CN109886729A - Method, apparatus, electronic equipment and the readable storage medium storing program for executing of probability are clicked in prediction - Google Patents
Method, apparatus, electronic equipment and the readable storage medium storing program for executing of probability are clicked in prediction Download PDFInfo
- Publication number
- CN109886729A CN109886729A CN201910024668.2A CN201910024668A CN109886729A CN 109886729 A CN109886729 A CN 109886729A CN 201910024668 A CN201910024668 A CN 201910024668A CN 109886729 A CN109886729 A CN 109886729A
- Authority
- CN
- China
- Prior art keywords
- displayed
- user
- history
- parameter information
- click
- 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
Abstract
The embodiment of the present application provides method, apparatus, electronic equipment and the readable storage medium storing program for executing of a kind of prediction click probability, to optimize click probability forecasting method in the related technology.This method comprises: obtaining the parameter information that object has been displayed, the parameter information that object has been displayed includes at least this and the type of object has been displayed, and the user's operation information that object has been displayed includes at least this and has been displayed whether object is clicked by user；According to the parameter information of object to be shown and Probabilistic Prediction Model is clicked, predicts the click probability of the object to be shown；Wherein, the desired value of historical user's operation information under the parameter information of object has been displayed using each history as reward function for the click Probabilistic Prediction Model, it is inputted using the parameter information that object has been displayed in each history as state, the click probability of the parameter information of object is had been displayed as action output with each history.
Description
Technical field
The invention relates to technical field of data processing more particularly to it is a kind of prediction click probability method, apparatus,
Electronic equipment and readable storage medium storing program for executing.
Background technique
With the popularity of the internet, more and more content suppliers (Content Provider, CP) and electric business platform
Occur in people's lives, brings convenience for people's lives.Such as: user can be clear by the website of content supplier
Look at various information, user can also choose the commodity of oneself needs online by electric business platform.Due to content supplier and electric business
Platform can attract huge flow, so most businessman's selections launch advertisement by content supplier and electric business platform, to increase
The audient of more advertisements optimizes the effect of publicity of advertisement.
In general, the display terminal that content supplier and electric business platform are used in a manner of information flow to user pushes away
It delivers letters breath, and is inserted into the advertisement of businessman in information flow, to pass through advertisement net income increase.Businessman is imitated to optimize the publicity of advertisement
Fruit needs to consider that audient to the click probability of advertisement, gives advertising display to the click higher audient of probability.
In the related technology, in order to predict the click probability of advertisement, firstly, the historical behavior data to user count,
The advertisement whether clicked to user in information flow counts.Then, it is established and is clicked according to the historical behavior data of user
Probabilistic Prediction Model.Finally, being predicted using Probabilistic Prediction Model is clicked the click probability of advertisement.
As it can be seen that in the related technology, the premise accurately predicted the click probability of advertisement is: being based on history abundant
Behavioral data has constructed the higher click Probabilistic Prediction Model of prediction accuracy.If the negligible amounts of historical behavior data
(for example, user saw that the number of the information flow comprising advertisement is seldom or the seldom click information stream of user in advertisement), then
The higher click Probabilistic Prediction Model of prediction accuracy can not be constructed, thus also can not accurately to the click probability of advertisement into
Row prediction.Also, click Probabilistic Prediction Model in the related technology is constructed according to the historical behavior data of most users,
With universality, using the click Probabilistic Prediction Model, obtained prediction result is: different user is general to the click of same advertisement
Rate is identical.That is, click Probabilistic Prediction Model in the related technology have ignored user individualized feature (such as: difference use
Family is usually interested in different advertisements).As it can be seen that predicting that the technical solution of the click probability of advertisement needs to be changed in the related technology
Into.
Summary of the invention
The embodiment of the present application provides method, apparatus, electronic equipment and the readable storage medium storing program for executing of a kind of prediction click probability, with
The click probability forecasting method of optimization in the related technology.
The embodiment of the present application first aspect provides a kind of method that probability is clicked in prediction, which comprises
The parameter information that object has been displayed is obtained, the parameter information that object has been displayed includes at least this and object has been displayed
Type, the user's operation information that object has been displayed include at least this have been displayed whether object is clicked by user；
According to the parameter information of object to be shown and Probabilistic Prediction Model is clicked, predicts that the click of the object to be shown is general
Rate；
Wherein, the operation of the historical user under the parameter information of object has been displayed with each history for the click Probabilistic Prediction Model
The desired value of information is reward function, is inputted using the parameter information that object has been displayed in each history as state, is respectively gone through with described
The click probability of the parameter information of object has been displayed as action output in history.
