Advertisement placement and selection method and its device, equipment, medium, and product

Through pre-trained adoption rate, click-through rate and conversion rate prediction models, recommendation scores are calculated and recommendation lists for advertising placement are constructed, which solves the problem of scarce product placement data in independent sites and improves the effectiveness and user experience of advertising placement.

CN114663155BActive Publication Date: 2024-11-12GUANGZHOU HUADUO NETWORK TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210348720.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-04-01
Publication Date
2024-11-12
Estimated Expiration
2042-04-01

AI Technical Summary

Technical Problem

In cross-border e-commerce service platforms, it is difficult for independent website management users to determine products with sales potential to advertise, resulting in difficult to reflect the return results, mainly due to the scarcity of product advertising data and low reference value.

Method used

Using pre-trained adoption rate, click-through rate and conversion rate prediction models, obtain product adoption rate, click-through rate and conversion rate data, calculate recommendation scores, and build a list of recommended products for advertising, helping management users to select suitable products for advertising. Delivery.

Benefits of technology

By accurately predicting product adoption rate and effectiveness data, quantitative recommendation scores are provided to help manage users make quick decisions, select products suitable for delivery, and improve the effectiveness of advertising. In particular, it provides effective product selection for independent sites with limited resources Auxiliary.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114663155B_ABST
    Figure CN114663155B_ABST
Patent Text Reader

Abstract

The present application discloses a method for selecting products for advertising placement and its device, equipment, medium, and product. The method includes: obtaining advertising prediction data corresponding to each product in a product database of an independent site, the advertising prediction data including adoption rate and effectiveness data, the adoption rate representing the predicted probability of the product being advertised, and the effectiveness data including the conversion probability that may be obtained after the product is advertised; calculating the recommendation score corresponding to each product based on the adoption rate and effectiveness data of the product; selecting products in the product database as advertising placement products based on the recommendation score to construct an advertising placement product selection recommendation list. The present application can accurately match advertising placement products that meet user preferences and achieve good advertising placement results.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of e-commerce information technology, and in particular to a method for selecting products for advertisement placement and its corresponding apparatus, computer equipment, computer-readable storage medium, and computer program product. Background Art

[0002] In the cross-border e-commerce service platform, basic technical services are provided for a large number of independent sites. Each independent site usually has an independent domain name. In order to activate the e-commerce transactions of the independent sites, the management users of the independent sites often need to recommend products to consumer users. For this purpose, they need to be able to select products on their platform and identify products with sales potential that are suitable for advertising, thereby promoting terminal sales or configuring the products for listing.

[0003] For the sales potential of goods, the management users of independent sites generally have relevant market trend considerations. For example, newly launched goods, goods in a certain category, low-priced goods, etc. have good corresponding market trends and have certain market competitiveness, that is, sales potential. Market trend considerations require the analysis of rich product advertising data. However, at the current stage, product advertising data is sparsely distributed. Since most independent sites have limited funding resources, the investment in product advertising is limited or even very little. Therefore, there are only a few independent sites with rich product advertising data, and the product advertising data of the vast majority of independent sites is scarce. More importantly, these scarce product advertising data are limited by limited investment costs and insufficient exposure, making it difficult to reflect the corresponding return results.

[0004] The scarcity of data and the low reference value of data make it difficult for most independent site management users to identify products with sales potential. So far, there is no better solution to this problem in the existing technology.

[0005] In order to solve the deficiencies in the prior art, the applicant has made corresponding explorations. Summary of the invention

[0006] The primary purpose of the present application is to solve at least one of the above problems and to provide an advertising selection method and its corresponding device, computer equipment, computer-readable storage medium, and computer program product.

[0007] In order to meet the various objectives of this application, this application adopts the following technical solutions:

[0008] A method for selecting products for advertisement placement provided to meet one of the purposes of this application includes the following steps:

[0009] Acquire advertising prediction data corresponding to each product in the product database of the independent site, wherein the advertising prediction data includes adoption rate and effectiveness data, wherein the adoption rate represents the predicted probability of the product being advertised, and the effectiveness data includes the conversion probability that may be obtained after the product is advertised;

[0010] Calculate the recommendation score corresponding to each product based on the adoption rate and effectiveness data of the product;

[0011] According to the recommendation score, commodities in the commodity database are selected as selection items for advertisement placement to construct a recommendation list of selection items for advertisement placement.

[0012] In a further embodiment, before the step of obtaining the advertisement prediction data corresponding to each commodity in the commodity database of the independent site, the following steps are included:

[0013] Determine the adoption rate of each product in the product database of the independent site using the adoption rate prediction model that has been pre-trained to a convergent state;

[0014] Determine the click rate of each product in the product database of the independent site using a click rate prediction model that has been pre-trained to a convergent state;

[0015] Determine the conversion rate of each product in the product database of the independent site using a conversion rate prediction model that has been pre-trained to a convergence state;

[0016] The adoption rate and effectiveness data of each product are stored in the product database corresponding to each product, and the effectiveness data is the product of the click rate and the conversion rate associated with the same product.

[0017] In a further embodiment, in the step of using an adoption rate prediction model pre-trained to a convergent state to determine the adoption rate of each product in the product database of the independent site, the training process of the adoption rate prediction model includes the following steps:

[0018] A training sample is called from a data set, wherein the training sample includes positive samples and negative samples, and each training sample is provided with a supervision label representing whether a product has been advertised. The positive sample includes product information and statistical information of advertised products in the product database of the e-commerce platform, and the negative sample includes product information and statistical information of non-advertised products in the product database of the e-commerce platform. The product information includes product listing time, product category, and product price, and the statistical information includes click-through rate, purchase rate, and repurchase rate;

[0019] Inputting the training sample into the adoption rate prediction model to obtain a commodity feature vector and a statistical feature vector corresponding to the commodity information and statistical information in the training sample;

[0020] The product feature vector and the statistical feature vector are concatenated to obtain a feature fusion vector, which is input into a fully connected layer and mapped to a preset classification space to obtain a predicted adoption rate;

[0021] The loss value corresponding to the cross entropy loss of the predicted adoption rate is calculated according to the supervised label, and it is determined whether the loss value reaches a preset threshold. When it reaches the preset threshold, the training is terminated; otherwise, the model is gradient updated according to the loss value, and the next training sample in the data set is called to continue iterative training of the model.

[0022] In a further embodiment, in the step of using a click-through rate prediction model pre-trained to a convergent state to determine the click-through rate of each product in the product database of an independent site, the training process of the click-through rate prediction model includes the following iterative execution steps:

[0023] Obtain training samples from a data set, wherein the training samples include product information of products that have been advertised and clicked by users in a product database of an e-commerce platform, and a supervision label corresponding to whether the advertisement for the product has been clicked by the user;

[0024] Inputting the training sample into the click rate prediction model to extract corresponding deep semantic features and obtain a feature vector;

[0025] Using a classifier to classify and map the feature vectors, and predicting the corresponding click rate;

[0026] A cross entropy loss value corresponding to the click rate predicted by the classifier is calculated according to the supervised label, and a gradient update is performed on the model according to the cross entropy loss value until the model converges.

[0027] In a further embodiment, in the step of using a conversion rate prediction model pre-trained to a convergent state to determine the conversion rate of each commodity in the commodity database of the independent site, the training process of the conversion rate prediction model includes the following iterative execution steps:

[0028] Obtain training samples from a data set, wherein the training samples include product information of products in a product database of an e-commerce platform that have been advertised and have generated relevant user behaviors after being clicked by users, and supervision labels corresponding to whether the advertisements for the products have generated relevant user behaviors after being clicked by users;

[0029] Inputting the training sample into the conversion rate prediction model to extract corresponding deep semantic features and obtain a feature vector;

[0030] Using a classifier to classify and map the feature vector to predict the corresponding conversion rate;

[0031] A cross entropy loss value corresponding to the conversion rate predicted by the classifier is calculated according to the supervised label, and a gradient update is performed on the model according to the cross entropy loss value until the model converges.

[0032] In an extended embodiment, selecting products in the product database as advertising selection products according to the recommendation scores to construct an advertising selection recommendation list includes the following steps:

[0033] Sorting the products in the product database according to their recommendation scores to build a recommended list of products for advertising;

[0034] In response to the online store's advertising recommendation request, the advertising selection recommendation list is pushed to the management user of the independent site.

