CN107194729A - Advertisement price competing method, device, electronic installation and computer-readable medium - Google Patents
Advertisement price competing method, device, electronic installation and computer-readable medium Download PDFInfo
- Publication number
- CN107194729A CN107194729A CN201710363782.9A CN201710363782A CN107194729A CN 107194729 A CN107194729 A CN 107194729A CN 201710363782 A CN201710363782 A CN 201710363782A CN 107194729 A CN107194729 A CN 107194729A
- Authority
- CN
- China
- Prior art keywords
- bid
- request
- advertisement
- bidding
- real
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0273—Determination of fees for advertising
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0283—Price estimation or determination
Landscapes
- Business, Economics & Management (AREA)
- Development Economics (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Strategic Management (AREA)
- Finance (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Economics (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The disclosure provides a kind of advertisement price competing method, device, electronic installation and computer-readable medium, and the advertisement price competing method includes：It is modeled based on enhancing learning algorithm, obtains the Competitive Bidding Model of relation between bid request, bid and income；The Competitive Bidding Model is optimized, the bidding strategy optimized；For real-time bid request by the Competitive Bidding Model and the bidding strategy of the optimization, the bid corresponding with the real-time bid request is provided.It is modeled based on enhancing learning algorithm, obtains the relation between bid request, income and bid, without carrying out substantial amounts of artificial parameter adjustment, reduces complexity.
Description
Technical field
The disclosure relates in general to Internet technical field, in particular to a kind of advertisement price competing method, device, electronics
Device and computer-readable medium.
Background technology
Party in request's platform (Demand-side platform, abbreviation DSP) is a kind of advertisement putting for serving advertiser
Platform, receives the request that advertiser delivers advertisement in multiple advertisement transaction platforms (Ad exchange, abbreviation ADX), participates in real-time
Advertisement bidding, buy target audience.Similar with paid search, DSP allows advertiser to be based on KPI Key Performance Indicator
(Performance Indicator, abbreviation KPI) optimizes advertisement serving policy, such as thousand times clicks spend (effective
Cost Per Click, abbreviation eCPC) etc..Real time bid (Real-time bidding, abbreviation RTB) is one kind with single exhibition
It is now unit, the sequencing bidding fashion of dealing advertisement stock (media advertisement position).
By taking the common application scenarios as an example, advertisement launching platform purchase media show chance, win after this bids, extensively
Announcement can be presented in the advertisement position of media, and user clicks on advertisement link, brings flow, media are to advertisement launching platform charging；User
Click on the advertisement of advertiser's dispensing and reach after Freight Basis, system is to advertiser's charging.Therefore, for showing chance every time
Bidding strategy, determines the rate of return on investment of advertiser and the profitability of platform.
Based on above-mentioned, at least there are the following problems in prior art：
Due to there is every manual intervention coefficient, the optimization aim for model of bidding and the target of Ask-Bid System are not consistent, are
System effect need it is a large amount of manually adjust parameter could realize it is optimal.Therefore, technical scheme of the prior art also exists and needs to be changed
Enter part.
Above- mentioned information is only used for strengthening the understanding of background of this disclosure, therefore it disclosed in the background section
It can include not constituting the information to prior art known to persons of ordinary skill in the art.
The content of the invention
The disclosure provides a kind of advertisement price competing method, device, electronic installation and computer-readable medium, solves above-mentioned technology
Problem.
Other characteristics and advantage of the disclosure will be apparent from by following detailed description, or partially by the disclosure
Practice and acquistion.
According to the one side of the disclosure there is provided a kind of price competing method, including：
It is modeled based on enhancing learning algorithm, obtains the Competitive Bidding Model of relation between bid request, bid and income；
The Competitive Bidding Model is optimized, the bidding strategy optimized；
For real-time bid request by the Competitive Bidding Model and the bidding strategy of the optimization, provide and the reality
When the corresponding bid of bid request.
In one embodiment of the disclosure, it is described based on enhancing learning algorithm be modeled including：
Respond the bid request to be bidded, obtain the bid；
Sliding-model control is carried out to the bid, multiple bid centrifugal pumps are obtained；
Calculated respectively according to the multiple bid centrifugal pump if bidding successfully and obtain the income, wherein the income
It is party in request's platform to the difference of the charging of advertiser and advertisement transaction platform to the charging of party in request's platform.
In one embodiment of the disclosure, before the response bid request, in addition to：
The bid request is transformed to natural language.
In one embodiment of the disclosure, obtain it is multiple bid centrifugal pumps after, in addition to：
When each described bid centrifugal pump is higher than real price, a real-time stream is replicated；
The real-time stream is (x, b, r, p), and wherein x is the bid request, and b is the bid, and r receives to be described
Benefit, p is the real price.
In one embodiment of the disclosure, in modeling process, exposure data, click logs and billing log are pressed
Spliced in real time according to time sequencing, obtain the real-time stream, wherein the exposure data is for the bid request
Bid and be successfully presented to the data of user, the click logs are that user clicks on the daily record data generated during media advertisement position, institute
State billing log and click on charging and advertisement transaction of the media advertisement Wei Shi parties in request platform to advertiser for the user
Daily record data of the platform to the charging of party in request's platform.
