CN114663155A

CN114663155A - Advertisement putting and selecting method and device, equipment, medium and product thereof

Info

Publication number: CN114663155A
Application number: CN202210348720.1A
Authority: CN
Inventors: 郭志伟
Original assignee: Guangzhou Huaduo Network Technology Co Ltd
Current assignee: Guangzhou Huaduo Network Technology Co Ltd
Priority date: 2022-04-01
Filing date: 2022-04-01
Publication date: 2022-06-24
Anticipated expiration: 2042-04-01

Abstract

The application discloses a method for advertisement putting and selecting, a device, equipment, a medium and a product thereof, wherein the method comprises the following steps: acquiring advertisement prediction data corresponding to each commodity in a commodity database of an independent site, wherein the advertisement prediction data comprise an acquisition rate and effect data, the acquisition rate represents the prediction probability of the commodity for advertisement delivery, and the effect data comprise a conversion probability representing possible harvest of the commodity after advertisement delivery; calculating a recommendation score corresponding to each commodity according to the acquisition rate and the effect data of the commodity; and selecting commodities in the commodity database as advertisement putting options according to the recommendation scores to construct an advertisement putting option recommendation list. The method and the device can accurately match the advertisement putting selections which accord with the preference of the user, and can obtain good advertisement putting effect.

Description

Advertisement putting and selecting method and device, equipment, medium and product thereof

Technical Field

The present application relates to the field of e-commerce information technologies, and in particular, to an advertisement delivery selection method, and a corresponding apparatus, computer device, computer-readable storage medium, and computer program product.

Background

In the cross-border e-commerce service platform, basic technical services are provided for massive independent sites in a centralized manner, each independent site usually has an independent domain name, in order to acquire e-commerce transactions of active independent sites, management users of the independent sites often need to recommend commodities to consumer users, and therefore commodities in the platform can be selected according to the needs, and commodities which have sales potential and are suitable for advertisement putting are determined, so that terminal sales or configuration putting on shelves of the commodities is promoted.

With respect to the sales potential of the commodities, management users of the independent sites generally have related market trend considerations, for example, a newly marketed commodity, a commodity of a certain category, a commodity with low price and the like have good market trends and certain market competitiveness, namely sales potential, and the market trend consideration needs to be analyzed by means of rich commodity advertisement putting data, but at the present stage, the commodity advertisement putting data are sparsely distributed, because most independent sites have limited financial resources and limited or even little cost for putting advertisements in commodities, the independent sites with abundant commodity advertisement putting data are only a few, most of the commodity advertisement data of independent sites are scarce, and more importantly, the scarce commodity advertisement data are limited by limited investment cost and underexposure, so that the corresponding return effect is difficult to embody.

The scarcity of data and the low reference value of data lead to the difficulty of determining the goods with sales potential by most management users of independent sites, and no better solution exists for the prior art so far.

In order to solve the defects of the prior art, the applicant makes corresponding researches.

Disclosure of Invention

A primary object of the present application is to solve at least one of the above problems and provide an advertisement delivery selection method and a corresponding apparatus, computer device, computer readable storage medium, and computer program product.

In order to meet various purposes of the application, the following technical scheme is adopted in the application:

the advertisement putting and selecting method adapted to one of the purposes of the application comprises the following steps:

acquiring advertisement prediction data corresponding to each commodity in a commodity database of an independent site, wherein the advertisement prediction data comprise an acquisition rate and effect data, the acquisition rate represents the prediction probability of the commodity for advertisement delivery, and the effect data comprise a conversion probability representing possible harvest of the commodity after advertisement delivery;

calculating a recommendation score corresponding to each commodity according to the acquisition rate and the effect data of the commodity;

and selecting commodities in the commodity database as advertisement putting options according to the recommendation scores to construct an advertisement putting option recommendation list.

In a further embodiment, before the step of obtaining advertisement prediction data corresponding to each commodity in the commodity database of the independent site, the method includes the following steps:

determining the adoption rate of each commodity in a commodity database of the independent station by adopting an adoption rate prediction model trained to a convergence state in advance;

determining the click rate of each commodity in a commodity database of the independent site by adopting a click rate prediction model trained to a convergence state in advance;

determining the conversion rate of each commodity in a commodity database of the independent station by adopting a conversion rate prediction model trained to a convergence state in advance;

storing the adoption rate and the effect data of each commodity in the commodity database corresponding to each commodity, wherein the effect data is the product of the click rate and the conversion rate related to the same commodity.

In a further embodiment, in the step of determining the adoption rate of each commodity in the commodity database of the independent site by using an adoption rate prediction model trained to a convergence state in advance, the training process of the adoption rate prediction model includes the following steps:

calling a training sample from a data set, wherein the training sample comprises a positive sample and a negative sample, each training sample is correspondingly provided with a supervision label for representing whether a commodity is put with an advertisement, the positive sample comprises commodity information and statistical information of the put-with advertisement commodity in a commodity database of an e-commerce platform, the negative sample comprises commodity information and statistical information of the put-with advertisement commodity in the commodity database of the e-commerce platform, the commodity information comprises commodity shelf time, commodity types and commodity prices, and the statistical information comprises click rate, purchase rate and repurchase rate;

inputting the training sample into the adoption rate prediction model to obtain commodity characteristic vectors and statistical characteristic vectors corresponding to the commodity information and the statistical information in the training sample;

splicing the commodity feature vector and the statistical feature vector to obtain a feature fusion vector, and mapping the feature fusion vector to a preset classification space by inputting the feature fusion vector into a full connection layer to obtain a prediction acceptance rate;

calculating a loss value corresponding to the cross entropy loss of the predicted acquisition rate according to the supervision label, judging whether the loss value reaches a preset threshold value, and stopping training when the loss value reaches the preset threshold value; otherwise, performing gradient updating on the model according to the loss value, and calling the next training sample in the data set to continue to perform iterative training on the model.

In a further embodiment, in the step of determining the click rate of each commodity in the commodity database of the independent site by using the click rate prediction model trained to the convergence state in advance, the training process of the click rate prediction model includes the following steps of iterative execution:

acquiring a training sample in a data set, wherein the training sample comprises commodity information of a commodity which is launched with an advertisement and clicked by a user in a commodity database of an e-commerce platform, and a supervision label which represents whether the launched advertisement of the commodity is clicked by the user or not;

inputting the training sample into the click rate prediction model to extract corresponding deep semantic features to obtain a feature vector;

classifying and mapping the characteristic vectors by adopting a classifier, and predicting the corresponding click rate;

and calculating a cross entropy loss value corresponding to the click rate predicted by the classifier according to the supervision label, and performing gradient updating on the model according to the cross entropy loss value until the model converges.

In a further embodiment, in the step of determining the conversion rate of each commodity in the commodity database of the independent site by using a conversion rate prediction model trained to a convergence state in advance, the training process of the conversion rate prediction model includes the following steps executed iteratively:

acquiring a training sample in a data set, wherein the training sample comprises commodity information of a commodity which is provided with an advertisement in a commodity database of an e-commerce platform and generates a relevant user behavior after being clicked by a user, and a supervision label which represents whether the provided advertisement of the commodity generates a corresponding relevant user behavior after being clicked by the user;

inputting the training samples into the conversion rate prediction model to extract corresponding deep semantic features to obtain feature vectors;

classifying and mapping the characteristic vectors by adopting a classifier to predict the corresponding conversion rate;

and calculating a cross entropy loss value corresponding to the conversion rate predicted by the classifier according to the supervision label, and performing gradient updating on the model according to the cross entropy loss value until the model converges.

