Product recommendation method and device, computer storage medium and electronic equipment

Abstract

The invention belongs to the technical field of product recommendation, and relates to a product recommendation method and device, a computer readable storage medium and electronic equipment. The method comprises the steps of obtaining to-be-recommended customer feature data, and obtaining a logistics product recommendation model corresponding to the to-be-recommended customer feature data; and predicting a target recommended product corresponding to the to-be-recommended customer feature data based on the logistics product recommendation model. According to the method and the device, the target recommended product corresponding to the feature data of the to-be-recommended customer can be predicted based on the logistics product recommendation model, so that the situation that the target recommended product is recommended to the customer according to subjective judgment of sales and sales experience in the prior art is avoided, and the prediction accuracy of the target recommended product is improved.

Description

Technical Field

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

[0002] With the development of logistics technology, logistics products have become more diverse. How to recommend logistics products suitable for customers has become a problem to be solved.

[0003] In related technologies, salespeople usually recommend logistics products to customers based on their own subjective judgment and sales experience, which results in the recommended logistics products not necessarily being suitable for customers, reducing the accuracy of product recommendations and, in turn, reducing customer experience.

[0004] In view of this, there is an urgent need in the art to develop a new product recommendation method and device.

[0005] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure, and therefore may include information that does not constitute the prior art known to ordinary technicians in the field. Summary of the invention

[0006] The purpose of the present disclosure is to provide a product recommendation method, a product recommendation device, a computer-readable storage medium and an electronic device, thereby overcoming the problem of low recommendation accuracy of logistics products caused by related technologies at least to a certain extent.

[0007] Other features and advantages of the present disclosure will become apparent from the following detailed description, or may be learned in part by the practice of the present disclosure.

[0008] According to a first aspect of an embodiment of the present invention, a product recommendation method is provided, the method comprising: obtaining customer feature data to be recommended, and obtaining a logistics product recommendation model corresponding to the customer feature data to be recommended; the logistics product recommendation model is trained using historical contract feature data as model samples; the historical contract feature data comprises historical contract customer feature data and historical contract product data corresponding to the historical contract customer feature data; based on the logistics product recommendation model, predicting a target recommended product corresponding to the customer feature data to be recommended.

[0009] In an exemplary embodiment of the present invention, before obtaining the logistics product recommendation model corresponding to the customer feature data to be recommended, the method also includes: obtaining historical contract feature data, and performing data processing on the historical contract feature data to construct a model sample; the model sample includes a training sample and a test sample; obtaining an initial logistics product recommendation model, using the training sample to train the initial logistics product recommendation model to obtain the trained initial logistics product recommendation model; using the test sample to test the trained initial logistics product recommendation model to obtain the logistics product recommendation model.

[0010] In an exemplary embodiment of the present invention, the historical contract characteristic data also includes historical sales characteristic data corresponding to the historical contract customer characteristic data.

[0011] In an exemplary embodiment of the present invention, the training samples include historical contracted products and historical contracted quantities corresponding to the historical contracted products; the initial product recommendation model is trained using the training samples to obtain the trained initial product recommendation model, including: obtaining the historical contracted quantities corresponding to the historical contracted products from the training samples respectively; obtaining a contracted quantity threshold corresponding to the historical contracted quantity, and determining a target historical contracted quantity from the historical contracted quantity based on the contracted quantity threshold; eliminating the target sample corresponding to the target historical contracted quantity from the training samples to obtain the eliminated training samples; and using the eliminated training samples to train the initial logistics product recommendation model to obtain the trained initial logistics product recommendation model.

[0012] In an exemplary embodiment of the present invention, the method further includes: training the initial logistics product recommendation model based on a focal loss function to obtain the trained initial logistics product recommendation model.

[0013] In an exemplary embodiment of the present invention, the initial logistics product recommendation model includes an initial product recommendation tree model, and the initial product recommendation tree model includes multiple regression trees; after the initial logistics product recommendation model is trained using the training sample to obtain the trained initial logistics product recommendation model, the method also includes: determining the frequency of each historical contract feature appearing in the multiple regression trees, and calculating the information gain corresponding to each historical contract feature based on the trained initial logistics product recommendation model; the training sample includes the historical contract feature; based on the frequency and the information gain, determining the target historical contract feature from the historical contract features; based on the target historical contract feature, optimizing the trained initial product recommendation tree model to obtain the logistics product recommendation tree model.

[0014] In an exemplary embodiment of the present invention, the logistics product recommendation model includes the logistics product recommendation tree model; the target recommended product corresponding to the customer feature data to be recommended is predicted based on the logistics product recommendation model, including: inputting the customer feature data to be recommended into the logistics product recommendation tree model to obtain multiple prediction values ​​corresponding to the multiple regression trees respectively; each of the prediction values ​​includes multiple first-type prediction values ​​corresponding to multiple product types respectively; the multiple prediction values ​​are calculated to obtain a target prediction value; the target prediction value includes multiple second-type prediction values ​​corresponding to the multiple product types respectively; based on the second-type prediction values, the target recommended product corresponding to the customer feature data to be recommended is predicted.

[0015] In an exemplary embodiment of the present invention, the logistics product recommendation model includes a multi-task learning model; the target recommended product corresponding to the customer feature data to be recommended is predicted based on the logistics product recommendation model, comprising: inputting the customer feature data to be recommended into the multi-task learning model to obtain a target estimated value; the target estimated value is used to represent the probability of products of different product categories being recommended; and based on the target estimated value, the target recommended product corresponding to the customer feature data to be recommended is predicted.

[0016] In an exemplary embodiment of the present invention, the logistics product recommendation model includes a causal inference model; the target recommended product corresponding to the customer characteristic data to be recommended is predicted based on the logistics product recommendation model, comprising: inputting the customer characteristic data to be recommended into the causal inference model to determine target historical contract characteristic data having a similar relationship with the customer characteristic data to be recommended; and predicting the target recommended product corresponding to the customer characteristic data to be recommended based on the historical contract products in the target historical contract characteristic data.

