CN110210882A - Promote position matching process and device, promotion message methods of exhibiting and device - Google Patents
Promote position matching process and device, promotion message methods of exhibiting and device Download PDFInfo
- Publication number
- CN110210882A CN110210882A CN201810236582.1A CN201810236582A CN110210882A CN 110210882 A CN110210882 A CN 110210882A CN 201810236582 A CN201810236582 A CN 201810236582A CN 110210882 A CN110210882 A CN 110210882A
- Authority
- CN
- China
- Prior art keywords
- promotion message
- popularization
- candidate
- value
- exposure
- 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
- G06Q30/0251—Targeted advertisements
-
- 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
- G06Q30/0277—Online advertisement
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Development Economics (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
This application involves a kind of popularization position matching process and devices, promotion message methods of exhibiting and device, promote position matching process and comprise determining that popularization position corresponding with user identifier and candidate popularization information;Obtain candidate popularization information estimates exposure clicking rate;Exposure clicking rate is estimated to be generated according to user characteristics corresponding with user identifier;Value is clicked according to the unit for estimating exposure clicking rate and candidate popularization information, determines the corresponding score of bidding of candidate popularization information;Score of bidding corresponding to screening from candidate popularization information meets the promotion message of popularization condition;Position will be promoted to match with promotion message source belonging to the promotion message that filters out;Position is promoted for showing the promotion message from matched promotion message source.Scheme provided by the present application enables popularization position resource to be fully utilized, and avoids wasting.
Description
Technical field
This application involves field of computer technology, more particularly to a kind of popularization position matching process, device and promotion message
Methods of exhibiting and device.
Background technique
With the development of computer technology, more and more promotion messages are presented to the user in online form.For example,
In line advertising display field, advertisement transaction platform would generally be according in DSP (party in request's platform, Demand-Side Platform)
The feature and demand of advertising information to be put distribute the SSP ad stream of (supplier's platform, Supply-Side Platform)
Resource is measured, the ad traffic resource to distribute to DSP can either bring advertisement value to DSP, and can bring to SSP accordingly
Ad traffic resource value.
It is that bit allocation will be promoted using fixed ratio mostly however, being directed to the popularization position of promotion message to be put at present
To each promotion message source, cause to promote the position wasting of resources.
Summary of the invention
Based on this, it is necessary to lead to the technology for promoting the position wasting of resources for the existing mode being allocated to popularization position
Problem provides a kind of popularization position matching process, device, computer readable storage medium and computer equipment.
A kind of popularization position matching process, which comprises
Determine popularization position corresponding with user identifier and candidate popularization information;
Obtain the candidate popularization information estimates exposure clicking rate;It is described estimate exposure clicking rate according to the user
Corresponding user characteristics are identified to generate;
Value is clicked according to the unit for estimating exposure clicking rate and the candidate popularization information, determines that the candidate pushes away
Guangxin ceases corresponding score of bidding;
Score of bidding corresponding to screening from the candidate popularization information meets the promotion message of popularization condition;
The popularization position is matched with promotion message source belonging to the promotion message that filters out;The popularization position is for opening up
Show the promotion message from matched promotion message source.
A kind of promotion message methods of exhibiting, which comprises
Determine popularization corresponding with user identifier position;
It receives and the matched promotion message in the popularization position;The promotion message from the popularization position is matched pushes away
Wide information source;The promotion message source is that corresponding with the user identifier and corresponding score of bidding meets popularization condition
Promotion message source belonging to candidate popularization information;The score of bidding is to estimate exposure point according to the candidate popularization information
The unit for hitting rate and the candidate popularization information clicks value determination;It is described estimate exposure clicking rate according to the user identifier
Corresponding user characteristics generate;
The matched promotion message of institute is shown in the popularization position.
A kind of popularization position coalignment, described device include:
Determining module, for determining popularization position corresponding with user identifier and candidate popularization information;
Module is obtained, estimates exposure clicking rate for obtain the candidate popularization information;It is described to estimate exposure clicking rate
It is generated according to user characteristics corresponding with the user identifier;
It bids score determining module, for estimating the unit point of exposure clicking rate and the candidate popularization information according to
Value is hit, determines the corresponding score of bidding of the candidate popularization information;
Screening module meets the popularization of popularization condition for score of bidding corresponding to the screening from the candidate popularization information
Information;
Matching module, for the popularization position to match with promotion message source belonging to the promotion message that filters out;Institute
It states and promotes position for showing the promotion message from matched promotion message source.
A kind of promotion message displaying device, described device include:
Position determining module is promoted, for determining popularization corresponding with user identifier position;
Promotion message receiving module, for receiving and the matched promotion message in the popularization position;The promotion message comes from
In with the matched promotion message source in popularization position;The promotion message source is corresponding and corresponding with the user identifier
Score of bidding meets promotion message source belonging to the candidate popularization information of popularization condition;The score of bidding is according to the time
Select the unit click value determination for estimating exposure clicking rate and the candidate popularization information of promotion message;It is described to estimate exposure point
Rate is hit to be generated according to user characteristics corresponding with the user identifier;
Display module, for showing the matched promotion message of institute in the popularization position.
A kind of computer readable storage medium is stored with computer program, when the computer program is executed by processor,
So that the step of processor executes above-mentioned popularization position matching process or promotes position methods of exhibiting.
A kind of computer equipment, including memory and processor, the memory are stored with computer program, the calculating
When machine program is executed by the processor, so that the processor executes above-mentioned popularization position matching process or promotes position methods of exhibiting
The step of.
Above-mentioned popularization position matching process, device, computer readable storage medium and computer equipment, promotion message displaying side
Method, device, computer readable storage medium and computer equipment, based on the candidate popularization information needle for belonging to different promotion message sources
Position will be promoted to the score of bidding for promoting position to match with promotion message source.Wherein, score is bidded according to candidate popularization information
Estimate exposure clicking rate determine, and estimate exposure clicking rate generated according to user characteristics corresponding with user identifier, it is pre- in this way
Candidate popularization information and the match condition of user can sufficiently be reflected by estimating exposure clicking rate.The unit of candidate popularization information clicks valence
Value can reflect value advantage of the promotion message in all candidate informations.It carries out promoting position matching based on score of bidding in this way
When, it can fully consider the valence of promotion message in demand and promotion message source of the user to promotion message in promotion message source
Value advantage enables popularization position resource to be fully utilized, avoids wasting.
Detailed description of the invention
Fig. 1 is the applied environment figure that position matching process is promoted in one embodiment;
Fig. 2 is the flow diagram that position matching process is promoted in one embodiment;
Fig. 3 is that position and the matched frame diagram in promotion message source will be promoted in one embodiment;
Fig. 4 is the discovery home tab of the App for downloading application program of the installation in one embodiment in terminal
Exemplary diagram;
Fig. 5 is that the unit of third type promotion message in one embodiment clicks the calculation method schematic diagram of value;
Fig. 6 is the flow diagram that the corresponding initial exposure weight in promotion message source is calculated in one embodiment;
Fig. 7 is the flow diagram that position matching process is promoted in a specific embodiment;
Fig. 8 is the flow diagram of promotion message methods of exhibiting in one embodiment;
Fig. 9 is the structural block diagram that position coalignment is promoted in one embodiment;
Figure 10 is the structural block diagram that position coalignment is promoted in another embodiment;
Figure 11 is the structural block diagram that position coalignment is promoted in another embodiment;
Figure 12 is the structural block diagram that promotion message shows device in one embodiment;
Figure 13 is the structural block diagram of computer equipment in one embodiment;
Figure 14 is the structural block diagram of computer equipment in one embodiment.
Specific embodiment
With reference to the accompanying drawings and embodiments, the application is further elaborated.
Fig. 1 is the applied environment figure that position matching process is promoted in one embodiment.Referring to Fig.1, the popularization position matching process
Applied to popularization position matching system.The popularization position matching system includes terminal 110 and server 120.Terminal 110 and server
120 pass through network connection.Terminal 110 specifically can be fixed terminal or mobile terminal, and mobile terminal specifically can be with mobile phone, plate
At least one of computer, laptop etc..Server 120 can be formed with the either multiple servers of independent server
Server cluster realize.
As shown in Fig. 2, in one embodiment, providing a kind of popularization position matching process.The present embodiment is mainly with the party
Method is applied to the server 120 in above-mentioned Fig. 1 to illustrate.Referring to Fig. 2, which specifically includes following step
It is rapid:
S202 determines popularization position corresponding with user identifier and candidate popularization information.
Wherein, promotion message is the online communication media for promoting object.Promotion message can be text, picture, keyword,
Ranking or video etc..Computer equipment can be used the forms such as character rendering, picture presentation or video presentation and to show institute to user
The object to be promoted.Promote object, for example can be application program to be downloaded, product to be purchased etc..
Candidate popularization information is the set of promotion message corresponding with user identifier.The corresponding time of different user identifiers
Select promotion message type can not identical, quantity can not be identical.In one embodiment, candidate popularization information belongs to difference
Promotion message source.Promotion message source is to promote position party in request.Promotion message source can have multiple types.For example, promotion message
Source includes advertisement side, game side, basic software side etc..
Position is promoted, is in the Internet media for showing the slot position of promotion message.Promoting position can be for downloading application
For showing the advertising slot position of each application program on the App (Application, computer applied algorithm) of program.For example, being used for
It downloads in the App home tab of application program for showing the recommendation list of " recommending application program ".Promoting position can also be society
The advertising slot position in application program is handed over, can be rolling advertisement column in portal website, context recommends tab bar, can be and search
Index the search result list etc. in holding up.
In one embodiment, terminal can be marked after detecting the exposure behavior of user by user identifier and with the user
Know corresponding popularization position to be allocated and be sent to server, server can obtain and the user after receiving the user identifier
Identify corresponding popularization position to be allocated and candidate popularization information.
The exposure behavior of user such as can be user and log in for downloading the behavior of the App of application program, or beat
Open the behavior of the App;It can also be that user browses the behavior of the social circle of social application program;It can also be the website that opens the doors
Webpage behavior, input behavior etc. of search key in a search engine.
User identifier is the character for unique identification user.For example it can be subscriber phone number or user account
Etc..In one embodiment, user identifier can also be terminal iidentification corresponding with any terminal.For example, for mobile whole
For end, user identifier can be mobile terminal IMEI (International Mobile Equipment Identity,
International mobile equipment identification number).
In one embodiment, server can get the exposure of user corresponding with user identifier from each the Internet media
User behavior characteristics are extracted in behavior, obtain the promotion message feature of each promotion message, it is special according to the user behavior extracted
Matching degree between the promotion message feature of each promotion message of seeking peace generates candidate popularization letter corresponding with the user identifier
Breath.
S204, obtain candidate popularization information estimates exposure clicking rate;Estimate exposure clicking rate according to user identifier pair
The user characteristics answered generate.
Wherein, estimating exposure clicking rate (Predicted Click Through Rate, PCTR) is candidate popularization information
The probability being clicked after being presented by popularization position corresponding with user identifier to user.
Specifically, server can be marked after getting candidate popularization information corresponding with user identifier according to the user
That knows that corresponding user characteristics obtain candidate popularization information estimates exposure clicking rate.Server can also be according to candidate popularization information
In each promotion message promotion message feature and the user characteristics come obtain candidate popularization information estimate exposure click
Rate.
User characteristics may include user's Figure Characteristics, user behavior characteristics and user characteristics and promotion message characteristic crossover
Assemblage characteristic.Wherein, user's Figure Characteristics include age of user corresponding with user identifier, gender, height, account information,
Network environment locating for the corresponding terminal of the user identifier, time environment etc..User behavior characteristics include the user identifier pair
The user answered exposes behavior in the history of candidate popularization information, and history exposes behavior such as the history point of candidate popularization information
Hit behavior, history corresponding with the object that candidate popularization information is promoted downloading behavior, historical search behavior and history consumption row
For etc..User characteristics and promotion message characteristic crossover assemblage characteristic refer between the candidate popularization information of excavation and user identifier
The composite attribute of each side's region feature, for example, the keyword of user's search with and the candidate popularization information content in word it is matched a
Number etc..It is generated due to estimating exposure clicking rate according to user characteristics corresponding with user identifier, can sufficiently reflect candidate popularization
The match condition of information and user.
