CN109978594A

CN109978594A - Order processing method, apparatus and medium

Info

Publication number: CN109978594A
Application number: CN201711471313.5A
Authority: CN
Inventors: 钟雨; 崔波
Original assignee: Beijing Jingdong Century Trading Co Ltd; Beijing Jingdong Shangke Information Technology Co Ltd
Current assignee: Beijing Jingdong Century Trading Co Ltd; Beijing Jingdong Shangke Information Technology Co Ltd
Priority date: 2017-12-28
Filing date: 2017-12-28
Publication date: 2019-07-05
Anticipated expiration: 2037-12-28
Also published as: CN109978594B

Abstract

Present disclose provides a kind of order processing method, apparatus and media.The described method includes: obtaining the order data of the article for meeting preset condition in the specific period, each order data corresponds to a kind of primary order of article；User advertising behavioral data is obtained, the user advertising behavioral data includes order user advertisement behavioral data and without order user advertisement behavioral data；Using the user advertising behavioral data as the input of neural network, and the analog value of value is actually generated as default answer with the corresponding order of each data in the user advertising behavioral data, the training neural network is completed with having predicted whether order generation until the consistency of prediction and the default answer of the neural network reaches predetermined standard time training；The neural network that output training is completed.The method also includes each series advertisements of neural network prediction completed using the training to the contribution rate of order and article total sales volume.

Description

Order processing method, apparatus and medium

Technical field

This disclosure relates to Internet technical field, more particularly, to a kind of order processing method, apparatus and medium.

Background technique

With the development of internet, the type (including dispensing channel, exhibition method etc.) of advertisement is more and more diversified.User Final buying behavior may be subjected to a certain or certain several influence in a plurality of types of advertisements, and different type The influence degree of advertisement may also be different.In the prior art in order to analyze various types of advertisements are how to influence user Final buying behavior and influence degree, usually pass through shallow Model (such as Markov model) or rule model (example Such as self-definition model, based on time decaying, Principle of Average Allocation etc.) it is analyzed.

During realizing present inventive concept, at least there are the following problems in the prior art: shallow-layer mould for inventor's discovery Type extracts weaker, expression for the deep layer linked character of the data between various types of advertisements and the final buying behavior of user Ability is limited；Rule model can not excavate in the data between various types of advertisements and the final buying behavior of user spy Sign, so that deeper data association information can not be provided.As it can be seen that in the prior art for various types of advertisements and user Final buying behavior between incidence relation analysis it is not accurate enough deep.

Summary of the invention

In view of this, present disclose provides one kind can in depth predict various types of advertisements and the final purchase of user The order processing method, apparatus and medium of incidence relation between behavior.

An aspect of this disclosure provides a kind of order processing method.The described method includes: obtaining in the specific period Meet the order data of the article of preset condition, each order data corresponds to a kind of primary order of article；Obtain user Advertisement behavioral data, the user advertising behavioral data include order user advertisement behavioral data and without order user advertisement row For data, in which: have described in each and correspond to an order in order user advertisement behavioral data, characterize placing an order for the order User consults the behavioral data of the advertisement of M seed type relevant to the article of the order；Each described no order user advertisement Behavioral data corresponds to one generated within the specific period without any order without order user, characterizes the no order User consults the behavioral data of the advertisement of relevant to the article for meeting the preset condition M seed type, M for more than or equal to 2 positive integer；Using the user advertising behavioral data as the input of neural network, and in the user advertising behavioral data The analog value that the corresponding order of each data actually generates value is default answer, and the training neural network is to have predicted whether Order generates, and completes until the consistency of prediction and the default answer of the neural network reaches predetermined standard time training； The neural network that output training is completed.

In accordance with an embodiment of the present disclosure, the method also includes being directed to each order, the advertisement of the M seed type is determined The contribution rate of the advertisement of middle each type.The advertisement for each type is specifically included, in the advertisement conduct of this type When type advertisements to be analyzed, execute following operation: by the corresponding user advertising behavioral data of the order with the class to be analyzed The relevant data of other kinds of advertisement except type advertisement are set as zero, obtain user's type advertisements behavioral data to be analyzed； User type advertisements behavioral data to be analyzed is input to the neural network that the training is completed；Obtain the mind that training is completed Output through network, wherein the output for the neural network that the training is completed corresponds to the type advertisements to be analyzed to described The contribution rate of order.

In accordance with an embodiment of the present disclosure, each in all orders of the article of preset condition the method also includes that will meet The contribution rate of the advertisement of each type of a order summarizes according to advertisement type correspondence, to obtain the wide of each type Accuse the sale contribution rate to the article for meeting the preset condition.

In accordance with an embodiment of the present disclosure, the lower single user for characterizing the order consults M relevant to the article of the order The behavioral data of the advertisement of seed type, including be not later than within the specific period on the day of the order generates and described specific Before period continuously in N days, exposure time of the lower single user in the different types of advertisement of M kind relevant to the article of the order Several and/or number of clicks, wherein N is the integer more than or equal to 0.

In accordance with an embodiment of the present disclosure, the article of the preset condition is consulted and met to the characterization no order user The behavioral data of the advertisement of relevant M seed type, it is N days continuous before being included in the specific period and the specific period It is interior, exposure frequency of the no order user in the different types of advertisement of M kind relevant to the article for meeting the preset condition And/or number of clicks, wherein N is the integer more than or equal to 0.

In accordance with an embodiment of the present disclosure, it is less than the no order in the data volume for having order user advertisement behavioral data When the data volume of user advertising behavioral data, reduce the data of the no order user advertisement behavioral data according to preset rules Amount, so that in the data of the data volume for having order user advertisement behavioral data and the no order user advertisement behavioral data Amount is suitable.

In accordance with an embodiment of the present disclosure, the neural network is convolutional neural networks.

In accordance with an embodiment of the present disclosure, the convolutional neural networks include multiple convolutional layers, the group for rectifying layer and pond layer The combination of conjunction and/or multiple full articulamentums and rectification layer.

Another aspect of the present disclosure additionally provides a kind of order processing device, including order data obtains module, Yong Huguang It accuses behavioral data and obtains module, neural metwork training module and output module.Order data obtains module for when obtaining specific The order data of the article for meeting preset condition in phase, each order data correspond to a kind of primary order of article.With Advertisement behavioral data in family obtains module for obtaining user advertising behavioral data, and the user advertising behavioral data includes order User advertising behavioral data and without order user advertisement behavioral data, in which: have order user advertisement behavior number described in each According to an order is corresponded to, the advertisement of the lower single user access M seed type relevant to the article of the order of the order is characterized Behavioral data；Each described no order user advertisement behavioral data, which corresponds to, to be produced within the specific period without any order Raw one characterizes the no order user and consults the M relevant to the article for meeting the preset condition without order user The behavioral data of the advertisement of seed type, M are the positive integer more than or equal to 2.Neural metwork training module is used for wide with the user Input of the behavioral data as neural network is accused, and real with the corresponding order of each data in the user advertising behavioral data The analog value of border generation value is default answer, and the training neural network is to have predicted whether order generation, until the nerve The consistency of prediction and the default answer of network reaches predetermined standard time training and completes.Output module has been trained for exporting At neural network.