In an expanded embodiment, selecting the commodities in the commodity database as advertisement putting candidates according to the recommendation scores to construct an advertisement putting candidate recommendation list, which includes the following steps:

sorting according to the recommendation scores corresponding to the commodities in the commodity database, and constructing an advertisement delivery option recommendation list;

and responding to an advertisement recommendation request of the online shop, and pushing the advertisement delivery option recommendation list to the management user of the independent site.

An advertisement placement option device adapted to one of the objects of the present application includes: the system comprises a data acquisition module, a score calculation module and a list construction module, wherein the data acquisition module is used for acquiring advertisement prediction data corresponding to each commodity in a commodity database of an independent site, the advertisement prediction data comprise an acquisition rate and effect data, the acquisition rate represents the prediction probability of the commodity to be advertised, and the effect data comprise the conversion probability representing the possible harvest of the commodity after the advertisement is advertised; the score calculating module is used for calculating the recommendation scores corresponding to the commodities according to the acquisition rate and the effect data of the commodities; and the list construction module is used for selecting the commodities in the commodity database as advertisement putting selections according to the recommendation scores to construct an advertisement putting selection recommendation list.

In a further embodiment, the data obtaining module includes: the acquisition rate determining module is used for determining the acquisition rate of each commodity in the commodity database of the independent station by adopting an acquisition rate prediction model which is trained to a convergence state in advance; the click rate confirmation module is used for determining the click rate of each commodity in the commodity database of the independent site by adopting a click rate prediction model which is trained to be in a convergence state in advance; the conversion rate confirmation module is used for determining the conversion rate of each commodity in the commodity database of the independent station by adopting a conversion rate prediction model which is trained to be in a convergence state in advance; and the data storage submodule is used for storing the adoption rate and the effect data of each commodity in the commodity database corresponding to each commodity, and the effect data is the product of the click rate and the conversion rate related to the same commodity.

In a further embodiment, in the adoption rate determination module, the training process of the adoption rate prediction model includes: the system comprises a training sample calling module, a commodity database and a training sample storage module, wherein the training sample calling module is used for calling a training sample called from a data set, the training sample comprises a positive sample and a negative sample, each training sample is correspondingly provided with a monitoring label for representing whether a commodity is put with an advertisement or not, the positive sample comprises commodity information and statistical information of a commodity with an advertisement put in the commodity database of an e-commerce platform, the negative sample comprises commodity information and statistical information of a commodity without an advertisement put in the commodity database of the e-commerce platform, the commodity information comprises commodity shelf time, commodity types and commodity prices, and the statistical information comprises click rate, purchase rate and repurchase rate; the characteristic extraction module is used for inputting the training sample into the adoption rate prediction model to obtain a commodity characteristic vector and a statistical characteristic vector corresponding to the commodity information and the statistical information in the training sample; the vector splicing module is used for splicing the commodity feature vector and the statistical feature vector to obtain a feature fusion vector, and inputting the feature fusion vector into a full connection layer to be mapped to a preset classification space to obtain a prediction acceptance rate; the iterative training module is used for calculating a loss value corresponding to the cross entropy loss of the predicted acceptance rate according to the supervision label, judging whether the loss value reaches a preset threshold value or not, and stopping training when the loss value reaches the preset threshold value; otherwise, performing gradient updating on the model according to the loss value, and calling the next training sample in the data set to continue to perform iterative training on the model.

In a further embodiment, in the click-through rate confirmation module, the iterative execution of the click-through rate prediction model in the training process includes: the training sample calling module is used for acquiring a training sample in a data set, wherein the training sample comprises commodity information of a commodity which is launched with an advertisement and clicked by a user in a commodity database of an e-commerce platform, and a supervision label which represents whether the launched advertisement of the commodity is clicked by the user or not; the feature extraction module is used for inputting the training samples into the click rate prediction model to extract corresponding deep semantic features so as to obtain feature vectors; the classification mapping module is used for performing classification mapping on the feature vectors by adopting a classifier and predicting the corresponding click rate; and the gradient updating module is used for calculating a cross entropy loss value corresponding to the click rate predicted by the classifier according to the supervision label, and performing gradient updating on the model according to the cross entropy loss value until the model converges.

In a further embodiment, in the conversion rate confirmation module, the iteratively executing a training process on the conversion rate prediction model includes: the training sample calling module is used for acquiring a training sample in a data set, wherein the training sample comprises commodity information of a commodity which is launched with an advertisement in a commodity database of an e-commerce platform and generates a related user behavior after being clicked by a user, and a supervision label which represents whether the launched advertisement of the commodity generates a corresponding related user behavior after being clicked by the user; the feature extraction module is used for inputting the training samples into the conversion rate prediction model to extract corresponding deep semantic features to obtain feature vectors; the classification mapping module is used for performing classification mapping on the feature vectors by adopting a classifier so as to predict the corresponding conversion rate; and the gradient updating module is used for calculating a cross entropy loss value corresponding to the conversion rate predicted by the classifier according to the supervision label, and performing gradient updating on the model according to the cross entropy loss value until the model converges.

In an extended embodiment, the list building module includes: the score sorting module is used for sorting according to the recommendation scores corresponding to the commodities in the commodity database and constructing an advertisement putting selection recommendation list; and the list pushing module is used for responding to an advertisement recommendation request of the on-line shop and pushing the advertisement delivery option recommendation list to the management user of the independent site.

The computer device comprises a central processing unit and a memory, wherein the central processing unit is used for calling and running a computer program stored in the memory to execute the steps of the advertisement putting option method.

A computer-readable storage medium, which stores a computer program implemented according to the method for selecting an advertisement delivery option in the form of computer-readable instructions, and when the computer program is called by a computer, executes the steps included in the method.

A computer program product provided to adapt another object of the present application comprises computer programs/instructions which, when executed by a processor, implement the steps of the method described in any of the embodiments of the present application.

As can be appreciated from the exemplary embodiments and the modified embodiments of the present application, the technical solution of the present application has various advantages, including but not limited to the following aspects:

firstly, the application calculates the corresponding recommendation score by obtaining the adoption rate and the effect data of the predicted commodity, and then construct the recommended list of the advertisement putting options to be pushed to the management user of the independent site according to the recommended scores, and the difference is that only the performance data of the predicted commodity is used as the reference index of the advertisement putting options in the prior art, the recommended scores obtained by fusing the characteristics of two dimensions, namely the corresponding acceptance rate and the performance data of the commodity, are used as the reference index of the advertisement putting options in the technical scheme, so that the probability of selecting and putting the advertisement on the commodity can be accurately predicted, and predicting the conversion probability of possible harvest after the commodity is advertised, and reflecting the degree of the commodity suitable for advertising more comprehensively by using a quantitative numerical value, so that a management user of an independent site can make a quick decision according to the recommendation score, preferably select the commodity suitable for advertising, and obtain good results after advertising.

Secondly, on one hand, most independent sites have limited financial resources and cannot have most abundant commodity advertisement putting data, so that a proper commodity cannot be decided for advertisement putting, and the advertisement putting recommendation list can effectively assist management users of the independent sites to carry out advertisement putting and selecting on commodities of the independent sites, so that the actual pain points of the management users are solved; on the other hand, for novice users who do not have too much experience of advertisement putting choices for management users of independent sites, the advertisement putting recommendation list can effectively assist the novice users in optimizing the appropriate commodities in the independent sites, so that the users can put advertisements, the user experience is improved, and the user stickiness is increased.

In addition, the technical scheme of the application is simple and easy to implement, the complexity of model calculation time is low, and large-scale parallelization can be realized in engineering.