Promotion message feature may include the content characteristic of promotion message, and content characteristic includes but is not limited to promotion message
Promotion message source belonging to caption text, price, sales volume, the attribute for promoting object, the classification for promoting object and promotion message
Etc..
In one embodiment, when the corresponding popularization object of candidate popularization information is application program to be downloaded, service
The candidate popularization information that device obtains estimates exposure clicking rate, can be used for estimating user corresponding with the user identifier and clicks displaying
Promotion message and download the probability of the application program.
S206 clicks value according to the unit for estimating exposure clicking rate and candidate popularization information, determines candidate popularization information
Corresponding score of bidding.
Wherein, it is candidate popularization information corresponding with user identifier that unit, which clicks value (Cost Per Click, CPC),
It promotes and primary institute's energy bring value is clicked by user after showing in position.The value for promoting position can be clicked with number of clicks and unit
The product of value carrys out quantization means, it may be assumed that
Income=n × cpc, wherein n indicates to promote the number that promotion message shown in position is clicked, cpc expression exhibition
The unit of the promotion message shown clicks value, and income indicates to promote the value of position.
In one embodiment, the corresponding unit of candidate popularization information for belonging to different promotion message sources clicks setting for value
Calibration will definitely be with not identical.The candidate popularization information for belonging to identical promotion message source, in the popularization object difference promoted, respectively
Self-corresponding unit clicks value can not also be identical.The unit click value of candidate popularization information can reflect promotion message and exist
Value advantage in all candidate informations.
It bids score, is that candidate popularization information popularization position corresponding with user identifier is subjected to matched reference value.It is candidate
The score of bidding of promotion message can be the product for estimating exposure clicking rate and unit click value of candidate popularization information, may be used also
Be estimate exposure clicking rate and unit click the linear weighted function that is calculated under respective weighting coefficient of value and.Based on competing
When valence score carries out promoting position matching, demand and popularization of the user to promotion message in promotion message source can be fully considered
The value advantage of promotion message in information source enables popularization position resource to be fully utilized, avoids wasting.
Specifically, server can obtain candidate popularization letter after getting candidate popularization information corresponding with user identifier
Cease it is corresponding single click on value, according to candidate popularization information estimate exposure clicking rate and unit clicks value, determine candidate
The score of bidding of promotion message.
In one embodiment, server can determine that candidate pushes away after getting the candidate popularization information with user identifier
The standard of value is clicked, according to the unit for the setting of different promotion message sources to determine in promotion message source belonging to the breath of Guangxin
The unit of each candidate popularization information clicks value.
In one embodiment, it includes: when candidate popularization is believed that the unit of candidate popularization information, which clicks the determination step of value,
When breath belongs to first kind promotion message, then the bid value that candidate popularization information single clicks on is obtained;Bid value is determined as waiting
The unit of promotion message is selected to click value.
Wherein, the promotion message source of first kind promotion message can be advertisement side.Specifically, when server determines candidate
When promotion message source belonging to promotion message is advertisement side, then obtaining advertisement side is the bid that the candidate popularization information single clicks on
Value, the unit that the bid value that will acquire is determined as candidate popularization information click value.For example, promoted in candidate popularization information
When popularization object is online shopping application program to be downloaded, it is advertisement side to the application on site that corresponding unit, which clicks value,
Program is downloaded the bid value once paid.
In one embodiment, it includes: when candidate popularization is believed that the unit of candidate popularization information, which clicks the determination step of value,
When breath belongs to Second Type promotion message, then obtains candidate popularization information and be clicked receipts of the rear downloaded application in specified duration
Benefit value mean value;Determine that the unit of candidate popularization information clicks value according to financial value mean value.
Wherein, the promotion message source of Second Type promotion message can be game side.Specifically, when server determines candidate
When promotion message source belonging to promotion message is game side, the popularization object that candidate popularization information is promoted is game to be downloaded
When application program, then obtaining candidate popularization information, to be clicked financial value of the rear institute's download games application program in specified duration equal
Value, the unit that the income mean value that will acquire is determined as the candidate popularization information click value.For example, server can be based on user's needle
Life Cycle forward price of the game application after being downloaded in 14 days is estimated to the history pay records of the game application
It is worth (Life-Time-Value 14, LTV14).M*LTV14 (m=1,2,3 ..., n) can be also used as the candidate popularization by server
The unit of information clicks value.
S208, score of bidding corresponding to screening from candidate popularization information meet the promotion message of popularization condition.
Specifically, server is after the score of bidding that candidate popularization information corresponding with user identifier has been determined, from candidate
The promotion message that corresponding score of bidding meets popularization condition is screened in promotion message.What the promotion message filtered out met pushes away
Wide condition can be score highest of bidding, and can be score of bidding and meets preset condition (ranking forefront) and estimate exposure rate most
Height can also be that score of bidding meets preset condition and unit click value is most high.
S210 will promote position and match with promotion message source belonging to the promotion message that filters out;Position is promoted for showing
Promotion message from matched promotion message source.
Specifically, server is in the promotion message for filtering out score of bidding from candidate popularization information and meeting popularization condition
Afterwards, it determines promotion message source belonging to the promotion message filtered out, popularization position corresponding with user identifier is pushed away with what is filtered out
Promotion message source belonging to the breath of Guangxin matches, and popularization position corresponding with user identifier can be shown believes from matched popularization
The promotion message in breath source matches to realize and will promote position based on score of bidding with promotion message source.
In one embodiment, it is further comprising the steps of to promote position matching process: receiving from matched promotion message
The promotion message in source;Received promotion message meets the sortord of bidding in matched promotion message source;Push institute is received to push away
Guangxin breath, the promotion message of push are used in the matched popularization position of institute show.
In the present embodiment, it receives and shows that the promotion message in matched popularization position is not necessarily bidding of filtering out
The highest promotion message of score, but by promotion message source according to the promotion message of internal sortord selection of bidding, thus
The separation of position matching mechanisms and promotion message source internal sort mechanism will be promoted by realizing.
In one embodiment, popularization position corresponding with user identifier can be promoted condition with the satisfaction filtered out by server
Promotion message matched so that promoting position for showing the promotion message filtered out.
As shown in figure 3, for position and the matched frame diagram in promotion message source will be promoted in one embodiment.Wherein, position is promoted
For the slot position of the recommendation list in the home tab of the App for downloading application program, it will be understood that show in recommendation list
Promotion message in slot position is for promoting corresponding App.Promotion message source includes three service access sides, is advertising business respectively
Access side, game service access side and basic software service access side.Each corresponding promotion message in service access side is promoted
Popularization object be advertisement App, game App and basic software App respectively.It opens a terminal in slot position distribution service detection to user
After the home tab of the App for downloading application program of upper installation, for each slot in the recommendation list in home tab
Position, slot position distribute service acquisition candidate popularization information corresponding with user identifier, calculate each candidate popularization information and calculate accordingly
Score of bidding, service access side belonging to the highest promotion message of score of bidding is filtered out from candidate popularization information, by slot
Bit allocation is to the service access side.Service access can distribute the slot position to any popularization according to oneself internal ordering mechanism
Information, terminal shows the promotion message in slot position, for promoting corresponding App, such as A, B and C.
As shown in figure 4, the discovery homepage of the App for downloading application program for the installation in terminal in one embodiment
The exemplary diagram of label.Referring to Fig. 4, illustrated in exemplary diagram by the multiple slot positions of recommendation list, each slot position is corresponding for showing
Promotion message, with the corresponding App of popularization.
Above-mentioned popularization position matching process is directed to based on the candidate popularization information for belonging to different promotion message sources and promotes the competing of position
Valence score will be promoted position and be matched with promotion message source.Wherein, score of bidding according to candidate popularization information estimates exposure point
It hits rate to determine, and estimates exposure clicking rate and generated according to user characteristics corresponding with user identifier, estimate exposure clicking rate in this way
It can sufficiently reflect candidate popularization information and the match condition of user.The unit of candidate popularization information is clicked to be worth can reflect and be pushed away
Guangxin ceases the value advantage in all candidate informations.When carrying out promoting position matching based on score of bidding in this way, can sufficiently it examine
The value advantage for considering promotion message in demand and promotion message source of the user to promotion message in promotion message source, so that pushing away
Wide position resource can be fully utilized, and avoid wasting.
In one embodiment, obtain candidate popularization information estimates exposure clicking rate, comprising: obtains and user identifier pair
The user characteristics answered;Extract the promotion message feature of candidate popularization information;According to user characteristics and promotion message feature, generates and wait
Select the feature vector of promotion message;Mould is predicted by exposure clicking rate corresponding with the affiliated promotion message source of candidate popularization information
Type, and the feature vector according to candidate popularization information, determine candidate popularization information estimates exposure clicking rate.
Wherein, the explanation about user characteristics corresponding with user identifier, the promotion message feature of candidate popularization information is asked
With reference to explanation above-mentioned, details are not described herein.Feature vector is between candidate popularization information user corresponding with user identifier
It contacts, the mathematical expression form of each feature of the candidate popularization information.The dimension of feature vector and referenced user characteristics,
The number of promotion message feature is related.
Specifically, server is in the promotion message for getting corresponding with user identifier user characteristics and candidate popularization information
After feature, by each Feature Mapping to the distributed space vector (distributed representation), generate with
Feature vector is input to belonging to trained and candidate popularization information by the corresponding feature vector of candidate popularization information
In the corresponding exposure clicking rate prediction model in promotion message source, by calculating, export the candidate popularization information estimates exposure point
Hit rate.
In one embodiment, exposure clicking rate prediction model can be machine learning model.Such as logistic regression
(Logistic Regression, LR) model.
In the present embodiment, user characteristics corresponding with user identifier are expressed by way of feature vector, promote letter
Feature is ceased, can sufficiently reflect candidate popularization information and the match condition of user, be carried out by trained machine learning model
Predict that is obtained estimates exposure clicking rate, it will be able to which being truly reflected the candidate popularization information may be by the use with the user identifier
The probability that family is clicked.
In one embodiment, the training step for exposing clicking rate prediction model includes: to obtain to belong to promotion message source
Historical user's feature, history promotion message feature corresponding to history promotion message and sample label;According to historical user's feature
With history promotion message feature, generation belongs to history feature vector corresponding to the promotion message in promotion message source;By history spy
Vector input exposure clicking rate prediction model is levied, history corresponding with history feature vector is obtained and exposes clicking rate;When basis is gone through
History exposes clicking rate and sample label determines when being unsatisfactory for training stop condition, and the model of adjustment exposure clicking rate prediction model is joined
It counts and continues to train, until deconditioning when meeting training stop condition.
Specifically, for each promotion message source, server obtains the history promotion message for belonging to corresponding promotion message source,
It is pre- to be obtained by model training for the corresponding exposure clicking rate in each promotion message source using history promotion message as sample data
Survey model.Wherein, history promotion message includes positive sample and negative sample, and positive sample is the sample being clicked, and negative sample is not
The sample being clicked.Positive sample and negative sample have corresponded to different sample labels.
In one embodiment, server is after the history promotion message for getting corresponding promotion message source, from history
Historical user's feature, history promotion message feature are extracted in promotion message, to generate the popularization letter for belonging to each promotion message source
The history feature vector is input to exposure clicking rate prediction model, obtains history exposure by the corresponding history feature vector of breath
Clicking rate.