In accordance with an embodiment of the present disclosure, described device further includes the advertisement contribution rate determining module of order.The advertisement of order Contribution rate determining module is used to be directed to each order, determines the tribute of the advertisement of each type in the advertisement of the M seed type Offer rate.Specifically include the advertisement for each type, when the advertisement of this type is as type advertisements to be analyzed, execute with Lower operation: by the corresponding user advertising behavioral data of the order with it is other kinds of wide except the type advertisements to be analyzed It accuses relevant data and is set as zero, obtain user's type advertisements behavioral data to be analyzed；By user type advertisements to be analyzed Behavioral data is input to the neural network that the training is completed；Obtain the output for the neural network that training is completed, wherein the instruction The output for practicing the neural network completed corresponds to the type advertisements to be analyzed to the contribution rate of the order.

In accordance with an embodiment of the present disclosure, described device further includes the advertisement contribution rate computing module always sold.Always sell The each type of each order in all orders for the article that advertisement contribution rate computing module is used to that preset condition will to be met The contribution rate of advertisement summarize according to advertisement type correspondence, to obtain the advertisement of each type to meeting the default item The sale contribution rate of the article of part.

Another aspect of the present disclosure provides a kind of order processing device, including one or more processors, and storage Device.The memory is for storing one or more programs.Wherein, when one or more of programs are one or more of When processor executes, so that one or more of processors realize order processing method as described above.

Another aspect of the present disclosure provides a kind of computer-readable medium, is stored thereon with executable instruction, the instruction Processor is set to realize order processing method as described above when being executed by processor.

Another aspect of the present disclosure provides a kind of computer program, and the computer program, which includes that computer is executable, to be referred to It enables, described instruction is when executed for realizing order processing method as described above.

In accordance with an embodiment of the present disclosure, can at least be partially solved can not accurately analyse in depth user in the prior art The problem of relevance between advertisement behavioral data and order data, and pass through a large amount of user advertising of neural network deep learning Incidence relation between behavioral data and order data may be implemented more precisely to predict to use according to user advertising behavioral data The technical effect whether family can finally be bought.Also, pass through the deep learning of neural network and the conclusion based on mass data Analysis, so that more accurate prediction can be provided to the analysis more elasticity, flexibility of user's buying behavior.

According to some embodiments of the present disclosure, each can be analyzed based in part on user advertising behavioral data The advertisement of type is how to influence the generation and corresponding influence degree of each order.

According to some embodiments of the present disclosure, various types of advertisements can be at least partly analyzed in the regular period Meet the contribution situation of the sales volume of the article of preset condition.To, for operator marketing methods assessment with budget allocation, And the prediction of marketing all has very great help, and greatly improves the accuracy and completeness of advertising results assessment.

Detailed description of the invention

By referring to the drawings to the description of the embodiment of the present disclosure, the above-mentioned and other purposes of the disclosure, feature and Advantage will be apparent from, in the accompanying drawings:

Fig. 1 diagrammatically illustrates the systemic framework of order processing method, apparatus and medium according to the embodiment of the present disclosure；

Fig. 2 diagrammatically illustrates the flow chart of the order processing method according to the embodiment of the present disclosure；

Fig. 3 diagrammatically illustrates the design schematic diagram of the order processing method according to the embodiment of the present disclosure；

Fig. 4 diagrammatically illustrates the flow chart of the order processing method according to another embodiment of the disclosure；

Fig. 5 diagrammatically illustrates the advertisement contribution rate that order is determined in the order processing method according to the embodiment of the present disclosure Method flow diagram；

Fig. 6 diagrammatically illustrates the flow chart of the order processing method according to disclosure another embodiment；

Fig. 7 diagrammatically illustrates the block diagram of the order processing device according to the embodiment of the present disclosure；And

Fig. 8 diagrammatically illustrates the block diagram of the order processing computer system according to the embodiment of the present disclosure.

Specific embodiment

Hereinafter, will be described with reference to the accompanying drawings embodiment of the disclosure.However, it should be understood that these descriptions are only exemplary , and it is not intended to limit the scope of the present disclosure.In addition, in the following description, descriptions of well-known structures and technologies are omitted, with Avoid unnecessarily obscuring the concept of the disclosure.

Term as used herein is not intended to limit the disclosure just for the sake of description specific embodiment.It uses herein The terms "include", "comprise" etc. show the presence of the feature, step, operation and/or component, but it is not excluded that in the presence of Or add other one or more features, step, operation or component.

There are all terms (including technical and scientific term) as used herein those skilled in the art to be generally understood Meaning, unless otherwise defined.It should be noted that term used herein should be interpreted that with consistent with the context of this specification Meaning, without that should be explained with idealization or excessively mechanical mode.

It, in general should be according to this using statement as " at least one in A, B and C etc. " is similar to Field technical staff is generally understood the meaning of the statement to make an explanation (for example, " system at least one in A, B and C " Should include but is not limited to individually with A, individually with B, individually with C, with A and B, with A and C, have B and C, and/or System etc. with A, B, C).Using statement as " at least one in A, B or C etc. " is similar to, generally come Saying be generally understood the meaning of the statement according to those skilled in the art to make an explanation (for example, " having in A, B or C at least One system " should include but is not limited to individually with A, individually with B, individually with C, with A and B, have A and C, have B and C, and/or the system with A, B, C etc.).It should also be understood by those skilled in the art that substantially arbitrarily indicating two or more The adversative conjunction and/or phrase of optional project shall be construed as either in specification, claims or attached drawing A possibility that giving including one of these projects, either one or two projects of these projects.For example, phrase " A or B " should A possibility that being understood to include " A " or " B " or " A and B ".

Embodiment of the disclosure provides a kind of order processing method, apparatus and medium.The described method includes: obtaining specific The order data of the article for meeting preset condition in period, each order data correspond to a kind of primary order of article； User advertising behavioral data is obtained, which includes order user advertisement behavioral data and without order user Advertisement behavioral data；Using the user advertising behavioral data as the input of neural network, and in the user advertising behavioral data The analog value that the corresponding order of each data actually generates value is default answer, and the training neural network is to have predicted whether to order It is single to generate, it is completed until the consistency of the default answer of prediction and this of the neural network reaches predetermined standard time training；Output instruction Practice the neural network completed.Wherein, this has each of order user advertisement behavioral data data to correspond to an order, table Levy the behavioral data of the advertisement of the lower single user access M seed type relevant to the article of the order of the order；This is used without order Each data of family advertisement behavioral data correspond to be used without one that any order generates without order within the specific period Family characterizes the behavioral data of the advertisement that the M seed type relevant to the article for meeting the preset condition is consulted without order user, M is the positive integer more than or equal to 2.

