plan can also include the customer's basic information (such as name, gender, birthday, etc.), fixed recommended products and recommendation reasons (for example, "If the customer needs money recently, T+0 financial management can be recommended"), recent activity information, recommendation channels, etc., to form a complete natural language recommendation plan. For example, the recommendation plan generated for this target customer can be as follows Table 1:

[0106] Table 1

[0107]

[0108] After the recommendation plan is generated, the recommendation plan can be displayed in the display interface so that the recommendation plan can be used as a reference for offline marketers when making offline product recommendations.

[0109] The product recommendation method provided in the embodiment of the present application can determine the target recommended product for the target customer from the products to be recommended held by the reference customer by determining the similarity between the target customer and the reference customer, and generate a recommendation plan for the target customer through the recommendation conditions corresponding to the target recommended product, so as to assist offline marketing personnel in recommending the target recommended product to the target customer. The embodiment of the present application generates a recommendation plan for the target customer to assist marketing personnel in offline product marketing through the recommendation plan, that is, the recommendation plan instructs marketing personnel to recommend the target recommended product to the target customer when the target customer meets the recommendation conditions. The present application integrates online marketing and offline marketing through the recommendation plan, which can improve the flexibility of product recommendations, enrich product recommendation methods, and improve the accuracy of product recommendations.

[0110] In one embodiment, Figure 2 As shown, in step 104, based on the similarity between the target customer and each reference customer, a target recommended product is determined from the products to be recommended held by each reference customer, including:

[0111] Step 202 : determining a target reference customer from among the reference customers based on the similarity between the target customer and the reference customers.

[0112] In the embodiments of the present application, a target reference customer is a reference customer whose similarity to the target customer meets a preset condition. The preset condition can be set by a person skilled in the art as needed. For example, a reference customer meeting the preset condition can be a reference customer with a high similarity ranking, or a reference customer with a similarity greater than a threshold.

[0113] Step 204 : for any product to be recommended held by any target reference customer, determine the intimacy between the target customer and the product to be recommended. The intimacy is used to represent the customer's preference for the product.

[0114] Step 206 : Determine a target recommended product from the products to be recommended held by each target reference customer based on the intimacy.

[0115] In this embodiment of the present application, intimacy is used to characterize a customer's preference for a product. A higher intimacy indicates a customer's willingness to hold the product or a stronger preference for the product. For example, the target customer's preference for each recommended product held by the target reference customer can be determined by determining the target reference customer's preference for the recommended product, and then determining the target customer's intimacy for each recommended product held by the target reference customer based on the similarity between the target reference customer and the target customer.

[0116] After determining the target reference customer, the embodiment of the present application can further determine the intimacy between the target customer and each target reference customer’s held products to be recommended based on the similarity between the target customer and each target reference customer. After obtaining the intimacy between the target customer and each product to be recommended, multiple target recommended products can be determined from each product to be recommended based on the intimacy corresponding to each product to be recommended. For example, each product to be recommended can be sorted from high to low according to the intimacy corresponding to each product to be recommended, and the products can be determined as target recommended products in order until the number of target recommended products meets the threshold. The threshold is a pre-set numerical value, and its specific value can be set by those skilled in the art as needed. For example, when there are more products to be recommended, a higher threshold can be set. When there are fewer products to be recommended, a lower threshold can be set. Examples of target recommended products obtained are shown in Table 2 below:

[0117] Table 2

[0118] Customer Number First priority target recommended products Second priority target recommended products The third recommended product 04************A XXX net value financial products XXX Fund XXXX Whole Life Insurance (5-year payment) 04************B XXXX Insurance (Participating Type) XXX Fund XXX net value financial products 04************C XXXX no fixed term financial products XXXX Whole Life Insurance (10-year payment) XXX Fund

[0119] In the product recommendation method provided in the embodiment of the present application, the target recommended product is determined from the products to be recommended held by the target reference customer through the intimacy between the target customer and the products to be recommended held by each target reference customer. That is, among the products to be recommended held by the target reference customers who are relatively similar to the target customer, the products to be recommended that the target reference customer likes more are used as the target recommended products. Therefore, the probability of the target recommended product being liked by the target customer can be increased, and the accuracy of product recommendation can be improved.

[0120] In one embodiment, Figure 3 As shown, in step 204, for any product to be recommended held by any target reference customer, determining the intimacy between the target customer and the product to be recommended includes:

[0121] Step 302 : For any target reference customer, determine the reference intimacy between the target reference customer and each to-be-recommended product held by the target reference customer.

[0122] Step 304 : For any product to be recommended held by the target reference customer, determine the intimacy between the target customer and the product to be recommended based on the reference intimacy and the similarity between the target customer and the target reference customer.

[0123] In an embodiment of the present application, the reference intimacy between the target reference customer and each product to be recommended held by the target reference customer can be determined, and the intimacy between the target customer and the product to be recommended can be determined based on the similarity between the target customer and the target reference customer. For example, for any target reference customer, the product of the similarity between the target customer and the target reference customer and the reference intimacy between the target reference customer and each product to be recommended held by the target reference customer can be used as the intimacy between the target customer and each product to be recommended held by the target reference customer. After repeating the above steps for each target reference customer, the intimacy between the target customer and the products to be recommended held by all target reference customers can be obtained.