In one embodiment, the history of acquisition can be exposed clicking rate and is compared with preset threshold by server, then will
Comparison result is compared with the sample label of promotion message, to determine whether history exposure clicking rate meets truth.
For example, preset threshold is 0.5, the true sample label of positive sample is 1, and the true sample label of negative sample is 0, when exposure is clicked
When history corresponding with the history promotion message exposure clicking rate of rate prediction model output is less than 0.5, then the history is promoted and believed
The sample label of breath prediction is 0, which is compared with true sample label, identical, is judged to predicting
Correctly, different then be determined as prediction error;Similarly, when exposure clicking rate prediction model output it is corresponding with history promotion message
History exposure clicking rate be greater than or equal to 0.5 when, then to the history promotion message prediction sample label be 1, by the prediction
Sample label is compared with true sample label, identical, be judged to predicting it is correct, it is different then be determined as prediction error.
In one embodiment, the statistics available history promotion message in sample data of server carries out clicking rate prediction
Accuracy rate adjusts model parameter, until then deconditioning, obtains pushing away with each history when accuracy rate meets deconditioning condition
The exposure clicking rate prediction model of wide information source.
In the above-described embodiments, by obtaining the history promotion message in each promotion message source, from true historical data
In extract historical user's feature and history promotion message feature carrys out training pattern, it is ensured that the exposure in each promotion message source
The accuracy rate that clicking rate prediction model is estimated.
In one embodiment, it includes: when candidate popularization is believed that the unit of candidate popularization information, which clicks the determination step of value,
When breath belongs to third type promotion message, it is determined that the estimation unit of third type promotion message clicks value;Estimate unit point
Hitting value is determined when the financial value total value that the financial value total value of the first popularization page and second promote the page is equal;First pushes away
The popularization position of the wide page is only used for showing third type promotion message;Second promotes the page including showing third type promotion message
The popularization position promoted position and show non-third type promotion message;Estimation unit is clicked into value and is determined as candidate popularization information
Unit click value.
Wherein, the promotion message source of third type promotion message can be basic software side.Specifically, when server determines
Promotion message source belonging to candidate popularization information is basic software side, and the popularization object that candidate popularization information is promoted is to be downloaded
Base application when, which does not have unit to click value, but there are users to click the candidate popularization information
The case where downloading promoted application program afterwards.When the candidate popularization information is illustrated in the popularization position for promoting the page, meeting
The value for the popularization position for showing non-third type promotion message in the page is caused to be lost, so that it may according to this portion lost
Value calculation is divided to click value from the estimation unit of the third type promotion message of basic software side.
Specifically, the popularization position of the first popularization page is only used for showing third type promotion message, then first promotes page
The financial value total value in face is the sum of the value of all popularization positions for showing third type promotion message in the page;Second promotes page
The popularization position in face is used to show the popularization position of third type promotion message promoted position and show non-third type promotion message, that
Second promote the page financial value total value be the page in it is all show third type promotion messages popularization positions values with
The sum of the value of all popularization positions for showing non-third type promotion message.In the financial value total value and second that first promotes the page
When the financial value total value of the popularization page is equal, so that it may be calculated in the second popularization page and show non-third type promotion message
Popularization position lose value, value is lost according to this estimation unit of third type promotion message can be obtained and click value, clothes
The estimation unit can be clicked the unit click value that value is determined as candidate popularization information by business device.
In the present embodiment, by obtaining non-third type promotion message displaying in the popularization page of identical financial value total value
In value lose, can express third type promotion message in the popularization page promote position in show after potential valence
Value.
In one embodiment, it includes: when candidate popularization is believed that the unit of candidate popularization information, which clicks the determination step of value,
When breath belongs to third type promotion message, then the first the first number of clicks for promoting third type promotion message in the page is obtained;
The popularization position of the first popularization page is only used for showing third type promotion message;It obtains second and promotes third type popularization in the page
The financial value of second number of clicks of information and non-third type promotion message;The second popularization page includes showing that third type pushes away
The popularization position of Guangxin breath promoted position and show non-third type promotion message;The financial value that will acquire is divided by the first number of clicks
With the difference of the second number of clicks, the unit for obtaining candidate popularization information clicks value.
Specifically, it can derive that the unit of third type promotion message clicks value by following formula:
Click1×cpc3=Click2×cpc3+ Income (formula 1)
1 left side of formula represents the financial value that position is promoted in the first popularization page, and the right represents promotes in the second popularization page
The financial value of position.Click1Indicate the first number of clicks;cpc3Indicate that the unit of third type promotion message clicks value;
Income indicates the financial value that the popularization position of non-third type promotion message is shown in the second popularization page;Click2Indicate second
Number of clicks.It can derive that the unit of third type promotion message clicks value according to formula 1:
In one embodiment, when the promoted object of third type popularization information is application program to be downloaded, then formula 2
In Click1Indicate the first download time that each application program to be downloaded is downloaded in the first popularization page;Income
Indicate the financial value that the popularization position of non-third type promotion message is shown in the second popularization page;Click2It indicates to promote second
The second number of clicks that each application program to be downloaded is downloaded in the page.It can be obtained by by formula 2 and belong to third class
The promotion message of type is corresponding when promoting application program to be downloaded to single click on value.
In one embodiment, the unit click value for belonging to third type promotion message can be identical.Server can be preparatory
The unit that storage belongs to the candidate popularization information of third type promotion message clicks value, is determining belonging to candidate popularization information
Promotion message source after, be directly obtained corresponding unit and click value.
As shown in figure 5, clicking the calculation method signal of value for the unit of third type promotion message in one embodiment
Figure.Referring to Fig. 5, the object that the promotion message of two popularization pages is promoted is application program to be downloaded;One pushes away wherein
It is the pure flow for being distributed as guiding in the wide page, indicates that the popularization page is only used for showing and belong to third type promotion message, only
Bring the download of application program to be downloaded;Another is promoted in the page, is the flow that net income is guiding, is indicated the popularization page
Face can show the download and income for belonging to non-third type promotion message, while bringing application program to be downloaded;Two popularizations
The Flow Value conservation of the page, the financial value for representing the popularization position for the displaying promotion message that two are promoted in the page are equal;This two
A corresponding Flow Value of the page of promoting can be by the download and estimation unit point of the application program to be downloaded of popularization
Value is hit to determine, derives that the unit for belonging to third type promotion message clicks valence so as to click value according to estimation unit
Value.
In the present embodiment, the principle equal according to the same financial value for promoting the page, can determine after transformation and belongs to
The acquisition modes of the candidate popularization information of third type promotion message can determine phase according to the number of clicks of candidate popularization information
The unit answered clicks value.
In one embodiment, score of bidding is that basis estimates exposure clicking rate, unit clicks value and candidate popularization letter
The exposure weight in promotion message source belonging to ceasing determines.
For example, the score of bidding of candidate popularization information, which can be estimating for candidate popularization information, exposes clicking rate, unit point
The product for hitting the exposure weight in value and the affiliated promotion message source of candidate popularization information can also be and estimate exposure clicking rate, list
Site hit value, the linear weighted function that the exposure weight in affiliated promotion message source is calculated under respective weighting coefficient and.
In one embodiment, the determination step of the exposure weight in the affiliated promotion message source of candidate popularization information includes: system
Count the exposure accounting in the affiliated promotion message source of candidate popularization information after previous popularization position matches;According to the exposure accounting of statistics, really
The settled exposure weight for carrying out promoting candidate's promotion message affiliated promotion message source when position matches.
Wherein, exposure accounting is to promote promotion message shown in the page and belong to the popularization position in the same promotion message source to account for
All ratios for promoting position.Specifically, server can count each promotion message source and push away previous after previous popularization position matches
Exposure accounting after wide position matching promote when position matches belonging to candidate's promotion message using previous exposure accounting as when secondary
The exposure weight in promotion message source.
For example, promoting popularization position corresponding with user identifier in the page has 10, and when previous popularization position matches, needle
To this 10 popularization positions, server can all obtain candidate popularization information corresponding with user identifier, and according to candidate popularization information
Estimate exposure clicking rate, unit click value and exposure weight product, the score of bidding of candidate popularization information is obtained, from candidate
The promotion message that score of bidding meets popularization condition is screened in promotion message, by each popularization position and the promotion message institute filtered out
The promotion message source of category is matched.For example there are 3 to be matched to belonging to first kind promotion message in this 10 popularization positions and push away
Wide information source has 2 popularization positions to be matched to promotion message source belonging to Second Type promotion message, has 5 popularization positions to be matched to
Promotion message source belonging to third type promotion message, then the corresponding exposure accounting in each promotion message source is 0.3,0.2
With 0.5.When secondary popularization position matches, the exposure weight for the promotion message for belonging to each promotion message source can be used previous affiliated
The exposure accounting in promotion message source.And so on, when promoting position matching next time, the exposure of the promotion message in each promotion message source
Weight can be used when time exposure accounting in affiliated promotion message source.
In the present embodiment, score of bidding is also related with exposure weight, and exposes weight and push away with promotion message source previous
Exposure accounting after wide bit allocation is related, can be realized each promotion message source belonging to candidate popularization information in the popularization page
Promote the fair competition of position.For example, only candidate information estimate exposure clicking rate and unit click value get a promotion it
Afterwards, the exposure accounting in affiliated promotion message source can just get a promotion, higher so as to obtain in popularization next time position competition
Exposure weight, realize the benign competition in each promotion message source.
In one embodiment, different promotion message sources are more uniformly distributed to make to promote position, and show from
The promotion message in different promotion message sources, server can obtain the corresponding initial exposure weight in each promotion message source, each
The initial exposure weight in promotion message source promoting the position matched initial stage, and promoting position can be uniformly distributed to not
With promotion message source.
In one embodiment, the determination step of the exposure weight in the affiliated promotion message source of candidate popularization information further include:
Obtain default exposure weight;By the exposure weight in the affiliated promotion message source of candidate popularization information from default exposure weight iteration tune
It is whole, and carried out after promoting position matching according to the exposure weight of current iteration, obtain the affiliated promotion message source of candidate popularization information
Accounting is exposed, stops iteration adjustment when exposing accounting and meeting balanced conditions of exposure, obtains promoting belonging to candidate popularization information
The initial exposure weight of information source.
Specifically, presetting exposure weight can be the exposure weight in each promotion message source of random initializtion.It is each to promote
The sum of corresponding default exposure weight of information source is 1.Server can calculate each history candidate popularization from default exposure weight
The score of bidding of information counts each promotion message under current exposure weight after carrying out popularization bit allocation based on the score of bidding
The exposure accounting in source adjusts preset exposure weight with predetermined gradient if exposure accounting is unsatisfactory for balanced conditions of exposure
It is whole, until the exposure accounting in each promotion message source meets balanced conditions of exposure, using preset exposure weight adjusted as
The initial exposure weight in promotion message source.
As shown in fig. 6, to calculate the process signal of the corresponding initial exposure weight in promotion message source in one embodiment
Figure.Server can the corresponding history of candidate popularization information to history promote sample and sample.History is promoted sample and be can be
Popularization data when the candidate popularization information is demonstrated in other popularization pages.In the pre- of each promotion message source of initialization
If calculating the exposure accounting in each promotion message source under exposure weight, until it is relatively stable by the exposure weight of iteration adjustment,
And corresponding exposure accounting meets balanced popularization condition, then will terminate to adjust, using exposure weight adjusted as each popularization
The initial exposure weight of information source.
In the present embodiment, it is united based on the corresponding popularization data of the candidate popularization information of history to preset exposure weight
Meter exposure accounting to adjust exposure accounting according to predetermined gradient obtains that the position matched initial stage can be being promoted equably
Initial exposure weight of the bit allocation to different promotion message sources will be promoted.
As shown in fig. 7, to promote the flow diagram of position matching process in a specific embodiment.Popularization position matching
Method specifically includes the following steps:
S702 determines popularization position corresponding with user identifier and candidate popularization information.