In one embodiment of the disclosure, also include before the Competitive Bidding Model modeling：
Initial bid strategy is obtained based on the optimization of existing bid information.
According to the another further aspect of the disclosure there is provided a kind of advertisement bidding device, including：
Modeling module, for being modeled based on enhancing learning algorithm, obtains relation between bid request, bid and income
Competitive Bidding Model；
Optimization module, for being optimized to the Competitive Bidding Model, the bidding strategy optimized；
Bid module, for passing through the Competitive Bidding Model and the bid plan of the optimization for real-time bid request
Slightly, the bid corresponding with the real-time bid request is provided.
In one embodiment of the disclosure, the modeling module includes：
Submodule is responded, is bidded for responding the bid request, obtains the bid；
Discrete submodule, for carrying out sliding-model control to the bid, obtains multiple bid centrifugal pumps；
Calculating sub module, the receipts are obtained for being calculated respectively according to the multiple bid centrifugal pump if bidding successfully
Benefit, wherein the income is charging and advertisement transaction platform charging to the party in request platform of party in request's platform to advertiser
Difference.
According to the another aspect of the disclosure, there is provided a kind of electronic installation, including processor；Memory, is stored for described
The instruction that processor control is operated as described above.
According to another aspect of the present disclosure there is provided a kind of computer-readable medium, the executable finger of computer is stored thereon with
Order, it is characterised in that the executable instruction realizes advertisement price competing method as described above when being executed by processor.
Advertisement price competing method, device, electronic installation and the computer-readable medium provided according to the embodiment of the present disclosure, wherein
Advertisement price competing method is based on enhancing learning algorithm and is modeled, and the relation between bid request, income and bid is obtained, without entering
The substantial amounts of artificial parameter adjustment of row, reduces complexity.
It should be appreciated that the general description of the above and detailed description hereinafter are only exemplary, this can not be limited
It is open.
Brief description of the drawings
Its example embodiment is described in detail by referring to accompanying drawing, above and other target, feature and the advantage of the disclosure will
Become more fully apparent.
Fig. 1 shows a kind of step flow chart of the advertisement price competing method provided in the embodiment of the disclosure one.
Fig. 2 shows the step flow chart of step S10 in the embodiment of the disclosure one.
Fig. 3 shows bid process request transfer schematic diagram in the embodiment of the present disclosure.
Fig. 4 shows to complete the Organization Chart of advertisement bidding in the embodiment of the present disclosure.
Fig. 5 shows the schematic diagram that user interacts with advertisement transaction platform and party in request's platform in the embodiment of the present disclosure.
Fig. 6 shows the process schematic of Competitive Bidding Model and environment successive optimization of bidding in the embodiment of the present disclosure.
Fig. 7 shows a kind of schematic diagram of the advertisement bidding device provided in another embodiment of the disclosure.
Fig. 8 shows the computer for being suitable to be used for realizing the electronic installation of the embodiment of the present application that the embodiment of the disclosure one is provided
The structural representation of system.
Embodiment
Example embodiment is described more fully with referring now to accompanying drawing.However, example embodiment can be with a variety of shapes
Formula is implemented, and is not understood as limited to example set forth herein；On the contrary, thesing embodiments are provided so that the disclosure will more
Fully and completely, and by the design of example embodiment those skilled in the art is comprehensively conveyed to.Accompanying drawing is only the disclosure
Schematic illustrations, be not necessarily drawn to scale.Identical reference represents same or similar part in figure, thus
Repetition thereof will be omitted.
Implement in addition, described feature, structure or characteristic can be combined in any suitable manner one or more
In mode.In the following description there is provided many details so as to provide fully understanding for embodiment of this disclosure.So
And, it will be appreciated by persons skilled in the art that the technical scheme of the disclosure can be put into practice and one in the specific detail is omitted
Or more, or can be using other methods, constituent element, device, step etc..In other cases, it is not shown in detail or describes
Known features, method, device, realization, material are operated to avoid that a presumptuous guest usurps the role of the host so that each side of the disclosure becomes mould
Paste.
Some block diagrams shown in accompanying drawing are functional entitys, not necessarily must be with physically or logically independent entity phase
Correspondence.These functional entitys can be realized using software form, or in one or more hardware modules or integrated circuit in fact
These existing functional entitys, or realize that these functions are real in heterogeneous networks and/or processor device and/or microcontroller device
Body.
By taking advertisement bidding as an example, average every thousand times of commodity in use show required cost (effective cost
Per mille, abbreviation ECPM) it is used as bid.ECPM=PCTR*ECPC, wherein ECPC are average thousand clicks institute bands of commodity
Come cost or estimate cost, PCTR is clicking rate discreet value of the commodity in request target advertisement position.In actual applications, system
Bid also has the adjustable coefficient of other manual interventions, control bid.End article is estimated using enhancing study at certain once
(Sponsored Search MDP, search engine markov decision process, Microsoft proposes for the bid of request, such as SS-MDP
It is a kind of by search engine account optimize the abstract mode for MDP) or RLB (Reinforcement to bid, one kind based on increasing
The account optimization method learnt by force), but existing enhancing Learning Scheme, account bid optimization is used to, in known users pair
In the case of the bidding strategy or user balance budget that Current ad is asked, the bid of some moment ad-request is estimated,
The target of system is often maximization clicking rate, for that can not obtain the scene of bidding strategy or remaining sum budget and not apply to.