[0124] It should be noted that if there are multiple target reference customers holding the same product to be recommended, that is, for a product to be recommended, there are multiple intimacy values ​​obtained by multiplying the similarity and the reference intimacy, the intimacy between the target customer and the product to be recommended can be determined by any of the following methods: taking the maximum value among the values ​​and using the maximum value as the intimacy between the target customer and the product to be recommended; taking the minimum value among the values ​​and using the minimum value as the intimacy between the target customer and the product to be recommended; taking the average value of the values ​​and using the average value as the intimacy between the target customer and the product to be recommended, etc. This embodiment of the application does not specifically limit this.

[0125] The product recommendation method provided in the embodiment of the present application determines the target recommended product from the products to be recommended held by the target reference customer through the intimacy between the target customer and the products to be recommended held by each target reference customer, as well as the similarity between the target customer and each target reference customer. That is, among the products to be recommended held by target reference customers who are relatively similar to the target customer, the products to be recommended that the target reference customer likes more are used as target recommended products. Therefore, the probability of the target recommended product being liked by the target customer can be increased, and the accuracy of product recommendation can be improved.

[0126] In one embodiment, Figure 4 As shown, in step 302, determining the reference intimacy between the target reference customer and each to-be-recommended product held by the target reference customer includes:

[0127] Step 402: For any product to be recommended, obtain product holding data corresponding to various product holding characteristics of the target reference customer.

[0128] Step 404 , for any product holding feature, determine the feature value of the target reference customer for the product holding feature based on the ranking of the product holding data corresponding to the product holding feature among the reference customers. The feature value is used to represent the importance of the product holding feature.

[0129] Step 406 : Determine the reference intimacy between the target reference customer and the product to be recommended based on the feature value of each product held by the target reference customer and the weight corresponding to each product held feature.

[0130] In an embodiment of the present application, the product holding feature is used to characterize the status type of the customer holding the product, such as the holding amount, holding time, etc., and the product holding data is used to characterize the status data corresponding to the product holding feature, such as: amount data, holding time data, etc.

[0131] For any product to be recommended, embodiments of the present application can sort the product holding data corresponding to each product holding characteristic of all reference customers holding the product to be recommended, thereby obtaining multiple product holding characteristic ranking sequences. For example, if product holding characteristics include holding amount, holding amount as a percentage of assets, holding time, and transaction frequency, after sorting all reference customers holding the product, four product holding characteristic ranking sequences can be obtained.

[0132] The ranking of the reference customer in the feature ranking sequence of each product can correspond to a feature value. The feature value is a pre-set numerical value, and the specific value can be set according to actual needs. In the embodiment of the present application, since it is necessary to achieve the purpose that when the customer prefers the product more, the customer's intimacy with the product is higher, the embodiment of the present application assigns a high feature value to a high ranking and a low feature value to a low ranking. For example, a feature value of 10 can be assigned to rankings 1-10, a feature value of 9 can be assigned to rankings 11-20... a feature value of 1 can be assigned to rankings 91-100, and a feature value of 0 can be assigned to rankings other than 100.

[0133] It should be noted that the above-mentioned method of setting the feature value is only an example in the embodiment of the present application. In fact, there is no specific limitation on the method of setting the feature value in the embodiment of the present application. Any method that can achieve high ranking and high feature value, and low ranking and low feature value is applicable to the embodiment of the present application.

[0134] Furthermore, based on the characteristic value corresponding to the target reference customer's ranking in the ranking sequence of each product holding feature, and the weight corresponding to each product holding feature, the reference intimacy between the target reference customer and the product to be recommended can be determined, wherein the reference intimacy is positively correlated with the characteristic value of the target reference customer for the product holding feature. Exemplarily, the characteristic value can be weighted and summed by the weight corresponding to each product holding feature to obtain the intimacy of the target reference customer towards the recommended product. The weight corresponding to the product holding feature is a pre-set value, which can be selected by those skilled in the art according to actual needs. For example, when the product holding feature contributes more to the judgment of whether the customer likes the product, the weight corresponding to the product holding feature can be set higher. When the product holding feature contributes less to the judgment of whether the customer likes the product, the weight corresponding to the product holding feature can be set lower.

[0135] For example, if the target reference customer ranks 15th in the ranking sequence of the amount held for a certain product to be recommended (corresponding to eigenvalue 9), ranks 21st in the ranking sequence of the proportion of the amount held to assets (corresponding to eigenvalue 8), ranks 72nd in the ranking sequence of the holding time (corresponding to eigenvalue 3), and ranks 3rd in the ranking sequence of the transaction frequency (corresponding to eigenvalue 10), and the weight of the amount held is 0.1, the weight of the proportion of the amount held to assets is 0.3, the weight of the holding time is 0.3, and the weight of the transaction frequency is 0.3, then the reference intimacy of the target reference customer for the product to be recommended is: 0.1×9+0.3×8+0.3×3+0.3×10=7.2.

[0136] The product recommendation method provided in the embodiment of the present application can quantify the degree to which the target reference customer likes each to-be-recommended product through a reference intimacy, so that the intimacy between the target reference customer and each to-be-recommended product can be determined through the reference intimacy between the target reference customer and each to-be-recommended product, and then the intimacy can be used as an indicator to predict the degree to which the target customer is likely to like each to-be-recommended product. In other words, among the to-be-recommended products held by target reference customers who are relatively similar to the target customer, the to-be-recommended product that the target reference customer likes more is used as the target recommended product, thereby increasing the probability that the target recommended product will be liked by the target customer, and improving the accuracy of product recommendations.