In accordance with an embodiment of the present disclosure, it can at least be partially solved that can not accurately to analyse in depth user in the prior art wide The problem of accusing the relevance between behavioral data and order data, and pass through a large amount of user advertising row of neural network deep learning Incidence relation between data and order data may be implemented more precisely to predict user according to user advertising behavioral data The technical effect that whether can finally buy.In accordance with an embodiment of the present disclosure, summarizing and analyzing based on mass data and deep learning, More elasticity, flexibility and more accurate prediction result are provided to analyze the buying behavior of user.

Fig. 1 diagrammatically illustrates the systemic framework of order processing method, apparatus and medium according to the embodiment of the present disclosure. It should be noted that being only the example that can apply the system architecture of the embodiment of the present disclosure shown in Fig. 1, to help art technology Personnel understand the technology contents of the disclosure, but are not meant to that the embodiment of the present disclosure may not be usable for other equipment, system, environment Or scene.

As shown in Figure 1, system architecture 100 may include terminal device 101,102,103, network according to this embodiment 104 and server 105.Network 104 between terminal device 101,102,103 and server 105 to provide communication link Medium.Network 104 may include various connection types, such as wired, wireless communication link or fiber optic cables etc..

User can be used terminal device 101,102,103 and be interacted by network 104 with server 105, to receive or send out Send message etc..Various telecommunication customer end applications, such as the application of shopping class, net can be installed on terminal device 101,102,103 (merely illustrative) such as the application of page browsing device, searching class application, instant messaging tools, mailbox client, social platform softwares.

Terminal device 101,102,103 can be the various electronic equipments with display screen and supported web page browsing, packet Include but be not limited to smart phone, tablet computer, pocket computer on knee and desktop computer etc..

Server 105 can be to provide the server of various services, such as utilize terminal device 101,102,103 to user The website browsed provides the back-stage management server (merely illustrative) supported.Back-stage management server can be to the use received The data such as family request analyze etc. processing, and by processing result (such as according to user's request or the webpage of generation, believe Breath or data etc.) feed back to terminal device.

In some embodiments, user can be used terminal device 101,102,103 and consult (such as browsing or click) packet Include the webpage, mail or search engine etc. of various types of advertisements.In further embodiments, user can pass through terminal Equipment 101,102,103 places an order.

In some embodiments, server 105 can analyze, handle and records the transmission of terminal device 101,102,103 The data of included advertisement type in user's request data, and can correspond to and generate user advertising behavioral data.In some implementations In example, server 105 can analyze, handle and record and wrapped in user's request data of the transmission of terminal device 101,102,103 The various order datas included.

It should be noted that order processing method provided by the embodiment of the present disclosure can generally be executed by server 105. Correspondingly, order processing device provided by the embodiment of the present disclosure generally can be set in server 105.The embodiment of the present disclosure Provided order processing method can also by be different from server 105 and can with terminal device 101,102,103 and/or clothes The server or server cluster that business device 105 communicates execute.Correspondingly, order placing equipment provided by the embodiment of the present disclosure can also be with Be set to different from server 105 and the server that can be communicated with terminal device 101,102,103 and/or server 105 or In server cluster.

It should be understood that the number of terminal device, network and server in Fig. 1 is only schematical.According to realization need It wants, can have any number of terminal device, network and server.

Fig. 2 diagrammatically illustrates the flow chart of the order processing method according to the embodiment of the present disclosure.

As shown in Fig. 2, including operation S210~operation S240 according to the order processing method of the embodiment of the present disclosure.

In operation S210, the order data of the article for meeting preset condition in the specific period, each order numbers are obtained According to the primary order for corresponding to a kind of article.

Meet preset condition article can be the article of for example a certain particular brand or the article of certain industry or The article etc. of a certain electric business platform of person.Specifically can according to actual analysis it needs to be determined that.

The specific period, which can be, needs previously selected one period according to analysis.

Each order data corresponds to a kind of primary order of article.Specifically, such as by the different moneys of same brand Article is considered as article not of the same race.In this case, for example, user has purchased Huawei's hand of multiple same moneys in once placing an order Machine only generates an order data, i.e. this Huawei's mobile phone a order in this case.In another example user is primary In placing an order while Huawei's mobile phone and Haier's refrigerator of two different moneys are had purchased, three orders will be generated in this case Data, i.e., Huawei's mobile phone one order data of each correspondence of two different moneys, Haier's refrigerator then correspond to another order data.

In operation S220, user advertising behavioral data is obtained, which includes order user advertisement Behavioral data and without order user advertisement behavioral data.

Wherein, each has order user advertisement behavioral data to correspond to an order, characterizes and is applied alone under the order Consult the behavioral data of the advertisement of M seed type relevant to the article of the order in family.Each this without order user advertisement behavior Data correspond within the specific period without any order generate one without order user, characterize this without order user consult The behavioral data of the advertisement of the M seed type relevant to the article for meeting the preset condition, M are the positive integer more than or equal to 2.

Specifically, the advertisement of M seed type is classified according to different modes or different analyses.

In some embodiments, when determining advertisement type, such as can be first by all advertisements according to the presentation to user Mode, which is divided into, shows advertisement (for example, user browses the advertisement shown on webpage when webpage) and search advertisements (for example, user exists The advertisement found after keyword search is carried out in search box) two major classes.Then on this basis further according to the dispensing canal of advertisement It further segments in road, wherein launching channel for example can be portal website, social networks, mail, navigation page etc..

In accordance with an embodiment of the present disclosure, each has data in order user advertisement behavioral data to order corresponding to one It is single, characterize the behavioral data of the advertisement of the lower single user access M seed type relevant to the article of the order of the order.Specifically Ground may include being not later than on the day of the order generates within the specific period and before the specific period in continuous N days, this Exposure frequency and/or number of clicks of the lower single user in the different types of advertisement of M kind relevant to the article of the order, wherein N For the integer more than or equal to 0.For example, as shown in table 1.

Table 1, which illustrates one, order user advertisement behavioral data.