S704 obtains user characteristics corresponding with user identifier.
S706 extracts the promotion message feature of candidate popularization information.
S708 generates the feature vector of candidate popularization information according to user characteristics and promotion message feature.
S710, by exposure clicking rate prediction model corresponding with the affiliated promotion message source of candidate popularization information, and foundation
The feature vector of candidate popularization information, determine candidate popularization information estimates exposure clicking rate.
S712 then obtains candidate popularization information and single clicks on when candidate popularization information belongs to first kind promotion message
Bid value;The unit that bid value is determined as candidate popularization information is clicked into value.
S714 is then obtained after candidate popularization information is clicked when candidate popularization information belongs to Second Type promotion message
Financial value mean value of the downloaded application in specified duration;Determine that the unit of candidate popularization information clicks valence according to financial value mean value
Value.
S716 then obtains first and promotes third class in the page when candidate popularization information belongs to third type promotion message
First number of clicks of type promotion message;The popularization position of the first popularization page is only used for showing third type promotion message;It obtains
Second promotes the second number of clicks of third type promotion message and the financial value of non-third type promotion message in the page;Second
Promoting the page includes the popularization position promoted position and show non-third type promotion message for showing third type promotion message;It will obtain
Divided by the difference of the first number of clicks and the second number of clicks, the unit for obtaining candidate popularization information is clicked to be worth the financial value taken.
S718 counts the previous exposure accounting for promoting the affiliated promotion message source of candidate popularization information after position matches.
S720 is determined according to the exposure accounting of statistics and promote popularization belonging to candidate's promotion message when position matches when secondary
The exposure weight of information source.
S722 exposes belonging to clicking rate, the unit of candidate popularization information click value and candidate popularization information according to estimating
Promotion message source exposure weight, determine the corresponding score of bidding of candidate popularization information.
S724, score of bidding corresponding to screening from candidate popularization information meet the promotion message of popularization condition.
S726 will promote position and match with promotion message source belonging to the promotion message that filters out.
S728 receives the promotion message from matched promotion message source;Received promotion message, which meets to match, to be pushed away
The sortord of bidding of wide information source.
S730, the received promotion message of push institute, the promotion message of push are used in the matched popularization position of institute show.
Above-mentioned popularization position matching process is directed to based on the candidate popularization information for belonging to different promotion message sources and promotes the competing of position
Valence score will be promoted position and be matched with promotion message source.Wherein, score of bidding according to candidate popularization information estimates exposure point
It hits rate to determine, and estimates exposure clicking rate and generated according to user characteristics corresponding with user identifier, estimate exposure clicking rate in this way
It can sufficiently reflect candidate popularization information and the match condition of user.The unit of candidate popularization information is clicked to be worth can reflect and be pushed away
Guangxin ceases the value advantage in all candidate informations.When carrying out promoting position matching based on score of bidding in this way, can sufficiently it examine
The value advantage for considering promotion message in demand and promotion message source of the user to promotion message in promotion message source, so that pushing away
Wide position resource can be fully utilized, and avoid wasting.
Fig. 7 is the flow diagram that position matching process is promoted in one embodiment.Although should be understood that the stream of Fig. 7
Each step in journey figure is successively shown according to the instruction of arrow, but these steps are not inevitable according to the suitable of arrow instruction
Sequence successively executes.Unless expressly stating otherwise herein, there is no stringent sequences to limit for the execution of these steps, these steps
It can execute in other order.Moreover, at least part step in Fig. 7 may include multiple sub-steps or multiple ranks
Section, these sub-steps or stage are not necessarily to execute completion in synchronization, but can execute at different times, this
The execution sequence in a little step perhaps stage be also not necessarily successively carry out but can be with other steps or other steps
Sub-step or at least part in stage execute in turn or alternately.
As shown in figure 8, in one embodiment, providing a kind of promotion message methods of exhibiting.The present embodiment is mainly with this
Method is applied to the terminal 110 in above-mentioned Fig. 1 to illustrate.Referring to Fig. 8, which specifically includes as follows
Step:
S802 determines popularization corresponding with user identifier position.
Wherein, position is promoted, is in the Internet media for showing the slot position of promotion message.Promoting position can be for downloading
For showing the advertising slot position of each application program on the App (Application, computer applied algorithm) of application program.For example,
For showing the recommendation list of " recommending application program " in App home tab for downloading application program.Promoting position can be with
It is the advertising slot position in social application program, can be rolling advertisement column in portal website, context recommends tab bar, it can be with
It is the search result list etc. in search engine.
In one embodiment, terminal can after detecting the exposure behavior of user, obtain it is corresponding with user identifier to
The popularization position of distribution.User identifier and popularization position to be allocated corresponding with the user identifier are sent to server, server
Corresponding with the user identifier popularization position and candidate popularization information to be allocated can be obtained after receiving the user identifier.
The exposure behavior of user such as can be user and log in for downloading the behavior of the App of application program, or beat
Open the behavior of the App;It can also be that user browses the behavior of the social circle of social application program;It can also be the website that opens the doors
Webpage behavior, input behavior etc. of search key in a search engine.
S804 is received and is promoted the matched promotion message in position;Promotion message from promote the matched promotion message in position
Source;Promotion message source is the candidate popularization information institute that corresponding with user identifier and corresponding score of bidding meets popularization condition
The promotion message source of category;It bids score, is the list for estimating exposure clicking rate and candidate popularization information according to candidate popularization information
Site is hit value and is determined;Exposure clicking rate is estimated to be generated according to user characteristics corresponding with user identifier.
Wherein, promotion message is the online communication media for promoting object.Promotion message can be text, picture, keyword,
Ranking or video etc..Computer equipment can be used the forms such as character rendering, picture presentation or video presentation and to show institute to user
The object to be promoted.Promote object, for example can be application program to be downloaded, product to be purchased etc..
Candidate popularization information is the set of promotion message corresponding with user identifier.The corresponding time of different user identifiers
Select promotion message type can not identical, quantity can not be identical.In one embodiment, candidate popularization information belongs to difference
Promotion message source.Promotion message source is to promote position party in request.Promotion message source can have multiple types.For example, promotion message
Source includes advertisement side, game side, basic software side etc..
Estimate exposure clicking rate (Predicted Click Through Rate, PCTR) be candidate popularization information by with
User identifier is corresponding to promote the probability being clicked after position is presented to user.
It is candidate popularization information in popularization corresponding with user identifier that unit, which clicks value (Cost Per Click, CPC),
Primary institute's energy bring value is clicked by user after showing in position.The value for promoting position can be clicked with number of clicks and unit to be worth
Product carry out quantization means, it may be assumed that
Income=n × cpc, wherein n indicates to promote the number that promotion message shown in position is clicked, cpc expression exhibition
The unit of the promotion message shown clicks value, and income indicates to promote the value of position.
It bids score, is that candidate popularization information popularization position corresponding with user identifier is subjected to matched reference value.It is candidate
The score of bidding of promotion message can be the product for estimating exposure clicking rate and unit click value of candidate popularization information, may be used also
Be estimate exposure clicking rate and unit click the linear weighted function that is calculated under respective weighting coefficient of value and.Based on competing
When valence score carries out promoting position matching, demand and popularization of the user to promotion message in promotion message source can be fully considered
The value advantage of promotion message in information source enables popularization position resource to be fully utilized, avoids wasting.
Specifically, after obtaining corresponding with the user identifier popularization position and candidate popularization information to be allocated, each time is calculated
The score of bidding for selecting promotion message determines according to each candidate score of bidding for providing information and promotes the matched promotion message in position
Source, and receive from matched promotion message source promotion message.
S806 shows the matched promotion message of institute in promoting position.
In one embodiment, the determination step for estimating exposure clicking rate of candidate popularization information includes: acquisition and user
Identify corresponding user characteristics;Extract the promotion message feature of candidate popularization information;According to user characteristics and promotion message feature,
Generate the feature vector of candidate popularization information;It is pre- by exposure clicking rate corresponding with the affiliated promotion message source of candidate popularization information
Model, and the feature vector according to candidate popularization information are surveyed, determine candidate popularization information estimates exposure clicking rate.
In one embodiment, the training step for exposing clicking rate prediction model includes: to obtain to belong to promotion message source
Historical user's feature, history promotion message feature corresponding to history promotion message and sample label;According to historical user's feature
With history promotion message feature, generation belongs to history feature vector corresponding to the promotion message in promotion message source;By history spy
Vector input exposure clicking rate prediction model is levied, history corresponding with history feature vector is obtained and exposes clicking rate;When basis is gone through
History exposes clicking rate and sample label determines when being unsatisfactory for training stop condition, and the model of adjustment exposure clicking rate prediction model is joined
It counts and continues to train, until deconditioning when meeting training stop condition.
In one embodiment, it includes: when candidate popularization is believed that the unit of candidate popularization information, which clicks the determination step of value,
When breath belongs to first kind promotion message, then the bid value that candidate popularization information single clicks on is obtained;Bid value is determined as waiting
The unit of promotion message is selected to click value.
In one embodiment, it includes: when candidate popularization is believed that the unit of candidate popularization information, which clicks the determination step of value,
When breath belongs to Second Type promotion message, then obtains candidate popularization information and be clicked receipts of the rear downloaded application in specified duration
Benefit value mean value;Determine that the unit of candidate popularization information clicks value according to financial value mean value.
In one embodiment, it includes: when candidate popularization is believed that the unit of candidate popularization information, which clicks the determination step of value,
When breath belongs to third type promotion message, it is determined that the estimation unit of third type promotion message clicks value;Estimate unit point
Hitting value is determined when the financial value total value that the financial value total value of the first popularization page and second promote the page is equal;First pushes away
The popularization position of the wide page is only used for showing third type promotion message;Second promotes the page including showing third type promotion message
The popularization position promoted position and show non-third type promotion message;Estimation unit is clicked into value and is determined as candidate popularization information
Unit click value.
In one embodiment, it includes: when candidate popularization is believed that the unit of candidate popularization information, which clicks the determination step of value,
When breath belongs to third type promotion message, it is financial value divided by the first number of clicks that the unit of candidate popularization information, which clicks value,
With the difference of the second number of clicks;Financial value is the financial value of non-third type promotion message in the second popularization page;First clicks
Number is the number of clicks of third type promotion message in the first popularization page;Second number of clicks is in the second popularization page
The number of clicks of third type promotion message;The popularization position of the first popularization page is only used for showing third type promotion message;The
The two popularization pages include the popularization position promoted position and show non-third type promotion message for showing third type promotion message.
In one embodiment, score of bidding is that basis estimates exposure clicking rate, unit clicks value and candidate popularization letter
The exposure weight in promotion message source belonging to ceasing determines;The determination step of the exposure weight in the affiliated promotion message source of candidate popularization information
It include: the previous exposure accounting for promoting the affiliated promotion message source of candidate popularization information after position matches of statistics;According to the exposure of statistics
Accounting is determined when the secondary exposure weight for carrying out promoting candidate's promotion message affiliated promotion message source when position matches.
In one embodiment, the determination step of the exposure weight in the affiliated promotion message source of candidate popularization information further include:
Obtain default exposure weight;By the exposure weight in the affiliated promotion message source of candidate popularization information from default exposure weight iteration tune
It is whole, and carried out after promoting position matching according to the exposure weight of current iteration, obtain the affiliated promotion message source of candidate popularization information
Accounting is exposed, stops iteration adjustment when exposing accounting and meeting balanced conditions of exposure, obtains promoting belonging to candidate popularization information
The initial exposure weight of information source.