Optionally, the method also includes:
Determine that the cumulative award desired value of object has been displayed in each history respectively, wherein each history has been displayed pair
The cumulative award desired value of elephant is that the user's operation of object has been displayed according to multiple history including object currently has been displayed
What information determined；
It is greater than the cumulative award phase that object has been displayed in any history in the cumulative award desired value of the current display object
In the case where prestige value, using the parameter information of the current display object and user's operation information as training sample, to the click
Probabilistic Prediction Model is trained update.
Optionally it is determined that the cumulative award desired value of the current display object, comprising:
The user's operation information that multiple history including the current display object are had been displayed to object, inputs respectively
Preset reward function, to obtain the reward value that object has been displayed in multiple history including the current display object；
The reward value that object has been displayed according to multiple history including the current display object, determines described current
Show the cumulative award desired value of object.
Optionally, the user's operation information that object has been displayed further include: the object that has been displayed is in default viewing area
The time persistently exposed in domain；
The user's operation information that multiple history including the current display object are had been displayed to object, inputs respectively
Preset reward function, to obtain the reward value that object has been displayed in multiple history including the current display object, comprising:
Each history in object has been displayed for multiple history including the current display object to have been displayed pair
As determining that the reward value of object has been displayed in the history according to following formula:
R=R1+aR2
Wherein, R indicates that the reward value of object, R has been displayed in the history1Indicate that object has been displayed whether by user's point in the history
It hits, a indicates default weight, R2Indicate that the history has been displayed object samples and shows that object persistently exposes in the default display area
The time of light.
Optionally, the method also includes:
The click probability predicted the click probability of multiple objects to be shown, and prediction is obtained is greater than default
The object to be shown of probability is sent to display terminal.
Optionally, the object to be shown that the click probability that prediction obtains is greater than predetermined probabilities is sent to display terminal, wrapped
It includes:
The multiple object to be shown is sent to the display terminal according to probability descending is clicked.
The embodiment of the present application second aspect provides a kind of device of prediction click probability, and described device includes:
Module is obtained, for obtaining the parameter information that object has been displayed, the parameter information that object has been displayed at least is wrapped
Include the type that object has been displayed, the user's operation information that object has been displayed include at least this have been displayed object whether by with
It clicks at family；
Prediction module, for described to aobvious according to the parameter information of object to be shown and click Probabilistic Prediction Model, prediction
Show the click probability of object；
Wherein, the operation of the historical user under the parameter information of object has been displayed with each history for the click Probabilistic Prediction Model
The desired value of information is reward function, is inputted using the parameter information that object has been displayed in each history as state, is respectively gone through with described
The click probability of the parameter information of object has been displayed as action output in history.
Optionally, described device further include:
Determining module, for determining that the cumulative award desired value of object has been displayed in each history respectively, wherein described each
The cumulative award desired value that object has been displayed in history is to be had been displayed according to multiple history including object currently has been displayed pair
What the user's operation information of elephant determined；
Object has been displayed for being greater than any history in the cumulative award desired value of the current display object in training module
Cumulative award desired value in the case where, with it is described it is current display object parameter information and user's operation information be training sample
This, is trained update to the click Probabilistic Prediction Model.
Optionally, the determining module includes:
Submodule is obtained, the user for object to be had been displayed in multiple history including the current display object grasps
Make information, input preset reward function respectively, with obtain include it is described it is current show object including multiple history have been displayed pair
The reward value of elephant；
First determines submodule, for object to have been displayed according to multiple history including the current display object
Reward value determines the cumulative award desired value of the current display object.
Optionally, the user's operation information that object has been displayed further include: the object that has been displayed is in default viewing area
The time persistently exposed in domain；
The acquisition submodule includes:
Second determines submodule, for having been displayed in object for multiple history including the current display object
Each history object has been displayed, according to following formula, determine that the reward value of object has been displayed in the history:
R=R1+aR2
Wherein, R indicates that the reward value of object, R has been displayed in the history1Indicate that object has been displayed whether by user's point in the history
It hits, a indicates default weight, R2Indicate that the history has been displayed object samples and shows that object persistently exposes in the default display area
The time of light.