As shown in table 1, which is selected as certain on November 1 ,~November 11, and N value is 30.This has order The targeted order of user advertising behavioral data (such as order a) was generated November 9.To which this has order user advertisement row It include the lower single user of the same day on November 1 to November 9 and October 2 to the order a between October 31 for data In the exposure frequency and/or number of clicks of the different types of advertisement of M kind relevant to the article of order a.Specifically, 11 in table 1 The data of the same day on the moon 1 to November 9 and October 2 into the table between October 31 for example can be and be applied alone under this Exposure frequency (as shown in table 1) of the family in the advertisement of corresponding types.Certainly, the exposure frequency that table 1 is lifted is only a kind of example, should There is order user advertisement behavioral data to be also possible to the number that the lower single user clicks the advertisement of corresponding types, or is also possible to The lower single user is in the sum of the exposure frequency of the advertisement of corresponding types and number of clicks.

Lower single date of order a is November 9 in table 1, therefore any advertisement behavior of the lower single user after November 9 Data do not have any contribution to the order.Therefore, corresponding this of order a has November 10 in order user advertisement behavioral data With November 11 without data (indicating the data in the table for sky with long short-term in table).

Table 1 has order user advertisement behavioral data

It will be appreciated, of course, that when the same user has the generation of multiple orders within the specific period, for multiple Dan Huixiang There should be order user advertisement behavioral data there are multiple.

In accordance with an embodiment of the present disclosure, each corresponds within the specific period without order user advertisement behavioral data There is no one of any order generation without order user, characterizes the article phase consulted without order user and meet the preset condition The behavioral data of the advertisement for the M seed type closed, including before the specific period and the specific period in continuous N days, this Exposure frequency and/or click of the no order user in the different types of advertisement of M kind relevant to the article for meeting the preset condition Number, wherein N is the integer more than or equal to 0.Such as shown in table 2.

Table 2 illustrates one without order user advertisement behavioral data.

It is specific as shown in table 2, for ease of description, the specific period is still selected as certain on November 1 ,~November 11, N Value is 30.It is not have any order to produce between 1~November 11 November that this is targeted without order user advertisement behavioral data Raw one is without order user (such as user W).This includes November 1 to November 11 without order user advertisement behavioral data User W between (i.e. the specific period) and October 2 to October 31 (before the specific period 30 days) with satisfaction The exposure frequency of the relevant different types of advertisement of M kind of article (for example, article of a certain particular brand) of the preset condition and/ Or number of clicks.

Specifically, the advertisement behavioral data of user W include user W within the specific period on November 1 to November 11, And the advertisement behavioral data (being shown in table with ellipsis) of every day in October 2 to October 31 all number of days.

Similar, the advertisement behavioral data of user W can be exposure frequency (such as table 2 of advertisement of the user W in corresponding types It is shown).Certainly, the exposure frequency that table 2 is lifted is only a kind of example, this is also possible to user W without single user advertisement behavioral data The number of the advertisement of corresponding types is clicked, or is also possible to exposure frequency and click of the user W in the advertisement of corresponding types The sum of number.

The corresponding no order user advertisement behavioral data of 2 user W of table

In operation S230, using the user advertising behavioral data as the input of neural network, and with the user advertising behavior It is default answer that the corresponding order of each data, which actually generates the analog value of value, in data, and the training neural network is to predict to be It is no to have order generation, it is completed until the consistency of the default answer of prediction and this of the neural network reaches predetermined standard time training.

The neural network can be any one neural network, such as Recognition with Recurrent Neural Network, convolutional neural networks or BP Neural network etc..

For example, can set each has order user advertisement behavioral data corresponding default in training neural network Answer is 1, each is 0 without the corresponding default answer of order user advertisement behavioral data.By operating, the institute that S220 is obtained is useful Family advertisement behavioral data (include order user advertisement behavioral data and without order user advertisement behavior) is input to neural network Repetition training is carried out to the neural network, until the prediction output of the neural network and the consistency of preset answer reach predetermined Standard (such as the neural network prediction output reached a certain level with the consistent probability of preset answer, such as 90%) when, The neural metwork training is completed.By this method, allow neural network depth learn user advertising behavioral data with most throughout one's life At order data between relationship.

In accordance with an embodiment of the present disclosure, this have order user advertisement behavioral data data volume be less than this without order user When the data volume of advertisement behavioral data, reduce the data volume of the no order user advertisement behavioral data according to preset rules, with Make have the data volume of order user advertisement behavioral data suitable with the data volume without order user advertisement behavioral data at this.

For example, obtaining user advertising behavioral data under some scenes, such as from a store.At this time, it is possible in spy The user that order is generated in the timing phase is only limitted to a few users, most likely results in order user advertisement behavioral data in this way Data volume of the data volume much smaller than no order user advertisement behavioral data.However, there is the data of order user advertisement behavioral data It measures again most important for the relationship between analysis user advertising behavioral data and the order data ultimately generated.At this point, in order to The accuracy rate for improving neural network learning prediction avoids largely excessively diluting user without order user advertisement behavioral data wide The relationship between behavioral data and the order data ultimately generated is accused, it is wide the no order user can be reduced according to preset rules Accuse the data volume of behavioral data.For example, being taken out from all no order user advertisement behavioral datas by the method for random sampling Sample goes out that quite (such as the equal or order of magnitude is identical or reaches scheduled ratio with the data volume that has order user advertisement behavioral data Example value) without order user advertisement behavioral data.

Again alternatively, in another example, under other scenes, there is the data volume of order user advertisement behavioral data to be much smaller than without ordering The data volume of single user advertisement behavioral data is extreme to be lacked, on the one hand less than the data volume of no order user advertisement behavioral data, On the other hand it can not meet the needs of data analysis (for example, needing what is obtained to have order user advertisement if analysis is estimated The data volume of behavioral data, which reaches 100,000, can meet analysis needs, and may there was only 80,000 data in practice).This When, it can also be by expanding the data for having order user advertisement behavioral data according to corresponding regular (such as random reproduction etc.) Amount, so that this has the data volume of order user advertisement behavioral data to meet analysis demand, and makes this have order user advertisement The data volume of behavioral data is suitable with the data volume without order user advertisement behavioral data.

Then, in operation S240, the neural network that training is completed is exported.The neural network that the training is completed can be used for root Whether can finally be bought according to the advertisement behavioral data prediction user of user.

In accordance with an embodiment of the present disclosure, it can at least be partially solved that can not accurately to analyse in depth user in the prior art wide The problem of accusing the relevance between behavioral data and order data, and pass through a large amount of user advertising row of neural network deep learning Relationship between data and order data, so as to more precisely whether predict user according to user advertising behavioral data It can finally buy, provide more elasticity, flexibility and more accurate conclusion and prediction to analyze the buying behavior of user As a result.

In accordance with an embodiment of the present disclosure, which is convolutional neural networks.

In accordance with an embodiment of the present disclosure, which includes multiple convolutional layers, the group for rectifying layer and pond layer The combination of conjunction and/or multiple full articulamentums and rectification layer.