[0017] In an exemplary embodiment of the present invention, after predicting the target recommended product corresponding to the characteristics of the customer to be recommended, the method further includes: sending the target recommended product to a sales terminal to obtain sales characteristic data; updating the historical contract characteristic data according to the sales characteristic data to obtain the updated historical contract characteristic data; constructing a model sample corresponding to the updated historical contract characteristic data, and using the model sample to train the logistics product recommendation model.

[0018] According to a second aspect of an embodiment of the present invention, a product recommendation device is provided, the device comprising: an acquisition module, configured to acquire customer feature data to be recommended, and to acquire a logistics product recommendation model corresponding to the customer feature data to be recommended; the logistics product recommendation model is trained using historical contract feature data as model samples; the historical contract feature data comprises historical contract customer feature data and historical contract product data corresponding to the historical contract customer feature data; a prediction module, configured to predict a target recommended product corresponding to the customer feature data to be recommended based on the logistics product recommendation model.

[0019] According to a third aspect of an embodiment of the present invention, there is provided an electronic device, comprising: a processor and a memory; wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the product recommendation method of any of the above exemplary embodiments is implemented.

[0020] According to a fourth aspect of an embodiment of the present invention, there is provided a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the product recommendation method in any of the above exemplary embodiments is implemented.

[0021] It can be seen from the above technical solutions that the product recommendation method, product recommendation device, computer storage medium and electronic device in the exemplary embodiments of the present invention have at least the following advantages and positive effects:

[0022] In the method and apparatus provided in the exemplary embodiments of the present disclosure, based on a logistics product recommendation model, a target recommended product corresponding to the customer feature data to be recommended is predicted. The logistics product recommendation model is trained using historical contract feature data as model samples. The historical contract feature data includes historical contract customer feature data and historical contract product data corresponding to the historical contract customer feature data. On the one hand, this avoids the situation in the prior art where target recommended products can only be predicted for customers based on the subjective judgment and sales experience of sales personnel. On the other hand, since the model samples used to train the logistics product recommendation model include historical contract customer feature data and historical contract product data corresponding to the historical contract customer feature data, based on the logistics product recommendation model, the compatibility between the predicted target recommended product and the customer is improved, and the prediction accuracy of the target recommended product is improved.

[0023] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The accompanying drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the specification are used to explain the principles of the present disclosure. Obviously, the accompanying drawings described below are only some embodiments of the present disclosure, and for ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without creative work.

[0025] Figure 1 A schematic diagram schematically shows a flow chart of a product recommendation method in an embodiment of the present disclosure;

[0026] Figure 2 The schematic diagram of the process of training the initial logistics product recommendation model in the product recommendation method in the embodiment of the present disclosure is schematically shown;

[0027] Figure 3 A schematic diagram of a process of training an initial product recommendation model in a product recommendation method in an embodiment of the present disclosure is schematically shown;

[0028] Figure 4 A schematic diagram of the process of training an initial product recommendation model in the product recommendation method in an embodiment of the present disclosure is schematically shown;

[0029] Figure 5 A schematic diagram of a process for predicting a target recommended product corresponding to the characteristic data of a customer to be recommended based on a logistics product recommendation model in a product recommendation method in an embodiment of the present disclosure is shown;

[0030] Figure 6 A schematic diagram of a process for predicting a target recommended product corresponding to the characteristic data of a customer to be recommended based on a logistics product recommendation model in a product recommendation method in an embodiment of the present disclosure is shown;

[0031] Figure 7 A schematic diagram of a process for predicting a target recommended product corresponding to the characteristic data of a customer to be recommended based on a logistics product recommendation model in a product recommendation method in an embodiment of the present disclosure is shown;

[0032] Figure 8 A schematic diagram of the process after predicting a target recommended product corresponding to a customer to be recommended in the product recommendation method in the embodiment of the present disclosure is schematically shown;

[0033] Fig. 9 The flowchart of the product recommendation method in the application scenario in the embodiment of the present disclosure is schematically shown;

[0034] Fig.10 A schematic diagram schematically shows the structure of a product recommendation device in an embodiment of the present disclosure;

[0035] Fig.11An electronic device for a product recommendation method in an embodiment of the present disclosure is schematically shown;

[0036] Fig.12 A computer-readable storage medium for a product recommendation method in an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0037] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as being limited to the examples set forth herein; on the contrary, these embodiments are provided so that the present disclosure will be more comprehensive and complete, and the concepts of the example embodiments are fully conveyed to those skilled in the art. The described features, structures, or characteristics may be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure may be practiced while omitting one or more of the specific details, or other methods, components, devices, steps, etc. may be adopted. In other cases, known technical solutions are not shown or described in detail to avoid obscuring various aspects of the present disclosure.

[0038] The terms "a", "an", "the" and "said" are used in this specification to indicate the presence of one or more elements / components / etc.; the terms "including" and "having" are used to express an open-ended inclusion and mean that additional elements / components / etc. may exist in addition to the listed elements / components / etc.; the terms "first" and "second" etc. are used only as labels and are not intended to limit the quantity of their objects.

[0039] In addition, the accompanying drawings are only schematic illustrations of the present disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings represent the same or similar parts, and their repeated descriptions will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities.

[0040] In view of the problems existing in the related art, the present disclosure proposes a product recommendation method. Figure 1 A flow chart of the product recommendation method is shown, Figure 1 As shown, the product recommendation method includes at least the following steps:

[0041] Step S110. Obtain the customer characteristic data to be recommended, and obtain the logistics product recommendation model corresponding to the customer characteristic data to be recommended; the logistics product recommendation model is trained using the historical contract characteristic data as a model sample; the historical contract characteristic data includes the historical contract customer characteristic data and the historical contract product data corresponding to the historical contract customer characteristic data.