[0137] In one embodiment, Figure 5 As shown, in step 102, determining the similarity between the target customer and each reference customer includes:

[0138] Step 502: Acquire characteristic data of target customers and characteristic data of reference customers based on the characteristics of each customer.

[0139] Step 504 : For any reference customer, determine the similarity between the target customer and the reference customer based on the characteristic data of the target customer and the characteristic data of the reference customer.

[0140] In this embodiment of the present application, characteristic data of a target customer and characteristic data of each reference customer can be obtained based on customer characteristics, and the similarity between the target customer and each reference customer can be determined. Customer characteristics are used to represent a customer's basic information, transaction behavior, and asset status. For example, basic customer information such as gender and age, as well as customer transaction information, customer asset information, customer in-bank product holdings, and customer behavior information can be included.

[0141] It should be noted that the embodiments of the present application do not specifically limit the method for determining the similarity between the target customer and the reference customer. Any method that can determine the similarity between the target customer and the reference customer based on the characteristic data of the target customer and the characteristic data of the reference customer is applicable to the embodiments of the present application.

[0142] For example, if the customer characteristics are: income, asset status, and risk rating, the similarity between the target customer and each reference customer can be determined based on the corresponding data of the target customer on income, asset status, and risk rating, as well as the corresponding data of each reference customer on income, asset status, and risk rating.

[0143] The product recommendation method provided in the embodiment of the present application can determine the similarity between the target customer and each reference customer based on customer characteristics, so as to determine the target reference customer from the reference customers based on the similarity, and then can determine the target recommended product from the products to be recommended held by the target reference customers who are relatively similar to the target customer. Therefore, the probability of the target recommended product being liked by the target customer can be increased, and the accuracy of product recommendation can be improved.

[0144] In one embodiment, Figure 6 As shown, in step 502, based on the characteristics of each customer, characteristic data of the target customer and characteristic data of each reference customer are obtained, including:

[0145] Step 602: Determine a target product field, where the target product field corresponds to at least one product to be recommended.

[0146] In the embodiments of the present application, the target product field refers to the category of products, such as fund products, insurance products, etc. The specific products to be recommended included in each target product field can be set by those skilled in the art according to actual needs.

[0147] In actual applications, the target product area can be selected by offline marketers based on marketing needs. For example, when an offline marketer needs to recommend fund products to a target customer, they can select fund products as the target product area. In this case, the generated target recommended products will all be fund products. The offline marketer can then recommend fund products to the target customer based on the target recommended products and the recommendation plan generated based on the target recommended products. Alternatively, offline marketers often have promotion authority for a specific product area. Therefore, the product area to which the marketer belongs can be determined based on the marketer's account information logged into the current system, and this product area can be used as the target product area.

[0148] Step 604 : determining a reference customer based on the target product field, wherein at least one product to be recommended held by the reference customer belongs to the target product field.

[0149] Step 606 : Acquire characteristic data of the target customer and characteristic data of each reference customer based on the customer characteristics corresponding to the target product field.

[0150] In an embodiment of the present application, characteristic data of the target customer and characteristic data of each reference customer can be obtained based on the customer characteristics corresponding to the target product field, and the similarity between the target customer and the reference customer can be determined based on the characteristic data of the target customer and the characteristic data of each reference customer to determine the target recommended product from the products to be recommended held by the reference customer.

[0151] The customer characteristics corresponding to the target product area may be characteristics that have a significant impact on whether a customer purchases the recommended product in that target product area. For example, if the target recommendation area is fund products, the customer characteristics related to fund products may be characteristics such as the customer's income, asset status, and risk rating; if the target recommendation area is insurance products, the customer characteristics related to insurance products may be characteristics such as the customer's age and asset status. The customer characteristics corresponding to each target product area may be the same or different, and this embodiment of the application does not specifically limit this.

[0152] After obtaining the customer characteristics corresponding to the target product field, the embodiment of the present application can further determine the similarity between the target customer and each reference customer based on the characteristic data of the target customer and the characteristic data of each reference customer, so as to determine the target recommended product from the products to be recommended held by the reference customer.

[0153] The product recommendation method provided in the embodiment of the present application can determine the similarity between the target customer and each reference customer through the customer characteristics corresponding to the target product field after determining the target product field, so as to determine the target recommended product from the products to be recommended held by the reference customer. The embodiment of the present application can complete the recommendation of products in different target product fields by presetting different customer characteristics for different target product fields, thereby enhancing the versatility of the method. Moreover, the embodiment of the present application only calculates the similarity between the target customer and each reference customer through the customer characteristics that have a greater impact on the target product field, so there is no need to involve all customer characteristics in the similarity calculation, thereby simplifying the steps of similarity calculation and improving the generation efficiency of recommendation schemes.

[0154] In one embodiment, Figure 7 As shown, in step 408, for any reference customer, the similarity between the target customer and the reference customer is determined based on the characteristic data of the target customer and the characteristic data of the reference customer, including:

[0155] Step 702: Determine the information value of each customer feature, and use the information value of each customer feature as the weighted weight of each customer feature.

[0156] In the embodiments of this application, Information Value is a metric that characterizes the ability of a customer characteristic to predict whether a customer is a likely purchaser. The higher the Information Value, the more effective the customer characteristic is at predicting whether a customer is a likely purchaser; the lower the Information Value, the less effective the customer characteristic is at predicting whether a customer is a likely purchaser. Information Value can be derived based on the number of customers who purchased a product and the number of customers who did not purchase the product, corresponding to the customer characteristic.