[0035] An advertising product selection device provided to meet one of the purposes of the present application includes: a data acquisition module, a score calculation module, and a list construction module, wherein the data acquisition module is used to obtain advertising prediction data corresponding to each product in a product database of an independent site, the advertising prediction data including adoption rate and effectiveness data, the adoption rate representing the predicted probability of the product being advertised, and the effectiveness data including the conversion probability that may be obtained after the product is advertised; the score calculation module is used to calculate the recommendation score corresponding to each product based on the adoption rate and effectiveness data of the product; the list construction module is used to select the products in the product database as advertising products based on the recommendation score to construct a recommended list of advertising products.

[0036] In a further embodiment, the data acquisition module includes: an adoption rate determination module, which is used to determine the adoption rate of each commodity in the commodity database of the independent site by using an adoption rate prediction model pre-trained to a convergence state; a click-through rate confirmation module, which is used to determine the click-through rate of each commodity in the commodity database of the independent site by using a click-through rate prediction model pre-trained to a convergence state; a conversion rate confirmation module, which is used to determine the conversion rate of each commodity in the commodity database of the independent site by using a conversion rate prediction model pre-trained to a convergence state; and a data storage sub-module, which is used to store the adoption rate and effectiveness data of each commodity corresponding to each commodity in the commodity database, wherein the effectiveness data is the product of the click-through rate and the conversion rate associated with the same commodity.

[0037] In a further embodiment, in the adoption rate determination module, the training process of the adoption rate prediction model includes: a training sample calling module, which is used to call a training sample from a data set, wherein the training sample includes a positive sample and a negative sample, and each training sample is provided with a corresponding supervision label representing whether the product has been advertised, the positive sample includes the product information and statistical information of the advertised products in the product database of the e-commerce platform, and the negative sample includes the product information and statistical information of the non-advertised products in the product database of the e-commerce platform, the product information includes the product listing time, product category, and product price, and the statistical information includes click-through rate, purchase rate, and repurchase rate; a feature extraction module, which is used to extract the product information and statistical information of the product database of the e-commerce platform. The training sample is input into the adoption rate prediction model to obtain a product feature vector and a statistical feature vector corresponding to the product information and statistical information in the training sample; a vector splicing module is used to splice the product feature vector and the statistical feature vector to obtain a feature fusion vector, and input the feature fusion vector into a fully connected layer to map it to a preset classification space to obtain a predicted adoption rate; an iterative training module is used to calculate a loss value corresponding to the cross entropy loss of the predicted adoption rate according to the supervision label, and determine whether the loss value reaches a preset threshold. When it reaches the preset threshold, the training is terminated; otherwise, the model is gradient updated according to the loss value, and the next training sample in the data set is called to continue iterative training of the model.

[0038] In a further embodiment, in the click-through rate confirmation module, the iterative execution training process of the click-through rate prediction model includes: a training sample calling module, which is used to obtain training samples in a data set, and the training samples include product information of products that have been advertised and clicked by users in the product database of the e-commerce platform, and corresponding supervisory labels that characterize whether the advertisements for the products are clicked by users; a feature extraction module, which is used to input the training samples into the click-through rate prediction model to extract corresponding deep semantic features and obtain feature vectors; a classification mapping module, which is used to classify and map the feature vectors using a classifier to predict the corresponding click-through rate; and a gradient update module, which is used to calculate the cross-entropy loss value corresponding to the click-through rate predicted by the classifier based on the supervisory label, and implement gradient update on the model based on the cross-entropy loss value until the model converges.

[0039] In a further embodiment, in the conversion rate confirmation module, the iterative execution training process of the conversion rate prediction model includes: a training sample calling module, which is used to obtain training samples in a data set, and the training samples include product information of products in the product database of the e-commerce platform that have been advertised and have generated relevant user behaviors after being clicked by users, and supervisory labels corresponding to whether the advertisements for the products have generated relevant user behaviors after being clicked by users; a feature extraction module, which is used to input the training samples into the conversion rate prediction model to extract corresponding deep semantic features and obtain feature vectors; a classification mapping module, which is used to classify and map the feature vectors using a classifier to predict the corresponding conversion rate; a gradient update module, which is used to calculate the cross entropy loss value corresponding to the conversion rate predicted by the classifier based on the supervisory labels, and implement gradient updates on the model based on the cross entropy loss value until the model converges.

[0040] In an extended embodiment, the list construction module includes: a rating sorting module, which is used to sort the recommended ratings corresponding to the products in the product database to construct a recommended list of product selections for advertising; and a list push module, which is used to respond to advertising recommendation requests from online stores and push the recommended list of product selections for advertising to the management user of the independent site.

[0041] A computer device provided to meet one of the purposes of the present application includes a central processing unit and a memory, wherein the central processing unit is used to call and run a computer program stored in the memory to execute the steps of the advertising placement and product selection method described in the present application.

[0042] A computer-readable storage medium is provided to meet another purpose of the present application, which stores a computer program implemented according to the advertising selection method in the form of computer-readable instructions. When the computer program is called and executed by a computer, the steps included in the method are executed.

[0043] A computer program product provided to meet another purpose of the present application includes a computer program / instruction, which, when executed by a processor, implements the steps of the method described in any embodiment of the present application.

[0044] It can be known from the typical embodiments and alternative embodiments of the present application that the technical solution of the present application has many advantages, including but not limited to the following aspects:

[0045] First, the present application calculates the corresponding recommendation score by obtaining the adoption rate and effectiveness data of the predicted products, and then constructs a recommendation list of advertising product selection based on the recommendation score and pushes it to the management users of independent sites. Different from the prior art that only relies on the effectiveness data of predicted products as a reference indicator for advertising product selection, the technical solution here integrates the recommendation score obtained by integrating the characteristics of the two dimensions of the adoption rate and effectiveness data of the products as a reference indicator for advertising product selection. It can accurately predict the probability of a product being selected for advertising, and predict the conversion probability that may be obtained after the product is advertised. It uses quantitative values ​​to more comprehensively reflect the degree to which a product is suitable for advertising, so that the management users of independent sites can make quick decisions based on the recommendation score, and select products that are suitable for advertising and can achieve good results after advertising.

[0046] Secondly, on the one hand, most independent sites have limited financial resources and are unable to have rich product advertising data, resulting in an inability to decide on suitable products for advertising. The advertising recommendation list can effectively assist the management users of such independent sites to select products for advertising on their own independent sites, thereby solving the real pain points of the management users; on the other hand, for novices who manage independent sites and do not have much experience in advertising selection, the advertising recommendation list can more effectively assist such novice users to select suitable products for their own independent sites, thereby advertising them, improving user experience, and increasing user stickiness.

[0047] In addition, the technical solution of the present application is simple and easy to implement, has low model calculation time complexity, and can be parallelized on a large scale in engineering. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0049] Figure 1 A flowchart of a typical embodiment of the method for selecting products for advertisement placement of the present application;

[0050] Figure 2 A schematic diagram of a process for pre-building product adoption rate and effectiveness data in an embodiment of the present application;

[0051] Figure 3 A flowchart of the training process of the adoption rate prediction model described in the embodiments of the present application;

[0052] Figure 4 A flowchart of the training process of the click-through rate prediction model described in the embodiments of the present application;

[0053] Figure 5Schematic diagram of the process of training the conversion rate prediction model described in the embodiments of the present application;

[0054] Figure 6 A schematic diagram of a process for constructing an advertisement placement product selection recommendation list for push in an embodiment of the present application;

[0055] Figure 7 This is a principle block diagram of the advertising selection device of this application;

[0056] Figure 8 A schematic diagram of the structure of a computer device used in this application. DETAILED DESCRIPTION

[0057] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be interpreted as limiting the present application.

[0058] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "one", "said", and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present application refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be an intermediate element. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The term "and / or" used herein includes all or any unit and all combinations of one or more associated listed items.

[0059] It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as those generally understood by those skilled in the art to which this application belongs. It should also be understood that terms such as those defined in general dictionaries should be understood to have meanings consistent with those in the context of the prior art, and will not be interpreted with idealized or overly formal meanings unless specifically defined as here.