[0042] Step S120. Based on the logistics product recommendation model, predict the target recommended product corresponding to the characteristic data of the customer to be recommended.

[0043] In the method and apparatus provided in the exemplary embodiments of the present disclosure, based on a logistics product recommendation model, a target recommended product corresponding to the customer feature data to be recommended is predicted. The logistics product recommendation model is trained using historical contract feature data as model samples. The historical contract feature data includes historical contract customer feature data and historical contract product data corresponding to the historical contract customer feature data. On the one hand, this avoids the situation in the prior art where target recommended products can only be predicted for customers based on the subjective judgment and sales experience of sales personnel. On the other hand, since the model samples used to train the logistics product recommendation model include historical contract customer feature data and historical contract product data corresponding to the historical contract customer feature data, based on the logistics product recommendation model, the compatibility between the predicted target recommended product and the customer is improved, and the prediction accuracy of the target recommended product is improved.

[0044] The following is a detailed description of each step of the product recommendation method.

[0045] In step S110, the customer characteristic data to be recommended is obtained, and a logistics product recommendation model corresponding to the customer characteristic data to be recommended is obtained; the logistics product recommendation model is trained using the historical contract characteristic data as a model sample; the historical contract characteristic data includes historical contract customer characteristic data and historical contract product data corresponding to the historical contract customer characteristic data.

[0046] In the exemplary embodiment of the present disclosure, the characteristic data of the customer to be recommended refers to data related to the characteristics of the customer. Specifically, the characteristic data of the customer to be recommended may be the industry to which the customer belongs, the level of the customer, the region where the customer is located, the data of the customer competitors who are in competition with the customer, the channel through which the customer sells logistics products, the sales method adopted by the customer (online or offline), the customer's requirements for the price of logistics products, the customer's requirements for the timeliness of logistics products, other customized requirements of the customer, etc. This exemplary embodiment does not specifically limit this. It is worth noting that the characteristic data of the customer to be recommended is obtained after authorization by the customer, and at the same time, the data involving customer privacy in the characteristic data of the customer to be recommended is desensitized.

[0047] The logistics product recommendation model refers to a recommendation model that is trained using historical contract feature data as model samples. The historical contract feature data includes historical contract customer feature data and historical contract product data. It is worth noting that both historical contract customer feature data and historical contract product data can only be obtained with the authorization of historical customers, and the data involving customer privacy is desensitized when it is obtained.

[0048] The historical contracted customer feature data describes the relevant features of customers who have successfully signed contracts in the past period of time, and the historical contracted product data describes the products signed by historically contracted customers.

[0049] For example, historical contracted customer feature data can be for existing customers or new contracted customers. If the historical contracted customer feature data is for existing customers, then the historical contracted customer feature data can be the customer's industry, customer level, contracted region, data on competitors competing with the customer, the customer's channel for selling logistics products, the customer's sales method (online or offline), the logistics required by the customer, the customer's requirements for logistics product prices, the customer's requirements for physical product timeliness, other customized requirements of the customer, etc.; if the historical contracted customer feature data is for new contracted customers, then the historical contracted customer feature data can be the new customer's input data, the new customer's estimated order quantity, the type of items consigned by the new customer, etc.

[0050] The historical contracted product data corresponds to the historical contracted customer characteristic data, which may be the historical contracted product itself, the target market of the historical contracted product, the selling points of the historical contracted product, the price of the historical contracted product, the timeliness of the historical contracted product, the number of historical contracts, and the types of products, etc.

[0051] Precisely because the logistics product recommendation model is trained using historical contract feature data as model samples, the logistics product recommendation model can be used to recommend target recommended products with higher accuracy for customers to be recommended based on the feature data of the customers to be recommended.

[0052] In an alternative embodiment, Figure 2 The flowchart of training the initial logistics product recommendation model in the product recommendation method is shown as follows: Figure 2 As shown, the method includes at least the following steps: in step S210, historical contract feature data is obtained, and data processing is performed on the historical contract feature data to construct a model sample; the model sample includes a training sample and a test sample.

[0053] The initial logistics product recommendation model refers to an untrained logistics product recommendation model. In order to train the initial logistics product recommendation model, it is necessary to obtain data corresponding to logistics products that have been successfully signed in the past (i.e., historical contract feature data). The so-called successful contract refers to the sales recommending a logistics product to a customer. If the customer accepts the logistics product, the contract is considered successful.

[0054] The historical contract feature data may specifically include historical contract customer feature data, that is, data related to the characteristics of customers who have successfully signed a contract for a logistics product. Customers who have successfully signed a contract for a logistics product may include existing customers and new customers. New customers refer to customers whose contract time is closer to the present than that of existing customers.

[0055] In addition, historical contract feature data may also include historical contract product data, that is, data describing the characteristics of the contracted products. Historical contract feature data may also include historical sales feature data, that is, data describing characteristics related to sales. It is worth noting that the sales refer to recommending logistics products to customers and enabling customers to successfully sign contracts.

[0056] After obtaining the above historical contract feature data, it is necessary to process the above historical contract feature data. For example, there are some continuous features (such as customer delivery orders, order weight, order volume, etc.) in the historical contract feature data, as well as some discrete features (such as consignment types, contract locations, target markets for logistics products, etc.), some identification features (such as customer levels, logistics product categories, etc.), and some statistical features (such as contract time, customer order frequency in weeks, months, quarters, and years, customer order quantity, customer order weight, etc.). For continuous features, they can be normalized; for discrete features, they can be encoded with one bit of validity; for identification features, they can be represented by identification numbers; for statistical features, they can be statistically analyzed. After performing corresponding data processing on different features, model samples can be constructed based on the pipeline mechanism (specifically, the pipeline mechanism) using the above data-processed data.

[0057] It is worth noting that the model samples can be split into training samples and test samples in a ratio of 8:2.