[0157] After calculating the information value of each customer feature, the embodiment of the present application can further use the information value of each customer feature as the weighted weight of each customer feature.

[0158] Step 704 : For any reference customer, determine the distance between the target customer and the reference customer in terms of various customer characteristics.

[0159] Step 706 : For any reference customer, based on the weighted weights of each customer feature, the distances between each target customer and the reference customer in each customer feature are fused to obtain the similarity between the target customer and the reference customer.

[0160] In this embodiment of the present application, the similarity between the target customer's feature data for each customer feature and the reference customer's feature data for each customer feature can be weighted and summed using the information value corresponding to the customer feature to obtain the similarity between the reference customer and the target customer. For example, the weighted Euclidean distance can be used to calculate the similarity between the reference customer and the target customer (see formula (1)):

[0161]

[0162] Where d is the similarity between the reference customer and the target customer, n is the total number of customer features, IV i is the IV value of the i-th customer feature, x 1i is the characteristic data of the reference customer on the i-th customer feature, x 2i It is the characteristic data of the target customer on the i-th customer feature.

[0163] The product recommendation method provided in the embodiment of the present application can obtain the similarity between the reference customer and the target customer by weighted summing the distance between the value of the target customer on each customer feature and the value of the reference customer on each customer feature. Among them, the weighted weight of the customer feature can be the information value of the customer feature. The embodiment of the present application uses the information value that characterizes the predictive ability of the customer feature as the weighted weight, which can make the customer features with high predictive ability contribute more to the similarity, and the customer features with low predictive ability contribute less to the similarity, so as to improve the accuracy of the similarity between the reference customer and the target customer.

[0164] In one embodiment, Figure 8 As shown, in step 702, determining the information value of each customer characteristic includes:

[0165] Step 802: binning the customer features to obtain multiple customer feature bins.

[0166] Step 804, for any customer feature bin, determine the evidence weight of the customer feature bin according to the number of customers who purchased the product corresponding to the customer feature bin, the number of customers who did not purchase the product corresponding to the customer feature bin, the number of customers who purchased the product corresponding to the customer feature, and the number of customers who did not purchase the product corresponding to the customer feature.

[0167] Step 806: Determine the information value of the customer feature based on the evidence weight of each customer feature bin.

[0168] In this embodiment of the present application, the customer characteristics can be first binned, and then the weight of evidence (WoE) of each bin can be calculated. The weight of evidence of each bin can then be weighted and summed to determine the information value of the customer characteristics. The WoE can be calculated by calculating the difference between the ratio of the number of customers who did not purchase the product in the bin to the total number of customers who did not purchase the product, and the ratio of the number of customers who purchased the product in the bin to the total number of customers who purchased the product (see formula (2)):

[0169]

[0170] Among them, WOE i is the evidence weight of the ith bin, Bad i is the number of customers who did not purchase the product in the i-th bin, Bad T is the total number of customers who have not purchased the product, Good i is the number of customers who purchased the product in the i-th bin, Good T is the total number of customers who purchased the product.

[0171] By weighting and summing the evidence weights of each bin in the customer feature, the information value of the customer feature can be obtained (see formula (3)):

[0172]

[0173] Where IV is the information value of the customer feature and n is the total number of bins.

[0174] The product recommendation method provided in the embodiments of the present application can calculate the information value of each customer feature through evidence weighting, and then use the information value of the customer feature as a weighted weight. The embodiments of the present application use the information value that characterizes the predictive ability of the customer feature as a weighted weight, so that customer features with high predictive ability contribute more to the similarity, while customer features with low predictive ability contribute less to the similarity, thereby improving the accuracy of the similarity between the reference customer and the target customer.

[0175] In one embodiment, Figure 9 As shown, in step 106, a recommendation plan for the target customer is generated based on the target recommended product and the recommendation conditions corresponding to the target recommended product, including:

[0176] Step 902: Obtain the customer tag corresponding to the target customer.

[0177] Step 904: Obtain the recommendation conditions corresponding to each target recommended product.

[0178] In an embodiment of the present application, the customer label corresponding to the target customer is used to describe the behavioral characteristics and asset characteristics of the target customer. For example, if a target customer has transferred money from an account with the same name outside the bank to an account within the bank in the past month, and has regular consumption records at a maternal and child products store in the past month, and through the personal information and asset information of the target customer, it can be judged that the target customer has a low risk tolerance, then the customer label corresponding to the target customer can be: "Recent transfer from an account with the same name outside the bank", "Recent consumption of baby products", "Low risk tolerance".

[0179] For example, the customer tag corresponding to the target customer can be stored in the corresponding information system, and the customer tag corresponding to the target customer can be obtained by calling the corresponding internal interface of the information system. The recommendation conditions corresponding to the target recommended products are pre-set conditions and can be pre-stored in an information storage area. The information storage area can be used to store the recommended products and the recommendation conditions corresponding to the recommended products. That is, after determining the target recommended products, the embodiment of the present application can obtain the recommendation conditions corresponding to each target recommended product from the information storage area.

[0180] Step 906 : For any target recommended product, if the customer tag corresponding to the target customer meets the recommendation condition corresponding to the target recommended product, the target recommended product is used as the recommended product.

[0181] Step 908: Generate a recommendation plan for the target customer based on each recommended product and the recommendation conditions corresponding to each recommended product.