Above-mentioned promotion message methods of exhibiting is directed to based on the candidate popularization information for belonging to different promotion message sources and promotes position
Score of bidding will be promoted position and be matched with promotion message source.Wherein, score of bidding according to candidate popularization information estimates exposure
Clicking rate determines, and estimates exposure clicking rate and generated according to user characteristics corresponding with user identifier, estimates exposure in this way and clicks
Rate can sufficiently reflect candidate popularization information and the match condition of user.The unit of candidate popularization information is clicked value and be can reflect
Value advantage of the promotion message in all candidate informations.It, can be abundant when carrying out promoting position matching based on score of bidding in this way
Consider the value advantage of promotion message in demand and promotion message source of the user to promotion message in promotion message source, so that
Promoting position resource can be fully utilized, and avoid wasting.
As shown in figure 9, in one embodiment, providing a kind of popularization position coalignment 900, which includes: true
Cover half block 902 obtains module 904, score determining module 906 of bidding, screening module 908 and matching module 910, in which:
Determining module 902, for determining popularization position corresponding with user identifier and candidate popularization information.
Module 904 is obtained, estimates exposure clicking rate for obtain candidate popularization information;Estimate exposure clicking rate according to
The corresponding user characteristics of user identifier generate.
It bids score determining module 906, for clicking valence according to the unit for estimating exposure clicking rate and candidate popularization information
Value, determines the corresponding score of bidding of candidate popularization information.
Screening module 908 meets the popularization of popularization condition for score of bidding corresponding to the screening from candidate popularization information
Information.
Matching module 910 matches for will promote position with promotion message source belonging to the promotion message that filters out;It promotes
Position is for showing the promotion message from matched promotion message source.
In one embodiment, as shown in Figure 10, module 904 is obtained to specifically include user characteristics acquisition module 1002, push away
Guangxin ceases characteristic extracting module 1004, feature vector generation module 1006 and estimates exposure clicking rate prediction module 1008.Wherein,
User characteristics obtain module 1002 for obtaining user characteristics corresponding with user identifier;Promotion message characteristic extracting module 1004
For extracting the promotion message feature of candidate popularization information;Feature vector generation module 1006 is used for according to user characteristics and popularization
Information characteristics generate the feature vector of candidate popularization information;Estimate exposure clicking rate prediction module 1008 be used for by with candidate
The corresponding exposure clicking rate prediction model in the affiliated promotion message source of promotion message, and the feature vector according to candidate popularization information,
Determine candidate popularization information estimates exposure clicking rate.
In one embodiment, it estimates in exposure clicking rate prediction module 1008 and exposes clicking rate prediction model training step
Include: obtain belong to historical user's feature, history promotion message feature corresponding to the history promotion message in promotion message source and
Sample label;According to historical user's feature and history promotion message feature, the promotion message institute that generation belongs to promotion message source is right
The history feature vector answered;History feature vector is inputted into exposure clicking rate prediction model, is obtained corresponding to history feature vector
History expose clicking rate;When being unsatisfactory for training stop condition according to history exposure clicking rate and sample label are determining, adjustment
It exposes the model parameter of clicking rate prediction model and continues to train, until deconditioning when meeting training stop condition.
In one embodiment, module 904 is obtained to be also used to when candidate popularization information belongs to first kind promotion message,
Then obtain the bid value that candidate popularization information single clicks on;The unit that bid value is determined as candidate popularization information is clicked into value.
In one embodiment, module 904 is obtained to be also used to when candidate popularization information belongs to Second Type promotion message,
It then obtains candidate popularization information and is clicked financial value mean value of the rear downloaded application in specified duration;It is true according to financial value mean value
The unit for determining candidate popularization information clicks value.
In one embodiment, module 904 is obtained to be also used to when candidate popularization information belongs to third type promotion message,
Then determine that the estimation unit of third type promotion message clicks value;It is the receipts for promoting the page first that estimation unit, which clicks value,
It is determined when benefit value total value and the equal financial value total value of the second popularization page;The popularization position of the first popularization page is only used for displaying the
Three type promotion messages;The second popularization page includes showing the popularization position of third type promotion message and showing that non-third type pushes away
The popularization position of Guangxin breath;It will estimate that unit clicks the unit click value that value is determined as candidate popularization information.
In one embodiment, module 904 is obtained to be also used to when candidate popularization information belongs to third type promotion message,
Then obtain the first the first number of clicks for promoting third type promotion message in the page;The popularization position of the first popularization page is only used for
Show third type promotion message;Obtain the second the second number of clicks and non-third for promoting third type promotion message in the page
The financial value of type promotion message;The second popularization page includes popularization position and the non-third of displaying for showing third type promotion message
The popularization position of type promotion message;The financial value that will acquire is waited divided by the difference of the first number of clicks and the second number of clicks
The unit of promotion message is selected to click value.
In one embodiment, score of bidding is that basis estimates exposure clicking rate, unit clicks value and candidate popularization letter
The exposure weight in promotion message source belonging to ceasing determines;After score determining module 906 of bidding is also used to count previous popularization position matching
The exposure accounting in the affiliated promotion message source of candidate popularization information;According to the exposure accounting of statistics, determines and carry out promoting position when secondary
The exposure weight in the affiliated promotion message source of timing candidate popularization information.
In one embodiment, score determining module 906 of bidding is also used to obtain default exposure weight;Candidate popularization is believed
Cease belonging to promotion message source exposure weight from default exposure weight iteration adjustment, and according to the exposure weight of current iteration into
After row promotes position matching, the exposure accounting in the affiliated promotion message source of candidate popularization information is obtained, until exposure accounting meets equilibrium
Stop iteration adjustment when conditions of exposure, obtains the initial exposure weight in the affiliated promotion message source of candidate popularization information.
In one embodiment, as shown in figure 11, promoting position coalignment 900 further includes promotion message receiving module 1102
With promotion message pushing module 1104, in which: promotion message receiving module 1102 is for receiving from matched promotion message
The promotion message in source;Received promotion message meets the sortord of bidding in matched promotion message source;Promotion message pushes mould
The received promotion message for pushing of block 1104, the promotion message of push are used in the matched popularization position of institute show.
Above-mentioned popularization position coalignment 900 is directed to based on the candidate popularization information for belonging to different promotion message sources and promotes position
Score of bidding match position is promoted with promotion message source.Wherein, score of bidding according to candidate popularization information estimates exposure
Light clicking rate determines, and estimates exposure clicking rate and generated according to user characteristics corresponding with user identifier, estimates exposure point in this way
The rate of hitting can sufficiently reflect candidate popularization information and the match condition of user.The unit of candidate popularization information clicks value can be anti-
Reflect value advantage of the promotion message in all candidate informations.When carrying out promoting position matching based on score of bidding in this way, Neng Gouchong
Divide the value advantage for considering promotion message in demand and promotion message source of the user to promotion message in promotion message source, makes
Position resource, which must be promoted, to be fully utilized, and avoid wasting.
As shown in figure 12, in one embodiment, a kind of promotion message displaying device 1200 is provided, device 1200 wraps
It includes: promoting position determining module 1202, promotion message receiving module 1204 and display module 1206, in which:
Position determining module 1202 is promoted, for determining popularization corresponding with user identifier position.
Promotion message receiving module 1204, for receiving and promoting the matched promotion message in position;Promotion message from
Promote the matched promotion message source in position;Promotion message source is that corresponding with user identifier and corresponding score of bidding meets popularization
Promotion message source belonging to the candidate popularization information of condition;It bids score, is clicked according to the exposure of estimating of candidate popularization information
The unit of rate and candidate popularization information is clicked value and is determined;Exposure clicking rate is estimated according to user characteristics corresponding with user identifier
It generates.
Display module 1206, for showing the matched promotion message of institute in promoting position.
In one embodiment, when candidate popularization information belongs to third type promotion message, the list of candidate popularization information
Value is hit in site, is difference of the financial value divided by the first number of clicks and the second number of clicks;Financial value is in the second popularization page
The financial value of non-third type promotion message;First number of clicks is the point of third type promotion message in the first popularization page
Hit number;Second number of clicks is the number of clicks of third type promotion message in the second popularization page;First promotes the page
Position is promoted to be only used for showing third type promotion message;The second popularization page includes the popularization position for showing third type promotion message
With the popularization position for showing non-third type promotion message.
Above-mentioned promotion message shows device 1200, is directed to and is promoted based on the candidate popularization information for belonging to different promotion message sources
The score of bidding of position will be promoted position and be matched with promotion message source.Wherein, it bids score estimating according to candidate popularization information
It exposes clicking rate to determine, and estimates exposure clicking rate and generated according to user characteristics corresponding with user identifier, estimate exposure in this way
Clicking rate can sufficiently reflect candidate popularization information and the match condition of user.The unit of candidate popularization information clicks value can be with
Reflect value advantage of the promotion message in all candidate informations.It, can when carrying out promoting position matching based on score of bidding in this way
Fully consider the value advantage of promotion message in demand and promotion message source of the user to promotion message in promotion message source,
Popularization position resource is fully utilized, avoids wasting.
Figure 13 shows the internal structure chart of computer equipment in one embodiment.The computer equipment specifically can be figure
Server 120 in 1.As shown in figure 13, it includes being connected by system bus which, which includes the computer equipment,
Processor, memory, network interface.Wherein, memory includes non-volatile memory medium and built-in storage.The computer equipment
Non-volatile memory medium be stored with operating system, can also be stored with computer program, which is held by processor
When row, it may make processor is realized to promote position matching process.Computer program can also be stored in the built-in storage, the computer
When program is executed by processor, it may make processor executes to promote position matching process.
In one embodiment, position coalignment provided by the present application of promoting can be implemented as a kind of shape of computer program
Formula, computer program can be run in computer equipment as shown in fig. 13 that.Composition can be stored in the memory of computer equipment
Each program module of position coalignment is promoted, for example, determining module shown in Fig. 9 902, acquisition module 904, score of bidding are true
Cover half block 906, screening module 908 and matching module 910.The computer program that each program module is constituted executes processor
The step of each embodiment of the application described in this specification promoted in the matching process of position.
For example, computer equipment shown in Figure 13 can pass through cover half really in popularization position coalignment as shown in Figure 9
Block 902 executes step S202.Computer equipment can execute step S204 by obtaining module 904.Computer equipment can be by competing
Valence score determining module 906 executes step S206.Computer equipment can execute step S208 by screening module 908.Computer
Equipment can execute step S210 by matching module 910.
In one embodiment, a kind of computer equipment, including memory and processor are provided, memory is stored with meter
Calculation machine program, when computer program is executed by processor, so that processor executes following steps: determination is corresponding with user identifier
Promote position and candidate popularization information;Obtain candidate popularization information estimates exposure clicking rate;Estimate exposure clicking rate according to with
Family identifies corresponding user characteristics and generates;Value is clicked according to the unit for estimating exposure clicking rate and candidate popularization information, is determined
The corresponding score of bidding of candidate popularization information;Score of bidding corresponding to screening from candidate popularization information meets pushing away for popularization condition
Guangxin breath;Position will be promoted to match with promotion message source belonging to the promotion message that filters out;Promote position for show from
The promotion message in matched promotion message source.
In one embodiment, computer program, which is executed by processor, obtains candidate popularization information and estimates exposure clicking rate
Step when so that processing implement body executes following steps: obtaining corresponding with user identifier user characteristics;Extract candidate popularization
The promotion message feature of information;According to user characteristics and promotion message feature, the feature vector of candidate popularization information is generated;Pass through
Exposure clicking rate prediction model corresponding with the affiliated promotion message source of candidate popularization information, and the feature according to candidate popularization information
Vector, determine candidate popularization information estimates exposure clicking rate.
In one embodiment, the training step for exposing clicking rate prediction model includes: to obtain to belong to promotion message source
Historical user's feature, history promotion message feature corresponding to history promotion message and sample label;According to historical user's feature
With history promotion message feature, generation belongs to history feature vector corresponding to the promotion message in promotion message source;By history spy
Vector input exposure clicking rate prediction model is levied, history corresponding with history feature vector is obtained and exposes clicking rate;When basis is gone through
History exposes clicking rate and sample label determines when being unsatisfactory for training stop condition, and the model of adjustment exposure clicking rate prediction model is joined
It counts and continues to train, until deconditioning when meeting training stop condition.