Drawings

The foregoing and/or additional aspects and advantages of the present application will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which:

FIG. 1 is a schematic flow chart diagram of an exemplary embodiment of an advertisement placement selection method of the present application;

FIG. 2 is a schematic flow chart of the acquisition rate and the performance data of the pre-constructed commodity in the embodiment of the present application;

FIG. 3 is a schematic flow chart illustrating a training process of the adoption rate prediction model in the embodiment of the present application;

FIG. 4 is a flowchart illustrating a training process of the click-through rate prediction model according to an embodiment of the present application;

FIG. 5 is a schematic flow chart illustrating a training process of the conversion rate prediction model in the embodiment of the present application;

fig. 6 is a schematic flow chart of constructing an advertisement delivery option recommendation list for pushing in the embodiment of the present application;

FIG. 7 is a schematic block diagram of an advertisement delivery option device of the present application;

fig. 8 is a schematic structural diagram of a computer device used in the present application.

Detailed Description

Reference will now be made in detail to embodiments of the present application, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the accompanying drawings are exemplary only for explaining the present application and are not construed as limiting the present application.

As used herein, the singular forms "a", "an", "the" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and/or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. It will be understood that when an element is referred to as being "connected" or "coupled" to another element, it can be directly connected or coupled to the other element or intervening elements may also be present. Further, "connected" or "coupled" as used herein may include wirelessly connected or wirelessly coupled. As used herein, the term "and/or" includes all or any element and all combinations of one or more of the associated listed items.

It will be understood by those within the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.

As will be appreciated by those skilled in the art, "client," "terminal," and "terminal device" as used herein include both devices that are wireless signal receivers, which are devices having only wireless signal receivers without transmit capability, and devices that are receive and transmit hardware, which have receive and transmit hardware capable of two-way communication over a two-way communication link. Such a device may include: cellular or other communication devices such as personal computers, tablets, etc. having single or multi-line displays or cellular or other communication devices without multi-line displays; PCS (Personal Communications Service), which may combine voice, data processing, facsimile and/or data communication capabilities; a PDA (Personal Digital Assistant) that may include a radio frequency receiver, a pager, internet/intranet access, web browser, notepad, calendar, and/or GPS (Global Positioning System) receiver; a conventional laptop and/or palmtop computer or other device having and/or including a radio frequency receiver. As used herein, a "client," "terminal device" can be portable, transportable, installed in a vehicle (aeronautical, maritime, and/or land-based), or situated and/or configured to operate locally and/or in a distributed fashion at any other location(s) on earth and/or in space. The "client", "terminal Device" used herein may also be a communication terminal, a web terminal, a music/video playing terminal, such as a PDA, an MID (Mobile Internet Device) and/or a Mobile phone with music/video playing function, and may also be a smart tv, a set-top box, and the like.

The hardware referred to by the names "server", "client", "service node", etc. is essentially an electronic device with the performance of a personal computer, and is a hardware device having necessary components disclosed by the von neumann principle such as a central processing unit (including an arithmetic unit and a controller), a memory, an input device, an output device, etc., a computer program is stored in the memory, and the central processing unit calls a program stored in an external memory into the internal memory to run, executes instructions in the program, and interacts with the input and output devices, thereby completing a specific function.

It should be noted that the concept of "server" as referred to in this application can be extended to the case of a server cluster. According to the network deployment principle understood by those skilled in the art, the servers should be logically divided, and in physical space, the servers may be independent from each other but can be called through an interface, or may be integrated into one physical computer or a set of computer clusters. Those skilled in the art should understand this variation and should not be so constrained as to implement the network deployment of the present application.

One or more technical features of the present application, unless expressly specified otherwise, may be deployed to a server for implementation by a client remotely invoking an online service interface provided by a capture server for access, or may be deployed directly and run on the client for access.

Various data referred to in the present application may be stored in a server remotely or in a local terminal device unless specified in the clear text, as long as the data is suitable for being called by the technical solution of the present application.

The person skilled in the art will know this: although the various methods of the present application are described based on the same concept so as to be common to each other, they may be independently performed unless otherwise specified. In the same way, for each embodiment disclosed in the present application, it is proposed based on the same inventive concept, and therefore, concepts of the same expression and concepts of which expressions are different but are appropriately changed only for convenience should be equally understood.

The embodiments to be disclosed herein can be flexibly constructed by cross-linking related technical features of the embodiments unless the mutual exclusion relationship between the related technical features is stated in the clear text, as long as the combination does not depart from the inventive spirit of the present application and can meet the needs of the prior art or solve the deficiencies of the prior art. Those skilled in the art will appreciate variations therefrom.

The advertisement delivery selection method can be programmed into a computer program product, is deployed in a client or a server to run, and is generally deployed in the server to implement, for example, in an e-commerce platform application scenario of the application, so that the method can be executed by accessing an open interface after the computer program product runs and performing human-computer interaction with a process of the computer program product through a graphical user interface.

Referring to fig. 1, in an exemplary embodiment of the advertisement delivery selection method of the present application, the method includes the following steps:

step S1100, obtaining advertisement prediction data corresponding to each commodity in a commodity database of an independent site, wherein the advertisement prediction data comprise an acquisition rate and effect data, the acquisition rate represents the prediction probability of the commodity to be advertised, and the effect data comprises a conversion probability representing possible harvest of the commodity after the advertisement is advertised;

the technical scheme of the application uses the operation environment of the electric business service platform as the application environment, and the electric business service platform can be an electric business service platform for opening independent site service, typically, such as a cross-border electric business service platform. Such a platform allows a television provider service platform to serve a large number of such independent sites by configuring each merchant's store as an individual independent site, due to the need to take into account the network environment between locations around the world and the independence between merchants.

Each independent site is provided with a commodity database corresponding to commodities sold on the website, the commodity database comprises a large amount of commodity information used for describing each commodity sold on the shelf, the commodity information includes but is not limited to various types of data such as commodity pictures, commodity titles, commodity details, commodity prices, commodity sku (class), inventory, commodity shelf life and the like, in addition, in order to make the goods sold on the shelf more smoothly, the management user of the independent station can put the advertisement on the goods sold on the shelf in the electric business service platform, so that the goods sold on the shelf are exposed to the user, before the execution process of the advertisement putting of the sold commodity on the shelf, the advertisement prediction data is generated in advance to assist the management user of the independent site to carry out the optimization and then the advertisement putting on the sold commodity on the shelf, therefore, the advertisement prediction data is stored in the commodity database in association with the corresponding sold commodity. Therefore, it can be understood that when the management user of the independent site triggers the advertisement prediction data acquisition request, the e-commerce service platform is called to provide a corresponding data access interface for the commodity database of each independent site in response to the request, so that the access operation is performed on the commodity database of the independent site, and the advertisement prediction data corresponding to each commodity in the commodity database can be obtained.

The advertisement performance data includes an acceptance rate and performance data, the acceptance rate characterizes the predicted probability of the commodity being advertised, and can be easily understood, the high or low acceptance rate depends on whether the commodity information of the commodity has characteristics suitable for advertisement placement, the adoption rate is high when the characteristics suitable for advertisement putting are provided, otherwise the adoption rate is low when the characteristics are not provided, the characteristics comprise commodity characteristics and statistical characteristics, the commodity characteristics comprise short commodity shelf-loading time, commodity class which is a hot-market commodity class of the current commodity sales market, low commodity price and the like, the statistical characteristics comprise high click rate, high repurchase rate, high sales volume, high purchase rate and the like after the advertisement is put on the commodity, therefore, the characteristics accord with the market trend consideration of the management user of the independent site on the selected commodity when the advertisement is put into the selected product, and the commodity with the characteristics has certain market competitiveness, namely sales potential.

The result data comprise conversion probabilities representing possible harvest of the commodity after the commodity is placed with the advertisement, the conversion probabilities comprise click rates corresponding to the click of the commodity after the advertisement is placed by a user and conversion rates corresponding to other user behaviors generated after the commodity is clicked by the user, the other user behaviors comprise but are not limited to behaviors of adding a shopping cart, purchasing, repeated purchasing and the like, understanding is easy, the result data reflect effects obtained after the commodity is placed with the advertisement, high conversion efficiency corresponding to the result data represents that good results can be obtained after the commodity is placed with the advertisement, otherwise, the obtained results are not good, and the good results can reflect interest of the user in the commodity placed with the advertisement and even purchase of the commodity needing the commodity to a certain extent.