[0182] In this embodiment of the present application, if any customer tag corresponding to a target customer satisfies any recommendation condition corresponding to a target recommended product, the target recommended product can be used as a recommended product, and a recommendation plan for the target customer is generated based on the recommended product and the recommendation condition corresponding to the recommended product. In other words, in this embodiment of the present application, before generating a recommendation plan, the target recommended products are first screened, and only the target recommended products whose customer tags corresponding to the target customer meet the recommendation condition are used as recommended products. In the recommendation plan, only the screened recommended products and the recommendation reasons corresponding to each recommended product are displayed.

[0183] For example, if a target customer's corresponding customer tags are: Age 3X, Recent Transfer from an External Account with the Same Name, and Recent Purchase of Baby Products, the target recommended products for this target customer are: XXX Net Value Wealth Management Product, XXX Fund, and XXXX Growth Insurance. The recommendation criteria for the XXX Net Value Wealth Management Product are: Low Risk Tolerance, Recent Transfer from an External Account with the Same Name, and Monthly Income Exceeding a Fixed Threshold. If the target customer meets the recommendation criteria of Recent Transfer from an External Account with the Same Name, this target recommended product can be selected as a recommended product and the recommendation plan will display the recommended product and the corresponding recommendation reason, for example: "Recent transfer from an external account with the same name to an internal account; XXXX Net Value Wealth Management Product is recommended."

[0184] If, however, the recommendation criteria for the XXX Fund are: monthly income exceeding a fixed threshold, high risk tolerance, or assets exceeding a fixed threshold, and the target customer does not meet any of these criteria, then the target recommended product is not considered a recommended product and may not be displayed in the recommendation plan. Similarly, if the target customer meets the "recent consumption of baby products" recommendation criteria for XXXX Growth Insurance, then XXXX Growth Insurance can also be recommended and displayed in the recommendation plan along with the reasoning behind the recommendation.

[0185] For example, the recommendation generated for this target customer might be: "There was a recent transfer from an external bank account with the same name to the bank, so we can recommend the XXXX net-worth wealth management product. Additionally, the customer has a history of regular purchases at maternity and baby stores, so we can try recommending XXXX growth insurance."

[0186] The product recommendation method provided in the embodiment of the present application first filters the target recommended products before generating a recommendation plan, and only recommends target recommended products whose customer tags of the target customers meet the recommendation conditions. In the recommendation plan, only the filtered recommended products and the corresponding recommendation reasons for each recommended product are displayed. Therefore, after obtaining the recommendation plan, offline marketers do not need to perform manual screening and judgment, and can directly recommend products to target customers based on the recommendation plan, thereby improving the accuracy of product recommendations.

[0187] In one embodiment, in step 106, a recommendation plan for the target customer is generated based on the target recommended product and the recommendation conditions corresponding to the target recommended product, including:

[0188] Get the recommendation conditions corresponding to each target recommended product.

[0189] Generate a recommendation plan for target customers based on each target recommended product and the recommendation conditions corresponding to each target recommended product.

[0190] In this embodiment of the present application, after obtaining the recommendation conditions corresponding to each target recommended product, a recommendation plan for the target customer can be directly generated based on each target recommended product and the recommendation conditions corresponding to each target recommended product. In other words, in this embodiment of the present application, the target recommended products are not filtered. Instead, all target recommended products and the recommendation conditions corresponding to the target recommended products are displayed in the recommendation plan. Offline marketers can then determine whether to recommend the target recommended products to the target customer based on the target recommended products and the recommendation conditions corresponding to the target recommended products.

[0191] For example, if the target recommended products for a target customer are: XXX net asset value wealth management product, XXX fund, and XXXX growth insurance, and the recommendation criteria for the XXX net asset value wealth management product are: low risk tolerance, recent transfers from an external account with the same name, and monthly income exceeding a fixed threshold; the recommendation criteria for the XXX fund are: monthly income exceeding a fixed threshold, high risk tolerance, and assets exceeding a fixed threshold; and the recommendation criteria for the XXXX growth insurance is: recent purchase of baby products, then the recommendation plan generated for the target customer might be: "If the customer has a low risk tolerance, or has recently transferred from an external account with the same name, or has a monthly income exceeding XXX, we can try recommending the XXXX net asset value wealth management product. If the customer has a monthly income exceeding XXX, a high risk tolerance, or in-bank assets exceeding XXX, we can try recommending the XXX fund. If the customer has recently purchased baby products, we can try recommending the XXXX growth insurance."

[0192] The product recommendation method provided in the embodiment of the present application does not screen the target recommended products before generating the recommendation plan. Instead, the offline marketing personnel determines whether the target customers meet the recommendation conditions of the target recommended products. Therefore, after obtaining the recommendation plan, the offline marketing personnel can determine for themselves whether to recommend the target recommended products to the target customers, thereby improving the flexibility of product recommendations.

[0193] In order to enable those skilled in the art to better understand the embodiments of the present application, the embodiments of the present application are described below with reference to specific examples.

[0194] For example, Figure 10 , which shows a flow chart of a product recommendation method.