Optionally, described device further include:
The point that sending module is predicted for the click probability to multiple objects to be shown, and prediction is obtained
The object to be shown that probability is hit greater than predetermined probabilities is sent to display terminal.
Optionally, the sending module includes:
Sending submodule, for the multiple object to be shown to be sent to the display eventually according to probability descending is clicked
End.
The embodiment of the present application third aspect provides a kind of computer readable storage medium, is stored thereon with computer program,
The step in the method as described in the application first aspect is realized when the program is executed by processor.
The embodiment of the present application fourth aspect provides a kind of electronic equipment, including memory, processor and is stored in memory
Computer program that is upper and can running on a processor, the processor realize method described in the application first aspect when executing
The step of.
Using click probability forecasting method provided by the embodiments of the present application, under the parameter information that object has been displayed with each history
Historical user's operation information desired value be reward function, using each history have been displayed object parameter information be state input,
The click probability of the parameter information of object has been displayed as action output with each history, obtains clicking Probabilistic Prediction Model, thus
To click Probabilistic Prediction Model consider that the parameter information that object has been displayed and user's operation information reflect the interest of user
Point, so clicking the point of interest that Probabilistic Prediction Model is more bonded user, clicking Probabilistic Prediction Model becomes for user individual
The model of customization, therefore using Probabilistic Prediction Model prediction user is clicked to the click probability of object to be shown, it can be improved general
The accuracy of rate prediction.
Detailed description of the invention
Technical solution in ord to more clearly illustrate embodiments of the present application, below by institute in the description to the embodiment of the present application
Attached drawing to be used is needed to be briefly described, it should be apparent that, the accompanying drawings in the following description is only some implementations of the application
Example, for those of ordinary skill in the art, without any creative labor, can also be according to these attached drawings
Obtain other attached drawings.
Fig. 1 is the flow chart for the method that probability is clicked in the prediction that one embodiment of the application proposes；
Fig. 2 is the flow chart of the method for the transmission object to be shown that one embodiment of the application provides；
Fig. 3 is the flow chart for the method that probability is clicked in the prediction that another embodiment of the application provides；
Fig. 4 is the flow chart for the method that probability is clicked in the prediction that another embodiment of the application provides；
Fig. 5 is the schematic diagram for the device that probability is clicked in the prediction that one embodiment of the application proposes；
Fig. 6 is the schematic diagram of the device for the transmission object to be shown that one embodiment of the application provides；
Fig. 7 is the schematic diagram for the electronic equipment that one embodiment of the application provides.
Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present application, technical solutions in the embodiments of the present application carries out clear, complete
Site preparation description, it is clear that described embodiment is some embodiments of the present application, instead of all the embodiments.Based on this Shen
Please in embodiment, every other implementation obtained by those of ordinary skill in the art without making creative efforts
Example, shall fall in the protection scope of this application.
It is the flow chart that the method for probability is clicked in the prediction that one embodiment of the application proposes with reference to Fig. 1, Fig. 1.Such as Fig. 1 institute
Show, method includes the following steps:
Step S11: obtaining the parameter information that object has been displayed, and the parameter information that object has been displayed has included at least this
Show that the type of object, the user's operation information that object has been displayed include at least this and object have been displayed whether by user's point
It hits.
In this example it is shown that object may include the various objects being shown on the display terminal that user uses, example
Such as: advertisement, news, link.According to whether showing on the display terminal that user uses, display object, which can be divided into, have been shown
Show object and object to be shown.It is the object shown on the display terminal that user uses that object, which has been displayed, to be shown right
As if the object not shown on the display terminal that user uses also.Object to be shown is shown on the display terminal that user uses
It becomes later and object has been displayed.No matter show that object or object to be shown has been displayed in object, display object can divide
For different type.For showing that object is advertisement, multiple types can be divided into: advertisement relevant to cuisines, related with tourism
Advertisement, advertisement relevant to hotel.