[0058] In step S220, an initial logistics product recommendation model is obtained, and the initial logistics product recommendation model is trained using training samples to obtain a trained initial logistics product recommendation model.

[0059] The training samples are input into the initial logistics product recommendation model, and the model parameters in the initial logistics product recommendation model are trained to obtain the trained initial logistics product recommendation model.

[0060] In step S230, the trained initial logistics product recommendation model is tested using the test sample to obtain a logistics product recommendation model.

[0061] Among them, after the initial logistics product recommendation model is trained, it is necessary to use test samples to test the training effect of the initial logistics product recommendation model, so as to adjust the trained initial logistics product recommendation model, and finally obtain a logistics product recommendation model that can be used to recommend target recommended products.

[0062] In this exemplary embodiment, the initial logistics product recommendation model is trained and tested using historical contract feature data to obtain a logistics product recommendation model, which greatly improves the prediction accuracy of the logistics product recommendation model.

[0063] In the present exemplary embodiment, the historical contract characteristic data also includes historical sales characteristic data corresponding to the historical contract customer characteristic data.

[0064] Among them, the historical contract characteristic data specifically includes three types of data, namely historical contract customer characteristic data, historical contract product data and historical sales characteristic data. It is worth noting that no matter whether it is historical contract customer characteristic data, historical sales characteristic data or historical contract product data, it can only be obtained with the authorization of the historical customer, and at the same time, the data involving customer privacy is desensitized.

[0065] For example, historical contracted customer feature data can be for existing customers or new contracted customers. If the historical contracted customer feature data is for existing customers, then the historical contracted customer feature data can be the customer's industry, customer level, contracted region, data on competitors competing with the customer, the customer's channel for selling logistics products, the customer's sales method (online or offline), the logistics required by the customer, the customer's requirements for logistics product prices, the customer's requirements for physical product timeliness, other customized requirements of the customer, etc.; if the historical contracted customer feature data is for new contracted customers, then the historical contracted customer feature data can be the new customer's input data, the new customer's estimated order quantity, the type of items consigned by the new customer, etc.

[0066] The historical contract product data corresponds to the historical contract customer feature data, which can specifically include the historical contract product itself, the target market of the historical contract product, the selling point of the historical contract product, the price of the historical contract product, the timeliness of the historical contract product, the number of historical contracts, and the type of product. The historical sales feature data corresponds to the historical contract customer feature data, which can specifically include sales channels, sales departments, and sales visit data. If the historical sales feature data corresponds to the existing customers, the historical sales feature data also includes the sales contract time, etc.

[0067] In this exemplary embodiment, the historical contract feature data includes historical contract customer feature data, historical contract product data, and historical sales feature data. That is, the historical contract feature data includes data in the above three dimensions, which helps to improve the training effect of the initial logistics product recommendation model, and further improve the prediction accuracy of the logistics product recommendation model.

[0068] In an alternative embodiment, Figure 3 FIG. 1 shows a flow chart of training the initial product recommendation model in the product recommendation method, wherein the training samples include historical contract products and historical contract quantities corresponding to the historical contract products, such as Figure 3 As shown, the method includes at least the following steps: In step S310, historical contract quantities corresponding to historical contract products are respectively obtained from training samples.

[0069] The historical contract feature data includes historical contract product data, and the historical contract product data includes the historical contract product itself and the historical contract quantity. Based on this, since the training samples are obtained after data processing of the historical contract feature data, the training samples also contain historical contract products (corresponding to the historical contract product itself) and historical contract quantities (corresponding to the historical contract quantity).

[0070] For example, in the training samples, the historical contract volumes corresponding to different historical contract products are obtained.

[0071] In step S320, a contract volume threshold corresponding to the historical contract volume is obtained, and based on the contract volume threshold, a target historical contract volume is determined from the historical contract volumes.

[0072] It is worth noting that there are some samples with low historical contract signing volume in the training samples, which have no training value for the initial product recommendation model. In order to improve the training effect of the initial product recommendation model, it is necessary to remove these training samples with low historical contract signing volume.

[0073] The signing volume threshold is the critical value for measuring whether the historical signing volume is low. When the historical signing volume is less than or equal to the signing volume threshold, the historical signing volume is determined to be the target historical signing volume.

[0074] In step S330, target samples corresponding to the target historical contract signing amount are removed from the training samples to obtain removed training samples.

[0075] After the target historical contract signing volume is determined, the target sample corresponding to the target historical contract signing volume is deleted from the training sample, thereby obtaining the training sample from which the target sample is eliminated.

[0076] In step S340, the initial logistics product recommendation model is trained using the eliminated training samples to obtain a trained initial logistics product recommendation model.

[0077] Among them, there are no target samples whose historical contract volume is less than or equal to the contract volume threshold in the eliminated training samples. Therefore, using the eliminated training samples to train the initial logistics product recommendation model can improve the training effect of the initial logistics product recommendation model.

[0078] In this exemplary embodiment, the target samples corresponding to the target historical contract quantity are eliminated from the training samples to obtain the eliminated training samples, and then the initial logistics product recommendation model is trained using the eliminated training samples. This can improve the training effect of the initial logistics product recommendation model, so as to further improve the prediction accuracy of the subsequent logistics product recommendation model.

[0079] In this exemplary embodiment, the method further includes: training the initial logistics product recommendation model based on the focal loss function to obtain a trained initial logistics product recommendation model.

[0080] Among them, from the sample distribution of training samples, the historical contract volume of different types of logistics products varies greatly. In order to solve the problem of uneven historical contract volume of different types of logistics products, the focal loss function (i.e., FocalLoss loss function) is introduced.

[0081] Based on this, the loss function of the initial logistics product recommendation model is set to the focal loss function to train the initial logistics product recommendation model to obtain the trained initial logistics product recommendation model.