[0195] The embodiment of the present application generates customized product recommendation solutions for target customers through a collaborative filtering algorithm. After offline marketing personnel determine the target customers and target product areas, they can obtain customer characteristics corresponding to the target product area, as well as reference customers who hold at least one product to be recommended belonging to the target product area. Customer characteristics are constructed based on a large amount of asset and behavior information in the data lake. The constructed customer characteristics can be basic customer information such as age and gender, or customer transaction information, customer asset information, customer in-bank product holding information, customer behavior information, etc. The embodiment of the present application divides each customer feature into 10 boxes to determine the information value corresponding to each customer feature, and uses the information value as a weighted weight. After determining the similarity between the target customer and each reference customer through weighted Euclidean distance, the embodiment of the present application can sort each reference customer from high to low according to the similarity, and determine the top 5 reference customers as target reference customers.

[0196] The embodiment of the present application can further determine the reference intimacy between each target reference customer and the product to be recommended held by each target reference customer. Since there are few customers in the banking industry who evaluate products, the embodiment of the present application needs to customize the calculation method of intimacy through product holding characteristics to measure the customer's preference or liking for the product. Taking the product holding characteristics including the amount held, the proportion of the amount held to assets, the holding time, and the transaction frequency as an example, for any product to be recommended, the product holding data corresponding to each product holding characteristic of all reference customers holding the product to be recommended can be sorted to obtain multiple ranking sequences to determine the ranking of each target reference customer in each product holding characteristic ranking sequence. For any target reference customer, according to the characteristic value corresponding to the ranking of the target reference customer in each product holding characteristic ranking sequence, and the weight corresponding to each product holding characteristic, the reference intimacy between the target reference customer and the product to be recommended can be obtained (see formula (IV)):

[0197] y=β1×Rank+β2×Ratio+β3×Hold+β4×Freq Formula (4)

[0198] Where y is the reference intimacy, Rank is the holding amount, β1 is the weight corresponding to the holding amount, Ratio is the ratio of the holding amount to the assets, β2 is the weight corresponding to the ratio of the holding amount to the assets, Hold is the holding time, β3 is the weight corresponding to the holding time, Freq is the transaction frequency, and β4 is the weight corresponding to the transaction frequency.

[0199] After repeating the above steps for each target reference customer and each product to be recommended held by each target reference customer, the reference intimacy between each target reference customer and each product to be recommended held by each target reference customer can be obtained.

[0200] Furthermore, for any target reference customer, the similarity between the target customer and the target reference customer can be multiplied by the reference affinity between the target reference customer and each product to be recommended held by the target reference customer. This affinity between the target customer and each product to be recommended held by the target reference customer can be calculated. By repeating the above steps for each target reference customer, the affinity between the target customer and each product to be recommended held by each target reference customer can be obtained.

[0201] The embodiment of the present application can further sort the products to be recommended from high to low according to the intimacy, and take the top 3 products to be recommended as target recommended products.

[0202] When generating a recommendation plan, the embodiment of the present application can display only the target recommended products whose customer tags corresponding to the target customers meet the recommendation conditions in the recommendation plan based on the customer tags corresponding to the target customers and the recommendation conditions corresponding to the target recommended products; it can also display all target recommended products and the recommendation conditions corresponding to the target recommended products, and offline marketing personnel can judge whether it is necessary to recommend the target recommended products to the target customers based on the target recommended products and the recommendation conditions corresponding to the target recommended products.

[0203] The product recommendation method provided in the embodiment of the present application, by establishing a customer feature pool, only needs to set different customer features corresponding to each target product field for different target product fields, so as to complete the recommendation of products in different business fields. Therefore, a universal product recommendation framework can be established to enhance the applicability of the framework in different scenarios. In addition, the embodiment of the present application can also determine multiple target recommended products at the same time and make product combination recommendations. Compared with single product recommendations, product combination recommendations can optimize customer asset allocation, enhance product linkage, and improve customer experience. The embodiment of the present application can also form a full-link closed-loop marketing plan from target customer demand analysis to recommendation plan reaching customers. After determining the target recommended product, different marketing methods and means of reaching customers can be combined, supplemented by customer feature overviews, application scenarios, product demonstration prototypes, typical cases, etc., to generate a complete recommendation plan. The recommendation plan is highly practical, easy to understand, and can be directly used for marketing.

[0204] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0205] Based on the same inventive concept, the present application also provides a product recommendation device for implementing the aforementioned product recommendation method. The solution provided by this device is similar to the solution described in the aforementioned method. Therefore, the specific limitations of one or more product recommendation device embodiments provided below can be found in the above-mentioned limitations of the product recommendation method and will not be repeated here.

[0206] In one embodiment, Figure 11 As shown, a product recommendation device is provided, including a first determination module 1102, a second determination module 1104, and a generation module 1106, wherein:

[0207] A first determining module 1102 is configured to determine the similarity between the target customer and each reference customer, wherein the reference customer is a customer who holds at least one product to be recommended;

[0208] The second determining module 1104 is configured to determine a target recommended product from the products to be recommended held by each reference customer based on the similarity between the target customer and each reference customer;

[0209] The generation module 1106 is used to generate a recommendation plan for the target customer based on the target recommended product and the recommendation conditions corresponding to the target recommended product. The recommendation plan is used to indicate that the target recommended product is recommended to the target customer when the target customer meets the recommendation conditions.