Following defect can so be brought：
1. in existing maximization ECPM schemes because there is every manual intervention coefficient, the optimization aim for model of bidding with
The target of Ask-Bid System is not consistent, it is impossible to ensure that current bidding strategy is optimal, and system effect needs a large amount of artificial adjust to join
It can realize optimal.Secondly, manual intervention coefficient is relatively subjective, it is difficult to find optimal value, realizes best effect.In addition, when number
Changed according to distribution, parameter needs to readjust.
2. the scheme of existing enhancing study, by reducing the space of bid, to solve to ask some bid datas are sparse
Topic, causes the action space of bid relatively small.Therefore for successful advertisement of bidding, system bid often focuses on some small
Interval, and for substantial amounts of bid behavior, it is possible to use data and few, this can cause the poor fitting of enhancing learning model to be asked
Topic.
3.SS-MDP methods, discreet value is unrelated with request contexts information, it is impossible to fully using user, media and commodity
Information etc..
4.RLB methods still need to estimate clicking rate, it is necessary to build clicking rate prediction model and bid model, system simultaneously
Realize more complicated, time complexity increase.
5. existing technical scheme was offline T+1 models mostly, using the data modeling of first T days, it was used to bid at T+1 days
Estimate, it is impossible to simulate environment of bidding in real time.
Understood based on above-mentioned, existing enhancing learning model is to that can not be directly realized by optimal and openness and real-time all
It is poor.
Fig. 1 shows a kind of step flow chart of the advertisement price competing method provided in the embodiment of the disclosure one, including following step
Suddenly：
As shown in figure 1, in step slo, being modeled based on enhancing learning algorithm, bid request, bid are obtained with receiving
The Competitive Bidding Model of relation between benefit.That is, being that (Deep Q-Network, one kind is by Google based on DQN in the present embodiment
The deep learning intelligent algorithm of DeepMind exploitations) come what is be modeled.
As shown in figure 1, in step S20, being optimized to Competitive Bidding Model, the bidding strategy optimized.
As shown in figure 1, in step s 30, Competitive Bidding Model and the bid plan of optimization are passed through for real-time bid request
Slightly, the bid corresponding with real-time bid request is provided.
The advertisement price competing method is based on enhancing learning algorithm framework and redefines advertisement bidding problem, and Q- is learnt using enhancing
Learning algorithms construct the Competitive Bidding Model of relation between bid request, bid and income, using bid request as input, profit
The discreet value of Current ad request bid is provided with the bidding strategy of Competitive Bidding Model and optimization, to provide optimal bid plan
Slightly, complexity is simplified.
Fig. 2 shows the step flow chart that step S10 is modeled based on enhancing learning algorithm in the present embodiment, including following
Step：
As shown in Fig. 2 in step s 11, response bid request is bidded, and is bid.
Comprising the information of advertisement position, commodity and user for participating in bidding in bid request, party in request's platform for from
The bid request x of the advertisement of advertisement transaction platform provides bid b, it is necessary to convert bid request in explanation, the present embodiment
For natural language, that is, bid request is processed into the natural language of special format, such Competitive Bidding Model can directly understand
The bid request of advertisement, in addition to being asked with natural language description, is not required to any feature extraction work.
For example：For the user Wang Yu in 9 points of access sports.sina.com.cn of the morning, the business of user will be presented to
Product are iphone7, ps4 and macbook, and this solicited message is changed into following English describes：
User wang2yu4 is accessing Sina Sports at 9:00am,we’ll display
iphone7,ps4,and macbook to him。
Next the characteristics of describing user, website, commodity each, including but not limited to user's purchasing history, user browse
The data such as history, age of user, user's sex, commodity price, commodity favorable comment number.
As shown in Fig. 2 in step s 12, carrying out sliding-model control to bid, obtaining multiple bid centrifugal pumps.
Assuming that the least unit of the bid of party in request's platform is b, the bid of system is discretized into 0, b, 2b ..., (n-1)
b,nb；Wherein nb is the maximum bid that party in request's platform allows.If for example, system is bid, the upper limit is 50 yuan, can be according to every
One mao, as a bid, is separated into altogether 0,1,2 ..., 500 501 bids altogether.
As shown in Fig. 2 in step s 13, being calculated and being received respectively according to multiple bid centrifugal pumps if bidding successfully
Benefit, wherein income are party in request's platform to the difference of the charging of advertiser and advertisement transaction platform to the charging of party in request's platform.
For each bid request x, party in request's platform provides multiple bids, and each bid is b integral multiple, if competing
Valency success, will produce corresponding income r, and obtain the real price p of advertisement position, and then request x+1 is reached next time；Such as
Not successful then income of really bidding is exactly 0.