The parameter information that object has been displayed includes at least this and the type of object has been displayed.In one embodiment, it has shown
The parameter information for showing object includes but is not limited to: this have been displayed object temperature, this clicking rate of object has been displayed.It has been displayed pair
The user's operation information of elephant includes at least this and has been displayed whether object is clicked by user.In one embodiment, it has been displayed pair
The user's operation information of elephant can also include: that the time that object persistently exposes in default display area has been displayed in this.Illustratively,
In the case where display screen display of the object in the display terminal that user uses has been displayed, object is had been displayed in display screen in this
Middle position continues residence time.In another embodiment, the user's operation information that object has been displayed can also include:
In the case where display screen display of the object in the display terminal that user uses has been displayed, whether the wifi switch of display terminal
It is opened by user.In another embodiment, the user's operation information that object has been displayed can also include: that object has been displayed in this
When it is exposed in default display area.Illustratively, object is having been displayed on the display screen for the display terminal that user uses
In the case where display, the middle position when object is shown in display screen is had been displayed in this.
Step S12: according to the parameter information of object to be shown and Probabilistic Prediction Model is clicked, predicts the object to be shown
Click probability；
Wherein, the operation of the historical user under the parameter information of object has been displayed with each history for the click Probabilistic Prediction Model
The desired value of information is reward function, is inputted using the parameter information that object has been displayed in each history as state, is respectively gone through with described
The click probability of the parameter information of object has been displayed as action output in history.
In the present embodiment, it is adopted for predicting individually to show that the click Probabilistic Prediction Model of the click probability of object can be
The model (in the case where executing step S12 for the first time) obtained with the relevant technologies, is also possible to updated model (non-first
In the case where executing step S12).
Wherein, updated model obtains in the following way:
Determine that the cumulative award desired value of object has been displayed in each history respectively, wherein each history has been displayed pair
The cumulative award desired value of elephant is that the user's operation of object has been displayed according to multiple history including object currently has been displayed
What information determined；
It is greater than the cumulative award phase that object has been displayed in any history in the cumulative award desired value of the current display object
In the case where prestige value, using the parameter information of the current display object and user's operation information as training sample, to the click
Probabilistic Prediction Model is trained update.
In the present embodiment, it is contemplated that the parameter information and user's operation information that object has been displayed actually characterize close
The interested display object of phase user or the uninterested display object of recent user.That is, for being shown in the recent period
Display object on the display terminal that user uses, if user, to the display subject interests, user would generally click this
Show object；If user loses interest in the display object, user will not usually click the display object.
Using click probability forecasting method provided by the embodiments of the present application, under the parameter information that object has been displayed with each history
Historical user's operation information desired value be reward function, using each history have been displayed object parameter information be state input,
The click probability of the parameter information of object has been displayed as action output with each history, obtains clicking Probabilistic Prediction Model, thus
To click Probabilistic Prediction Model consider that the parameter information that object has been displayed and user's operation information reflect the interest of user
Point, so clicking the point of interest that Probabilistic Prediction Model is more bonded user, clicking Probabilistic Prediction Model becomes for user individual
The model of customization, therefore using Probabilistic Prediction Model prediction user is clicked to the click probability of object to be shown, it can be improved general
The accuracy of rate prediction.
It in another embodiment, further include that object has been displayed default in this in the user's operation information that object has been displayed
In the case where the time persistently exposed in display area, for the display pair being shown on the display terminal that user uses in the recent period
As if user, to the display subject interests, user would generally click the display object, alternatively, even if user does not click on
The display object, the time which persistently exposes in default display area are usually longer；If user is to the display
Object is lost interest in, then user will not usually click the display object or user does not click on the display object, and the display pair
As the time persistently exposed in default display area is shorter.
In order to which the accuracy of probabilistic forecasting is gradually increased, the present embodiment is proposed according to the parameter information and use that object has been displayed
Family operation information is updated, so that updated click Probabilistic Prediction Model is more bonded user to Probabilistic Prediction Model is clicked
Point of interest, in this way, be directed to different users, updated clicks Probabilistic Prediction Model difference, updated click probability is pre-
Surveying model is the model for the customization of each user individual.Parameter information and the user behaviour of object have been displayed by making full use of
Make information, be updated to Probabilistic Prediction Model is clicked, so that updated Probabilistic Prediction Model of clicking is with the interest of user
Point change and adjust in due course, using updated clicks Probabilistic Prediction Model progress probabilistic forecasting, improve probabilistic forecasting
Accuracy.