Fig. 3 diagrammatically illustrates the design schematic diagram of the order processing method according to the embodiment of the present disclosure.

As shown in figure 3, according in the order processing method of the embodiment of the present disclosure, by the advertisement behavioral data of user (including There are order user advertisement behavioral data, such as data shown in table 1, and without order user advertisement behavioral data, such as shown in table 2 Data) be input to the input layer of convolutional neural networks.Then by transmitting layer by layer inside the convolutional neural networks, the volume is obtained The prediction output of the output layer of product neural network.When the prediction of the convolutional neural networks is exported with the consistency of preset answer not Meet predetermined standard time, the convolutional neural networks be trained repeatedly, until the convolutional neural networks prediction output with it is pre- If the consistency of answer reach preassigned, convolutional neural networks training is completed.Hereafter, so that it may utilize the convolutional Neural Network predicts whether user can finally buy according to the advertisement behavioral data of user.

As shown in figure 3, the convolutional neural networks by input layer A, convolutional layer B, rectification layer C, pond layer D, full articulamentum E, Rectify layer F, depletion layer G, output layer H composition.

Input layer A main purpose is to read matrix character, is directly output to convolutional layer B.

Convolutional layer B can be used for being arranged Convolution Filter, and Convolution Filter parameter provides after being optimized by back-propagation algorithm.

Rectification layer C can be rectified using non-linear Sigmoid function, can effectively extract user advertising behavioral data With a variety of nonlinear relationships between order data.

Pond layer D can carry out data cleansing, avoid over-fitting, control parameter quantity etc..

In some embodiments, the combination of convolutional layer B, rectification layer C, pond layer D can repeat repeatedly.In some realities It applies in example, 3~4 groups of repetitions can be used, wherein first group of combination is generally used for extracting rudimentary local feature, in subsequent combination Convolutional layer B can be used for extracting global feature.

Full articulamentum E can be used for doing linear transformation to the input of pond layer D, for example, each neuron of full articulamentum E It may be coupled to all matrix dots of pond layer D.

It is different from the rectification function of layer C to rectify layer F.The purpose for rectifying layer F is mainly non-to the output progress of full articulamentum E Linear change (such as can equally use Sigmoid function).

In some embodiments, it can repeat that multiple full articulamentum E are arranged and rectify the combination of layer F.In some embodiments In, the combination of full articulamentum E and rectification layer F can be two.

Depletion layer G is the last layer of convolutional neural networks shown in Fig. 3, for punishing the predicted value of convolutional neural networks With the difference of default answer.In some embodiments, the prediction result of convolutional neural networks is indicated with Probability Forms.In this feelings Under condition, depletion layer G uses Sigmoid cross entropy loss function.

Output layer H is the output valve of G depletion layer, is the prediction result of the convolutional neural networks.

Fig. 4 diagrammatically illustrates the flow chart of the order processing method according to another embodiment of the disclosure.

As shown in figure 4, the order processing method in addition to operating S210~operation S240, further includes operation S250.

The contribution of the advertisement of each type in the advertisement of the M seed type is determined for each order in operation S250 Rate.

In accordance with an embodiment of the present disclosure, operation S250 includes the advertisement for each type, in the advertisement of this type When as type advertisements to be analyzed, executes and refer to operating process shown in fig. 5.

Fig. 5 diagrammatically illustrates the advertisement contribution rate that order is determined in the order processing method according to the embodiment of the present disclosure Method flow diagram.

As shown in figure 5, being directed to each order in operation S250, each type in the advertisement of the M seed type is determined The contribution rate of advertisement include advertisement for each type, held when the advertisement of this type is as type advertisements to be analyzed Row operation S251~operation S253.

In operation S251, by the corresponding user advertising behavioral data of the order with its except the type advertisements to be analyzed The relevant data of the advertisement of his type are set as zero, obtain user's type advertisements behavioral data to be analyzed.

For example, when advertisement type to be analyzed be table 1 shown in " advertisement 1: email advertisement " when, can be by removing in table 1 Other except " advertisement 1: email advertisement " this column have the table of data to be disposed as zero, to obtain that " advertisement 1: mail is wide The corresponding user's type advertisements behavioral data to be analyzed of announcement ".

In operation S252, user type advertisements behavioral data to be analyzed is input to the neural network of training completion.

Operation S253, obtain training complete neural network output, wherein the training complete neural network it is defeated Correspond to the type advertisements to be analyzed out to the contribution rate of the order.

For example, " advertisement 1: email advertisement " corresponding user type advertisements behavior number to be analyzed that the example above is obtained According to, be input to the training completion neural network, and get the training completion neural network output layer H output.? In some embodiments, the output of neural network which completes can be showed with Probability Forms, so as to by the probability value As " advertisement 1: email advertisement " to the contribution rate of order a.In further embodiments, the neural network of training completion is defeated It can express otherwise out, then can carry out corresponding data processing method as the case may be can be obtained " advertisement 1: postal Contribution rate of the part advertisement " to order a.

Fig. 6 diagrammatically illustrates the flow chart of the order processing method according to disclosure another embodiment.

As shown in fig. 6, the order processing method in addition to operating S210~operation S250, further includes operation S260.

In operation S260, the wide of each type of each order in all orders of the article of preset condition will be met The contribution rate of announcement summarizes according to advertisement type correspondence, to obtain the advertisement of each type to the article for meeting the preset condition Sale contribution rate.

For example, the article of Haier's brand produces 100,000 orders within the specific period.It is right in operation S250 Advertisement type " advertisement 1: email advertisement " should have been analyzed to the contribution rate of each order, then by " advertisement in operation S260 1: email advertisement " it adds up one by one to the contribution rate of 100,000 orders, available " advertisement 1: email advertisement " and Haier's brand Article between associated data.It in some embodiments, can be by the associated data of the corresponding M seed type of 100,000 orders After probability is normalized, sale contribution rate of the advertisement to the article for meeting the preset condition of each type is obtained.

Fig. 7 diagrammatically illustrates the block diagram of the order processing device according to the embodiment of the present disclosure.

As shown in fig. 7, including that order data obtains module 710, use according to the order processing device 700 of the embodiment of the present disclosure Advertisement behavioral data in family obtains module 720, neural metwork training module 730 and output module 740.

It can be used to implement the order processing side with reference to Fig. 2~Fig. 6 according to the order processing device 700 of the embodiment of the present disclosure Method.

Order data obtains the order data that module 710 is used to obtain the article for meeting preset condition in the specific period, Each order data corresponds to a kind of primary order of article.