[0082] In this exemplary embodiment, the initial logistics product recommendation model is trained based on the focal loss function to solve the problem of uneven historical contract volumes of different types of logistics products, thereby improving the training effect of the initial logistics product recommendation model and helping subsequent logistics product recommendation models to recommend more accurate target recommended products.

[0083] In an alternative embodiment, Figure 4The schematic diagram of the process of training the initial product recommendation model in the product recommendation method is shown. The initial logistics product recommendation model includes an initial product recommendation tree model. The initial product recommendation tree model includes multiple regression trees, such as Figure 4 As shown, the method includes at least the following steps: in step S410, determine the frequency of occurrence of each historical contract feature in multiple regression trees, and calculate the information gain corresponding to each historical contract feature based on the trained initial logistics product recommendation model; the training sample includes the historical contract feature.

[0084] The initial logistics product recommendation model may specifically be an initial product recommendation tree model, which may specifically include multiple regression trees.

[0085] Based on this, the training samples are input into the initial product recommendation tree model to train the initial product recommendation tree model, and the frequency of occurrence of different historical contract features in the initial product recommendation tree model can be determined. For example, the historical contract feature of logistics product price appears once in each of the five regression trees of the product recommendation tree model, and the frequency of occurrence of logistics product price in multiple regression trees is 5. In addition, the sum of the information gain of the historical contract feature as a split node in the initial product recommendation tree model can also be obtained.

[0086] In step S420, target historical contract features are determined from the historical contract features based on frequency and information gain.

[0087] Through the frequency of each historical contract feature and the sum of the information gain of each historical contract feature, the quotient of the sum of the information gain and the frequency of the same historical contract feature can also be calculated. Furthermore, based on the quotient and the frequency, one or more target historical contract features that have the greatest impact on the training effect of the initial logistics product recommendation model can be determined from the historical contract features.

[0088] In step S430, based on the target historical contract features, the trained initial product recommendation tree model is optimized to obtain a logistics product recommendation tree model.

[0089] Among them, after determining one or more target historical contract features that have the greatest impact on the training effect of the initial logistics product recommendation model, higher model parameters can be set for the target historical contract features to further optimize the trained initial logistics product recommendation model and obtain a logistics product recommendation tree model with better training effect.

[0090] In this exemplary embodiment, based on frequency and information gain, the target historical contract features are determined from the historical contract features, so as to optimize the trained initial logistics product recommendation model through the target historical contract features. Based on this, the resulting logistics product recommendation tree model will predict more accurate target recommended products.

[0091] In an alternative embodiment, Figure 5 The flowchart of the product recommendation method for predicting the target recommended product corresponding to the characteristic data of the customer to be recommended based on the logistics product recommendation model is shown. The logistics product recommendation model includes a logistics product recommendation tree model, such as Figure 5 As shown, the method includes at least the following steps: in step S510, the customer feature data to be recommended is input into the logistics product recommendation tree model to obtain multiple prediction values ​​corresponding to multiple regression trees respectively; each prediction value includes multiple first type prediction values ​​corresponding to multiple product types respectively.

[0092] Among them, based on the target historical contract characteristics, the trained initial product recommendation tree model was optimized to obtain the logistics product recommendation tree model.

[0093] At this time, after the feature data of the customer to be recommended is input into the logistics product recommendation tree model, since the logistics product recommendation tree model includes multiple regression trees, each regression tree will generate a prediction value corresponding to the feature data of the customer to be recommended. It is worth noting that each prediction value includes multiple first-type prediction values, and the multiple first-type prediction values ​​correspond to multiple product types respectively.

[0094] In step S520, a plurality of prediction values ​​are calculated to obtain a target prediction value; the target prediction value includes a plurality of second type prediction values ​​corresponding to the plurality of product types.

[0095] The sum calculation result is obtained by summing up the multiple prediction values, and each value in the sum calculation result is mapped in the interval [0,1] through a normalization function to obtain a target prediction value. It is worth noting that the target prediction value includes multiple second-type prediction values, which correspond to multiple product types respectively and are used to describe the probability of recommending physical products corresponding to the product types to customers.

[0096] In step S530, based on the second type prediction value, a target recommended product corresponding to the to-be-recommended customer feature data is predicted.

[0097] Among them, since the second type prediction value is used to describe the probability of recommending a logistics product corresponding to the product type to the customer, the second type prediction values ​​can be sorted to determine one or more second type prediction values ​​with the highest sorting, and the logistics products corresponding to these one or more second type prediction values ​​can be used as the predicted target recommended products corresponding to the customer characteristic data to be recommended.

[0098] For example, the target prediction value includes 10 second-type prediction values. By sorting these 10 second-type prediction values, it can be determined that the largest second-type prediction value is 0.7. The logistics product A corresponding to the largest second-type prediction value 0.7 is determined to be the predicted target recommended product corresponding to the feature data of the customer to be recommended.

[0099] In this exemplary embodiment, a more accurate target recommended product corresponding to the feature data of the customer to be recommended can be predicted through the logistics product recommendation tree model.

[0100] In an alternative embodiment, Figure 6 The flowchart of the product recommendation method is shown in FIG. 1 , which is based on the logistics product recommendation model and predicts the target recommended product corresponding to the feature data of the customer to be recommended. The logistics product recommendation model includes a multi-task learning model, such as Figure 6 As shown, the method includes at least the following steps: in step S610, the feature data of the customer to be recommended is input into the multi-task learning model to obtain a target estimate; the target estimate is used to represent the probability of products of different product categories being recommended.

[0101] The logistics product recommendation model also includes a multi-task learning model. Specifically, the multi-task learning model can be a PLE (progressive layered extraction) model. The multi-task learning model can predict multiple targets, for example, it can predict product categories, it can predict order rates, and it can also predict estimated revenue after signing. Therefore, compared with the logistics product recommendation tree model, the prediction target of the task learning model is no longer limited to product types, thereby expanding the range of predictable targets.