In one embodiment, step S12 the following steps are included:
According to the parameter information of object to be shown and updated click Probabilistic Prediction Model, the object to be shown is predicted
Click probability.
In the present embodiment, it during obtaining updated click Probabilistic Prediction Model, is utilized and object has been displayed
Parameter information, thus, in updated click Probabilistic Prediction Model, the parameter information that object has been displayed is updated click
One model parameter of Probabilistic Prediction Model has corresponding weighted value, will in the click probabilistic forecasting to object to be shown
The parameter information of the object to be shown is input in updated click Probabilistic Prediction Model, is finally predicted user and is waited showing to this
Show the click probability of object.
In the present embodiment, it is contemplated that parameter information and user's operation information that object has been displayed reflect the interest of user
Point is updated so making full use of the parameter information and user's operation information that object has been displayed to Probabilistic Prediction Model is clicked,
So that the updated point of interest clicked Probabilistic Prediction Model and be more bonded user, updated click Probabilistic Prediction Model become needle
To the model of user individual customization, therefore using updated click Probabilistic Prediction Model prediction user to object to be shown
Probability is clicked, the accuracy of probabilistic forecasting can be improved.
In conjunction with above embodiments, in another embodiment of the application, the quantity of object to be shown be it is multiple, to multiple to aobvious
Show that the click probability of each object to be shown in object is predicted, then the click probability of multiple objects to be shown is arranged
Sequence, and the display terminal used based on ranking results to user sends one or more of multiple objects to be shown.Thus, ginseng
Fig. 2 is examined, Fig. 2 is the flow chart of the method for the transmission object to be shown that one embodiment of the application provides.As shown in Fig. 2, this method
It is further comprising the steps of in addition to including step S11- step S12:
Step S13: predicting the click probability of multiple objects to be shown, and the click probability that prediction is obtained
Object to be shown greater than predetermined probabilities is sent to display terminal.
In one embodiment, the object to be shown that the click probability that prediction obtains is greater than predetermined probabilities is sent to aobvious
Show terminal, comprising:
The multiple object to be shown is sent to the display terminal according to probability descending is clicked.
In the present embodiment, it in the case where the quantity of object to be shown is multiple, for each object to be shown, executes
Step S11- step S12, obtains the click probability of each object to be shown.Then multiple respective clicks of objects to be shown are general
Rate is screened out from it and clicks the object to be shown that probability is greater than predetermined probabilities respectively compared with predetermined probabilities, namely filter out with
The compactness of the point of interest of user is higher than the object to be shown of default compactness, is then sent to the object to be shown filtered out
The display terminal that user uses.Wherein, predetermined probabilities are pre-set, and predetermined probabilities are arranged bigger, and being sent to user makes
The compactness of the point of interest of the object to be shown and user of display terminal is higher；Predetermined probabilities are arranged smaller, are sent to
The compactness of the point of interest of the object to be shown and user for the display terminal that user uses is lower.
Since the compactness for being sent to the object to be shown for the display terminal that user uses and the point of interest of user is higher,
So the probability that the object to be shown is clicked by user is higher, thus object to be shown is that content supplier and electric business platform are brought
The probability of income is higher, improves the probability by the traffic transformation of content supplier and electric business platform for income.Also, due to hair
The compactness for giving the object to be shown for the display terminal that user uses and the point of interest of user is higher, so optimizing user's body
It tests.
In conjunction with above each embodiment, in another embodiment of the application, it includes current display object that object, which has been displayed,.?
In this case, another embodiment of the application provides a kind of method that probability is clicked in prediction.It is that the application is another with reference to Fig. 3, Fig. 3
The flow chart of the method for probability is clicked in the prediction that embodiment provides.As shown in figure 3, the method comprising the steps of S11, step S12 and
Following steps:
Step S12 ': using the parameter information of the current display object and user's operation information as training sample, to described
It clicks Probabilistic Prediction Model to be trained, to obtain updated click Probabilistic Prediction Model.
In the present embodiment, become the model customized for each user individual to make to click Probabilistic Prediction Model,
The thought of intensified learning is applied to in the renewal process for clicking Probabilistic Prediction Model.Specifically, due to currently showing object
Parameter information and user's operation information, reflect the current point of interest of user in real time, thus will current display object parameter
Information and user's operation information are input to and click in Probabilistic Prediction Model, to the click Probabilistic Prediction Model as training sample
It is trained, the click Probabilistic Prediction Model after training is updated click Probabilistic Prediction Model.It is understood that such as
Described previously, which can be the model obtained using the relevant technologies and (is executing step for the first time
In the case where rapid S11- step S12 '), it is also possible to the last updated mould for executing and obtaining after step S11- step S12 '
Type (in the case where non-first execution step S11- step S12 ').