[0060] It will be understood by those skilled in the art that the "client", "terminal" and "terminal device" used herein include both devices with wireless signal receivers, which are devices with only wireless signal receivers without transmission capabilities, and devices with receiving and transmitting hardware, which are devices with receiving and transmitting hardware capable of two-way communication on a two-way communication link. Such devices may include: cellular or other communication devices such as personal computers, tablet computers, which have single-line displays or multi-line displays or cellular or other communication devices without multi-line displays; PCS (Personal Communications Service, personal communication system), which can combine voice, data processing, fax and / or data communication capabilities; PDA (Personal Digital Assistant, personal digital assistant), which may include a radio frequency receiver, pager, Internet / intranet access, web browser, notepad, calendar and / or GPS (Global Positioning System, global positioning system) receiver; conventional laptop and / or palmtop computers or other devices, which have and / or include a conventional laptop and / or palmtop computer or other device with and / or including a radio frequency receiver. The "client", "terminal" and "terminal device" used herein may be portable, transportable, installed in a vehicle (air, sea and / or land), or suitable for and / or configured to run locally, and / or in a distributed form, at any other location on the earth and / or in space. The "client", "terminal" and "terminal device" used herein may also be a communication terminal, an Internet terminal, a music / video playing terminal, for example, a PDA, a MID (Mobile Internet Device) and / or a mobile phone with a music / video playing function, or a smart TV, a set-top box and other devices.

[0061] The hardware referred to by the names such as "server", "client", and "service node" in this application is essentially an electronic device with the equivalent capabilities of a personal computer. It is a hardware device with the necessary components revealed by the von Neumann principle, such as a central processing unit (including an arithmetic unit and a controller), a memory, an input device, and an output device. The computer program is stored in its memory, and the central processing unit calls the program stored in the external memory into the internal memory for execution, executes the instructions in the program, and interacts with the input and output devices to complete specific functions.

[0062] It should be pointed out that the concept of "server" referred to in this application can also be extended to the case of server clusters. According to the network deployment principle understood by those skilled in the art, the servers should be logically divided. In physical space, these servers can be independent of each other but can be called through interfaces, or integrated into a physical computer or a set of computer clusters. Those skilled in the art should understand this flexibility, and should not use it to restrict the implementation of the network deployment method of this application.

[0063] Unless expressly specified, one or more technical features of the present application can be deployed on a server for implementation and accessed by a client through a remote call to obtain an online service interface provided by the server, or can be directly deployed and run on a client for access.

[0064] Unless explicitly specified, the various data involved in this application can be stored remotely on a server or on a local terminal device, as long as it is suitable for being called by the technical solution of this application.

[0065] Those skilled in the art should be aware that, although the various methods of the present application are described based on the same concept and thus present commonality to each other, unless otherwise specified, these methods can be independently executed. Similarly, for each embodiment disclosed in the present application, they are all proposed based on the same inventive concept, therefore, concepts with the same expression, and concepts that are appropriately changed for convenience despite different expressions, should be understood as equivalent.

[0066] Unless the mutually exclusive relationship between the embodiments to be disclosed in this application is explicitly stated, the relevant technical features involved in each embodiment can be cross-combined to flexibly construct a new embodiment, as long as such combination does not deviate from the creative spirit of this application and can meet the needs of the prior art or solve certain deficiencies in the prior art. Those skilled in the art should be aware of this flexibility.

[0067] An advertisement placement and product selection method of the present application can be programmed as a computer program product and deployed in a client or server for execution. For example, in the e-commerce platform application scenario of the present application, it is generally deployed in a server for implementation, whereby the method can be executed by accessing an interface opened after the computer program product is running and performing human-computer interaction with the process of the computer program product through a graphical user interface.

[0068] See also Figure 1 In a typical embodiment, the advertising product selection method of the present application includes the following steps:

[0069] Step S1100: Acquire advertising prediction data corresponding to each product in the product database of the independent site, wherein the advertising prediction data includes adoption rate and effectiveness data, wherein the adoption rate represents the predicted probability of the product being advertised, and the effectiveness data includes the conversion probability that may be obtained after the product is advertised;

[0070] The technical solution of the present application uses the operating environment of the e-commerce service platform as its application environment. The e-commerce service platform may be an e-commerce service platform that opens independent site services, typically, such as a cross-border e-commerce service platform. Such platforms need to consider the network environment between different parts of the world and the independence between merchants, and by configuring each merchant's store as an independent site, the e-commerce service platform serves a large number of such independent sites.

[0071] Each of the independent sites has a product database corresponding to the products on sale on its own website, and the product database contains a large amount of product information for describing each product on sale, and the product information includes but is not limited to various types of data such as product pictures, product titles, product details, product prices, product sku (category), inventory, product listing time, etc. In addition, in order to make the products on sale more popular, the management user of the independent site will advertise the products on sale in the e-commerce service platform, so that the products on sale are exposed to users, and before the advertising delivery execution process of the products on sale, the advertising prediction data is pre-generated to assist the management user of the independent site to select the products on sale and then deliver the advertisements, so the advertising prediction data is associated with the corresponding products on sale and stored in the product database. Based on this, it is not difficult to understand that when the management user of the independent site triggers the advertising prediction data acquisition request, the response to the request can be to provide a corresponding data access interface to the product database of each independent site by calling the e-commerce service platform, so as to perform an access operation on the product database of the independent site, and obtain the advertising prediction data corresponding to each product in the product database.

[0072] The advertising effectiveness data includes adoption rate and effectiveness data. The adoption rate represents the predicted probability of the product being advertised. It is not difficult to understand that the high or low adoption rate depends on whether the product information of the product has features suitable for advertising. The product with features suitable for advertising has a high adoption rate, and vice versa. The features include product characteristics and statistical features. The product characteristics include short shelf time, hot-selling product categories in the current product sales market, and low product prices. The statistical features include high click-through rate, high repurchase rate, high sales volume, and high purchase rate after the product is advertised. It can be seen that this feature meets the market trend considerations of the management users of the independent site when selecting products for advertising. Therefore, products with this feature have certain market competitiveness, that is, sales potential.

[0073] The effectiveness data includes a characterization of the conversion probability that may be obtained after the product is advertised. The conversion probability includes the predicted click-through rate corresponding to the user clicking on the product after the advertisement is placed, and the predicted conversion rate corresponding to other user behaviors generated after the user clicks on the product advertisement. The other user behaviors include but are not limited to adding to a shopping cart, purchasing, repeat purchasing and the like. It is not difficult to understand that the effectiveness data reflects the effect obtained after the product is advertised. A high conversion efficiency corresponding to the effectiveness data indicates that the product can achieve good results after the advertisement is placed, otherwise the effect may be poor. The good results can to a certain extent reflect that the user is interested in the advertised product and even purchases it out of need.

[0074] Step S1200: Calculate the recommendation score corresponding to each product according to the adoption rate and effectiveness data of the product;

[0075] Since the adoption rate and effectiveness data of the product are reference indicators that can assist the management users of independent sites to select products for advertising, in order to obtain a more comprehensive and intuitive numerical representation, the number of reference indicators is reduced, and the reference indicators of the two dimensions of current adoption rate and effectiveness data are combined to be reduced to a single reference indicator represented by the numerical representation. In one embodiment, the value is obtained by multiplying the purchase rate of the product, the click-through rate and the conversion rate corresponding to the effectiveness data, and the value range is 0-100%, so that the value is used as the recommendation score. For example, if the purchase rate is 80%, the click-through rate is 85%, and the conversion rate is 95%, then the recommendation score corresponding to the multiplication of the three is 64.5%. It is not difficult to understand that the higher the recommendation score, the higher the sales potential of the product, and the better the effect achieved after the corresponding product is advertised.

[0076] Step S1300: Select products in the product database as advertising selection products according to the recommendation scores to construct an advertising selection recommendation list.