[0102] After inputting the feature data of the customer to be recommended into the multi-task learning model, the target estimate of the multi-task learning model for the specified target will be obtained. For example, if the specified estimated target is the estimated income after signing the contract, the target estimate is the estimated income after signing the contract.

[0103] In step S620, the target recommended product corresponding to the feature data of the customer to be recommended is predicted based on the target estimated value.

[0104] Assuming that the target estimated value is the estimated income after signing the contract, after obtaining the target estimated value, the estimated income that the customer can get by signing different products can be known. Furthermore, based on the target estimated value, the logistics product with the highest estimated income after signing the contract can be used as the predicted target recommended product corresponding to the characteristic data of the customer to be recommended.

[0105] In this exemplary embodiment, compared with the logistics product recommendation tree model, the prediction target of the multi-task learning model is no longer limited to the product type. Therefore, it not only improves the accuracy of the predicted target recommended products, but also expands the scope of the targets that can be predicted.

[0106] In an alternative embodiment, Figure 7 The flowchart of the product recommendation method for predicting the target recommended product corresponding to the characteristic data of the customer to be recommended based on the logistics product recommendation model is shown. The logistics product recommendation model includes a causal inference model, such as Figure 7 As shown, the method at least includes the following steps: In step S710, the feature data of the customer to be recommended is input into the causal inference model to determine the target historical contract feature data having a similar relationship with the feature data of the customer to be recommended.

[0107] Among them, by inputting the feature data of the customer to be recommended into the causal inference model, the target historical contract feature data having a similar relationship with the feature of the customer to be recommended can be determined. It is worth noting that the so-called similar relationship means: by training the causal inference model using the historical contract feature number, it can be determined that because of the features in the historical contract feature data, it is decided to recommend which physical product to the customer. For example, because the contract area in the historical contract feature data 1 is at D, the logistics product H is recommended to the customer; because the customer level in the historical contract feature data 2 is 3, the customer requires the price of the logistics product to be 200 to 270, and the category of the logistics product belongs to L-1, the logistics product G is recommended to the customer. On this basis, if the contract area in the feature data of the customer to be recommended is D, the target historical contract feature data corresponding to the feature data of the customer to be recommended is the historical contract feature data 1.

[0108] In step S720, based on the historical contract products in the target historical contract feature data, the target recommended product corresponding to the to-be-recommended customer feature data is predicted.

[0109] Among them, after determining the target historical contract feature data, the historical contract products in the target historical contract feature data are used to determine the predicted target recommended products corresponding to the to-be-recommended customer feature data.

[0110] For example, if the contract area in the to-be-recommended customer feature data is D, the target historical contract feature data corresponding to the to-be-recommended customer feature data is historical contract feature data 1. Based on this, the predicted target recommended product corresponding to the to-be-recommended customer feature data is logistics product H.

[0111] In this exemplary embodiment, the customer characteristic data to be recommended is input into the causal inference model, which can determine the target historical contract characteristic data that has a similar relationship with the customer characteristic data to be recommended, and then the historical contract products in the target historical contract characteristic data can be determined as the predicted target recommended products corresponding to the customer characteristic data to be recommended, thereby greatly improving the accuracy of the predicted target recommended products.

[0112] In an alternative embodiment, Figure 8 FIG. 1 shows a flow chart of the process after predicting the target recommended product corresponding to the customer to be recommended in the product recommendation method, such as Figure 8 As shown, the method includes at least the following steps: In step S810, the target recommended product is sent to a sales terminal to obtain sales feature data.

[0113] The sales terminal refers to the terminal used by the salesperson. After the target recommended product is determined, it is sent to the sales terminal.

[0114] It is worth noting that the target recommended product can be one or more. After the target recommended products are sent to the sales terminal, all the target recommended products can be displayed at the sales terminal, or only a part of the target recommended products can be displayed at the sales terminal. For example, only the target recommended products that meet the sales characteristics can be displayed, and the sales characteristics refer to the characteristics of the salesperson corresponding to the sales terminal.

[0115] Based on this, sales personnel can recommend target products to customers. If the customer successfully signs the contract, a corresponding piece of historical contract feature data (i.e., sales feature data) can be generated.

[0116] In step S820, the historical contract characteristic data is updated according to the sales characteristic data to obtain updated historical contract characteristic data.

[0117] Among them, the sales terminal can recommend target products to customers. If the customer successfully signs the contract, a historical contract feature data (i.e., sales feature data) can be generated accordingly. Correspondingly, the historical contract feature data needs to be updated according to the newly generated sales feature data. Based on this, the updated historical contract feature data includes the newly generated sales feature data, the contract customer feature data corresponding to the newly generated sales feature data, and the contract product data corresponding to the newly generated sales feature data.

[0118] In step S830, a model sample corresponding to the updated historical contract feature data is constructed, and the logistics product recommendation model is trained using the model sample.

[0119] Among them, based on the updated historical contract feature data, the model samples are reconstructed, and the updated model samples are used to continue training the logistics product recommendation model to improve the prediction accuracy of the logistics product recommendation model.

[0120] In addition to sending the target recommended products to the sales terminal to show the target recommended products to the sales staff, the target recommended products can also be shown to the sales staff offline. In addition, a recommendation monitoring dashboard can also be built. After the target recommended product is recommended to the customer, the customer can choose to sign a contract. At this time, the target recommended products that have been signed can be counted according to different feature dimensions, and the statistical results can be displayed in the recommendation monitoring dashboard. For example, the target recommended products that have been signed can be counted according to the two feature dimensions of product type and contract area, and the statistical results and the corresponding feature dimensions can be displayed in the recommendation monitoring dashboard.

[0121] In this exemplary embodiment, a model sample corresponding to the updated historical contract feature data is constructed, and the logistics product recommendation model is trained using the model sample to further improve the prediction accuracy of the logistics product recommendation model.