Fig. 3 shows bid process request transfer schematic diagram, it is assumed that request x+1 process is transferred to from request x, income r is only
It is relevant with request x, and request before x is separate.That is, bid process meets Markov attribute, obtain
One markov decision process (Markov decision processe, abbreviation MDP).MDP provides a kind of for modeling
The mathematical framework of decision process, the income of decision-making each time is only relevant with current decision-making, unrelated with state before,
MDP problems can use Dynamic Programming or enhancing study to solve.
When each bid centrifugal pump is higher than real price p, a real-time stream is replicated.Real-time stream for (x,
B, r, p), wherein x is bid request, and b is bid, and r is income, and p is real price.
It should be noted that being used as the training number of Competitive Bidding Model in the present embodiment using real-time stream (x, b, p, r)
According to.MDP time sequencing is critically important, therefore in order to safeguard that the number of user is namely bidded successfully and be presented to exposure data (
According to) time sequencing, the present embodiment is suitable according to the time to exposure data, click logs and billing log in modeling process
Sequence is spliced in real time, obtains real-time stream, and wherein exposure data is successfully presented to user's to be bidded for bid request
Data, click logs are that user clicks on the daily record data generated during media advertisement position, and billing log is that user clicks on media advertisement
Wei Shi parties in request platform is to the daily record data of the charging of advertiser and advertisement transaction platform to the charging of party in request's platform.
Fig. 4 show in the present embodiment complete advertisement bidding Organization Chart, as shown in figure 4, including service layer, model training with
And data Layer.Wherein data Layer include ad-request daily record (x, b), bid successfully daily record (r, p) and based on kafka splicing
Real-time stream (x, b, r, p).Service layer includes the bid request x of advertisement, model loading and bid b.
In Fig. 4, in service layer, ad-request DSP is bidded, and bid is provided by Competitive Bidding Model, records Request Log；ADX
Bid request is received, if bidding success, advertisement is presented to user, and when user clicks on media advertisement position, ADX is generated and returned
The information such as charging, DSP records successfully daily record of bidding；When user clicks on the advertisement that advertiser delivers, DSP to advertiser's charging, this
Shi Shengcheng click logs.
Fig. 5 shows the schematic diagram that user interacts with advertisement transaction platform and party in request's platform in the embodiment of the present disclosure, interaction
What the main body in advertisement bidding involved by process, i.e. user, media advertisement position, ADX, DSP, advertiser and advertiser delivered
Advertisement.
As shown in figure 5, first, user's browsing media, media just ask advertisement to ADX, and ADX sends advertisement bidding to DSP
Ask x, DSP to return to the b that bids and return to advertisement to media to ADX, ADX, media just show advertisement to user.Next, user's point
Hit media advertisement position, media start charging and be sent to ADX, i.e. ADX record medias the information real price p that deducts fees.It
Afterwards, if user clicks on the advertisement that advertiser delivers, DSP is just to advertiser's charging.
Because training data is respectively present in different daily records, based on kafka to exposure, click and billing log data
Stream, does real-time splicing.Existence time is poor between exposure data, click data and metering data, therefore, is counted in real time in splicing generation
When being trained according to stream (x, b, r, p), exposure data needs to wait click data to generate with metering data, therefore, bid b and energy
It was observed that feedback (r, p) between have certain delay, data flow can only accomplish quasi real time.Prior art is all offline mould
Type, it is impossible to which reaction in real time is bidded the change of environment, the present embodiment, can be with using near-realtime data (in delay 1 hour) training
Solve real time problems.
By using user click data quasi real time etc., profit and loss and the change of clicking rate can be perceived in real time,
For profit flow, profit maximization can be realized as early as possible；For loss flow, can it stop loss in time, it is ensured that platform is held
Continuous sexual development.
In the present embodiment Fig. 4, model training is generally divided into two stages：Training stage in off-line training step and line.
What in general DQN was used is all model free (being used to bid without using other strategies, bid model is used as using only DQN)
The mode trained in real time.But, the real-time training of model is progress simultaneously with service, in order to be able to when model is reached the standard grade for the first time,
With regard to that can have relatively good effect, it is necessary to which before real-time training, DQN models are optimized based on existing bid information (x, b, r, p)
Obtain initialization strategy.Real-time training stage, model provides rational bidding strategy to request on line, it was observed that corresponding feedback
After, based on real-time training data, bidding strategy is continued to optimize, until realizing optimal.As Fig. 6 is shown in the embodiment of the present disclosure
The process schematic of Competitive Bidding Model and environment successive optimization of bidding, that is, Competitive Bidding Model provide bid b into environment of bidding, so
Constantly optimize Competitive Bidding Model further according to income r and real price p afterwards.
So, Competitive Bidding Model is trained using real-time stream, continues to optimize current strategies, the mould of model training generation
On type file real-time synchronization to line, as bidding strategies, the real-time update of implementation model, so as to ensure that model can be anti-in real time
Should be bidded the change of environment.