In one embodiment, it includes: when candidate popularization is believed that the unit of candidate popularization information, which clicks the determination step of value,
When breath belongs to first kind promotion message, then the bid value that candidate popularization information single clicks on is obtained;Bid value is determined as waiting
The unit of promotion message is selected to click value.
In one embodiment, it includes: when candidate popularization is believed that the unit of candidate popularization information, which clicks the determination step of value,
When breath belongs to Second Type promotion message, then obtains candidate popularization information and be clicked receipts of the rear downloaded application in specified duration
Benefit value mean value;Determine that the unit of candidate popularization information clicks value according to financial value mean value.
In one embodiment, it includes: when candidate popularization is believed that the unit of candidate popularization information, which clicks the determination step of value,
When breath belongs to third type promotion message, it is determined that the estimation unit of third type promotion message clicks value;Estimate unit point
Hitting value is determined when the financial value total value that the financial value total value of the first popularization page and second promote the page is equal;First pushes away
The popularization position of the wide page is only used for showing third type promotion message;Second promotes the page including showing third type promotion message
The popularization position promoted position and show non-third type promotion message;Estimation unit is clicked into value and is determined as candidate popularization information
Unit click value.
In one embodiment, it includes: when candidate popularization is believed that the unit of candidate popularization information, which clicks the determination step of value,
When breath belongs to third type promotion message, then the first the first number of clicks for promoting third type promotion message in the page is obtained;
The popularization position of the first popularization page is only used for showing third type promotion message;It obtains second and promotes third type popularization in the page
The financial value of second number of clicks of information and non-third type promotion message;The second popularization page includes showing that third type pushes away
The popularization position of Guangxin breath promoted position and show non-third type promotion message;The financial value that will acquire is divided by the first number of clicks
With the difference of the second number of clicks, the unit for obtaining candidate popularization information clicks value.
In one embodiment, score of bidding is that basis estimates exposure clicking rate, unit clicks value and candidate popularization letter
The exposure weight in promotion message source belonging to ceasing determines;The determination step of the exposure weight in the affiliated promotion message source of candidate popularization information
It include: the previous exposure accounting for promoting the affiliated promotion message source of candidate popularization information after position matches of statistics;According to the exposure of statistics
Accounting is determined when the secondary exposure weight for carrying out promoting candidate's promotion message affiliated promotion message source when position matches.
In one embodiment, the determination step of the exposure weight in the affiliated promotion message source of candidate popularization information further include:
Obtain default exposure weight;By the exposure weight in the affiliated promotion message source of candidate popularization information from default exposure weight iteration tune
It is whole, and carried out after promoting position matching according to the exposure weight of current iteration, obtain the affiliated promotion message source of candidate popularization information
Accounting is exposed, stops iteration adjustment when exposing accounting and meeting balanced conditions of exposure, obtains promoting belonging to candidate popularization information
The initial exposure weight of information source.
In one embodiment, when computer program is executed by processor, so that processor also executes following steps: receiving
Promotion message from matched promotion message source;Received promotion message meets the sequence of bidding in matched promotion message source
Mode;The received promotion message of push institute, the promotion message of push are used in the matched popularization position of institute show.
Above-mentioned computer equipment, based on belonging to the candidate popularization information in different promotion message sources for promoting bidding point for position
Number will be promoted position and be matched with promotion message source.Wherein, score of bidding according to candidate popularization information estimates exposure clicking rate
It determines, and estimates exposure clicking rate and generated according to user characteristics corresponding with user identifier, estimating exposure clicking rate in this way can
The sufficiently match condition of reflection candidate popularization information and user.The unit of candidate popularization information, which clicks value, can reflect popularization letter
Cease the value advantage in all candidate informations.When carrying out promoting position matching based on score of bidding in this way, use can be fully considered
The value advantage of promotion message in demand and promotion message source of the family to promotion message in promotion message source, so that promoting position
Resource can be fully utilized, and avoid wasting.
In one embodiment, a kind of computer readable storage medium is provided, computer program is stored thereon with, is calculated
When machine program is executed by processor, so that processor executes following steps: determining popularization position corresponding with user identifier and candidate
Promotion message;Obtain candidate popularization information estimates exposure clicking rate;Exposure clicking rate is estimated according to corresponding with user identifier
User characteristics generate;Value is clicked according to the unit for estimating exposure clicking rate and candidate popularization information, determines candidate popularization information
Corresponding score of bidding;Score of bidding corresponding to screening from candidate popularization information meets the promotion message of popularization condition;It will push away
Wide position matches with promotion message source belonging to the promotion message filtered out;It promotes position and believes for displaying from popularization is matched
The promotion message in breath source.
In one embodiment, computer program, which is executed by processor, obtains candidate popularization information and estimates exposure clicking rate
Step when so that processing implement body executes following steps: obtaining corresponding with user identifier user characteristics;Extract candidate popularization
The promotion message feature of information;According to user characteristics and promotion message feature, the feature vector of candidate popularization information is generated;Pass through
Exposure clicking rate prediction model corresponding with the affiliated promotion message source of candidate popularization information, and the feature according to candidate popularization information
Vector, determine candidate popularization information estimates exposure clicking rate.
In one embodiment, the training step for exposing clicking rate prediction model includes: to obtain to belong to promotion message source
Historical user's feature, history promotion message feature corresponding to history promotion message and sample label;According to historical user's feature
With history promotion message feature, generation belongs to history feature vector corresponding to the promotion message in promotion message source;By history spy
Vector input exposure clicking rate prediction model is levied, history corresponding with history feature vector is obtained and exposes clicking rate;When basis is gone through
History exposes clicking rate and sample label determines when being unsatisfactory for training stop condition, and the model of adjustment exposure clicking rate prediction model is joined
It counts and continues to train, until deconditioning when meeting training stop condition.
In one embodiment, it includes: when candidate popularization is believed that the unit of candidate popularization information, which clicks the determination step of value,
When breath belongs to first kind promotion message, then the bid value that candidate popularization information single clicks on is obtained;Bid value is determined as waiting
The unit of promotion message is selected to click value.
In one embodiment, it includes: when candidate popularization is believed that the unit of candidate popularization information, which clicks the determination step of value,
When breath belongs to Second Type promotion message, then obtains candidate popularization information and be clicked receipts of the rear downloaded application in specified duration
Benefit value mean value;Determine that the unit of candidate popularization information clicks value according to financial value mean value.
In one embodiment, it includes: when candidate popularization is believed that the unit of candidate popularization information, which clicks the determination step of value,
When breath belongs to third type promotion message, it is determined that the estimation unit of third type promotion message clicks value;Estimate unit point
Hitting value is determined when the financial value total value that the financial value total value of the first popularization page and second promote the page is equal;First pushes away
The popularization position of the wide page is only used for showing third type promotion message;Second promotes the page including showing third type promotion message
The popularization position promoted position and show non-third type promotion message;Estimation unit is clicked into value and is determined as candidate popularization information
Unit click value.
In one embodiment, it includes: when candidate popularization is believed that the unit of candidate popularization information, which clicks the determination step of value,
When breath belongs to third type promotion message, then the first the first number of clicks for promoting third type promotion message in the page is obtained;
The popularization position of the first popularization page is only used for showing third type promotion message;It obtains second and promotes third type popularization in the page
The financial value of second number of clicks of information and non-third type promotion message;The second popularization page includes showing that third type pushes away
The popularization position of Guangxin breath promoted position and show non-third type promotion message;The financial value that will acquire is divided by the first number of clicks
With the difference of the second number of clicks, the unit for obtaining candidate popularization information clicks value.
In one embodiment, score of bidding is that basis estimates exposure clicking rate, unit clicks value and candidate popularization letter
The exposure weight in promotion message source belonging to ceasing determines;The determination step of the exposure weight in the affiliated promotion message source of candidate popularization information
It include: the previous exposure accounting for promoting the affiliated promotion message source of candidate popularization information after position matches of statistics;According to the exposure of statistics
Accounting is determined when the secondary exposure weight for carrying out promoting candidate's promotion message affiliated promotion message source when position matches.
In one embodiment, the determination step of the exposure weight in the affiliated promotion message source of candidate popularization information further include:
Obtain default exposure weight;By the exposure weight in the affiliated promotion message source of candidate popularization information from default exposure weight iteration tune
It is whole, and carried out after promoting position matching according to the exposure weight of current iteration, obtain the affiliated promotion message source of candidate popularization information
Accounting is exposed, stops iteration adjustment when exposing accounting and meeting balanced conditions of exposure, obtains promoting belonging to candidate popularization information
The initial exposure weight of information source.
In one embodiment, when computer program is executed by processor, so that processor also executes following steps: receiving
Promotion message from matched promotion message source;Received promotion message meets the sequence of bidding in matched promotion message source
Mode;The received promotion message of push institute, the promotion message of push are used in the matched popularization position of institute show.
Above-mentioned computer readable storage medium is directed to based on the candidate popularization information for belonging to different promotion message sources and promotes position
Score of bidding match position is promoted with promotion message source.Wherein, score of bidding according to candidate popularization information estimates exposure
Light clicking rate determines, and estimates exposure clicking rate and generated according to user characteristics corresponding with user identifier, estimates exposure point in this way
The rate of hitting can sufficiently reflect candidate popularization information and the match condition of user.The unit of candidate popularization information clicks value can be anti-
Reflect value advantage of the promotion message in all candidate informations.When carrying out promoting position matching based on score of bidding in this way, Neng Gouchong
Divide the value advantage for considering promotion message in demand and promotion message source of the user to promotion message in promotion message source, makes
Position resource, which must be promoted, to be fully utilized, and avoid wasting.
Figure 14 shows the internal structure chart of computer equipment in one embodiment.The computer equipment specifically can be figure
Terminal 110 in 1.As shown in figure 14, it includes the place connected by system bus which, which includes the computer equipment,
Manage device, memory, network interface, input unit and display screen.Wherein, memory includes non-volatile memory medium and interior storage
Device.The non-volatile memory medium of the computer equipment is stored with operating system, can also be stored with computer program, the computer
When program is executed by processor, processor may make to realize promotion message methods of exhibiting.Meter can also be stored in the built-in storage
Calculation machine program when the computer program is executed by processor, may make processor to execute promotion message methods of exhibiting.Computer is set
Standby display screen can be liquid crystal display or electric ink display screen, and the input unit of computer equipment can be display screen
The touch layer of upper covering is also possible to the key being arranged on computer equipment shell, trace ball or Trackpad, can also be external
Keyboard, Trackpad or mouse etc..
It will be understood by those skilled in the art that structure shown in Figure 13 and 14, only relevant to application scheme
The block diagram of part-structure does not constitute the restriction for the computer equipment being applied thereon to application scheme, specific to calculate
Machine equipment may include perhaps combining certain components or with different components than more or fewer components as shown in the figure
Arrangement.
In one embodiment, promotion message provided by the present application shows that device can be implemented as a kind of computer program
Form, computer program can be run in computer equipment as shown in figure 14.Group can be stored in the memory of computer equipment
Each program module of device is shown at promotion message, for example, popularization position determining module 1202, promotion message shown in Figure 12 connect
Receive module 1204 and display module 1206.The computer program that each program module is constituted executes processor in this specification
Step in the promotion message methods of exhibiting of each embodiment of the application of description.
For example, computer equipment shown in Figure 14 can show the determination in device by promotion message as shown in figure 12
Module 1202 executes step S1102.Computer equipment can execute step S1104 by promotion message receiving module 1204.It calculates
Machine equipment can execute step S1106 by display module 1206.