Step S1200, calculating recommendation scores corresponding to the commodities according to the acquisition rate and the effect data of the commodities;

since the adoption rate and the performance data of the commodity are reference indexes capable of assisting management users of independent sites to select the commodity used for advertising, accordingly, the number of the reference indexes is reduced for obtaining more comprehensive and intuitive numerical representation, and the reference indexes combining two dimensions of the current adoption rate and the performance data are reduced into the reference index represented by a single numerical value, in one embodiment, the numerical value is obtained by calculating the multiplication of the purchase rate of the commodity, the click rate and the conversion rate corresponding to the performance data, and the numerical value ranges from 0% to 100%, so that the numerical value is used as the recommendation score, for example, the purchase rate is 80%, the click rate is 85%, and the conversion rate is 95%, and then the recommendation score corresponding to the multiplication of the three is 64.5%. It is understood that higher recommendation scores represent higher sales potential of the product and better results obtained after advertising the product.

And step S1300, selecting the commodities in the commodity database as advertisement putting selections according to the recommendation scores to construct an advertisement putting selection recommendation list.

Further, according to the recommendation scores corresponding to the commodities in the commodity database, selecting a plurality of commodities with the highest recommendation scores as the advertisement putting candidates, the advertisement putting selections are sorted in the order of the recommendation scores corresponding to the advertisement putting selections from high to low, further constructing an advertisement putting choice recommendation list by the ordered advertisement putting choices, wherein the number of the specifically selected commodities can be flexibly changed by the technicians in the field according to the number of the commodities allowed to be displayed in the advertisement putting choice recommendation list under the actual conditions, thereby, the advertisement delivery option recommendation list is pushed to the administrator user of the independent site, so that the administrator user can make a decision quickly according to the advertisement putting choice recommendation list, and accurately and preferably selects commodities with high sales potential and good effect obtained after the commodities are put into the advertisement for advertisement putting.

As can be appreciated from the exemplary embodiments of the present application, the technical solution of the present application has various advantages, including but not limited to the following aspects:

firstly, the application calculates the corresponding recommendation score by obtaining the adoption rate and the effect data of the predicted commodity, and then construct the recommended list of the advertisement putting options to be pushed to the management user of the independent site according to the recommended scores, and the difference is that only the performance data of the predicted commodity is used as the reference index of the advertisement putting options in the prior art, the recommended scores obtained by fusing the characteristics of two dimensions, namely the corresponding acceptance rate and the performance data of the commodity, are used as the reference index of the advertisement putting options in the technical scheme, so that the probability of selecting and putting the advertisement on the commodity can be accurately predicted, and predicting the conversion probability of possible harvest after the commodity is advertised, and reflecting the degree of the commodity suitable for advertising more comprehensively by using a quantitative numerical value, so that a management user of an independent site can make a quick decision according to the recommendation score, preferably selects the commodity suitable for advertising, and can obtain good effect after advertising as an advertising option.

Referring to fig. 2, in a further embodiment, before the step of obtaining advertisement prediction data corresponding to each commodity in the commodity database of the independent site in step S1100, the method includes the following steps:

step S1000, adopting a pre-trained to convergence state acceptance rate prediction model to determine the acceptance rate of each commodity in a commodity database of an independent site;

and calling a data access interface corresponding to the commodity database of the independent site provided by the E-commerce service platform, acquiring commodity information corresponding to each commodity from the commodity database, and inputting the commodity information into an acceptance rate prediction model trained to be in a convergence state in advance, so that the model determines the acceptance rate of each commodity.

The commodity information includes, but is not limited to, commodity pictures, commodity titles, commodity details, commodity prices, commodity sku (category), inventory, and commodity shelf life. The acquisition rate prediction model is trained to a convergence state in advance to have the capability of predicting the acquisition rate corresponding to each commodity according to the commodity information of the commodity, specifically, the model can extract commodity characteristics and statistical characteristics corresponding to the commodity advertisement launched by a management user of an independent site from the commodity information of the commodity to determine the acquisition rate, wherein the commodity characteristics comprise but are not limited to short commodity shelf life, current commodities of hot sales categories of commodity sales markets and low commodity price, and the statistical characteristics comprise but are not limited to high click rate, high repeated purchase rate, high sales volume and high purchase rate after the commodity advertisement is launched.

Calling the pre-trained to convergent adoption rate prediction model to extract the characteristics of each commodity in the commodity database of the independent site, extracting the commodity characteristics and the statistical characteristics to obtain two paths of output characteristic vectors, wherein the two paths of output characteristic vectors are respectively a commodity characteristic vector and a statistical characteristic vector, further fusing the commodity characteristic vector and the characteristic vector to obtain a fused characteristic vector, mapping the fused characteristic vector to a classification space, and then performing probability calculation on the classification space by adopting a classifier constructed by Softmax or Sigmod to obtain the probability that the fused characteristic vector is mapped to the classification space corresponding to the commodity selected and put the advertisement, namely the adoption rate.

The process of training the purchase rate prediction model to convergence is further disclosed in the subsequent embodiments of the present application, and the step is temporarily pressed.

Step S1010, determining the click rate of each commodity in the commodity database of the independent site by adopting a click rate prediction model which is trained to a convergence state in advance;

and calling a data access interface corresponding to the commodity database of the independent site provided by the E-commerce service platform, acquiring commodity information corresponding to each commodity from the commodity database, and inputting the commodity information into a click rate prediction model trained to be in a convergence state in advance, so that the click rate of each commodity is determined by the model.

The commodity information includes, but is not limited to, commodity pictures, commodity titles, commodity details, commodity prices, commodity sku (category), inventory, and commodity shelf life. The click rate prediction model is trained to a convergence state in advance, so that the click rate prediction model has the capability of prompting a user to click on corresponding commodity features after the commodity information of the commodity is provided with the advertisement, and the click rate of the commodity is determined, wherein the commodity features comprise picture features and text features.

Calling the click rate prediction model trained to be convergent in advance to perform feature extraction on commodity information corresponding to each commodity in a commodity database of the independent site, extracting commodity features corresponding to the commodity advertisement launched by the user from the commodity information of the commodity, obtaining a feature vector output by the model, mapping the feature vector to a classification space, then performing probability calculation on the classification space by adopting a classifier constructed by Softmax or Sigmod, and obtaining the probability that the feature vector is mapped to the classification space where the commodity is clicked by the user after the advertisement is launched, namely the click rate.

The process from training of the click-through rate prediction model to convergence is further disclosed in the subsequent embodiments of the present application, and this step is temporarily not performed.

Step S1020, determining the conversion rate of each commodity in the commodity database of the independent site by adopting a conversion rate prediction model trained to a convergence state in advance;

and calling a data access interface corresponding to the commodity database of the independent station provided by the E-commerce service platform, acquiring commodity information corresponding to each commodity from the commodity database, inputting the commodity information to a conversion rate prediction model trained to a convergence state in advance, and determining the conversion rate of each commodity by the model.

The commodity information includes, but is not limited to, commodity pictures, commodity titles, commodity details, commodity prices, commodity sku (category), inventory, and commodity shelf life.

The conversion rate prediction model is trained to a convergence state in advance, so that the conversion rate prediction model has the capability of prompting generation of commodity characteristics corresponding to other user behaviors and determining the conversion rate of the commodity according to the commodity advertisement in the commodity information of the commodity after being clicked by a user, wherein the commodity characteristics comprise picture characteristics and text characteristics.