[0210] The product recommendation device provided in the embodiment of the present application can determine the target recommended product for the target customer from the products to be recommended held by the reference customer by determining the similarity between the target customer and the reference customer, and generate a recommendation plan for the target customer through the recommendation conditions corresponding to the target recommended product, so as to assist offline marketing personnel in recommending the target recommended product to the target customer. The embodiment of the present application generates a recommendation plan for the target customer to assist marketing personnel in offline product marketing through the recommendation plan, that is, the recommendation plan instructs marketing personnel to recommend the target recommended product to the target customer when the target customer meets the recommendation conditions. The present application integrates online marketing and offline marketing through the recommendation plan, which can improve the flexibility of product recommendations, enrich product recommendation methods, and improve the accuracy of product recommendations.

[0211] In one embodiment, the second determining module 1104 is further configured to:

[0212] determining a target reference customer from among the reference customers based on the similarity between the target customer and the reference customers;

[0213] For any of the to-be-recommended products held by any of the target reference customers, determining the degree of intimacy between the target customer and the to-be-recommended product, wherein the intimacy is used to represent the customer's preference for the product;

[0214] According to the intimacy, a target recommended product is determined from the products to be recommended held by each target reference customer.

[0215] In one embodiment, the second determining module 1104 is further configured to:

[0216] For any of the target reference customers, determining a reference intimacy between the target reference customer and each of the to-be-recommended products held by the target reference customer;

[0217] For any of the to-be-recommended products held by the target reference customer, the intimacy between the target customer and the to-be-recommended product is determined based on the reference intimacy and the similarity between the target customer and the target reference customer.

[0218] In one embodiment, the second determining module 1104 is further configured to:

[0219] For any of the products to be recommended, obtaining product holding data corresponding to various product holding characteristics of the target reference customer;

[0220] For any of the product holding characteristics, determining a characteristic value of the target reference customer for the product holding characteristic based on the ranking of the product holding data corresponding to the product holding characteristic of the target reference customer among the reference customers, wherein the characteristic value is used to represent the importance of the product holding characteristic;

[0221] The reference intimacy between the target reference customer and the product to be recommended is determined based on the feature value of the target reference customer for each feature held by the product and the weight corresponding to the feature held by each product.

[0222] In one embodiment, the first determining module 1102 is further configured to:

[0223] Acquiring characteristic data of the target customer and characteristic data of each reference customer based on the characteristics of each customer;

[0224] For any of the reference customers, the similarity between the target customer and the reference customer is determined based on the feature data of the target customer and the feature data of the reference customer.

[0225] In one embodiment, the first determining module 1102 is further configured to:

[0226] Determining a target product field, where the target product field corresponds to at least one of the products to be recommended;

[0227] Determining a reference customer based on the target product field, wherein at least one of the products to be recommended held by the reference customer belongs to the target product field;

[0228] According to the customer characteristics corresponding to the target product field, characteristic data of the target customer and characteristic data of each of the reference customers are obtained.

[0229] In one embodiment, the first determining module 1102 is further configured to:

[0230] Determining the information value of each of the customer characteristics, and using the information value of each of the customer characteristics as a weighted weight of each of the customer characteristics;

[0231] For any of the reference customers, determining the distance between the target customer and the reference customer in terms of each of the customer characteristics;

[0232] For any of the reference customers, the distances between the target customers and the reference customers in terms of the customer characteristics are fused according to the weighted weights of the customer characteristics to obtain the similarity between the target customer and the reference customer.

[0233] In one embodiment, the first determining module 1102 is further configured to:

[0234] Performing binning processing on the customer features to obtain multiple customer feature bins;

[0235] For any of the customer feature bins, determine the weight of evidence for the customer feature bin based on the number of customers who purchased the product corresponding to the customer feature bin, the number of customers who did not purchase the product corresponding to the customer feature bin, the number of customers who purchased the product corresponding to the customer feature, and the number of customers who did not purchase the product corresponding to the customer feature;

[0236] The information value of the customer feature is determined based on the evidence weight of each of the customer feature bins.

[0237] In one embodiment, the generating module 1106 is further configured to:

[0238] Obtaining a customer tag corresponding to the target customer;

[0239] Obtaining the recommendation conditions corresponding to each of the target recommended products;

[0240] For any of the target recommended products, if the customer tag corresponding to the target customer meets the recommendation condition corresponding to the target recommended product, the target recommended product is used as the recommended product;

[0241] A recommendation plan for the target customer is generated based on each of the recommended products and the recommendation conditions corresponding to each of the recommended products.

[0242] Each module in the above-mentioned product recommendation device can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in the computer device in the form of software, so that the processor can call and execute the corresponding operations of each module.

[0243] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 12 As shown. The computer device includes a processor, a memory, and a network interface connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a product recommendation method is implemented.

[0244] Those skilled in the art will understand that Figure 12 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0245] In one embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0246] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0247] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0248] It should be noted that the customer information (including but not limited to customer device information, customer personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the customer or fully authorized by all parties.

[0249] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processor involved in the various embodiments provided herein may be, but are not limited to, a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic unit, a data processing logic unit based on quantum computing, and the like.