Because a large amount of bids are often focused in some minizones or certain several bid, remaining bid data amount is rare, just
Sparse sex chromosome mosaicism can be caused.In order to solve the sparse sex chromosome mosaicism of data, it can be obtained for successful bid request of bidding
Corresponding real price, when bidding above real price, bids with regard to that can win this, and concluded price is constant, and profit is not
Become.Therefore to each bid higher than concluded price, single exposure data are all replicated, therefore exposure data is replicated (maximum
Bid-p) it is secondary.Such as, exposure data (x, b, r, p), be respectively after replicating：(x,int(p)+i,r,p).Wherein, int (p)
Expression is rounded to p, and int (p)+i<P_max, P_max represent the maximum bid allowed, that is, in bid centrifugal pump
nb。
In summary, the Competitive Bidding Model built based on DQN, then based on KPI Key Performance Indicator (Key Performance
Indicator, abbreviation KPI) optimize, the relation between Direct Modeling income (r) and bid (b).Used because only that working as
Behavior reaches Freight Basis after family is clicked on, and ad system ability charging simultaneously produces income, and this time request could get a profit, system optimization
Target taken into account the indexs such as system sustainability (profit) and clicking rate simultaneously, Direct Modeling is bid and the pass between KPI
System can be higher consumption and profit.Wherein Freight Basis refers to refer to behavior of the user in advertisement in Internet advertising business
Reach a certain standard ability charging, such as click on click charging (be usually 1 click charging or 2 click charging) of the depth for n,
And for example conversion charging (user buys/sent the charging of the conversion behavior ability such as order).The Competitive Bidding Model using bid request as
Input, for the purpose of maximum gain (income or clicking rate), directly gives the discreet value of Current ad request bid, not only
System complexity is simplified, also as the relation of Direct Modeling request and bid, it is ensured that current bidding strategy is exactly
Optimal solution.
In addition, the relation of Direct Modeling bid, request and profit, eliminates CTR (Click Through Rate, point
Hit percent of pass, Internet advertising term, equal to clicking on divided by show) estimate and the intermediate steps such as ECPC is estimated, also eliminate
Substantial amounts of artificial parameter, not only system realization is simpler, and manual intervention is smaller.System effect is ensure that from principle most
It is excellent, in the parameter that presence can be manually adjusted, it is difficult to ensure that system is optimal.
Fig. 7 shows a kind of schematic diagram for advertisement bidding device that another embodiment of the disclosure is provided, as shown in fig. 7, this is wide
Announcement device 100 of bidding includes：Modeling module 110, optimization module 120 and bid module 130.
Modeling module 110 is used to be modeled based on enhancing learning algorithm, obtains closing between bid request, bid and income
The Competitive Bidding Model of system.Optimization module 120 is used to optimize Competitive Bidding Model, the bidding strategy optimized.Bid module 130
For, by Competitive Bidding Model and the bidding strategy of optimization, providing relative with real-time bid request for real-time bid request
The bid answered.
Wherein modeling module 110 includes：Submodule, discrete submodule and calculating sub module are responded, response submodule is used for
Response bid request is bidded, and is bid；Discrete submodule is used to carry out sliding-model control to bid, obtains multiple bids
Centrifugal pump；Calculating sub module, which is used to be calculated respectively according to multiple bid centrifugal pumps if bidding successfully, obtains income, wherein receiving
Benefit is party in request's platform to the difference of the charging of advertiser and advertisement transaction platform to the charging of party in request's platform.
The advertisement bidding device can realize the advertisement price competing method identical technique effect provided such as above-mentioned embodiment, this
Place is repeated no more.
On the other hand, the disclosure additionally provides a kind of electronic installation, including processor and memory, and memory storage is used for
Above-mentioned processor controls the instruction of following operation：
It is modeled based on enhancing learning algorithm, obtains the Competitive Bidding Model of relation between bid request, bid and income；It is right
Competitive Bidding Model is optimized, the bidding strategy optimized；Pass through Competitive Bidding Model and optimization for real-time bid request
Bidding strategy, provides the bid corresponding with real-time bid request.
Below with reference to Fig. 8, it illustrates suitable for for the computer system 800 for the electronic installation for realizing the embodiment of the present application
Structural representation.Electronic installation shown in Fig. 8 is only an example, to the function of the embodiment of the present application and should not use model
Shroud carrys out any limitation.
As shown in figure 8, computer system 800 includes CPU (CPU) 801, it can be read-only according to being stored in
Program in memory (ROM) 802 or be loaded into program in random access storage device (RAM) 803 from storage part 808 and
Perform various appropriate actions and processing.In RAM 803, the system that is also stored with 800 operates required various programs and data.
CPU 801, ROM 802 and RAM 803 are connected with each other by bus 804.Input/output (I/O) interface 805 is also connected to always
Line 804.
I/O interfaces 805 are connected to lower component：Importation 806 including keyboard, mouse etc.；Penetrated including such as negative electrode
The output par, c 807 of spool (CRT), liquid crystal display (LCD) etc. and loudspeaker etc.；Storage part 808 including hard disk etc.；
And the communications portion 809 of the NIC including LAN card, modem etc..Communications portion 809 via such as because
The network of spy's net performs communication process.Driver 810 is also according to needing to be connected to I/O interfaces 805.Detachable media 811, such as
Disk, CD, magneto-optic disk, semiconductor memory etc., are arranged on driver 810, in order to read from it as needed
Computer program be mounted into as needed storage part 808.