User advertising behavioral data obtains module 720 and is used to obtain user advertising behavioral data, the user advertising behavior number According to including order user advertisement behavioral data and without order user advertisement behavioral data, in which: each has order user Advertisement behavioral data corresponds to an order, and the lower single user for characterizing the order consults M type relevant to the article of the order The behavioral data of the advertisement of type；Each this correspond to without order user advertisement behavioral data it is not any within the specific period One of order generation characterizes this without order user and consults the M relevant to the article for meeting the preset condition without order user The behavioral data of the advertisement of seed type, M are the positive integer more than or equal to 2.

Neural metwork training module 730 is used for the input using the user advertising behavioral data as neural network, and with this It is default answer, the training nerve that the corresponding order of each data, which actually generates the analog value of value, in user advertising behavioral data Network is to have predicted whether order generation, until the consistency of the default answer of prediction and this of the neural network reaches preassigned Shi Xunlian is completed.

Output module 740 is used to export the neural network of training completion.

In accordance with an embodiment of the present disclosure, it can at least be partially solved that can not accurately to analyse in depth user in the prior art wide The problem of accusing the relevance between behavioral data and order data, and pass through a large amount of user advertising row of neural network deep learning Relationship between data and order data may be implemented more precisely whether to predict user according to user advertising behavioral data The technical effect that can finally buy provides more elasticity, flexibility and more accurate conclusion and prediction result.

In accordance with an embodiment of the present disclosure, which further includes the advertisement contribution rate determining module of order 750。

The advertisement contribution rate determining module 750 of order is used to be directed to each order, determines every in the advertisement of the M seed type The contribution rate of a type of advertisement.It specifically includes, the advertisement for each type, is used as in the advertisement of this type wait divide When analysing type advertisements, executes and refer to operating process shown in fig. 5.

In accordance with an embodiment of the present disclosure, which further includes the advertisement contribution rate computing module always sold 760。

It is each in all orders for the article that the advertisement contribution rate computing module 760 always sold is used to that preset condition will to be met The contribution rate of the advertisement of each type of a order summarizes according to advertisement type correspondence, to obtain the advertisement of each type To the sale contribution rate for the article for meeting the preset condition.

In accordance with an embodiment of the present disclosure, each has order user advertisement behavioral data to can refer to table 1.This has order user The lower single user that behavioral data can characterize the order consults the behavior number of the advertisement of M seed type relevant to the article of the order According to, including be not later than within the specific period on the day of the order generates and before the specific period it is N days continuous in, this places an order Exposure frequency and/or number of clicks of the user in the different types of advertisement of M kind relevant to the article of the order, wherein N is big In the integer for being equal to 0.

In accordance with an embodiment of the present disclosure, each can refer to table 2 without order user advertisement behavioral data.This is used without no order Advertisement behavioral data in family can characterize this and consult the wide of M seed type relevant to the article for meeting the preset condition without order user The behavioral data of announcement, comprising: before the specific period and the specific period in continuous N days, this without order user with expire Foot the preset condition the relevant different types of advertisement of M kind of article exposure frequency and/or number of clicks, wherein N be greater than Integer equal to 0.

In accordance with an embodiment of the present disclosure, this have order user advertisement behavioral data data volume be less than this without order user When the data volume of advertisement behavioral data, expand the data volume for having order user advertisement behavioral data according to preset rules, so that There is the data volume of order user advertisement behavioral data suitable with the data volume without order user advertisement behavioral data at this.

According to the embodiment of the present disclosure, which is convolutional neural networks.

According to the embodiment of the present disclosure, which includes: the combination of multiple convolutional layers, rectification layer and pond layer, And/or the combination of multiple full articulamentums and rectification layer.

It is understood that order data obtains module 710, user advertising behavioral data obtains module 720, neural network Training module 730, output module 740, the advertisement contribution rate determining module 750 of order and the advertisement contribution rate meter always sold It calculates module 760 and may be incorporated in a module and realize or any one module therein can be split into multiple modules. Alternatively, at least partly function of one or more modules in these modules can mutually be tied at least partly function of other modules It closes, and is realized in a module.According to an embodiment of the invention, order data obtains module 710, user advertising behavioral data Obtain module 720, neural metwork training module 730, output module 740, order advertisement contribution rate determining module 750 and At least one of advertisement contribution rate computing module 760 always sold can at least be implemented partly as hardware circuit, such as Field programmable gate array (FPGA) programmable logic array (PLA), system on chip, the system on substrate, in encapsulation is System, specific integrated circuit (ASIC), or can be to carry out the hardware such as any other rational method that is integrated or encapsulating to circuit Or firmware realizes, or is realized with software, the appropriately combined of hardware and firmware three kinds of implementations.Alternatively, order data Obtain module 710, user advertising behavioral data obtains module 720, neural metwork training module 730, output module 740, order Advertisement contribution rate determining module 750 and at least one of the advertisement contribution rate computing module 760 always sold can be at least It is implemented partly as computer program module, when the program is run by computer, the function of corresponding module can be executed.

Fig. 8 diagrammatically illustrates the block diagram of the order processing computer system according to the embodiment of the present disclosure.Shown in Fig. 8 Computer system is only an example, should not function to the embodiment of the present disclosure and use scope bring any restrictions.

As shown in figure 8, include processor 801 according to the computer system 800 of the embodiment of the present disclosure, it can be according to storage It is loaded into random access storage device (RAM) 803 in the program in read-only memory (ROM) 802 or from storage section 808 Program and execute various movements appropriate and processing.Processor 801 for example may include general purpose microprocessor (such as CPU), refer to Enable set processor and/or related chip group and/or special microprocessor (for example, specific integrated circuit (ASIC)), etc..Processing Device 801 can also include the onboard storage device for caching purposes.Processor 801 may include for executing with reference to Fig. 2~Fig. 6 Single treatment unit either multiple processing units of the different movements of the method flow according to the embodiment of the present disclosure of description.

In RAM 803, it is stored with system 800 and operates required various programs and data.Processor 801, ROM 802 with And RAM 803 is connected with each other by bus 804.Processor 801 is held by executing the program in ROM 802 and/or RAM 803 Various operations of the row above with reference to Fig. 2~Fig. 6 order processing method described.It is being removed it is noted that the program also can store In one or more memories other than ROM 802 and RAM 803.Processor 801 can also be stored in this by executing Or the program in multiple memories come execute above with reference to Fig. 2~Fig. 6 describe order processing method various operations.

In accordance with an embodiment of the present disclosure, system 800 can also include input/output (I/O) interface 808, input/output (I/O) interface 805 is also connected to bus 804.System 800 can also include be connected to I/O interface 805 with one in lower component Item is multinomial: the importation 806 including keyboard, mouse etc.；Including such as cathode-ray tube (CRT), liquid crystal display (LCD) Deng and loudspeaker etc. output par, c 807；Storage section 808 including hard disk etc.；And including such as LAN card, modulatedemodulate Adjust the communications portion 809 of the network interface card of device etc..Communications portion 809 executes communication process via the network of such as internet. Driver 610 is also connected to I/O interface 805 as needed.Detachable media 811, such as disk, CD, magneto-optic disk, semiconductor Memory etc. is mounted on as needed on driver 810, in order to be pacified as needed from the computer program read thereon It is packed into storage section 808.