[0250] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0251] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A product recommendation method, characterized in that: The method comprises: Determine the similarity between the target customer and each reference customer, wherein the reference customer is a customer who holds at least one product to be recommended; Determining a target recommended product from the products to be recommended held by each of the reference customers based on the similarity between the target customer and each of the reference customers; Generate a recommendation plan for the target customer based on the target recommended product and the recommendation conditions corresponding to the target recommended product, wherein the recommendation plan indicates that the target recommended product should be recommended to the target customer if the target customer meets the recommendation conditions; the recommendation plan includes basic information of the target customer, the target recommended product, reason for recommendation, recent activity information, and recommendation channel; Determining the similarity between the target customer and each reference customer includes: Acquiring characteristic data of the target customer and characteristic data of each reference customer based on the characteristics of each customer; Performing binning processing on the customer features to obtain multiple customer feature bins; For any of the customer feature bins, determine the weight of evidence for the customer feature bin based on the number of customers who purchased the product corresponding to the customer feature bin, the number of customers who did not purchase the product corresponding to the customer feature bin, the number of customers who purchased the product corresponding to the customer feature, and the number of customers who did not purchase the product corresponding to the customer feature; Determining the information value of the customer feature according to the weight of evidence of each of the customer feature bins, and using the information value of each of the customer features as a weighted weight of each of the customer features; For any of the reference customers, determining the distance between the target customer and the reference customer in terms of each of the customer characteristics; For any of the reference customers, based on the weighted weights of the customer characteristics, the distances between the target customers and the reference customers in the customer characteristics are fused to obtain the similarity between the target customer and the reference customer; The step of determining a target recommended product from the products to be recommended held by each reference customer based on the similarity between the target customer and each reference customer includes: determining a target reference customer from among the reference customers based on the similarity between the target customer and the reference customers; For any of the products to be recommended, obtaining product holding data corresponding to various product holding characteristics of the target reference customer; For any of the product holding characteristics, determining a characteristic value of the target reference customer for the product holding characteristic based on the ranking of the product holding data corresponding to the product holding characteristic of the target reference customer among the reference customers, wherein the characteristic value is used to represent the importance of the product holding characteristic; Determining a reference intimacy between the target reference customer and the product to be recommended based on the feature value of each product held by the target reference customer and the weight corresponding to each product held feature; For any of the to-be-recommended products held by the target reference customer, determining the intimacy between the target customer and the to-be-recommended product based on the reference intimacy and the similarity between the target customer and the target reference customer; According to the intimacy, a target recommended product is determined from the products to be recommended held by each target reference customer.

2. The method according to claim 1, characterized in that The step of obtaining the characteristic data of the target customer and the characteristic data of each reference customer based on the characteristics of each customer includes: Determining a target product field, where the target product field corresponds to at least one of the products to be recommended; Determining a reference customer based on the target product field, wherein at least one of the products to be recommended held by the reference customer belongs to the target product field; According to the customer characteristics corresponding to the target product field, characteristic data of the target customer and characteristic data of each of the reference customers are obtained.

3. The method according to claim 1, characterized in that Generating a recommendation plan for the target customer based on the target recommended product and the recommendation conditions corresponding to the target recommended product includes: Obtaining a customer tag corresponding to the target customer; Obtaining the recommendation conditions corresponding to each of the target recommended products; For any of the target recommended products, if the customer tag corresponding to the target customer meets the recommendation condition corresponding to the target recommended product, the target recommended product is used as the recommended product; A recommendation plan for the target customer is generated based on each of the recommended products and the recommendation conditions corresponding to each of the recommended products.

4. A product recommendation device, characterized in that: The device comprises: A first determination module is configured to determine the similarity between the target customer and each reference customer, wherein the reference customer is a customer who holds at least one product to be recommended; A second determining module is configured to determine a target recommended product from the products to be recommended held by each reference customer based on the similarity between the target customer and each reference customer; a generating module, configured to generate a recommendation plan for the target customer based on the target recommended product and the recommendation conditions corresponding to the target recommended product, wherein the recommendation plan indicates that the target recommended product should be recommended to the target customer if the target customer meets the recommendation conditions; The first determination module is specifically configured to obtain characteristic data of the target customer and characteristic data of each reference customer based on each customer characteristic; perform binning processing on the customer characteristics; obtain multiple customer characteristic bins; for any of the customer characteristic bins, determine the weight of evidence for the customer characteristic bin based on the number of customers who purchased the product corresponding to the customer characteristic bin, the number of customers who did not purchase the product corresponding to the customer characteristic bin, the number of customers who purchased the product corresponding to the customer characteristic, and the number of customers who did not purchase the product corresponding to the customer characteristic; determine the information value of the customer characteristic based on the weight of evidence for each of the customer characteristic bins, and use the information value of each of the customer characteristics as the weighted weight of each of the customer characteristics; for any of the reference customers, determine the distance between the target customer and the reference customer on each of the customer characteristics; for any of the reference customers, fuse the distance between the target customer and the reference customer on each of the customer characteristics based on the weighted weight of each of the customer characteristics to obtain the similarity between the target customer and the reference customer; The second determination module is specifically used to determine the target reference customer from each of the reference customers based on the similarity between the target customer and each of the reference customers; for any of the products to be recommended, obtain the product holding data corresponding to the target reference customer for each product holding feature; for any of the product holding features, determine the feature value of the target reference customer for the product holding feature based on the ranking of the product holding data corresponding to the product holding feature of the target reference customer among each of the reference customers, and the feature value is used to characterize the importance of the product holding feature; determine the reference intimacy between the target reference customer and the product to be recommended based on the feature value of the target reference customer for each of the product holding features and the weight corresponding to each of the product holding features; for any of the products to be recommended held by the target reference customer, determine the intimacy between the target customer and the product to be recommended based on the reference intimacy and the similarity between the target customer and the target reference customer; and determine the target recommended product from the products to be recommended held by each of the target reference customers based on the intimacy.

5. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 3 are implemented.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 3 are implemented.

7. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 3 are implemented.

Images