[0077] Furthermore, based on the recommendation scores corresponding to each product in the product database, multiple products with the highest recommendation scores are selected as the advertising product selections, and the advertising product selections are sorted in descending order according to the recommendation scores corresponding to the advertising product selections, and then a recommended advertising product selection list is constructed from the sorted advertising product selections. The specific number of selected products can be flexibly adjusted by technical personnel in this field according to the number of products allowed to be displayed in the recommended advertising product selection list according to actual conditions. Thus, the recommended advertising product selection list is pushed to the administrator user of the independent site, so that the administrator user can make decisions quickly based on the recommended advertising product selection list, and accurately select products with high sales potential and good results that can be achieved after the product advertisement is placed for advertising.

[0078] It can be known from the typical embodiments of the present application that the technical solution of the present application has many advantages, including but not limited to the following aspects:

[0079] First, the present application calculates the corresponding recommendation score by obtaining the adoption rate and effectiveness data of the predicted products, and then constructs a recommendation list of advertising product selections based on the recommendation score and pushes it to the management users of independent sites. Different from the prior art that only relies on the effectiveness data of predicted products as a reference indicator for advertising product selection, the technical solution here integrates the recommendation score obtained by integrating the characteristics of the two dimensions of the adoption rate and effectiveness data of the products as a reference indicator for advertising product selection. It can accurately predict the probability of a product being selected for advertising, and predict the conversion probability that may be obtained after the product is advertised. It uses quantitative values ​​to more comprehensively reflect the degree to which a product is suitable for advertising, so that the management users of independent sites can make quick decisions based on the recommendation score, and select products that are suitable for advertising and can obtain good results after the advertising as advertising product selections.

[0080] Secondly, on the one hand, most independent sites have limited financial resources and are unable to have rich product advertising data, resulting in an inability to decide on suitable products for advertising. The advertising recommendation list can effectively assist the management users of such independent sites to select products for advertising on their own independent sites, thereby solving the real pain points of the management users; on the other hand, for novices who manage independent sites and do not have much experience in advertising selection, the advertising recommendation list can more effectively assist such novice users to select suitable products for their own independent sites, thereby advertising them, improving user experience, and increasing user stickiness.

[0081] In addition, the technical solution of the present application is simple and easy to implement, has low model calculation time complexity, and can be parallelized on a large scale in engineering.

[0082] See also Figure 2 In a further embodiment, before step S1100, the step of obtaining the advertisement prediction data corresponding to each commodity in the commodity database of the independent site, the following steps are included:

[0083] Step S1000: using an adoption rate prediction model that has been pre-trained to a convergent state to determine the adoption rate of each product in the product database of the independent site;

[0084] The corresponding data access interface of the commodity database of the independent site provided by the e-commerce service platform is called, and the commodity information corresponding to each commodity is obtained from the commodity database and input into the adoption rate prediction model pre-trained to a convergence state, so that the adoption rate of each commodity is determined by the model.

[0085] The product information includes but is not limited to product images, product titles, product details, product prices, product sku (category), inventory, and product shelf time. The adoption rate prediction model is pre-trained to a convergence state so that it has the ability to predict the corresponding adoption rate of each product based on the product information of the product. Specifically, the model can extract the product characteristics and statistical characteristics from the product information of the product that prompt the management user of the independent site to select the product for advertising, and determine the adoption rate. The product characteristics include but are not limited to a short shelf time, a hot-selling product in the current product sales market, and a low price. The statistical characteristics include but are not limited to a high click-through rate, a high repurchase rate, a high sales volume, and a high purchase rate after the product is advertised.

[0086] The adoption rate prediction model that has been pre-trained to convergence is called to perform feature extraction on each product in the product database of the independent site, and the product features and the statistical features are extracted, so as to obtain two feature vectors of its output, which are respectively a product feature vector and a statistical feature vector. Further, the product feature vector and the feature vector are fused to obtain a fused feature vector, and the fused feature vector is mapped to a classification space. Then, a classifier constructed by Softmax or Sigmod is used to perform probability calculation on the classification space to obtain the probability that the fused feature vector is mapped to the classification space corresponding to the product being selected for advertisement placement, that is, the adoption rate.

[0087] The process of training the purchase rate prediction model to convergence is further disclosed in the subsequent embodiments of this application, and this step is not discussed here for the time being.

[0088] Step S1010: using a click rate prediction model that has been pre-trained to a convergent state to determine the click rate of each product in the product database of the independent site;

[0089] The corresponding data access interface of the product database of the independent site provided by the e-commerce service platform is called, and the product information corresponding to each product is obtained from the product database and input into the click-through rate prediction model that has been pre-trained to a convergence state, so that the click-through rate of each product is determined by the model.

[0090] The product information includes but is not limited to product images, product titles, product details, product prices, product SKUs (categories), inventory, and product shelf time. The click-through rate prediction model is pre-trained to a convergence state so that it has the ability to determine the click-through rate of a product by prompting users to click on the corresponding product features after the product is advertised in the product information, and the product features include image features and text features.

[0091] The click-through rate prediction model that has been pre-trained to convergence is called to perform feature extraction on the product information corresponding to each product in the product database of the independent site, and the product features that prompt users to click on the product advertisement in the product information are extracted to obtain the feature vector output by the model, and the feature vector is mapped to the classification space. Then, a classifier constructed by Softmax or Sigmod is used to perform probability calculation on the classification space to obtain the probability that the feature vector is mapped to the classification space and the product is clicked by the user after the advertisement is placed, that is, the click-through rate.

[0092] The process of training the click-through rate prediction model to convergence is further disclosed in the subsequent embodiments of this application, and this step is not discussed here for the time being.

[0093] Step S1020: using a conversion rate prediction model that has been pre-trained to a convergence state to determine the conversion rate of each commodity in the commodity database of the independent site;

[0094] The corresponding data access interface of the commodity database of the independent site provided by the e-commerce service platform is called, and the commodity information corresponding to each commodity is obtained from the commodity database and input into the conversion rate prediction model pre-trained to a convergence state, so that the conversion rate of each commodity is determined by the model.

[0095] The product information includes but is not limited to product pictures, product titles, product details, product prices, product sku (category), inventory, and product listing time.

[0096] The conversion rate prediction model is pre-trained to a convergence state so that it has the ability to determine the conversion rate of a product based on product features corresponding to other user behaviors generated after the product advertisement in the product information is clicked by the user, and the product features include image features and text features.

[0097] The conversion rate prediction model that has been pre-trained to convergence is called to perform feature extraction on the product information corresponding to each product in the product database of the independent site, and the product features corresponding to other user behaviors generated after the product advertisement is clicked by the user are extracted from the product information of the product, and the feature vector output by the model is obtained, and the feature vector is mapped to the classification space. Then, a classifier constructed by Softmax or Sigmod is used to perform probability calculation on the classification space to obtain the probability that the feature vector is mapped to the classification space of other user behaviors generated after the product advertisement is clicked by the user, that is, the conversion rate.

[0098] The process of training the conversion rate prediction model to convergence is further disclosed in the subsequent embodiments of this application, and this step is not discussed here for the time being.

[0099] Step S1030: storing the adoption rate and effectiveness data of each product in the product database corresponding to each product, wherein the effectiveness data is the product of the click rate and the conversion rate associated with the same product.

[0100] The performance data includes the click rate and conversion rate corresponding to the product.

[0101] For the adoption rate and effectiveness data of each product, both are time-sensitive. Therefore, a scheduled task can be set to execute steps S1000-S1020 at a specific time point. For example, the specific time point may be the early morning of each day. Since the product data in the product database changes less at this time, and the server of the e-commerce service platform has sufficient resources to allocate, it is very suitable for executing the scheduled task. Accordingly, the adoption rate and effectiveness data can be properly updated to make them more accurate, laying the foundation for the accuracy of the advertising selection in the subsequent steps.

[0102] In order to facilitate the subsequent call of the adoption rate and effectiveness data of each product, the two are mapped and associated with the corresponding products and stored in the database. For the effectiveness data, the product of the click rate and the conversion rate associated with the same product is used as the numerical representation of the effectiveness data.

[0103] In this embodiment, the purchase data and performance data corresponding to each commodity in the commodity database of the independent site are prepared in advance, so that subsequent instructions or requests for calling the purchase data and performance data can be quickly responded to, which greatly improves the response efficiency, reduces user waiting time, improves user experience, and increases user stickiness.