Calling the conversion rate prediction model trained to be convergent in advance to perform feature extraction on commodity information corresponding to each commodity in a commodity database of the independent site, extracting commodity information of the commodity, clicking a commodity advertisement delivered by a user, prompting generation of commodity features corresponding to other user behaviors, obtaining feature vectors output by the model, mapping the feature vectors to a classification space, then performing probability calculation on the classification space by adopting a classifier constructed by Softmax or Sigmod, and obtaining the conversion rate which is the probability that the feature vectors are mapped to the classification space of other user behaviors after the commodity is clicked by the user.

The process of training the conversion prediction model to converge is further disclosed in the examples of the subsequent part of the application, and the step is temporarily pressed.

Step S1030, storing the adoption rate and the effect data of each commodity in the commodity database corresponding to each commodity, wherein the effect data is the product of the click rate and the conversion rate associated with the same commodity.

The performance data includes click rate and conversion rate corresponding to the commodity.

For each commodity, the acquisition rate and the performance data both have timeliness, so a timing task can be set, and the timing task is set to execute steps S1000-S1020 at a specific time point, for example, the specific time point can be the early morning time of each day, and the timing task is very suitable for being executed because the commodity data in the commodity database at the time is less in variation and the resources which can be allocated by the server of the e-commerce service platform are sufficient, so that the acquisition rate and the performance data are properly updated, the acquisition rate and the performance data are both more accurate, and an accurate basis is laid for the advertisement putting selection of the subsequent steps.

In order to facilitate the subsequent calling of the adoption rate and the performance data of each commodity, the commodity corresponding to the mapping association of the adoption rate and the performance data is stored in the database, and for the performance data, the product of the click rate and the conversion rate associated with the same commodity is used as the numerical representation of the performance data.

In this embodiment, the purchase data and the performance data corresponding to each commodity in the commodity database of the independent site are prepared in advance, so that the instruction or the request for calling the purchase data and the performance data can be quickly responded subsequently, the response efficiency is greatly improved, the user waiting time is reduced, the user experience is improved, and the user stickiness is increased.

Referring to fig. 3, in 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 by using the adoption rate prediction model trained to the convergence state in advance, the training process of the adoption rate prediction model includes the following steps:

step S1001, a training sample is called from a data set, the training sample comprises a positive sample and a negative sample, each training sample is correspondingly provided with a supervision label for representing whether a commodity is put with an advertisement, the positive sample comprises commodity information and statistical information of a commodity with an advertisement put in a commodity database of an e-commerce platform, the negative sample comprises commodity information and statistical information of a commodity without an advertisement put in the commodity database of the e-commerce platform, the commodity information comprises commodity shelf time, commodity types and commodity prices, and the statistical information comprises click rate, purchase rate and repurchase rate;

in one embodiment, in order to facilitate the extraction of relevant features more accurately by the adoption rate prediction model, the training samples in the data set are subjected to text preprocessing operation, where the text preprocessing operation includes removing non-text portions in the training samples, for example, deleting some special non-chinese and english characters and punctuation marks by using a regular expression; the training samples are segmented, for English, the segmentation can be directly called a split () function, for Chinese, a mode of segmentation at a bus is adopted, and for the mode, a direct third-party platform can be called directly to call the method in the pre-packaged method function data packet; the stop word in the training sample is removed, and the removal of the stop word has no influence on the overall semantics, such as a large number of virtual words and pronouns in Chinese or nouns and verbs without specific meanings, and articles and pronouns in English.

The positive and negative sample ratios are flexibly set by a person skilled in the art according to prior knowledge or experimental experience, and are recommended to be 1: 10, the statistical information in the positive sample includes a click rate, a purchase rate, and a repurchase rate corresponding to the advertised product, and the statistical information in the negative sample is null.

Step S1002, inputting the training sample into the adoption rate prediction model to obtain commodity characteristic vectors and statistical characteristic vectors corresponding to the commodity information and the statistical information in the training sample;

inputting the training samples into the adoption rate prediction model for feature extraction, extracting commodity information and statistical information of the positive and negative sample commodities to prompt a management user of an independent site to select commodity features and statistical features corresponding to the commodity advertisement to obtain commodity feature vectors and statistical feature vectors output by the model, the commodity characteristics include but are not limited to short commodity shelf life, commodity of hot market of current commodity sales market, low commodity price, the statistical characteristics include, but are not limited to, high click rate, high repurchase rate, high sales volume and high purchase rate after the advertisement is placed on the commodity, one hot code is adopted for each characteristic in the corresponding commodity characteristic and the statistical characteristic, the code can also adopt a code module in Word2vec and Bert, thereby obtaining two paths of corresponding imbedding codes, namely the commodity feature vector and the statistical feature vector.

Step S1003, splicing the commodity feature vector and the statistical feature vector to obtain a feature fusion vector, and mapping the input full-connection layer to a preset classification space to obtain a prediction acceptance rate;

the stitching is vector stitching, and feature fusion is performed on the commodity feature vector and the statistical feature vector, and in general, in order to fuse the two feature vectors, the feature fusion vector is usually obtained by directly adding corresponding elements in the same dimension, if the two dimensions are different, the feature fusion vector can be converted into the vector with the same dimension through linear transformation, further, the feature fusion vector is input into a full connection layer to be linearly converted and is mapped to a binary space, the binary space comprises a positive space representing that the commodity is selected to put the advertisement and a negative space representing that the commodity is not selected to put the advertisement, and then a classifier is adopted to calculate the probability that the feature fusion vector is mapped to the corresponding binary space, in the process, the probability is obtained as the predicted acceptance rate through a Sigmod or Softmax method function.

Step S1004, calculating a loss value corresponding to the cross entropy loss of the predicted acceptance rate according to the supervision label, judging whether the loss value reaches a preset threshold value, and terminating training when the loss value reaches the preset threshold value; otherwise, performing gradient updating on the model according to the loss value, and calling the next training sample in the data set to continue to perform iterative training on the model.

Calling a preset loss function, wherein the preset loss function can be flexibly set by a person skilled in the art according to prior knowledge or experimental experience, calculating a loss value corresponding to the cross entropy loss of the predicted acquisition rate based on the supervision label, and when the loss value reaches a preset threshold value, indicating that the model is trained to a convergence state, so that model training can be terminated; and when the loss value does not reach the preset threshold value, indicating that the model is not converged, then performing gradient updating on the model according to the loss value, generally performing back propagation to correct the weight parameters of each link of the model so as to further approximate the model to be converged, and then continuing to call next sample data in the data set to perform iterative training on the model until the model is trained to be in a convergence state.

Those skilled in the art will appreciate that the adoption rate prediction model is supervised trained based on the data set of supervision labels and positive and negative examples until the model converges, the adoption rate prediction model has the capability of predicting the corresponding adoption rate of each commodity according to the commodity information of the commodity, specifically, the model can prompt the management user of the independent site to select the commodity characteristics and the statistical characteristics corresponding to the commodity advertisement according to the commodity information of the commodity extracted from the commodity database to obtain the corresponding acceptance rate of the commodity, the commodity characteristics include but are not limited to short shelf life of the commodity, hot commodity in the current commodity sales market, low commodity price, the statistical characteristics comprise high click rate, high repurchase rate, high sales volume, high purchase rate and the like corresponding to the commodities after advertisements are put on the commodities.

In the embodiment, the training process of the adoption rate prediction model based on the deep learning model is disclosed, and it can be seen that under the training of the sample data and the supervision label of the data set, the adoption rate prediction model extracts the commodity features and the statistical features corresponding to the commodity information of the commodity to perform reasoning, so that more accurate prediction capability can be obtained, and the follow-up prediction of the adoption rate corresponding to the commodity can be served.