Especially, in accordance with an embodiment of the present disclosure, the process described above with reference to flow chart may be implemented as computer
Software program.For example, embodiment of the disclosure includes a kind of computer program product, it includes being carried on computer-readable medium
On computer program, the computer program include be used for execution flow chart shown in method program code.In such reality
Apply in example, the computer program can be downloaded and installed by communications portion 809 from network, and/or from detachable media
811 are mounted.When the computer program is performed by CPU (CPU) 801, limited in the system for performing the application
Above-mentioned functions.
It should be noted that the computer-readable medium shown in the application can be computer-readable signal media or meter
Calculation machine computer-readable recording medium either the two any combination.Computer-readable medium for example may be-but not limited to-
Electricity, magnetic, optical, electromagnetic, system, device or the device of infrared ray or semiconductor, or it is any more than combination.It is computer-readable
The more specifically example of medium can include but is not limited to：Electrical connection with one or more wires, portable computer magnetic
Disk, hard disk, random access storage device (RAM), read-only storage (ROM), erasable programmable read only memory (EPROM or sudden strain of a muscle
Deposit), optical fiber, portable compact disc read-only storage (CD-ROM), light storage device, magnetic memory device or above-mentioned appoint
The suitable combination of meaning.In this application, computer-readable medium can be it is any include or storage program tangible medium, the journey
Sequence can be commanded execution system, device or device and use or in connection.And in this application, it is computer-readable
Signal media can include in a base band or as carrier wave a part propagate data-signal, can wherein carrying computer
The program code of reading.The data-signal of this propagation can take various forms, including but not limited to electromagnetic signal, optical signal or
Above-mentioned any appropriate combination.Computer-readable signal media can also be any calculating beyond computer-readable medium
Machine computer-readable recording medium, the computer-readable medium can be sent, propagated or be transmitted for by instruction execution system, device or device
Part is used or program in connection.The program code included on computer-readable medium can use any appropriate Jie
Matter is transmitted, and is included but is not limited to：Wirelessly, electric wire, optical cable, RF etc., or above-mentioned any appropriate combination.
Flow chart and block diagram in accompanying drawing, it is illustrated that according to the system of the various embodiments of the application, method and computer journey
Architectural framework in the cards, function and the operation of sequence product.At this point, each square frame in flow chart or block diagram can generation
The part of one module of table, program segment or code, a part for above-mentioned module, program segment or code is comprising one or more
Executable instruction for realizing defined logic function.It should also be noted that in some realizations as replacement, institute in square frame
The function of mark can also be with different from the order marked in accompanying drawing generation.For example, two square frames succeedingly represented are actual
On can perform substantially in parallel, they can also be performed in the opposite order sometimes, and this is depending on involved function.Also
It is noted that the combination of each square frame in block diagram or flow chart and the square frame in block diagram or flow chart, can use and perform rule
Fixed function or the special hardware based system of operation realize, or can use the group of specialized hardware and computer instruction
Close to realize.
Being described in unit involved in the embodiment of the present application can be realized by way of software, can also be by hard
The mode of part is realized.Described unit can also be set within a processor, for example, can be described as：A kind of processor bag
Include transmitting element, acquiring unit, determining unit and first processing units.Wherein, the title of these units is under certain conditions simultaneously
The restriction in itself to the unit is not constituted, for example, transmitting element is also described as " sending picture to the service end connected
Obtain the unit of request ".
On the other hand, the disclosure additionally provides a kind of computer-readable medium, and the computer-readable medium can be above-mentioned
Included in equipment described in embodiment；Can also be individualism, and without be incorporated the equipment in.Above computer can
Read medium and carry one or more program, when said one or multiple programs are performed by the equipment so that this
Equipment includes：It is modeled based on enhancing learning algorithm, obtains the Competitive Bidding Model of relation between bid request, bid and income；
Competitive Bidding Model is optimized, the bidding strategy optimized；Pass through Competitive Bidding Model and optimization for real-time bid request
Bidding strategy, provide the bid corresponding with real-time bid request.
It will be clearly understood that the present disclosure describes how forming and use particular example, but the principle of the disclosure is not limited to
Any details of these examples.On the contrary, the teaching based on disclosure disclosure, these principles can be applied to many other
Embodiment.
The illustrative embodiments of the disclosure are particularly shown and described above.It should be appreciated that the disclosure is not limited
In detailed construction described herein, set-up mode or implementation method；On the contrary, the disclosure is intended to cover included in appended claims
Spirit and scope in various modifications and equivalence setting.
Claims (10)
1. a kind of advertisement price competing method, it is characterised in that including：
It is modeled based on enhancing learning algorithm, obtains the Competitive Bidding Model of relation between bid request, bid and income；
The Competitive Bidding Model is optimized, the bidding strategy optimized；
For real-time bid request by the Competitive Bidding Model and the bidding strategy of the optimization, provide with it is described in real time
The corresponding bid of bid request.