[0104] See also Figure 3In a further embodiment, in step S1000, in the step of determining the adoption rate of each commodity in the commodity database of the independent site using the adoption rate prediction model pre-trained to a convergence state, the training process of the adoption rate prediction model includes the following steps:

[0105] Step S1001: call a training sample from a data set, wherein the training sample includes a positive sample and a negative sample, and each training sample is provided with a supervision label representing whether a product has been advertised. The positive sample includes product information and statistical information of advertised products in the product database of the e-commerce platform, and the negative sample includes product information and statistical information of non-advertised products in the product database of the e-commerce platform. The product information includes product listing time, product category, and product price, and the statistical information includes click-through rate, purchase rate, and repurchase rate;

[0106] In one embodiment, in order to facilitate the adoption rate prediction model to more accurately extract relevant features from the training samples in the data set, the training samples in the data set are subjected to text preprocessing operations, and the text preprocessing operations include removing non-text parts in the training samples, such as using regular expressions to delete some special non-Chinese and English characters and punctuation marks therein; segmenting the training samples, for English, the word segmentation can directly call the split() function, and for Chinese, the stammering word segmentation method is adopted, and the corresponding method can be called directly by calling the method function data package pre-packaged by the third-party platform to directly call the method therein; removing stop words in the training samples, and removing the stop words has no effect on the overall semantics, such as a large number of function words, pronouns or nouns and verbs without specific meanings in Chinese, and articles and pronouns in English.

[0107] The ratio of positive and negative samples can be flexibly set by those skilled in the art based on prior knowledge or experimental experience, and is recommended to be 1:10, wherein the statistical information in the positive samples includes the click-through rate, purchase rate, and repurchase rate corresponding to the advertised products, and the statistical information in the negative samples is empty.

[0108] Step S1002: input the training sample into the adoption rate prediction model to obtain a commodity feature vector and a statistical feature vector corresponding to the commodity information and statistical information in the training sample;

[0109] The training samples are input into the adoption rate prediction model for feature extraction, and the product features and statistical features corresponding to the product information and statistical information of the positive and negative sample products that prompt the management user of the independent site to select the product for advertising are extracted to obtain the product feature vector and statistical feature vector output by the model. The product features include but are not limited to a short shelf time, products of hot-selling categories in the current product sales market, and low prices. The statistical features include but are not limited to a high click-through rate, a high repurchase rate, a high sales volume, and a high purchase rate after the product is advertised. Each feature in the corresponding product features and statistical features is encoded using one hot coding. The coding here can also use the coding module in Word2vec and Bert, so as to obtain the output corresponding two-way embedding coding, namely the product feature vector and statistical feature vector.

[0110] Step S1003: concatenate the product feature vector and the statistical feature vector to obtain a feature fusion vector, input the feature fusion vector into a fully connected layer and map it to a preset classification space to obtain a predicted adoption rate;

[0111] The splicing is vector splicing, and feature fusion is performed on the product feature vector and the statistical feature vector. Generally speaking, in order to fuse the two feature vectors, the corresponding elements of the two are usually directly added in the same dimension to obtain the feature fusion vector. If the dimensions of the two are different, they can be converted into vectors of the same dimension through linear transformation. Further, the feature fusion vector is input into the fully connected layer for linear transformation to map it to a binary classification space. The binary classification space includes a positive space that represents that the product is selected for advertising, and a negative space that represents that the product is not selected for advertising. Then, a classifier is used to calculate the probability corresponding to the mapping of the feature fusion vector to the binary space. In this process, the probability is obtained as the predicted adoption rate through the Sigmod or Softmax method function.

[0112] Step S1004: calculate the loss value corresponding to the cross entropy loss of the predicted adoption rate according to the supervised label, and determine whether the loss value reaches a preset threshold. When it reaches the preset threshold, terminate the training; otherwise, implement a gradient update on the model according to the loss value, and call the next training sample in the data set to continue iterative training of the model.

[0113] A preset loss function is called, which can be flexibly set by technicians in this field according to prior knowledge or experimental experience. The loss value corresponding to the cross entropy loss of the predicted adoption rate is calculated based on the supervised label. When the loss value reaches the preset threshold, it indicates that the model has been trained to a convergence state, thereby terminating the model training; when the loss value does not reach the preset threshold, it indicates that the model has not converged, and then the model is gradient updated according to the loss value. Usually, the weight parameters of each link of the model are corrected by back propagation to make the model further close to convergence, and then, the next sample data in the data set is continued to be called to perform iterative training on the model until the model is trained to a convergence state.

[0114] Those skilled in the art should know that the adoption rate prediction model is supervised and trained based on the data set consisting of the supervision labels and positive and negative samples until the model converges, so that the adoption rate prediction model has the ability to predict the adoption rate corresponding to each commodity based on the commodity information of the commodity. Specifically, the model can obtain the adoption rate corresponding to the commodity based on the commodity characteristics and statistical characteristics extracted from the commodity information of the commodity in the commodity database that prompt the management user of the independent site to select the commodity for advertising. The commodity characteristics include but are not limited to a short shelf time, hot-selling categories in the current commodity sales market, and low commodity prices. The statistical characteristics include high click-through rate, high repurchase rate, high sales volume, high purchase rate, etc. corresponding to the commodity after the advertisement is placed.

[0115] In this embodiment, the training process of the adoption rate prediction model based on the deep learning model is disclosed. It can be seen that under the sample data and supervised label training of the data set, the adoption rate prediction model extracts the product features and statistical features corresponding to the product information of the product for inference, and can obtain more accurate prediction capabilities, thereby ensuring that it can subsequently serve the prediction of the adoption rate corresponding to the product.

[0116] See also Figure 4 In a further embodiment, in step S1010, in the step of determining the click rate of each commodity in the commodity database of the independent site using the click rate prediction model pre-trained to a convergent state, the training process of the click rate prediction model includes the following iterative execution steps:

[0117] Step S1011, obtaining training samples in the data set, wherein the training samples include product information of products that have been advertised and clicked by users in the product database of the e-commerce platform, and a supervision label corresponding to whether the advertisement of the product has been clicked by the user;

[0118] The product information includes but is not limited to product images, product titles, product details, product prices, product sku (category), inventory, and product listing time. In one embodiment, in order to facilitate the training samples in the data set to be more accurately extracted by the neural network model, the training samples in the data set are subjected to text preprocessing operations, and the text preprocessing operations include removing non-text parts in the training samples, such as using regular expressions to delete some special non-Chinese and English characters and punctuation marks; performing word segmentation on the training samples, for English, the word segmentation can directly call the split() function, for Chinese, the stuttering word segmentation method can be used, and the corresponding method can be called directly by the third-party platform to directly call the method in the pre-packaged method function data package; removing stop words in the training samples, removing the stop words has no effect on the overall semantics, such as a large number of function words, pronouns or nouns and verbs without specific meanings in Chinese, and articles and pronouns in English.

[0119] Step S1012: input the training sample into the click rate prediction model to extract corresponding deep semantic features and obtain a feature vector;

[0120] The training samples are input into the click-through rate prediction model for feature extraction, and the corresponding product features that prompt users to click on the product after the product is advertised are extracted from the product information of the product. In the process of extracting the features of the image features corresponding to the product image in the product information, the product image is first divided into multiple image units, and the size of each image unit is equal. Therefore, after the feature image feature extraction by the click-through rate prediction model, each image unit can obtain a corresponding single image feature vector, and then, the output of the click-through rate prediction model is converted into a high-dimensional feature vector, and the corresponding position information of the corresponding image elements of each high-dimensional feature vector in the product image is spliced ​​into an image coding vector; feature extraction is performed on the text information such as the product title and product details in the product information after the above preprocessing, and the corresponding text features are extracted, and one is used for each feature Hot coding, the coding module in Word2vec and Bert can also be used here to obtain the embedding coding corresponding to the model output, that is, the product text coding vector. Further, the image coding vector and the product text coding vector are vector-concatenated to perform feature fusion. Generally speaking, in order to fuse the two coding vectors, the corresponding elements of the two are usually directly added in the same dimension to obtain a fused coding vector. If the dimensions of the two are different, they can be converted into vectors of the same dimension through linear transformation.