Referring to fig. 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 by using the click rate prediction model trained to the convergence state in advance, a training process of the click rate prediction model includes the following iteratively executed steps:

step S1011, obtaining a training sample in a data set, wherein the training sample comprises commodity information of a commodity which is launched with an advertisement and clicked by a user in a commodity database of an e-commerce platform, and a supervision label which represents whether the launched advertisement of the commodity is clicked by the user or not;

the commodity information includes, but is not limited to, commodity pictures, commodity titles, commodity details, commodity prices, commodity sku (category), inventory, and commodity shelf life. In one embodiment, in order to facilitate the neural network model to extract relevant features more accurately from the training samples in the data set, a text preprocessing operation is performed on the training samples in the data set, where the text preprocessing operation includes removing non-text portions in the training samples, for example, deleting some special non-chinese and english characters and punctuation marks by using a regular expression; the training samples are segmented, for English, the segmentation can be directly called a split () function, for Chinese, a mode of segmentation at a bus is adopted, and for the mode, a direct third-party platform can be called directly to call the method in the pre-packaged method function data packet; the stop word in the training sample is removed, and the removal of the stop word has no influence on the overall semantics, such as a large number of virtual words and pronouns in Chinese or nouns and verbs without specific meanings, and articles and pronouns in English.

Step S1012, inputting the training samples into the click rate prediction model to extract corresponding deep semantic features, and obtaining feature vectors;

inputting the training sample into the click rate prediction model for feature extraction, extracting the commodity information of the commodity, and prompting a user to click the corresponding commodity feature after the commodity is advertised, in the feature extraction process of the picture feature corresponding to the commodity picture in the commodity information, firstly dividing the commodity picture into a plurality of picture units, wherein the size of each picture unit is equal to the size of the corresponding picture unit, so that after the feature picture feature extraction is carried out by the click rate prediction model, each picture unit can obtain a corresponding single-picture feature vector, then, converting the output of the click rate prediction model into a high-dimensional feature vector, and splicing the high-dimensional feature vector into picture coding vectors according to the corresponding position information of the graphic elements corresponding to the high-dimensional feature vectors in the commodity picture; extracting the characteristics of the preprocessed text information such as the title and the details of the product in the product information, extracting corresponding text characteristics, adopting one hot coding corresponding to each characteristic, wherein the coding can also adopt coding modules in Word2vec and Bert, so as to obtain embedding coding corresponding to model output, namely a product text coding vector, further performing vector splicing on the picture coding vector and the product text coding vector to perform characteristic fusion, generally speaking, in order to fuse the two coding vectors, the two coding vectors are directly added with corresponding elements in the same dimension to obtain a fused coding vector, and if the dimensions of the two coding vectors are different, the fused coding vector can be converted into a vector in the same dimension through linear transformation.

The click rate prediction model structure comprises a picture feature extraction module and a text feature extraction module, wherein the two modules can be taken from a neural network model which is realized based on the CNN and is suitable for carrying out deep semantic feature extraction on pictures, such as a corresponding feature extraction module in the models of Resnet, EffectientNet and the like, and a neural network model which is realized based on the RNN and is suitable for carrying out deep semantic feature extraction on texts, such as a corresponding feature extraction module in the models of Bert, Electra and the like.

Step S1013, a classifier is adopted to carry out classification mapping on the feature vectors, and the corresponding click rate is predicted;

further, the fusion coding vector is input to a full connection layer for linear conversion, and is mapped to a binary space, the binary space comprises a positive class space in which the delivered advertisement representing the commodity is clicked by a user, and a negative class space in which the delivered advertisement representing the commodity is not clicked by the user, a classifier is further adopted to calculate the probability corresponding to the fusion coding vector mapped to the secondary space, and the probability is obtained through a Sigmod or Softmax method function in the process and is used as the predicted click rate.

And S1014, calculating a cross entropy loss value corresponding to the click rate predicted by the classifier according to the supervision label, and performing gradient updating on the model according to the cross entropy loss value until the model converges.

Calling a preset loss function, wherein the preset loss function can be flexibly set by a person skilled in the art according to prior knowledge or experimental experience, calculating a loss value corresponding to the cross entropy loss of the predicted click rate based on the supervision label, and when the loss value reaches a preset threshold value, indicating that the model is trained to a convergence state, so that model training can be terminated; and when the loss value does not reach the preset threshold value, indicating that the model is not converged, then performing gradient updating on the model according to the loss value, generally performing back propagation to correct the weight parameters of each link of the model so as to further approximate the model to be converged, and then continuing to call next sample data in the data set to perform iterative training on the model until the model is trained to be in a convergence state.

It can be understood that the click rate prediction model is supervised and trained based on the supervision label until the model converges, so that the model can extract the commodity feature corresponding to the click of the commodity advertisement by the user from the commodity information of the commodity, and determine the click rate corresponding to the commodity, wherein the commodity feature comprises a text feature and an image feature.

In the embodiment, the training process of the click rate prediction model based on the deep learning model is disclosed, and it can be seen that under the training of the sample data and the supervision label of the data set, the click rate prediction model extracts the picture characteristics and the text characteristics in the commodity information of the commodity to carry out reasoning, so that more accurate prediction capability can be obtained, and the subsequent prediction of the click rate corresponding to the commodity can be served, so that the possibility that the commodity is clicked by a user after the advertisement is put in can be reflected more accurately by the click rate.

Referring to fig. 5, in a further embodiment, in the step S1020 of determining the conversion rate of each commodity in the commodity database of the independent site by using the conversion rate prediction model trained to the convergence state in advance, the training process of the conversion rate prediction model includes the following steps executed iteratively:

step S1021, obtaining a training sample in a data set, wherein the training sample comprises commodity information of a commodity which is launched with an advertisement in a commodity database of an e-commerce platform and generates a related user behavior after being clicked by a user, and a supervision label which represents whether the launched advertisement of the commodity generates a corresponding related user behavior after being clicked by the user;

the commodity information includes, but is not limited to, commodity pictures, commodity titles, commodity details, commodity prices, commodity sku (category), inventory, and commodity shelf life. The related user behaviors include, but are not limited to, joining a shopping cart, purchasing, and repurchasing.

In one embodiment, in order to facilitate the neural network model to extract relevant features more accurately from training samples in a data set, a text preprocessing operation is performed on the training samples in the data set, where the text preprocessing operation includes removing non-text portions in the training samples, for example, deleting some special non-chinese and english characters and punctuations in the training samples by using a regular expression; the training samples are segmented, for English, the segmentation can be directly called a split () function, for Chinese, a mode of segmentation at a bus is adopted, and for the mode, a direct third-party platform can be called directly to call the method in the pre-packaged method function data packet; the stop word in the training sample is removed, and the removal of the stop word has no influence on the overall semantics, such as a large number of virtual words and pronouns in Chinese or nouns and verbs without specific meanings, and articles and pronouns in English.

Step S1022, inputting the training sample into the conversion rate prediction model to extract corresponding deep semantic features, and obtaining a feature vector;

inputting the training sample into the conversion rate prediction model for feature extraction, extracting commodity features corresponding to other user behaviors after a commodity advertisement put in the commodity information of the commodity is clicked by a user, dividing the commodity picture into a plurality of picture units in the feature extraction process of the picture features corresponding to the commodity picture in the commodity information, wherein the size of each picture unit is equal to that of each picture unit, so that each picture unit can obtain a corresponding single-picture feature vector after the feature extraction is carried out by the conversion rate prediction model, then converting the output of the conversion rate prediction model into a high-dimensional feature vector, and splicing the high-dimensional feature vector into picture coding vectors according to corresponding position information of each high-dimensional feature vector corresponding to a picture element in the commodity picture; extracting the characteristics of the preprocessed text information such as the title and the details of the commodity in the commodity information, extracting corresponding text characteristics, adopting onehot coding corresponding to each characteristic, wherein the coding can also adopt coding modules in Word2vec and Bert, so as to obtain embedding coding corresponding to model output, namely a commodity text coding vector, further performing vector splicing on the picture coding vector and the commodity text coding vector to perform characteristic fusion, generally speaking, in order to fuse the two coding vectors, the two coding vectors are directly added with corresponding elements in the same dimension to obtain a fused coding vector, and if the dimensions of the two coding vectors are different, the fused coding vector can be converted into a vector in the same dimension through linear transformation.