In accordance with an embodiment of the present disclosure, it may be implemented as computer software journey above with reference to the method for flow chart description Sequence.For example, embodiment of the disclosure includes a kind of computer program product comprising carry meter on a computer-readable medium Calculation machine program, the computer program include the program code for method shown in execution flow chart.In such embodiments, The computer program can be downloaded and installed from network by communications portion 809, and/or be pacified from detachable media 811 Dress.When the computer program is executed by processor 801, the above-mentioned function of limiting in the system of the embodiment of the present disclosure is executed.Root According to embodiment of the disclosure, system as described above, unit, module, unit etc. can by computer program module come It realizes.

It should be noted that computer-readable medium shown in the disclosure can be computer-readable signal media or meter Calculation machine readable storage medium storing program for executing either the two any combination.Computer readable storage medium for example can be --- but not Be limited to --- electricity, magnetic, optical, electromagnetic, infrared ray or semiconductor system, device or device, or any above combination.Meter The more specific example of calculation machine readable storage medium storing program for executing can include but is not limited to: have the electrical connection, just of one or more conducting wires Taking formula computer disk, hard disk, random access storage device (RAM), read-only memory (ROM), erasable type may be programmed read-only storage Device (EPROM or flash memory), optical fiber, portable compact disc read-only memory (CD-ROM), light storage device, magnetic memory device, Or above-mentioned any appropriate combination.In the disclosure, computer readable storage medium can be it is any include or storage journey The tangible medium of sequence, the program can be commanded execution system, device or device use or in connection.And at this In open, computer-readable signal media may include in a base band or as the data-signal that carrier wave a part is propagated, Wherein carry computer-readable program code.The data-signal of this propagation can take various forms, including but unlimited In electromagnetic signal, optical signal or above-mentioned any appropriate combination.Computer-readable signal media can also be that computer can Any computer-readable medium other than storage medium is read, which can send, propagates or transmit and be used for By the use of instruction execution system, device or device or program in connection.Include on computer-readable medium Program code can transmit with any suitable medium, including but not limited to: wireless, electric wire, optical cable, RF etc. are above-mentioned Any appropriate combination.In accordance with an embodiment of the present disclosure, computer-readable medium may include above-described ROM 802 And/or one or more memories other than RAM 803 and/or ROM 802 and RAM 803.

Flow chart and block diagram in attached drawing are illustrated according to the system of the various embodiments of the disclosure, method and computer journey The architecture, function and operation in the cards of sequence product.In this regard, each box in flowchart or block diagram can generation A part of one module, program segment or code of table, a part of above-mentioned module, program segment or code include one or more Executable instruction for implementing the specified logical function.It should also be noted that in some implementations as replacements, institute in box The function of mark can also occur in a different order than that indicated in the drawings.For example, two boxes succeedingly indicated are practical On can be basically executed in parallel, they can also be executed in the opposite order sometimes, and this depends on the function involved.Also it wants It is noted that the combination of each box in block diagram or flow chart and the box in block diagram or flow chart, can use and execute rule The dedicated hardware based systems of fixed functions or operations is realized, or can use the group of specialized hardware and computer instruction It closes to realize.

As on the other hand, the disclosure additionally provides a kind of computer-readable medium, which can be Included in equipment described in above-described embodiment；It is also possible to individualism, and without in the supplying equipment.Above-mentioned calculating Machine readable medium carries one or more program, when said one or multiple programs are executed by the equipment, makes Equipment execution is obtained according to the order processing method of the embodiment of the present disclosure.

In accordance with an embodiment of the present disclosure, this method comprises: obtaining ordering for the article for meeting preset condition in the specific period Forms data, each order data correspond to a kind of primary order of article；Obtain user advertising behavioral data, the user advertising Behavioral data includes order user advertisement behavioral data and without order user advertisement behavioral data, in which: this has order user Each of advertisement behavioral data data correspond to an order, and the lower single user for characterizing the order is consulted and the object of the order The behavioral data of the advertisement of the relevant M seed type of product；Each data without order user advertisement behavioral data correspond to One generated in the specific period without any order characterizes this and consults and to meet this default without order user without order user The behavioral data of the advertisement of the relevant M seed type of the article of condition, M are the positive integer more than or equal to 2；With the user advertising row Input for data as neural network, and actually generated with the corresponding order of each data in the user advertising behavioral data The analog value of value is default answer, and the training neural network is to have predicted whether order generation, until the prediction of the neural network Reach predetermined standard time training with the consistency for presetting answer to complete；The neural network that output training is completed.

In accordance with an embodiment of the present disclosure, this method further includes determining every in the advertisement of the M seed type for each order The contribution rate of a type of advertisement.The advertisement for each type is specifically included, is used as in the advertisement of this type wait divide When analysing type advertisements, execute following operation: by the corresponding user advertising behavioral data of the order with the type advertisements to be analyzed Except the relevant data of other kinds of advertisement be set as zero, obtain user's type advertisements behavioral data to be analyzed；By the use Family type advertisements behavioral data to be analyzed is input to the neural network of training completion；Obtain the defeated of the neural network that training is completed Out, wherein the output for the neural network that the training is completed corresponds to the type advertisements to be analyzed to the contribution rate of the order.

In accordance with an embodiment of the present disclosure, this method further includes each in all orders for will meet the article of preset condition The contribution rate of the advertisement of each type of order summarizes according to advertisement type correspondence, to obtain the advertisement pair of each type Meet the sale contribution rate of the article of the preset condition.

In accordance with an embodiment of the present disclosure, the lower single user of the characterization order consults M kind relevant to the article of the order The behavioral data of the advertisement of type, including be not later than within the specific period on the day of the order generates and the specific period it In first continuous N days, the lower single user the different types of advertisement of M kind relevant to the article of the order exposure frequency and/or Number of clicks, wherein N is the integer more than or equal to 0.

In accordance with an embodiment of the present disclosure, the characterization this without order user consult it is relevant to the article for meeting the preset condition The behavioral data of the advertisement of M seed type, including before the specific period and the specific period in continuous N days, this is without order User the different types of advertisement of M kind relevant to the article for meeting the preset condition exposure frequency and/or number of clicks, In, N is the integer more than or equal to 0.