[0121] The click-through rate prediction model structure includes an image feature extraction module and a text feature extraction module. The two modules can be taken from a neural network model based on CNN that is suitable for deep semantic feature extraction of images, such as the corresponding feature extraction modules in models such as Resnet and EfficientNet, and a neural network model based on RNN that is suitable for deep semantic feature extraction of texts, such as the corresponding feature extraction modules in models such as Bert and Electra.

[0122] Step S1013: using a classifier to classify and map the feature vectors to predict corresponding click-through rates;

[0123] Furthermore, the fused coding vector is input into the fully connected layer for linear conversion and mapped to a binary classification space, which includes a positive space representing that the advertisements for the products are clicked by users, and a negative space representing that the advertisements for the products are not clicked by users. Then, a classifier is used to calculate the probability corresponding to the mapping of the fused coding vector to the binary space. In this process, the probability is obtained through the Sigmod or Softmax method function as the predicted click-through rate.

[0124] Step S1014: Calculate the cross entropy loss value corresponding to the click rate predicted by the classifier according to the supervision label, and perform gradient update on the model according to the cross entropy loss value until the model converges.

[0125] A preset loss function is called, which can be flexibly set by a technician in this field according to prior knowledge or experimental experience. The loss value corresponding to the cross entropy loss of the predicted click-through rate is calculated based on the supervised label. When the loss value reaches the preset threshold, it indicates that the model has been trained to a convergence state, so that the model training can be terminated; when the loss value does not reach the preset threshold, it indicates that the model has not converged, so the model is gradient updated according to the loss value, usually back-propagating to correct the weight parameters of each link of the model to make the model further close to convergence, and then, the next sample data in the data set is continued to be called to perform iterative training on the model until the model is trained to a convergence state.

[0126] It can be understood that the click-through rate prediction model is supervised and trained based on the supervisory labels until the model converges, so that the model can extract the product features in the product information that prompt the product advertisement to be clicked by users, and determine the click-through rate corresponding to the product, and the product features include text features and image features.

[0127] In this embodiment, the training process of the click-through rate prediction model implemented based on the deep learning model is disclosed. It can be seen that under the sample data and supervised label training of the data set, the click-through rate prediction model extracts the image features and text features in the product information of the product for inference, and can obtain more accurate prediction capabilities, thereby ensuring that it can subsequently serve the prediction of the click-through rate corresponding to the product, so as to more accurately reflect the possibility of the product being clicked by the user after the advertisement is released with the click-through rate.

[0128] See also Figure 5 In a further embodiment, in step S1020, in the step of determining the conversion rate of each commodity in the commodity database of the independent site using the conversion rate prediction model pre-trained to a convergence state, the training process of the conversion rate prediction model includes the following iterative execution steps:

[0129] Step S1021, obtaining training samples in the data set, wherein the training samples include product information of products in the product database of the e-commerce platform that have been advertised and have generated relevant user behaviors after being clicked by users, and supervision labels corresponding to whether the advertisements for the products have generated relevant user behaviors after being clicked by users;

[0130] The product information includes but is not limited to product images, product titles, product details, product prices, product SKUs (categories), inventory, and product launch time. The relevant user behaviors include but are not limited to adding to a shopping cart, purchasing, and repurchasing.

[0131] In one embodiment, in order to facilitate the neural network model to more accurately extract relevant features from the training samples in the data set, the training samples in the data set are subjected to text preprocessing operations, and the text preprocessing operations include removing non-text parts in the training samples, such as using regular expressions to delete some special non-Chinese and English characters and punctuation marks therein; segmenting the training samples, for English, the word segmentation can directly call the split() function, and for Chinese, the stammering word segmentation method is adopted, and the corresponding method can be directly called by calling the method function data package prepackaged by the third-party platform to directly call the method therein; removing stop words in the training samples, and removing the stop words has no effect on the overall semantics, such as a large number of function words, pronouns or nouns and verbs without specific meanings in Chinese, and articles and pronouns in English.

[0132] Step S1022: input the training sample into the conversion rate prediction model to extract corresponding deep semantic features and obtain a feature vector;

[0133] The training samples are input into the conversion rate prediction model for feature extraction, and the product features corresponding to other user behaviors generated after the product advertisement is clicked by the user in the product information of the product are extracted. In the process of extracting the features of the image features corresponding to the product image in the product information, the product image is first divided into a plurality of image units, and the size of each image unit is the same. Therefore, after the feature image feature is extracted by the conversion rate prediction model, each image unit can obtain a corresponding single image feature vector. Then, the output of the conversion rate prediction model is converted into a high-dimensional feature vector, and the image encoding vector is spliced ​​according to the corresponding position information of the image elements corresponding to each high-dimensional feature vector in the product image; The product title, product details and other text information in the product information are subjected to feature extraction after the above preprocessing, and the corresponding text features are extracted. Onehot encoding is used for each feature. The encoding here can also use the encoding module in Word2vec and Bert, so as to obtain the embedding encoding corresponding to the model output, that is, the product text encoding vector. Further, the image encoding vector and the product text encoding vector are vector spliced ​​to perform feature fusion. Generally speaking, in order to fuse the two encoding vectors, the corresponding elements of the two are usually directly added in the same dimension to obtain a fused encoding vector. If the dimensions of the two are different, they can be converted into vectors of the same dimension through linear transformation.

[0134] The conversion rate prediction model structure includes an image feature extraction module and a text feature extraction module. The two modules can be taken from a neural network model based on CNN that is suitable for deep semantic feature extraction of images, such as the corresponding feature extraction modules in models such as Resnet and EfficientNet, and a neural network model based on RNN that is suitable for deep semantic feature extraction of texts, such as the corresponding feature extraction modules in models such as Bert and Electra.

[0135] Step S1023: using a classifier to classify and map the feature vector to predict the corresponding conversion rate;

[0136] Furthermore, the fused coding vector is input into the fully connected layer for linear conversion and mapped to a binary classification space, which includes a positive space representing that the advertisement for the product generates relevant user behavior after being clicked by the user, and a negative space representing that the advertisement for the product does not generate relevant user behavior after being clicked by the user. Then, a classifier is used to calculate the probability corresponding to the mapping of the fused coding vector to the binary space. In this process, the probability is obtained through the Sigmod or Softmax method function as the predicted conversion rate.

[0137] Step S1024: Calculate the cross entropy loss value corresponding to the conversion rate predicted by the classifier according to the supervised label, and perform gradient update on the model according to the cross entropy loss value until the model converges.

[0138] A preset loss function is called, which can be flexibly set by technicians in this field according to prior knowledge or experimental experience. The loss value corresponding to the cross entropy loss of the predicted conversion rate is calculated based on the supervised label. When the loss value reaches the preset threshold, it indicates that the model has been trained to a convergence state, thereby terminating the model training; when the loss value does not reach the preset threshold, it indicates that the model has not converged, and then the model is gradient updated according to the loss value. Usually, the weight parameters of each link of the model are corrected by back propagation to make the model further close to convergence, and then, the next sample data in the data set is continued to be called to perform iterative training on the model until the model is trained to a convergence state.

[0139] It can be understood that the conversion rate prediction model is supervised and trained based on the supervisory labels until the model converges, so that the model can extract the product features corresponding to other user behaviors generated after the product advertisement is clicked by the user in the product information of the product, and determine the conversion rate corresponding to the product, wherein the product features include text features and image features.

[0140] In this embodiment, the training process of the conversion rate prediction model implemented based on the deep learning model is disclosed. It can be seen that under the sample data and supervised label training of the data set, the conversion rate prediction model extracts the image features and text features in the product information of the product for reasoning, and can obtain more accurate prediction capabilities, thereby ensuring that it can subsequently serve the prediction of the conversion rate corresponding to the product, so as to more accurately reflect the possibility of generating relevant user behavior after the product's advertisement is clicked by the user with the conversion rate.