In one embodiment, a kind of computer equipment, including memory and processor are provided, memory is stored with meter
Calculation machine program, when computer program is executed by processor, so that processor executes following steps: determination is corresponding with user identifier
Promote position;It receives and promotes the matched promotion message in position;Promotion message from promote the matched promotion message source in position;It promotes
Information source is that corresponding with user identifier and corresponding score of bidding meets and pushes away belonging to the candidate popularization information of popularization condition
Wide information source;It bids score, is clicked according to the unit for estimating exposure clicking rate and candidate popularization information of candidate popularization information
Value determines;Exposure clicking rate is estimated to be generated according to user characteristics corresponding with user identifier;It shows and is matched in promoting position
Promotion message.
In one embodiment, when candidate popularization information belongs to third type promotion message, the list of candidate popularization information
Value is hit in site, is difference of the financial value divided by the first number of clicks and the second number of clicks;Financial value is in the second popularization page
The financial value of non-third type promotion message;First number of clicks is the point of third type promotion message in the first popularization page
Hit number;Second number of clicks is the number of clicks of third type promotion message in the second popularization page;First promotes the page
Position is promoted to be only used for showing third type promotion message;The second popularization page includes the popularization position for showing third type promotion message
With the popularization position for showing non-third type promotion message.
Above-mentioned computer equipment, based on belonging to the candidate popularization information in different promotion message sources for promoting bidding point for position
Number will be promoted position and be matched with promotion message source.Wherein, score of bidding according to candidate popularization information estimates exposure clicking rate
It determines, and estimates exposure clicking rate and generated according to user characteristics corresponding with user identifier, estimating exposure clicking rate in this way can
The sufficiently match condition of reflection candidate popularization information and user.The unit of candidate popularization information, which clicks value, can reflect popularization letter
Cease the value advantage in all candidate informations.When carrying out promoting position matching based on score of bidding in this way, use can be fully considered
The value advantage of promotion message in demand and promotion message source of the family to promotion message in promotion message source, so that promoting position
Resource can be fully utilized, and avoid wasting.
In one embodiment, a kind of computer readable storage medium is provided, computer program is stored thereon with, is calculated
When machine program is executed by processor, so that processor executes following steps: determining popularization corresponding with user identifier position;Receive with
Promote the matched promotion message in position;Promotion message from promote the matched promotion message source in position;Promotion message source, be with
Family identifies corresponding and corresponding score of bidding and meets promotion message source belonging to the candidate popularization information of popularization condition;It bids
Score is to click value determination according to the unit for estimating exposure clicking rate and candidate popularization information of candidate popularization information;It estimates
Exposure clicking rate is generated according to user characteristics corresponding with user identifier;The matched promotion message of institute is shown in promoting position.
In one embodiment, when candidate popularization information belongs to third type promotion message, the list of candidate popularization information
Value is hit in site, is difference of the financial value divided by the first number of clicks and the second number of clicks;Financial value is in the second popularization page
The financial value of non-third type promotion message;First number of clicks is the point of third type promotion message in the first popularization page
Hit number;Second number of clicks is the number of clicks of third type promotion message in the second popularization page;First promotes the page
Position is promoted to be only used for showing third type promotion message;The second popularization page includes the popularization position for showing third type promotion message
With the popularization position for showing non-third type promotion message.
Above-mentioned computer readable storage medium is directed to based on the candidate popularization information for belonging to different promotion message sources and promotes position
Score of bidding match position is promoted with promotion message source.Wherein, score of bidding according to candidate popularization information estimates exposure
Light clicking rate determines, and estimates exposure clicking rate and generated according to user characteristics corresponding with user identifier, estimates exposure point in this way
The rate of hitting can sufficiently reflect candidate popularization information and the match condition of user.The unit of candidate popularization information clicks value can be anti-
Reflect value advantage of the promotion message in all candidate informations.When carrying out promoting position matching based on score of bidding in this way, Neng Gouchong
Divide the value advantage for considering promotion message in demand and promotion message source of the user to promotion message in promotion message source, makes
Position resource, which must be promoted, to be fully utilized, and avoid wasting.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with
Relevant hardware is instructed to complete by computer program, the program can be stored in a non-volatile computer and can be read
In storage medium, the program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein, provided herein
Each embodiment used in any reference to memory, storage, database or other media, may each comprise non-volatile
And/or volatile memory.Nonvolatile memory may include that read-only memory (ROM), programming ROM (PROM), electricity can be compiled
Journey ROM (EPROM), electrically erasable ROM (EEPROM) or flash memory.Volatile memory may include random access memory
(RAM) or external cache.Each technical characteristic of above embodiments can be combined arbitrarily, to make to describe
Succinctly, combination not all possible to each technical characteristic in above-described embodiment is all described, as long as however, these technologies
Contradiction is not present in the combination of feature, all should be considered as described in this specification.
The several embodiments of the application above described embodiment only expresses, the description thereof is more specific and detailed, but simultaneously
The limitation to the application the scope of the patents therefore cannot be interpreted as.It should be pointed out that for those of ordinary skill in the art
For, without departing from the concept of this application, various modifications and improvements can be made, these belong to the guarantor of the application
Protect range.Therefore, the scope of protection shall be subject to the appended claims for the application patent.
Claims (15)
1. a kind of popularization position matching process, comprising:
Determine popularization position corresponding with user identifier and candidate popularization information;
Obtain the candidate popularization information estimates exposure clicking rate;It is described estimate exposure clicking rate according to the user identifier
Corresponding user characteristics generate;
Value is clicked according to the unit for estimating exposure clicking rate and the candidate popularization information, determines the candidate popularization letter
Cease corresponding score of bidding;
Score of bidding corresponding to screening from the candidate popularization information meets the promotion message of popularization condition;
The popularization position is matched with promotion message source belonging to the promotion message that filters out;The popularization position is come for showing
From in the promotion message in matched promotion message source.
2. the method according to claim 1, wherein the acquisition candidate popularization information estimates exposure point
Hit rate, comprising:
Obtain user characteristics corresponding with the user identifier;
Extract the promotion message feature of the candidate popularization information;
According to the user characteristics and promotion message feature, the feature vector of the candidate popularization information is generated;
By exposure clicking rate prediction model corresponding with the affiliated promotion message source of the candidate popularization information, and according to the time
The feature vector for selecting promotion message, determine the candidate popularization information estimates exposure clicking rate.
3. according to the method described in claim 2, it is characterized in that, the training step packet of the exposure clicking rate prediction model
It includes:
Acquisition belongs to historical user's feature corresponding to the history promotion message in the promotion message source, history promotion message feature
And sample label;
According to historical user's feature and history promotion message feature, the promotion message institute for belonging to the promotion message source is generated
Corresponding history feature vector;
The history feature vector is inputted into the exposure clicking rate prediction model, is obtained corresponding with the history feature vector
History exposes clicking rate;
When being unsatisfactory for training stop condition according to history exposure clicking rate and the sample label are determining, the exposure is adjusted
The model parameter of light clicking rate prediction model simultaneously continues to train, until deconditioning when meeting training stop condition.
4. the method according to claim 1, wherein the unit of the candidate popularization information clicks the determination of value
Step includes:
When the candidate popularization information belongs to first kind promotion message, then
Obtain the bid value that the candidate popularization information single clicks on;
The unit that the bid value is determined as the candidate popularization information is clicked into value.
5. the method according to claim 1, wherein the unit of the candidate popularization information clicks the determination of value
Step includes:
When the candidate popularization information belongs to Second Type promotion message, then
It obtains the candidate popularization information and is clicked financial value mean value of the rear downloaded application in specified duration;
Determine that the unit of the candidate popularization information clicks value according to the financial value mean value.
6. the method according to claim 1, wherein the unit of the candidate popularization information clicks the determination of value
Step includes:
When the candidate popularization information belongs to third type promotion message, then
Determine that the estimation unit of third type promotion message clicks value;It is to promote page first that the estimation unit, which clicks value,
It is determined when the financial value total value in face and the equal financial value total value of the second popularization page;Described first promotes the popularization position of the page only
For showing third type promotion message;The second popularization page includes popularization position and the exhibition for showing third type promotion message
Show the popularization position of non-third type promotion message;
The estimation unit is clicked into the unit click value that value is determined as the candidate popularization information.
7. the method according to claim 1, wherein the unit of the candidate popularization information clicks the determination of value
Step includes:
When the candidate popularization information belongs to third type promotion message, then
Obtain the first the first number of clicks for promoting third type promotion message in the page;Described first promotes the popularization position of the page
It is only used for showing third type promotion message;
It obtains second and promotes the second number of clicks of third type promotion message and the receipts of non-third type promotion message in the page
Benefit value;The second popularization page includes popularization position and the non-third type promotion message of displaying for showing third type promotion message
Popularization position;
The financial value that will acquire obtains the list of the candidate popularization information divided by the difference of the first number of clicks and the second number of clicks
Value is hit in site.
8. the method according to claim 1, wherein the score of bidding is to estimate exposure according to click
Rate, the unit click value and the exposure weight in the affiliated promotion message source of the candidate popularization information determines;The candidate pushes away
Guangxin cease belonging to the determination step of exposure weight in promotion message source include:
Count the exposure accounting in the affiliated promotion message source of candidate popularization information after previous popularization position matches;
According to the exposure accounting of statistics, determines and promote candidate popularization information affiliated promotion message source when position matches when secondary
Exposure weight.
9. according to the method described in claim 8, it is characterized in that, the exposure in the affiliated promotion message source of the candidate popularization information
The determination step of weight further include:
Obtain default exposure weight;
By the exposure weight in the affiliated promotion message source of the candidate popularization information from default exposure weight iteration adjustment, and according to
The exposure weight of current iteration carries out after promoting position matching, and the exposure for obtaining the affiliated promotion message source of the candidate popularization information accounts for
Than stopping iteration adjustment when the exposure accounting meets balanced conditions of exposure, obtaining pushing away belonging to the candidate popularization information
The initial exposure weight of wide information source.
10. method according to any one of claim 1 to 9, which is characterized in that the method also includes:
Receive the promotion message from matched promotion message source;Received promotion message meets matched promotion message source
It bids sortord;
The received promotion message of push institute, the promotion message of push are used in the matched popularization position of institute show.
11. a kind of promotion message methods of exhibiting, which comprises
Determine popularization corresponding with user identifier position;
It receives and the matched promotion message in the popularization position;The promotion message is believed from the matched popularization in the popularization position
Breath source;The promotion message source is the candidate that corresponding with the user identifier and corresponding score of bidding meets popularization condition
Promotion message source belonging to promotion message;The score of bidding is to estimate exposure clicking rate according to the candidate popularization information
Value is clicked with the unit of the candidate popularization information to determine;The exposure clicking rate of estimating is according to corresponding with the user identifier
User characteristics generate;
The matched promotion message of institute is shown in the popularization position.
12. according to the method for claim 11, which is characterized in that promoted when the candidate popularization information belongs to third type
When information, it is financial value divided by the first number of clicks and the second number of clicks that the unit of the candidate popularization information, which clicks value,
Difference;The financial value is the financial value of non-third type promotion message in the second popularization page;First number of clicks,
It is the number of clicks of third type promotion message in the first popularization page;Second number of clicks is in the second popularization page
The number of clicks of third type promotion message;The popularization position of the first popularization page is only used for showing that third type promotes letter
Breath;The second popularization page includes popularization position and the non-third type promotion message of displaying for showing third type promotion message
Promote position.