In accordance with an embodiment of the present disclosure, this have order user advertisement behavioral data data volume be less than this without order user When the data volume of advertisement behavioral data, reduce the data volume without order user advertisement behavioral data according to preset rules, so that There is the data volume of order user advertisement behavioral data suitable with the data volume without order user advertisement behavioral data at this.

Embodiment of the disclosure is described above.But the purpose that these embodiments are merely to illustrate that, and It is not intended to limit the scope of the present disclosure.Although respectively describing each embodiment above, but it is not intended that each reality Use cannot be advantageously combined by applying the measure in example.The scope of the present disclosure is defined by the appended claims and the equivalents thereof.It does not take off From the scope of the present disclosure, those skilled in the art can make a variety of alternatives and modifications, these alternatives and modifications should all fall in this Within scope of disclosure.

Claims

1. a kind of order processing method, comprising:

The order data of the article for meeting preset condition in the specific period is obtained, each order data corresponds to a kind of article Primary order；

User advertising behavioral data is obtained, the user advertising behavioral data includes that order user advertisement behavioral data and nothing are ordered Single user advertisement behavioral data, in which: there is order user advertisement behavioral data to correspond to an order described in each, characterization should The lower single user of order consults the behavioral data of the advertisement of M seed type relevant to the article of the order；Each described nothing is ordered Single user advertisement behavioral data corresponds to one generated within the specific period without any order without order user, characterization The no order user consults the behavioral data of the advertisement of the M seed type relevant to the article for meeting the preset condition, M For the positive integer more than or equal to 2；

Using the user advertising behavioral data as the input of neural network, and with each in the user advertising behavioral data The analog value that the corresponding order of data actually generates value is default answer, and the training neural network is to have predicted whether that order is raw At until the consistency of prediction and the default answer of the neural network reaches predetermined standard time training completion；And

The neural network that output training is completed.

2. according to the method described in claim 1, further include:

For each order, the contribution rate of the advertisement of each type in the advertisement of the M seed type is determined, including for every A type of advertisement executes following operation when the advertisement of this type is as type advertisements to be analyzed:

By in the corresponding user advertising behavioral data of the order with the other kinds of advertisement except the type advertisements to be analyzed Relevant data are set as zero, obtain user's type advertisements behavioral data to be analyzed；

User type advertisements behavioral data to be analyzed is input to the neural network that the training is completed；And

Obtain the output for the neural network that training is completed, wherein the output for the neural network that the training is completed corresponds to described Contribution rate of the type advertisements to be analyzed to the order.

3. according to the method described in claim 2, further include:

To meet the contribution rate of the advertisement of each type of each order in all orders of the article of preset condition according to Advertisement type correspondence summarizes, and is contributed with obtaining the advertisement of each type the sale for the article for meeting the preset condition Rate.

4. method according to any one of claims 1 to 3, wherein it is described characterize the order lower single user consult with The behavioral data of the advertisement of the relevant M seed type of the article of the order, comprising:

It is not later than within the specific period on the day of the order generates and before the specific period in continuous N days, under described Exposure frequency and/or number of clicks of the single user in the different types of advertisement of M kind relevant to the article of the order, wherein N is Integer more than or equal to 0.

5. method according to any one of claims 1 to 3, wherein the characterization no order user is consulted and expired The behavioral data of the advertisement of the relevant M seed type of article of the foot preset condition, comprising:

Before the specific period and the specific period in continuous N days, the no order user with meet it is described pre- If the exposure frequency and/or number of clicks of the relevant different types of advertisement of M kind of the article of condition, wherein N is more than or equal to 0 Integer.

6. method according to any one of claims 1 to 3, in which:

It is less than the data of the no order user advertisement behavioral data in the data volume for having order user advertisement behavioral data When amount, reduce the data volume of the no order user advertisement behavioral data according to preset rules, so as to have order user described The data volume of advertisement behavioral data is suitable with the data volume of the no order user advertisement behavioral data.

7. method according to any one of claims 1 to 3, in which:

The neural network is convolutional neural networks.

8. according to the method described in claim 7, wherein, the convolutional neural networks include:

Multiple convolutional layers, the combination for rectifying layer and pond layer, and/or

The combination of multiple full articulamentums and rectification layer.

9. a kind of order processing device, comprising:

Order data obtains module, for obtaining the order data of the article for meeting preset condition in the specific period, each Order data corresponds to a kind of primary order of article；

User advertising behavioral data obtains module, for obtaining user advertising behavioral data, the user advertising behavioral data packet Order user advertisement behavioral data is included and without order user advertisement behavioral data, in which: have order user wide described in each It accuses and corresponds to an order in behavioral data, the lower single user for characterizing the order consults M type relevant to the article of the order The behavioral data of the advertisement of type；Each described no order user advertisement behavioral data corresponds to not to be had within the specific period The article phase consulted without order user, the characterization no order user and meet the preset condition that any order generates The behavioral data of the advertisement for the M seed type closed, M are the positive integer more than or equal to 2；

Neural metwork training module, for the input using the user advertising behavioral data as neural network, and with the use It is default answer, the training nerve that the corresponding order of each data, which actually generates the analog value of value, in family advertisement behavioral data Network is to have predicted whether order generation, until the consistency of the prediction and the default answer of the neural network reaches predetermined Training is completed when standard；And

Output module, for exporting the neural network of training completion.

10. device according to claim 9, further includes:

The advertisement contribution rate determining module of order determines each in the advertisement of the M seed type for being directed to each order The contribution rate of the advertisement of type is wide as type to be analyzed in the advertisement of this type including the advertisement for each type When announcement, following operation is executed:

11. device according to claim 10, further includes:

The advertisement contribution rate computing module always sold, each order in all orders of the article for preset condition will to be met The contribution rate of advertisement of each type summarize according to advertisement type correspondence, to obtain the advertisement of each type to full The sale contribution rate of the article of the foot preset condition.

12. according to device described in claim 9~10 any one, wherein the lower single user for characterizing the order is consulted The behavioral data of the advertisement of M seed type relevant to the article of the order, comprising:

13. according to device described in claim 9~10 any one, wherein the characterization no order user consult with Meet the behavioral data of the advertisement of the relevant M seed type of article of the preset condition, comprising:

14. according to device described in claim 9~10 any one, in which:

15. according to device described in claim 9~10 any one, in which:

The neural network is convolutional neural networks.

16. device according to claim 15, wherein the convolutional neural networks include:

The combination of multiple full articulamentums and rectification layer.

17. a kind of order processing device, comprising:

One or more processors；And

Memory, for storing one or more programs,

Wherein, when one or more of programs are executed by one or more of processors, so that one or more of Processor realizes method described in any one according to claim 1~8.

18. a kind of computer-readable medium, is stored thereon with executable instruction, which makes processor real when being executed by processor Now method described in any one according to claim 1~8.