2. advertisement price competing method according to claim 1, it is characterised in that described to be modeled based on enhancing learning algorithm
Including：
Respond the bid request to be bidded, obtain the bid；
Sliding-model control is carried out to the bid, multiple bid centrifugal pumps are obtained；
Calculated respectively according to the multiple bid centrifugal pump if bidding successfully and obtain the income, wherein the income is to need
The side's of asking platform is to the difference of the charging of advertiser and advertisement transaction platform to the charging of party in request's platform.
3. advertisement price competing method according to claim 2, it is characterised in that before the response bid request, in addition to：
The bid request is transformed to natural language.
4. advertisement price competing method according to claim 2, it is characterised in that obtain after multiple bid centrifugal pumps, also wrap
Include：
When each described bid centrifugal pump is higher than real price, a real-time stream is replicated；
The real-time stream is (x, b, r, p), and wherein x is the bid request, and b is the bid, and r is the income, and p is
The real price.
5. advertisement price competing method according to claim 4, it is characterised in that in modeling process, to exposure data, is clicked on
Daily record and billing log are spliced in real time sequentially in time, obtain the real-time stream, wherein the exposure data
The data of user are successfully presented to be bidded for the bid request, when the click logs are that user clicks on media advertisement position
The daily record data of generation, the billing log is that the user clicks on the media advertisement Wei Shi parties in request platform to advertiser's
The daily record data of charging and advertisement transaction platform to the charging of party in request's platform.
6. advertisement price competing method according to claim 1, it is characterised in that also include before the Competitive Bidding Model modeling：
Initial bid strategy is obtained based on the optimization of existing bid information.
7. a kind of advertisement bidding device, it is characterised in that including：
Modeling module, for being modeled based on enhancing learning algorithm, obtains the competing of relation between bid request, bid and income
Valency model；
Optimization module, for being optimized to the Competitive Bidding Model, the bidding strategy optimized；
Bid module, for, by the Competitive Bidding Model and the bidding strategy of the optimization, being given for real-time bid request
Go out the bid corresponding with the real-time bid request.
8. advertisement bidding system according to claim 7, it is characterised in that the modeling module includes：
Submodule is responded, is bidded for responding the bid request, obtains the bid；
Discrete submodule, for carrying out sliding-model control to the bid, obtains multiple bid centrifugal pumps；
Calculating sub module, the income is obtained for being calculated respectively according to the multiple bid centrifugal pump if bidding successfully,
Wherein described income be party in request's platform to the charging of advertiser and advertisement transaction platform to the charging of party in request's platform it
Difference.
9. a kind of electronic installation, it is characterised in that including：
Processor；
Memory, stores the instruction that the operation as described in claim any one of 1-6 is controlled for the processor.
10. a kind of computer-readable medium, is stored thereon with computer executable instructions, it is characterised in that the executable finger
The advertisement price competing method as described in claim any one of 1-6 is realized when order is executed by processor.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710363782.9A CN107194729A (en) | 2017-05-22 | 2017-05-22 | Advertisement price competing method, device, electronic installation and computer-readable medium |
| CN202010623473.2A CN111784400A (en) | 2017-05-22 | 2017-05-22 | Advertisement bidding method, device, electronic device and computer readable medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710363782.9A CN107194729A (en) | 2017-05-22 | 2017-05-22 | Advertisement price competing method, device, electronic installation and computer-readable medium |
Related Child Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010623473.2A Division CN111784400A (en) | 2017-05-22 | 2017-05-22 | Advertisement bidding method, device, electronic device and computer readable medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107194729A true CN107194729A (en) | 2017-09-22 |
Family
ID=59874193
Family Applications (2)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710363782.9A Pending CN107194729A (en) | 2017-05-22 | 2017-05-22 | Advertisement price competing method, device, electronic installation and computer-readable medium |
| CN202010623473.2A Pending CN111784400A (en) | 2017-05-22 | 2017-05-22 | Advertisement bidding method, device, electronic device and computer readable medium |
Family Applications After (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010623473.2A Pending CN111784400A (en) | 2017-05-22 | 2017-05-22 | Advertisement bidding method, device, electronic device and computer readable medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (2) | CN107194729A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108280682A (en) * | 2018-01-16 | 2018-07-13 | 深圳市和讯华谷信息技术有限公司 | Advertisement placement method, terminal and computer readable storage medium |
| CN109949073A (en) * | 2019-02-15 | 2019-06-28 | 北京达佳互联信息技术有限公司 | A kind of information method of adjustment, device, electronic equipment and storage medium |