13. a kind of popularization position coalignment, which is characterized in that described device includes:
Determining module, for determining popularization position corresponding with user identifier and candidate popularization information;
Module is obtained, estimates exposure clicking rate for obtain the candidate popularization information;It is described estimate exposure clicking rate according to
User characteristics corresponding with the user identifier generate;
It bids score determining module, the unit for estimating exposure clicking rate and the candidate popularization information according to clicks valence
Value, determines the corresponding score of bidding of the candidate popularization information;
Screening module meets the popularization letter of popularization condition for score of bidding corresponding to the screening from the candidate popularization information
Breath;
Matching module, for the popularization position to match with promotion message source belonging to the promotion message that filters out;It is described to push away
Wide position is for showing the promotion message from matched promotion message source.
14. a kind of computer readable storage medium is stored with computer program, when the computer program is executed by processor,
So that the processor is executed such as the step of any one of claims 1 to 12 the method.
15. a kind of computer equipment, including memory and processor, the memory is stored with computer program, the calculating
When machine program is executed by the processor, so that the processor is executed such as any one of claims 1 to 12 the method
Step.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810236582.1A CN110210882A (en) | 2018-03-21 | 2018-03-21 | Promote position matching process and device, promotion message methods of exhibiting and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810236582.1A CN110210882A (en) | 2018-03-21 | 2018-03-21 | Promote position matching process and device, promotion message methods of exhibiting and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110210882A true CN110210882A (en) | 2019-09-06 |
Family
ID=67779079
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810236582.1A Pending CN110210882A (en) | 2018-03-21 | 2018-03-21 | Promote position matching process and device, promotion message methods of exhibiting and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110210882A (en) |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110764860A (en) * | 2019-10-21 | 2020-02-07 | 北京三快在线科技有限公司 | Project configuration method, device, server and storage medium in user interface |
CN110969479A (en) * | 2019-11-07 | 2020-04-07 | 连尚(新昌)网络科技有限公司 | Method, equipment and computer storage medium for displaying promoted resources |
CN111385353A (en) * | 2020-02-28 | 2020-07-07 | 上海连尚网络科技有限公司 | Method and equipment for providing application presentation information |
CN112488766A (en) * | 2020-12-09 | 2021-03-12 | 广州品唯软件有限公司 | Page display diagram setting method and device, computer equipment and storage medium |
CN112541145A (en) * | 2019-09-20 | 2021-03-23 | 百度在线网络技术(北京)有限公司 | Page display method, device, equipment and storage medium |
CN112819500A (en) * | 2020-09-30 | 2021-05-18 | 腾讯科技(上海)有限公司 | Information display method, strategy network training method, equipment and storage medium |
CN113139115A (en) * | 2020-01-19 | 2021-07-20 | 北京沃东天骏信息技术有限公司 | Information recommendation method, search method, device, client, medium and equipment |
CN113239275A (en) * | 2021-05-28 | 2021-08-10 | 北京百度网讯科技有限公司 | Information pushing method and device, electronic equipment and storage medium |
CN113283930A (en) * | 2021-05-12 | 2021-08-20 | 建信金融科技有限责任公司 | Marketing activity operation method and device |
CN114240527A (en) * | 2021-10-12 | 2022-03-25 | 北京淘友天下科技发展有限公司 | Resource pushing method and device, electronic equipment, readable medium and computer program |
CN114710686A (en) * | 2022-03-11 | 2022-07-05 | 武汉斗鱼鱼乐网络科技有限公司 | Exposure method, device, medium and equipment for live broadcast room |
CN117372166A (en) * | 2023-10-26 | 2024-01-09 | 北京开科唯识技术股份有限公司 | Efficient tail-end distribution processing method, device and storage medium |
CN113139115B (en) * | 2020-01-19 | 2024-09-24 | 北京沃东天骏信息技术有限公司 | Information recommendation method, search method, device, client, medium and equipment |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090144133A1 (en) * | 2006-05-16 | 2009-06-04 | Kt Corporation | Context related advertisement/information exposure method and recommendation service system using the same |
CN104636940A (en) * | 2013-11-07 | 2015-05-20 | 深圳市腾讯计算机系统有限公司 | Advertisement delivery method, advertisement delivery system, advertisement screening method, advertisement screening device, and server |
WO2015124024A1 (en) * | 2014-02-24 | 2015-08-27 | 北京奇虎科技有限公司 | Method and device for promoting exposure rate of information, method and device for determining value of search word |
WO2017101734A1 (en) * | 2015-12-15 | 2017-06-22 | 腾讯科技(深圳)有限公司 | Method and device for recommending content item |
CN107222526A (en) * | 2017-05-16 | 2017-09-29 | 百度在线网络技术(北京)有限公司 | Push method, device, equipment and the computer-readable storage medium of promotion message |
US20170293934A1 (en) * | 2015-05-11 | 2017-10-12 | Tencent Technology (Shenzhen) Company Limited | Method for determining validity of delivering of promotion information, monitoring server and terminal |
CN107657486A (en) * | 2017-10-19 | 2018-02-02 | 厦门美柚信息科技有限公司 | A kind of advertisement placement method and device |
CN110532468A (en) * | 2019-08-26 | 2019-12-03 | 北京齐尔布莱特科技有限公司 | A kind of recommended method of site resource, device and calculate equipment |
-
2018
- 2018-03-21 CN CN201810236582.1A patent/CN110210882A/en active Pending
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090144133A1 (en) * | 2006-05-16 | 2009-06-04 | Kt Corporation | Context related advertisement/information exposure method and recommendation service system using the same |
CN104636940A (en) * | 2013-11-07 | 2015-05-20 | 深圳市腾讯计算机系统有限公司 | Advertisement delivery method, advertisement delivery system, advertisement screening method, advertisement screening device, and server |
WO2015124024A1 (en) * | 2014-02-24 | 2015-08-27 | 北京奇虎科技有限公司 | Method and device for promoting exposure rate of information, method and device for determining value of search word |
US20170293934A1 (en) * | 2015-05-11 | 2017-10-12 | Tencent Technology (Shenzhen) Company Limited | Method for determining validity of delivering of promotion information, monitoring server and terminal |
WO2017101734A1 (en) * | 2015-12-15 | 2017-06-22 | 腾讯科技(深圳)有限公司 | Method and device for recommending content item |
CN107222526A (en) * | 2017-05-16 | 2017-09-29 | 百度在线网络技术(北京)有限公司 | Push method, device, equipment and the computer-readable storage medium of promotion message |
CN107657486A (en) * | 2017-10-19 | 2018-02-02 | 厦门美柚信息科技有限公司 | A kind of advertisement placement method and device |
CN110532468A (en) * | 2019-08-26 | 2019-12-03 | 北京齐尔布莱特科技有限公司 | A kind of recommended method of site resource, device and calculate equipment |
Non-Patent Citations (5)
Title |
---|
PAVEL KIREYEV等: "Do Display ads Influence Search? Attribution and Dynamics in Online Advertising" * |
岳昆等: "基于概率图模型的互联网广告点击率预测", 《华东师范大学学报(自然科学版)》 * |
杨诚: "基于用户实时反馈的点击率预估算法", 《计算机应用》 * |
纪文迪等: "广告点击率估算技术综述", 《华东师范大学学报(自然科学版)》 * |
陈巧红等: "基于递归神经网络的广告点击率预估研究", 《浙江理工大学学报(自然科学版)》 * |
Cited By (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112541145A (en) * | 2019-09-20 | 2021-03-23 | 百度在线网络技术(北京)有限公司 | Page display method, device, equipment and storage medium |
CN112541145B (en) * | 2019-09-20 | 2023-10-17 | 百度在线网络技术(北京)有限公司 | Page display method, device, equipment and storage medium |
CN110764860A (en) * | 2019-10-21 | 2020-02-07 | 北京三快在线科技有限公司 | Project configuration method, device, server and storage medium in user interface |
CN110969479A (en) * | 2019-11-07 | 2020-04-07 | 连尚(新昌)网络科技有限公司 | Method, equipment and computer storage medium for displaying promoted resources |
CN113139115B (en) * | 2020-01-19 | 2024-09-24 | 北京沃东天骏信息技术有限公司 | Information recommendation method, search method, device, client, medium and equipment |
CN113139115A (en) * | 2020-01-19 | 2021-07-20 | 北京沃东天骏信息技术有限公司 | Information recommendation method, search method, device, client, medium and equipment |
CN111385353A (en) * | 2020-02-28 | 2020-07-07 | 上海连尚网络科技有限公司 | Method and equipment for providing application presentation information |
CN112819500B (en) * | 2020-09-30 | 2023-09-19 | 腾讯科技(上海)有限公司 | Information display method, strategy network training equipment and storage medium |
CN112819500A (en) * | 2020-09-30 | 2021-05-18 | 腾讯科技(上海)有限公司 | Information display method, strategy network training method, equipment and storage medium |
CN112488766A (en) * | 2020-12-09 | 2021-03-12 | 广州品唯软件有限公司 | Page display diagram setting method and device, computer equipment and storage medium |
CN113283930A (en) * | 2021-05-12 | 2021-08-20 | 建信金融科技有限责任公司 | Marketing activity operation method and device |
CN113239275B (en) * | 2021-05-28 | 2023-06-30 | 北京百度网讯科技有限公司 | Information pushing method, device, electronic equipment and storage medium |
CN113239275A (en) * | 2021-05-28 | 2021-08-10 | 北京百度网讯科技有限公司 | Information pushing method and device, electronic equipment and storage medium |
CN114240527A (en) * | 2021-10-12 | 2022-03-25 | 北京淘友天下科技发展有限公司 | Resource pushing method and device, electronic equipment, readable medium and computer program |
CN114710686A (en) * | 2022-03-11 | 2022-07-05 | 武汉斗鱼鱼乐网络科技有限公司 | Exposure method, device, medium and equipment for live broadcast room |
CN117372166A (en) * | 2023-10-26 | 2024-01-09 | 北京开科唯识技术股份有限公司 | Efficient tail-end distribution processing method, device and storage medium |
CN117372166B (en) * | 2023-10-26 | 2024-03-08 | 北京开科唯识技术股份有限公司 | Efficient tail-end distribution processing method, device and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110210882A (en) | Promote position matching process and device, promotion message methods of exhibiting and device | |
WO2019242331A1 (en) | User behavior prediction method and apparatus, and behavior prediction model training method and apparatus | |
US20200327449A1 (en) | User retention platform | |
US20190012683A1 (en) | Method for predicting purchase probability based on behavior sequence of user and apparatus for the same | |
CN105493057B (en) | The content selection controlled using accuracy | |
CN104281961B (en) | For the advertisement in on-line system and the quality score system of content | |
CN106407210B (en) | A kind of methods of exhibiting and device of business object | |
US20140188593A1 (en) | Selecting an advertisement for a traffic source | |
CN111143738B (en) | Resource display method and device, electronic equipment and storage medium | |
CN110992097B (en) | Processing method and device for price of revenue product, computer equipment and storage medium | |
US20160210656A1 (en) | System for marketing touchpoint attribution bias correction | |
CN110413868A (en) | Information recommendation method, device, system and storage medium | |
US20170061515A1 (en) | Systems and methods for setting allocations and prices for content in an online marketplace | |
CN104834641A (en) | Processing method of network medium information and relevant system | |
CN108777701A (en) | A kind of method and device of determining receiver | |
CN111798280B (en) | Multimedia information recommendation method, device and equipment and storage medium | |
KR101085943B1 (en) | Method and Server apparatus for providing advertizing service | |
CN110515904B (en) | Quality prediction model training method, quality prediction method and device for media file | |
CN111008335A (en) | Information processing method, device, equipment and storage medium | |
CN111552835A (en) | File recommendation method and device and server | |
US10304081B1 (en) | Yielding content recommendations based on serving by probabilistic grade proportions | |
CN112115354B (en) | Information processing method, device, server and storage medium | |
CN113919866A (en) | Model training method, advertisement putting method, device, equipment and storage medium | |
KR101083002B1 (en) | Method and Server apparatus for calculating User Conversion Rate | |
CN111507471B (en) | Model training method, device, equipment and storage medium |
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 |