[0141] See also Figure 6 In the extended embodiment, step S1300, selecting the products in the product database as the advertising selection products according to the recommendation score to construct the advertising selection product recommendation list, includes the following steps:

[0142] Step S1310: sort the commodities in the commodity database according to their recommendation scores, and construct a recommended list of commodities for advertisement placement;

[0143] According to the recommendation scores corresponding to the products in the product database, multiple products with the highest corresponding recommendation scores are selected as advertising product selections, and the products are sorted in order from high to low according to the recommendation scores corresponding to the advertising product selections, and then a recommended advertising product selection list is constructed based on the advertising product selections in this sorting. The number of selected products can be flexibly changed by technical personnel in this field according to the number of products displayed in the recommended advertising product selection list.

[0144] In an extended embodiment, three other sortings of three dimensions can be set for the recommended list of selected products for advertising placement, and the three sortings can be switched with the aforementioned recommendation score sorting. The recommended list of selected products for advertising placement can be set to be displayed in the aforementioned recommendation score sorting by default, and the three sortings can be switched to be displayed optionally. Specifically, the three sortings of the three dimensions are based on the product listing time, product category, and product price corresponding to the selected products for advertising placement, and the corresponding product listing time is sorted in order from early to late; the product category is sorted in order of the first letter of the pinyin of the first letter AZ, where the same type is classified together; the product price is sorted in order from low to high, and the three sortings are used as the optional switching display of the recommended list of selected products for advertising placement. As a result, it is convenient for users to quickly select suitable products from the recommended list of selected products for advertising placement based on the recommendation score sorting combined with the corresponding characteristics of the other three sortings. For example, users can select products ranked third in recommendation score and first in product price from the recommended list of selected products for advertising placement, that is, products with relatively the lowest price and higher estimated sales potential.

[0145] Step S1320: respond to the online store's advertising recommendation request, and push the advertising product recommendation list to the management user of the independent site.

[0146] The independent site runs its own store, namely an online store. The management user of the independent site can touch the advertisement recommendation button set up by the e-commerce service platform to provide advertisement recommendation service in the online store, thereby triggering the advertisement recommendation request to the server of the e-commerce service platform. The server receives the request and responds to it, and pushes the advertisement selection recommendation list to the management user of the independent site.

[0147] In this embodiment, a recommended list of products for advertising is constructed based on the recommendation scores corresponding to the products in the product database, and is pushed to the management user of the independent site, so that the management user can make quick decisions and accurately select products with certain sales potential and better advertising results.

[0148] See also Figure 7, an advertising placement and selection device provided to meet one of the purposes of the present application is a functional embodiment of the advertising placement and selection method of the present application, the device includes: a data acquisition module 1100, a score calculation module 1200, and a list construction module 1300, wherein the data acquisition module 1100 is used to obtain the advertising prediction data corresponding to each product in the product database of the independent site, the advertising prediction data includes adoption rate and effectiveness data, the adoption rate represents the predicted probability of the product being advertised, and the effectiveness data includes the conversion probability that may be obtained after the product is advertised; the score calculation module 1200 is used to calculate the recommendation score corresponding to each product according to the adoption rate and effectiveness data of the product; the list construction module 1300 is used to select the product in the product database as the advertising placement selection according to the recommendation score to construct a recommended list of advertising placement selection.

[0149] In a further embodiment, the data acquisition module 1100 includes: an adoption rate determination module, which is used to determine the adoption rate of each commodity in the commodity database of the independent site by using an adoption rate prediction model pre-trained to a convergence state; a click-through rate confirmation module, which is used to determine the click-through rate of each commodity in the commodity database of the independent site by using a click-through rate prediction model pre-trained to a convergence state; a conversion rate confirmation module, which is used to determine the conversion rate of each commodity in the commodity database of the independent site by using a conversion rate prediction model pre-trained to a convergence state; and a data storage sub-module, which is used to store the adoption rate and effectiveness data of each commodity corresponding to each commodity in the commodity database, wherein the effectiveness data is the product of the click-through rate and the conversion rate associated with the same commodity.

[0150] In a further embodiment, in the adoption rate determination module, the training process of the adoption rate prediction model includes: a training sample calling module, which is used to call a training sample from a data set, wherein the training sample includes a positive sample and a negative sample, and each training sample is provided with a corresponding supervision label representing whether the product has been advertised, the positive sample includes the product information and statistical information of the advertised products in the product database of the e-commerce platform, and the negative sample includes the product information and statistical information of the non-advertised products in the product database of the e-commerce platform, the product information includes the product listing time, product category, and product price, and the statistical information includes click-through rate, purchase rate, and repurchase rate; a feature extraction module, which is used to extract the product information and statistical information of the product database of the e-commerce platform. The training sample is input into the adoption rate prediction model to obtain a product feature vector and a statistical feature vector corresponding to the product information and statistical information in the training sample; a vector splicing module is used to splice the product feature vector and the statistical feature vector to obtain a feature fusion vector, and input the feature fusion vector into a fully connected layer to map it to a preset classification space to obtain a predicted adoption rate; an iterative training module is used to calculate a loss value corresponding to the cross entropy loss of the predicted adoption rate according to the supervision label, and determine whether the loss value reaches a preset threshold. When it reaches the preset threshold, the training is terminated; otherwise, the model is gradient updated according to the loss value, and the next training sample in the data set is called to continue iterative training of the model.

[0151] In a further embodiment, in the click-through rate confirmation module, the iterative training process of the click-through rate prediction model includes: a training sample calling module, which is used to obtain training samples in a data set, and the training samples include product information of products that have been advertised and clicked by users in the product database of the e-commerce platform, and corresponding supervisory labels that characterize whether the advertisements for the products are clicked by users; a feature extraction module, which is used to input the training samples into the click-through rate prediction model to extract corresponding deep semantic features and obtain feature vectors; a classification mapping module, which is used to classify and map the feature vectors using a classifier to predict the corresponding click-through rate; and a gradient update module, which is used to calculate the cross-entropy loss value corresponding to the click-through rate predicted by the classifier based on the supervisory label, and implement gradient update on the model based on the cross-entropy loss value until the model converges.

[0152] In a further embodiment, in the conversion rate confirmation module, the iterative training process of the conversion rate prediction model includes: a training sample calling module, which is used to obtain training samples in a data set, wherein the training samples include product information of products that have been advertised in the product database of the e-commerce platform and have generated relevant user behaviors after being clicked by users, and supervisory labels corresponding to whether the advertisements for the products have generated relevant user behaviors after being clicked by users; a feature extraction module, which is used to input the training samples into the conversion rate prediction model to extract corresponding deep semantic features and obtain feature vectors; a classification mapping module, which is used to classify and map the feature vectors using a classifier to predict the corresponding conversion rate; and a gradient update module, which is used to calculate the cross entropy loss value corresponding to the conversion rate predicted by the classifier based on the supervisory labels, and implement gradient updates on the model based on the cross entropy loss value until the model converges.

[0153] In an extended embodiment, the list construction module 1300 includes: a rating sorting module, which is used to sort the recommended ratings corresponding to the products in the product database to construct a recommended list of selected products for advertising; and a list push module, which is used to respond to advertising recommendation requests from online stores and push the recommended list of selected products for advertising to the management user of the independent site.

[0154] In order to solve the above technical problems, the present application also provides a computer device. Figure 8 As shown, a schematic diagram of the internal structure of a computer device. The computer device includes a processor, a computer-readable storage medium, a memory, and a network interface connected via a system bus. Among them, the computer-readable storage medium of the computer device stores an operating system, a database, and computer-readable instructions. The database may store a control information sequence. When the computer-readable instructions are executed by the processor, the processor can implement a method for selecting products for advertising. The processor of the computer device is used to provide computing and control capabilities to support the operation of the entire computer device. The memory of the computer device may store computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor can execute the method for selecting products for advertising. The network interface of the computer device is used to connect and communicate with a terminal. Those skilled in the art can understand that Figure 8 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 those shown in the figure, or combine certain components, or have a different arrangement of components.

[0155] In this embodiment, the processor is used to execute Figure 7The memory stores the program code and various data required to execute the above modules or submodules. The network interface is used to transmit data between user terminals or servers. The memory in this embodiment stores the program code and data required to execute all modules / submodules in the advertising selection device of this application, and the server can call the program code and data of the server to execute the functions of all submodules.