[0122] In the method and apparatus provided in the exemplary embodiments of the present disclosure, based on a logistics product recommendation model, a target recommended product corresponding to the customer feature data to be recommended is predicted. The logistics product recommendation model is trained using historical contract feature data as model samples. The historical contract feature data includes historical contract customer feature data and historical contract product data corresponding to the historical contract customer feature data. On the one hand, this avoids the situation in the prior art where target recommended products can only be predicted for customers based on the subjective judgment and sales experience of sales personnel. On the other hand, since the model samples used to train the logistics product recommendation model include historical contract customer feature data and historical contract product data corresponding to the historical contract customer feature data, based on the logistics product recommendation model, the compatibility between the predicted target recommended product and the customer is improved, and the prediction accuracy of the target recommended product is improved.

[0123] The following is a detailed description of the method for transporting objects and goods in the embodiment of the present disclosure in conjunction with an application scenario.

[0124] Fig. 9 The flowchart of the product recommendation method in the application scenario in the embodiment of the present disclosure is schematically shown. Fig. 9 As shown, process node 910 is the start node, and data 911 is the characteristic data of the customer to be recommended.

[0125] Process node 920 is a data processing node, model 931 is a logistics product recommendation tree model, model 932 is a multi-task learning model, and model 933 is a causal inference model.

[0126] Through model 931, model 932 and model 933, the target recommended product 940 corresponding to data 911 can be predicted. And the target recommended product 940 is displayed to salesperson 941. If the salesperson adopts the target recommended product 940, the target recommended product 940 will be recommended to the customer to obtain sales feature data 950. Based on this, model 931, model 932 and model 933 are retrained according to sales feature data 950; if the salesperson does not adopt the target recommended product 940, the above process ends. Process node 960 is the end node.

[0127] In this application scenario, based on the logistics product recommendation model, the target recommended product corresponding to the customer feature data to be recommended is predicted. The logistics product recommendation model is trained with historical contract feature data as model samples. The historical contract feature data includes historical contract customer feature data and historical contract product data corresponding to the historical contract customer feature data. On the one hand, it avoids the situation in the prior art that the target recommended product can only be predicted for the customer based on the subjective judgment and sales experience of the sales staff; on the other hand, because the model samples used to train the logistics product recommendation model include historical contract customer feature data and historical contract product data corresponding to the historical contract customer feature data, based on the logistics product recommendation model, the compatibility between the predicted target recommended product and the customer is improved, and the prediction accuracy of the target recommended product is improved.

[0128] In addition, in an exemplary embodiment of the present disclosure, a product recommendation device is also provided.

[0129] Fig.10 The structural diagram of the product recommendation device is shown in FIG. Fig.10 As shown, the product recommendation device 1000 may include: an acquisition module 1010 and a prediction module 1020. Among them:

[0130] The acquisition module 1010 is configured to acquire the customer characteristic data to be recommended, and to acquire a logistics product recommendation model corresponding to the customer characteristic data to be recommended; the logistics product recommendation model is trained using the historical contract characteristic data as a model sample; the historical contract characteristic data includes the historical contract customer characteristic data and the historical contract product data corresponding to the historical contract customer characteristic data; the prediction module 1020 is configured to predict the target recommended product corresponding to the customer characteristic data to be recommended based on the logistics product recommendation model.

[0131] The specific details of the above-mentioned product recommendation device 1000 have been described in detail in the corresponding product recommendation method, so they are not repeated here.

[0132] It should be noted that, although several modules or units of the product recommendation device 1000 are mentioned in the above detailed description, such division is not mandatory. In fact, according to an embodiment of the present disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.

[0133] In addition, in an exemplary embodiment of the present disclosure, an electronic device capable of implementing the above method is also provided.

[0134] Refer to the following Fig.11 1100 according to this embodiment of the present invention will be described. Fig.11 The electronic device 1100 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0135] like Fig.11 As shown, the electronic device 1100 is in the form of a general computing device. The components of the electronic device 1100 may include, but are not limited to: the at least one processing unit 1110, the at least one storage unit 1120, a bus 1130 connecting different system components (including the storage unit 1120 and the processing unit 1110), and a display unit 1140.

[0136] The storage unit stores program codes, which can be executed by the processing unit 1110, so that the processing unit 1110 executes the steps according to various exemplary embodiments of the present invention described in the above “Exemplary Method” section of this specification.

[0137] The storage unit 1120 may include a readable medium in the form of a volatile storage unit, such as a random access memory unit (RAM) 1121 and / or a cache memory unit 1122 , and may further include a read-only memory unit (ROM) 1123 .

[0138] The storage unit 1120 may also include a program / utility 1124 having a set (at least one) of program modules 1125, such program modules 1125 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include the reality of a network environment.

[0139] Bus 1130 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0140] The electronic device 1100 may also communicate with one or more external devices 1170 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 1100, and / or communicate with any device that enables the electronic device 1100 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface 1150. Furthermore, the electronic device 1100 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 1160. As shown, the network adapter 1160 communicates with other modules of the electronic device 1100 via a bus 1130. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 1100, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0141] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the embodiment of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the embodiment of the present disclosure.

[0142] In an exemplary embodiment of the present disclosure, a computer-readable storage medium is also provided, on which a program product capable of implementing the above method of the present specification is stored. In some possible embodiments, various aspects of the present invention can also be implemented in the form of a program product, which includes a program code, and when the program product is run on a terminal device, the program code is used to enable the terminal device to perform the steps according to various exemplary embodiments of the present invention described in the above "Exemplary Method" section of the present specification.

[0143] refer to Fig.12As shown, a program product 1200 for implementing the above method according to an embodiment of the present invention is described, which can adopt a portable compact disk read-only memory (CD-ROM) and include program code, and can be run on a terminal device, such as a personal computer. However, the program product of the present invention is not limited thereto. In this document, a readable storage medium can be any tangible medium containing or storing a program, which can be used by or in combination with an instruction execution system, an apparatus or a device.