The conversion rate prediction model structure comprises an image feature extraction module and a text feature extraction module, wherein the two modules can be taken from a neural network model which is realized based on the CNN and is suitable for carrying out deep semantic feature extraction on an image, such as a corresponding feature extraction module in the models of Resnet, EffectientNet and the like, and a neural network model which is realized based on the RNN and is suitable for carrying out deep semantic feature extraction on a text, such as a corresponding feature extraction module in the models of Bert, Electra and the like.

Step S1023, a classifier is adopted to carry out classification mapping on the feature vectors so as to predict the corresponding conversion rate;

further, the fusion coding vector is input to a full connection layer for linear conversion, and is mapped to a binary space, the binary space comprises a positive class space in which a relevant user behavior is generated after a delivered advertisement representing a commodity is clicked by a user, and a negative class space in which a relevant user behavior is not generated after a delivered advertisement representing a commodity is clicked by a user, a classifier is further adopted to calculate the probability corresponding to the fusion coding vector mapped to the second class space, and the probability is obtained as the prediction conversion rate through a Sigmod or Softmax method function in the process.

And step S1024, calculating a cross entropy loss value corresponding to the conversion rate predicted by the classifier according to the supervision label, and performing gradient updating on the model according to the cross entropy loss value until the model converges.

Calling a preset loss function, wherein the preset loss function can be flexibly set by a person skilled in the art according to prior knowledge or experimental experience, calculating a loss value corresponding to the cross entropy loss of the predicted conversion rate based on the supervision label, and when the loss value reaches a preset threshold value, indicating that the model is trained to a convergence state, so that model training can be stopped; and when the loss value does not reach the preset threshold value, indicating that the model is not converged, then performing gradient updating on the model according to the loss value, generally performing back propagation to correct the weight parameters of each link of the model so as to further approximate the model to be converged, and then continuing to call next sample data in the data set to perform iterative training on the model until the model is trained to be in a convergence state.

It can be understood that the conversion rate prediction model is supervised and trained based on the supervision label until the model converges, so that the model can extract the commodity characteristics corresponding to other user behaviors after the commodity advertisement delivered in the commodity information of the commodity is clicked by the user, and determine the conversion rate corresponding to the commodity, wherein the commodity characteristics comprise text characteristics and picture characteristics.

In the embodiment, a training process of the conversion rate prediction model based on the deep learning model is disclosed, and it can be seen that under training of the sample data and the supervision label of the data set, the conversion rate prediction model extracts the picture features and the text features in the commodity information of the commodity to perform reasoning, so that more accurate prediction capability can be obtained, and therefore, the prediction of the conversion rate corresponding to the commodity can be served subsequently, and the possibility that the delivered advertisement of the commodity is clicked by a user to generate the related user behavior can be reflected more accurately by the conversion rate.

Referring to fig. 6, in an expanded embodiment, in step S1300, selecting a product in the product database as an advertisement placement selection according to the recommendation score to construct an advertisement placement selection recommendation list, the method includes the following steps:

step S1310, sorting according to recommendation scores corresponding to commodities in the commodity database, and constructing an advertisement delivery option recommendation list;

and selecting a plurality of commodities with the highest corresponding recommendation scores as advertisement putting candidates according to the recommendation scores corresponding to the commodities in the commodity database, sequencing the commodities in the sequence of the recommendation scores corresponding to the advertisement putting candidates from high to low, and constructing an advertisement putting candidate recommendation list according to the advertisement putting candidates in the sequence, wherein the number of the selected commodities can be flexibly changed by a person skilled in the art according to the number of commodities displayed by the advertisement putting candidate recommendation list.

In an extended embodiment, three sorts of another three dimensions may be set for the advertisement delivery option recommendation list, and the three sorts may be switched with the recommendation score sorting, and the advertisement delivery option recommendation list may be set to default to the recommendation score sorting for display, and the three sorts of displays may be switched optionally, specifically, the three sorts of the three dimensions are sorted in order from the early to late of the commodity shelf establishment time according to the commodity shelf time, commodity category, and commodity price corresponding to the advertisement delivery option; the first phonetic letters A-Z of the categories of the commodities are ordered in sequence, wherein the categories of the same category are classified together; the commodity prices are ranked in order from low to high, and thus selectable switching display of the advertisement placement option recommendation list is performed in the three ranks. Therefore, the user can conveniently and quickly select the appropriate commodities from the advertisement putting choice recommendation list based on the recommendation score ranking and the characteristics corresponding to the other three rankings.

Step S1320, responding to the advertisement recommendation request of the online shop, and pushing the advertisement delivery option recommendation list to the management user of the independent site.

The independent site operates a store, namely an online store, and a management user of the independent site can trigger an advertisement recommendation button set by an e-commerce service platform for providing advertisement recommendation service on the online store, so that an advertisement recommendation request is correspondingly triggered to a server of the e-commerce service platform, the server receives and responds to the request, and the advertisement delivery option recommendation list is pushed to the management user of the independent site.

In this embodiment, an advertisement delivery option recommendation list is constructed according to the recommendation scores corresponding to the commodities in the commodity database, and is pushed to the management user of the independent site, so that the management user can make a quick decision conveniently, and a commodity which has a certain sales potential and can obtain a better advertisement delivery effect is accurately and preferably selected.

Referring to fig. 7, an advertisement delivery and selection device adapted to one of the purposes of the present application is a functional embodiment of the advertisement delivery and selection method of the present application, and the device includes: the system comprises a data acquisition module 1100, a score calculation module 1200 and a list construction module 1300, wherein the data acquisition module 1100 is used for acquiring advertisement prediction data corresponding to each commodity in a commodity database of an independent site, the advertisement prediction data comprises an acquisition rate and effect data, the acquisition rate represents the prediction probability of the commodity being advertised, and the effect data comprises a conversion probability representing the possible harvest of the commodity after the commodity is advertised; the score calculating module 1200 is configured to calculate, according to the acquisition rate and the effect data of the commodities, a recommendation score corresponding to each commodity; and the list construction module 1300 is configured to select the commodities in the commodity database as advertisement putting candidates according to the recommendation scores to construct an advertisement putting candidate recommendation list.

In a further embodiment, the data obtaining module 1100 includes: the acquisition rate determining module is used for determining the acquisition rate of each commodity in the commodity database of the independent station by adopting an acquisition rate prediction model which is trained to a convergence state in advance; the click rate confirmation module is used for determining the click rate of each commodity in the commodity database of the independent site by adopting a click rate prediction model which is trained to a convergence state in advance; the conversion rate confirmation module is used for determining the conversion rate of each commodity in the commodity database of the independent station by adopting a conversion rate prediction model which is trained to be in a convergence state in advance; and the data storage submodule is used for storing the adoption rate and the effect data of each commodity in the commodity database corresponding to each commodity, and the effect data is the product of the click rate and the conversion rate related to the same commodity.