[0156] The present application also provides a storage medium storing computer-readable instructions. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the advertising placement and product selection method of any embodiment of the present application.

[0157] The present application also provides a computer program product, including a computer program / instruction, which implements the steps of the method described in any embodiment of the present application when executed by one or more processors.

[0158] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments of the present application can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a computer-readable storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0159] Those skilled in the art will appreciate that the various operations, methods, steps, measures, and schemes in the processes discussed in this application may be alternated, altered, combined, or deleted. Further, other steps, measures, and schemes in the various operations, methods, and processes discussed in this application may also be alternated, altered, rearranged, decomposed, combined, or deleted. Further, the steps, measures, and schemes in the prior art that are similar to those disclosed in this application may also be alternated, altered, rearranged, decomposed, combined, or deleted.

[0160] To summarize, this application uses a purchase rate prediction model, a click-through rate prediction model, and a conversion rate prediction model to accurately predict the probability of a product being selected for advertising, the probability of a product being clicked by a user after the product advertisement is exposed, and the probability of a corresponding user behavior being generated after the product advertisement is clicked by a user. These probabilities accurately reflect the market competitiveness of the product, that is, its sales potential, as well as the results that can be achieved after the product is advertised. Based on this, a list of selected products for advertising is constructed to assist users in quickly making decisions on product selection for advertising based on the advertising list, and accurately selecting products that have certain sales potential and can achieve better advertising results.

[0161] The above description is only a partial implementation method of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for selecting products for advertising placement, characterized in that: The steps include: Determine the adoption rate of each product in the product database of the independent site using the adoption rate prediction model that has been pre-trained to a convergent state; Acquire advertising prediction data corresponding to each product in the product database of the independent site, wherein the advertising prediction data includes adoption rate and effectiveness data, wherein the adoption rate represents the predicted probability of the product being advertised, and the effectiveness data includes the conversion probability that may be obtained after the product is advertised; Calculate the recommendation score corresponding to each product based on the adoption rate and effectiveness data of the product; Selecting products in the product database as advertising selection products according to the recommendation scores to construct an advertising selection product recommendation list; In the step of using the adoption rate prediction model pre-trained to a convergent state to determine the adoption rate of each commodity in the commodity database of the independent site, the training process of the adoption rate prediction model includes the following steps: A training sample is called from a data set, wherein the training sample includes positive samples and negative samples, and each training sample is provided with a supervision label representing whether a product has been advertised. The positive sample includes product information and statistical information of advertised products in the product database of the e-commerce platform, and the negative sample includes product information and statistical information of non-advertised products in the product database of the e-commerce platform. The product information includes product listing time, product category, and product price, and the statistical information includes click-through rate, purchase rate, and repurchase rate; Inputting the training sample into the adoption rate prediction model to obtain a commodity feature vector and a statistical feature vector corresponding to the commodity information and statistical information in the training sample; The product feature vector and the statistical feature vector are concatenated to obtain a feature fusion vector, which is input into a fully connected layer and mapped to a preset classification space to obtain a predicted adoption rate; The loss value corresponding to the cross entropy loss of the predicted adoption rate is calculated according to the supervised label, and it is determined whether the loss value reaches a preset threshold. When it reaches the preset threshold, the training is terminated; otherwise, the model is gradient updated according to the loss value, and the next training sample in the data set is called to continue iterative training of the model.

2. The method for selecting products for advertisement placement according to claim 1, characterized in that: Before the step of obtaining the advertising prediction data corresponding to each product in the product database of the independent site, the following steps are included: Determine the click rate of each product in the product database of the independent site using a click rate prediction model that has been pre-trained to a convergent state; Determine the conversion rate of each product in the product database of the independent site using a conversion rate prediction model that has been pre-trained to a convergence state; The adoption rate and effectiveness data of each product are stored in the product database corresponding to each product, and the effectiveness data is the product of the click rate and the conversion rate associated with the same product.

3. The method for selecting products for advertisement placement according to claim 2, characterized in that: In the step of using a click rate prediction model pre-trained to a convergent state to determine the click rate of each product in the product database of the independent site, the training process of the click rate prediction model includes the following iterative execution steps: Obtain training samples from a data set, wherein the training samples include product information of products that have been advertised and clicked by users in a product database of an e-commerce platform, and a supervision label corresponding to whether the advertisement for the product has been clicked by the user; Inputting the training sample into the click rate prediction model to extract corresponding deep semantic features and obtain a feature vector; Using a classifier to classify and map the feature vectors, and predicting the corresponding click rate; A cross entropy loss value corresponding to the click rate predicted by the classifier is calculated according to the supervised label, and a gradient update is performed on the model according to the cross entropy loss value until the model converges.

4. The method for selecting products for advertisement placement according to claim 2, characterized in that: In the step of using a conversion rate prediction model pre-trained to a convergent state to determine the conversion rate of each commodity in the commodity database of the independent site, the training process of the conversion rate prediction model includes the following iterative execution steps: Obtain training samples from a data set, wherein the training samples include product information of products in a product database of an e-commerce platform that have been advertised and have generated relevant user behaviors after being clicked by users, and supervision labels corresponding to whether the advertisements for the products have generated relevant user behaviors after being clicked by users; Inputting the training sample into the conversion rate prediction model to extract corresponding deep semantic features and obtain a feature vector; Using a classifier to classify and map the feature vector to predict the corresponding conversion rate; A cross entropy loss value corresponding to the conversion rate predicted by the classifier is calculated according to the supervised label, and a gradient update is performed on the model according to the cross entropy loss value until the model converges.

5. The method for selecting products for advertisement placement according to claim 1, characterized in that: Selecting products in the product database as advertising selection products according to the recommendation scores to construct an advertising selection recommendation list includes the following steps: Sorting the products in the product database according to their recommendation scores to build a recommended list of products for advertising; In response to the online store's advertising recommendation request, the advertising selection recommendation list is pushed to the management user of the independent site.

6. An advertisement placement and product selection device, characterized in that: include: An adoption rate determination module, for determining the adoption rate of each product in the product database of the independent site using an adoption rate prediction model pre-trained to a convergence state; A data acquisition module, used to acquire advertising prediction data corresponding to each product in the product database of the independent site, wherein the advertising prediction data includes adoption rate and effectiveness data, wherein the adoption rate represents the predicted probability of the product being advertised, and the effectiveness data includes the conversion probability that may be obtained after the product is advertised; A rating calculation module, used to calculate the recommendation rating corresponding to each product based on the adoption rate and effectiveness data of the product; A list building module, used to select commodities in the commodity database as advertisement selections according to the recommendation scores to build a recommendation list of advertisement selections; The adoption rate determination module and the structure corresponding to the training process of the adoption rate prediction model include: A training sample calling module is used to call a training sample from a data set, wherein the training sample includes positive samples and negative samples, and each training sample is provided with a supervision label representing whether a product has been advertised. The positive sample includes the product information and statistical information of advertised products in the product database of the e-commerce platform, and the negative sample includes the product information and statistical information of non-advertised products in the product database of the e-commerce platform. The product information includes the product listing time, product category, and product price, and the statistical information includes click-through rate, purchase rate, and repurchase rate; A feature extraction module, used for inputting the training sample into the adoption rate prediction model to obtain a commodity feature vector and a statistical feature vector corresponding to the commodity information and statistical information in the training sample; A vector concatenation module is used to concatenate the product feature vector and the statistical feature vector to obtain a feature fusion vector, and input the vector into a fully connected layer to map it to a preset classification space to obtain a predicted adoption rate; The iterative training module is used to calculate the loss value corresponding to the cross entropy loss of the predicted adoption rate according to the supervised label, and determine whether the loss value reaches a preset threshold. When it reaches the preset threshold, the training is terminated; otherwise, the model is gradient updated according to the loss value, and the next training sample in the data set is called to continue iterative training of the model.

7. A computer device comprising a central processing unit and a memory, characterized in that: The central processing unit is used to call and run the computer program stored in the memory to execute the steps of the method according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that: It stores a computer program implemented according to the method described in any one of claims 1 to 5 in the form of computer-readable instructions, and when the computer program is called and executed by a computer, the steps included in the corresponding method are executed.

9. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method described in any one of claims 1 to 5 are implemented.