[0144] The program product may use any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0145] Computer readable signal media may include data signals propagated in baseband or as part of a carrier wave, in which readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Readable signal media may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0146] The program code embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the foregoing.

[0147] Program code for performing the operations of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., through the Internet using an Internet service provider).

[0148] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary technical means in the art that are not disclosed in the present disclosure. The specification and examples are to be considered exemplary only, and the true scope and spirit of the present disclosure are indicated by the claims.

Claims

1. A product recommendation method, characterized in that: The method comprises: Acquire the customer feature data to be recommended, and acquire the logistics product recommendation model corresponding to the customer feature data to be recommended; the logistics product recommendation model is trained using the historical contract feature data as a model sample; the historical contract feature data includes the historical contract customer feature data and the historical contract product data corresponding to the historical contract customer feature data; Based on the logistics product recommendation model, a target recommended product corresponding to the feature data of the customer to be recommended is predicted.

2. The product recommendation method according to claim 1, characterized in that: Before obtaining the logistics product recommendation model corresponding to the customer feature data to be recommended, the method further includes: Acquire historical contract feature data, and perform data processing on the historical contract feature data to construct a model sample; the model sample includes a training sample and a test sample; Acquire an initial logistics product recommendation model, and train the initial logistics product recommendation model using the training samples to obtain the trained initial logistics product recommendation model; The trained initial logistics product recommendation model is tested using the test sample to obtain a logistics product recommendation model.

3. The product recommendation method according to claim 2, characterized in that: The historical contract characteristic data also includes historical sales characteristic data corresponding to the historical contract customer characteristic data.

4. The product recommendation method according to claim 2, characterized in that: The training samples include historical contract products and historical contract volumes corresponding to the historical contract products; Using the training samples to train the initial product recommendation model to obtain the trained initial product recommendation model includes: From the training samples, respectively obtain the historical contract volumes corresponding to the historical contract products; Acquire a contract volume threshold value corresponding to the historical contract volume, and determine a target historical contract volume from the historical contract volume based on the contract volume threshold value; Eliminating the target sample corresponding to the target historical contract signing amount from the training samples to obtain the eliminated training sample; The initial logistics product recommendation model is trained using the eliminated training samples to obtain the trained initial logistics product recommendation model.

5. The product recommendation method according to claim 4, characterized in that: The method further comprises: Based on the focal loss function, the initial logistics product recommendation model is trained to obtain the trained initial logistics product recommendation model.

6. The product recommendation method according to claim 2, characterized in that: The initial logistics product recommendation model includes an initial product recommendation tree model, and the initial product recommendation tree model includes multiple regression trees; After the initial logistics product recommendation model is trained using the training samples to obtain the trained initial logistics product recommendation model, the method further includes: Determine the frequency of occurrence of each historical contract feature in the multiple regression trees, and calculate the information gain corresponding to each historical contract feature based on the trained initial logistics product recommendation model; the training sample includes the historical contract feature; Based on the frequency and the information gain, determining a target historical contract feature from the historical contract features; Based on the target historical contract features, the trained initial product recommendation tree model is optimized to obtain a logistics product recommendation tree model.

7. The product recommendation method according to claim 6, characterized in that: The logistics product recommendation model includes the logistics product recommendation tree model; The step of predicting a target recommended product corresponding to the feature data of the customer to be recommended based on the logistics product recommendation model includes: Inputting the customer feature data to be recommended into the logistics product recommendation tree model to obtain a plurality of prediction values ​​corresponding to the plurality of regression trees respectively; each of the prediction values ​​includes a plurality of first type prediction values ​​corresponding to a plurality of product types respectively; Calculating the multiple prediction values ​​to obtain a target prediction value; the target prediction value includes a plurality of second type prediction values ​​corresponding to the multiple product types respectively; Based on the second type of prediction value, a target recommended product corresponding to the to-be-recommended customer feature data is predicted.

8. The product recommendation method according to claim 1, characterized in that: The logistics product recommendation model includes a multi-task learning model; The step of predicting a target recommended product corresponding to the feature data of the customer to be recommended based on the logistics product recommendation model includes: Inputting the feature data of the customer to be recommended into the multi-task learning model to obtain a target estimated value; According to the target estimated value, a target recommended product corresponding to the feature data of the customer to be recommended is predicted.

9. The product recommendation method according to claim 2, characterized in that: The logistics product recommendation model includes a causal inference model; The step of predicting a target recommended product corresponding to the feature data of the customer to be recommended based on the logistics product recommendation model includes: Inputting the to-be-recommended customer characteristic data into the causal inference model to determine target historical contract characteristic data having a similar relationship with the to-be-recommended customer characteristic data; Based on the historical contract products in the target historical contract feature data, the target recommended products corresponding to the to-be-recommended customer feature data are predicted.

10. The product recommendation method according to claim 3, characterized in that: After predicting the target recommended product corresponding to the characteristics of the customer to be recommended, the method further includes: Sending the target recommended product to a sales terminal to obtain sales characteristic data; According to the sales characteristic data, the historical contract characteristic data is updated to obtain the updated historical contract characteristic data; A model sample corresponding to the updated historical contract feature data is constructed, and the logistics product recommendation model is trained using the model sample.

11. A product recommendation device, characterized in that: include: The acquisition module is configured to acquire the feature data of the customer to be recommended, and acquire the logistics product recommendation model corresponding to the feature data of the customer to be recommended; the logistics product recommendation model is trained with the historical contract feature data as the model sample; the historical contract feature data includes the historical contract customer feature data and the historical contract product data corresponding to the historical contract customer feature data; The prediction module is configured to predict the target recommended product corresponding to the feature data of the customer to be recommended based on the logistics product recommendation model.

12. An electronic device, characterized in that: include: processor; A memory, configured to store executable instructions of the processor; Wherein, the processor is configured to execute the product recommendation method described in any one of claims 1-10 by executing the executable instructions.

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

Images