| CN110009426A (en) * | 2019-04-09 | 2019-07-12 | 科达集团股份有限公司技术分公司 | A kind of advertisement bid optimization method based on eCPM conjugate gradient descent method fitting technique |
| CN110135904A (en) * | 2019-05-10 | 2019-08-16 | 四川合锐信息技术有限公司 | A kind of automatically bid optimization method and system based on streaming computing |
| CN110189181A (en) * | 2019-06-10 | 2019-08-30 | 深圳市云企汇财税顾问有限公司 | A kind of accounting system benefited for the advertisement browsing of Online Video APP |
| WO2020134012A1 (en) * | 2018-12-29 | 2020-07-02 | 北京字节跳动网络技术有限公司 | Value combination determination method and apparatus, storage medium and electronic device |
| CN111598616A (en) * | 2020-05-11 | 2020-08-28 | 百度在线网络技术（北京）有限公司 | Method, device and equipment for realizing object set screening and storage medium |
| CN112016947A (en) * | 2019-05-31 | 2020-12-01 | 百度在线网络技术（北京）有限公司 | Advertisement space distribution method, device, equipment and storage medium |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113592549B (en) * | 2021-07-30 | 2024-03-12 | 北京奇艺世纪科技有限公司 | Conversion rate prediction method and device and electronic equipment |
| CN113807891A (en) * | 2021-09-16 | 2021-12-17 | 北京沃东天骏信息技术有限公司 | Advertisement putting processing method and device |
| CN115131057B (en) * | 2022-05-20 | 2024-02-06 | 欧冶工业品股份有限公司 | Competitive and non-competitive purchase pricing strategy management information processing method and system |
-
2017
- 2017-05-22 CN CN201710363782.9A patent/CN107194729A/en active Pending
- 2017-05-22 CN CN202010623473.2A patent/CN111784400A/en active Pending
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108280682A (en) * | 2018-01-16 | 2018-07-13 | 深圳市和讯华谷信息技术有限公司 | Advertisement placement method, terminal and computer readable storage medium |
| WO2020134012A1 (en) * | 2018-12-29 | 2020-07-02 | 北京字节跳动网络技术有限公司 | Value combination determination method and apparatus, storage medium and electronic device |
| CN109949073A (en) * | 2019-02-15 | 2019-06-28 | 北京达佳互联信息技术有限公司 | A kind of information method of adjustment, device, electronic equipment and storage medium |
| CN110009426A (en) * | 2019-04-09 | 2019-07-12 | 科达集团股份有限公司技术分公司 | A kind of advertisement bid optimization method based on eCPM conjugate gradient descent method fitting technique |
| CN110135904A (en) * | 2019-05-10 | 2019-08-16 | 四川合锐信息技术有限公司 | A kind of automatically bid optimization method and system based on streaming computing |
| CN112016947A (en) * | 2019-05-31 | 2020-12-01 | 百度在线网络技术（北京）有限公司 | Advertisement space distribution method, device, equipment and storage medium |
| CN112016947B (en) * | 2019-05-31 | 2023-08-15 | 百度在线网络技术（北京）有限公司 | Advertisement space distribution method, device, equipment and storage medium |
| CN110189181A (en) * | 2019-06-10 | 2019-08-30 | 深圳市云企汇财税顾问有限公司 | A kind of accounting system benefited for the advertisement browsing of Online Video APP |
| CN111598616A (en) * | 2020-05-11 | 2020-08-28 | 百度在线网络技术（北京）有限公司 | Method, device and equipment for realizing object set screening and storage medium |
| CN111598616B (en) * | 2020-05-11 | 2023-08-01 | 百度在线网络技术（北京）有限公司 | Method, device, equipment and storage medium for realizing object set screening |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111784400A (en) | 2020-10-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107194729A (en) | Advertisement price competing method, device, electronic installation and computer-readable medium | |
| Miralles-Pechuán et al. | A novel methodology for optimizing display advertising campaigns using genetic algorithms | |
| JP5081269B2 (en) | System and method for managing advertising campaigns on a network | |
| US20170098236A1 (en) | Exploration of real-time advertising decisions | |
| CN106971321A (en) | Marketing message method for pushing, device, equipment and storage medium | |
| EP4242955A1 (en) | User profile-based object recommendation method and device | |
| CN111210255B (en) | Advertisement pushing method and device and electronic equipment | |
| US20180204250A1 (en) | Predictive attribution-adjusted bidding for electronic advertisements | |
| US11410203B1 (en) | Optimized management of online advertising auctions | |
| US20080154662A1 (en) | System and method for generating a maximum utility slate of advertisements for online advertisement auctions | |
| Varian | Use and abuse of network effects | |
| CN111382359A (en) | Service strategy recommendation method and device based on reinforcement learning and electronic equipment | |
| CN111179030B (en) | Advertisement bidding method and device and electronic equipment | |
| CN112163886A (en) | Real-time bidding advertisement resource allocation method based on reinforcement learning | |
| KR20030045765A (en) | Apparatus and method for service trading system based on knowledge search system using questions and answers | |
| WO2020104806A1 (en) | Real-time bidding | |
| KR102571415B1 (en) | Method for controlling a marketing system for online direct transaction | |
| CN114298728A (en) | Data processing method and related device | |
| CN111199454B (en) | Real-time user conversion evaluation method and device and electronic equipment | |
| CN107527128B (en) | Resource parameter determination method and equipment for advertisement platform | |
| CN103890802A (en) | Methods and apparatus related to billing and accounting to establish asset value | |
| US20230089895A1 (en) | System and methods for bid optimization in real-time bidding | |
| CN114418609A (en) | Deal probability estimation method, storage medium, and program product | |
| CN113034183A (en) | Pricing processing method and device, electronic equipment and storage medium | |
| Kang | Sponsored link auctions with consumer search |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170922 |
|
| RJ01 | Rejection of invention patent application after publication |