In a further embodiment, in the click-through rate confirmation module, the iterative training process of the click-through rate prediction model includes: the training sample calling module is used for acquiring a training sample in a data set, wherein the training sample comprises commodity information of a commodity which is launched with an advertisement and clicked by a user in a commodity database of an e-commerce platform, and a supervision label which represents whether the launched advertisement of the commodity is clicked by the user or not; the feature extraction module is used for inputting the training samples into the click rate prediction model to extract corresponding deep semantic features so as to obtain feature vectors; the classification mapping module is used for performing classification mapping on the feature vectors by adopting a classifier and predicting the corresponding click rate; and the gradient updating module is used for calculating a cross entropy loss value corresponding to the click rate predicted by the classifier according to the supervision label, and performing gradient updating on the model according to the cross entropy loss value until the model converges.

In a further embodiment, in the conversion rate confirmation module, the iterative training process of the conversion rate prediction model includes: the training sample calling module is used for acquiring a training sample in a data set, wherein the training sample comprises commodity information of a commodity which is launched with an advertisement in a commodity database of an e-commerce platform and generates a related user behavior after being clicked by a user, and a supervision label which represents whether the launched advertisement of the commodity generates a corresponding related user behavior after being clicked by the user; the feature extraction module is used for inputting the training samples into the conversion rate prediction model to extract corresponding deep semantic features to obtain feature vectors; the classification mapping module is used for performing classification mapping on the feature vectors by adopting a classifier so as to predict the corresponding conversion rate; and the gradient updating module is used for calculating a cross entropy loss value corresponding to the conversion rate predicted by the classifier according to the supervision label, and performing gradient updating on the model according to the cross entropy loss value until the model converges.

In an expanded embodiment, the list building module 1300 includes: the score sorting module is used for sorting according to the recommendation scores corresponding to the commodities in the commodity database and constructing an advertisement putting choice recommendation list; and the list pushing module is used for responding to an advertisement recommendation request of the on-line shop and pushing the advertisement delivery option recommendation list to the management user of the independent site.

In order to solve the technical problem, an embodiment of the present application further provides a computer device. As shown in fig. 8, the internal structure of the computer device is schematically illustrated. The computer device includes a processor, a computer-readable storage medium, a memory, and a network interface connected by a system bus. The computer readable storage medium of the computer device stores an operating system, a database and computer readable instructions, the database can store control information sequences, and the computer readable instructions, when executed by the processor, can cause the processor to implement an advertisement putting selection method. The processor of the computer device is used for providing calculation and control capability and supporting the operation of the whole computer device. The memory of the computer device may have stored therein computer readable instructions that, when executed by the processor, may cause the processor to perform the advertising selection method of the present application. The network interface of the computer device is used for connecting and communicating with the terminal. Those skilled in the art will appreciate that the architecture shown in fig. 8 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.

In this embodiment, the processor is configured to execute specific functions of each module and its sub-module in fig. 7, and the memory stores program codes and various data required for executing the modules or the sub-modules. The network interface is used for data transmission to and from a user terminal or a server. The memory in this embodiment stores program codes and data required for executing all modules/sub-modules in the advertisement delivery and selection device of the present application, and the server can call the program codes and data of the server to execute the functions of all sub-modules.

The present application also provides a storage medium having stored thereon computer-readable instructions which, when executed by one or more processors, cause the one or more processors to perform the steps of the method of advertising selection of any of the embodiments of the present application.

The present application also provides a computer program product comprising computer programs/instructions which, when executed by one or more processors, implement the steps of the method as described in any of the embodiments of the present application.

It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments of the present application can be implemented by a computer program, which can be stored in a computer-readable storage medium, and when the computer program is executed, the processes of the embodiments of the methods can be included. The storage medium may be a computer-readable storage medium such as a magnetic disk, an optical disk, a Read-Only Memory (ROM), or a Random Access Memory (RAM).

Those of skill in the art will appreciate that the various operations, methods, steps in the processes, acts, or solutions discussed in this application can be interchanged, modified, combined, or eliminated. Further, other steps, measures, or schemes in various operations, methods, or flows that have been discussed in this application can be alternated, altered, rearranged, broken down, combined, or deleted. Further, steps, measures, schemes in the prior art having various operations, methods, procedures disclosed in the present application may also be alternated, modified, rearranged, decomposed, combined, or deleted.

To sum up, the method accurately predicts the probability of the commodity being selected to launch the advertisement, the probability of the commodity launching the advertisement being clicked by the user after being exposed, and the probability of the corresponding user behavior generated after the commodity launching the advertisement is clicked by the user through the purchase rate prediction model, the click rate prediction model and the conversion rate prediction model, and the probabilities accurately reflect the market competitiveness of the commodity, namely the sales potential and the effect which can be obtained after the commodity launching the advertisement, so that the advertisement launching option list is constructed accordingly, the method is convenient for assisting the user to make a decision of the advertisement launching option according to the advertisement launching list, and the commodity which has certain sales potential and can obtain better advertisement launching effect is accurately and preferably selected.

The foregoing is only a partial embodiment of the present application, and it should be noted that, for those skilled in the art, several modifications and decorations can be made without departing from the principle of the present application, and these modifications and decorations should also be regarded as the protection scope of the present application.

Claims

1. An advertisement putting and selecting method is characterized by comprising the following steps:

2. The advertisement delivery selection method according to claim 1, wherein the step of obtaining advertisement prediction data corresponding to each commodity in the commodity database of the independent site is preceded by the steps of:

determining the adoption rate of each commodity in a commodity database of the independent station by adopting an adoption rate prediction model which is trained to a convergence state in advance;

determining the conversion rate of each commodity in a commodity database of the independent site by adopting a conversion rate prediction model trained to a convergence state in advance;

3. The method for selecting advertisement placement according to claim 2, wherein in the step of determining the adoption rate of each commodity in the commodity database of the independent site by using the adoption rate prediction model trained to the convergence state in advance, the training process of the adoption rate prediction model comprises the following steps:

splicing the commodity feature vector and the statistical feature vector to obtain a feature fusion vector, and inputting the feature fusion vector into a full connection layer to map the feature fusion vector to a preset classification space to obtain a prediction acceptance rate;

4. The method for selecting advertisement placement according to claim 2, wherein in the step of determining the click-through rate of each commodity in the commodity database of the independent site by using the click-through rate prediction model trained to the convergence state in advance, the training process of the click-through rate prediction model includes the following steps of iterative execution:

5. The method for selecting advertisement placement according to claim 2, wherein in the step of determining the conversion rate of each commodity in the commodity database of the independent site by using the conversion rate prediction model trained to the convergence state in advance, the training process of the conversion rate prediction model includes the following steps executed iteratively:

6. The method of claim 1, wherein selecting the commodities in the commodity database as advertisement putting candidates according to the recommendation scores to construct an advertisement putting selection recommendation list comprises the following steps:

7. A text translation apparatus, comprising:

the system comprises a data acquisition module, a storage module and a display module, wherein the data acquisition module is used for acquiring advertisement prediction data corresponding to each commodity in a commodity database of an independent site, the advertisement prediction data comprise an acquisition rate and effect data, the acquisition rate represents the prediction probability of the commodity to be advertised, and the effect data comprise the conversion probability representing the possible harvest of the commodity after the commodity is advertised;

the score calculating module is used for calculating the recommendation scores corresponding to the commodities according to the acquisition rate and the effect data of the commodities;

and the list construction module is used for selecting the commodities in the commodity database as advertisement putting selections according to the recommendation scores to construct an advertisement putting selection recommendation list.

8. A computer device comprising a central processor and a memory, characterized in that the central processor is adapted to invoke execution of a computer program stored in the memory to perform the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that it stores, in the form of computer-readable instructions, a computer program implemented according to the method of any one of claims 1 to 6, which, when invoked by a computer, performs the steps comprised by the corresponding method.

10. A computer program product comprising computer program/instructions, characterized in that the computer program/instructions, when executed by a processor, implement the steps of the method as claimed in any one of claims 1 to 6.