Click Probabilistic Prediction Model after training is updated click Probabilistic Prediction Model, above-mentioned training process
So that the updated individualized feature clicked Probabilistic Prediction Model and cover user, utilizes the updated click probabilistic forecasting
Model carries out probabilistic forecasting, improves the accuracy of probabilistic forecasting.
In conjunction with above each embodiment, in another embodiment of the application, it includes current display object that object, which has been displayed, and removes
Outside, object has been displayed in further include display order adjacent with the current display display order of object upper one.In the case, originally
Apply for that another embodiment provides a kind of method that probability is clicked in prediction.It is that another embodiment of the application provides with reference to Fig. 4, Fig. 4
Prediction click probability method flow chart.As shown in figure 4, the method comprising the steps of S11, step S12 and following steps:
Step S121: determine that the cumulative award desired value of object has been displayed in each history respectively, wherein each history
The cumulative award desired value that object has been displayed is that object has been displayed according to multiple history including object currently has been displayed
What user's operation information determined；
Step S122: it is greater than any history in the cumulative award desired value of the current display object and the tired of object has been displayed
It is right using the parameter information of the current display object and user's operation information as training sample in the case that bonuses distributed according to strict calculations encourages desired value
The click Probabilistic Prediction Model is trained update.
In one embodiment, step S121 the following steps are included:
The accumulative of object has been displayed in the cumulative award desired value for determining the current display object respectively and described upper one
Reward desired value, wherein a cumulative award desired value that object has been displayed be according to including this object has been displayed including it is more
What the user's operation information that object has been displayed in a history determined；
In one embodiment, step S122 the following steps are included:
It is greater than described upper one cumulative award that object has been displayed in the cumulative award desired value of the current display object
In the case where desired value, using the parameter information of the current display object and user's operation information as training sample, to the point
It hits Probabilistic Prediction Model to be trained, to obtain updated click Probabilistic Prediction Model.
In the present embodiment, a cumulative award desired value for having been displayed object, corresponding to including that this has been displayed object and exists
The expectation that object is the probability that content supplier and electric business platform bring income has been displayed in interior multiple history, also reflect including
The expectation that the compactness of object and the point of interest of user has been displayed in multiple history including object has been displayed in this.
In view of can all consume processor resource to each update for clicking Probabilistic Prediction Model, increase the processing of processor
Burden, in order to improve the utilization efficiency of processor resource, and reduces the processing load of processor, in the present embodiment, every time
Execute step S11 after, first determine whether this to click Probabilistic Prediction Model update whether it is necessary to if it is necessary, then
This update is executed, conversely, can then continue to continue to use last updated click Probabilistic Prediction Model.
Specifically, every time after execution step S11, it is first determined the cumulative award desired value of current display object, and really
The reward desired value that object has been displayed in upper one adjacent with the current display display order of object of display order is determined, then by two
A reward desired value compares, if currently the cumulative award desired value of display object is larger, this is to click probabilistic forecasting
The update of model is necessary, thus executes step S122 (or step S12 ')；If the currently cumulative award phase of display object
Prestige value is smaller, then can continue to continue to use last updated click Probabilistic Prediction Model.
It is understood that if currently object has been displayed in cumulative award desired value greatly thereon one of display object
Desired value is rewarded, then illustrates that rewarding desired value is also not up to maximum, thus it is necessary to pre- to last updated click probability
It surveys model to continue to update, to increase the subsequent display object for being sent to the display terminal that user uses as content supplier
The probability of income is brought with electric business platform, and improves the subsequent display object for being sent to the display terminal that user uses and user
Point of interest compactness.
In one embodiment, the cumulative award desired value of the current display object is determined, comprising:
The user's operation information that multiple history including the current display object are had been displayed to object, inputs respectively
Preset reward function, to obtain the reward value that object has been displayed in multiple history including the current display object；
The reward value that object has been displayed according to multiple history including the current display object, determines described current
Show the cumulative award desired value of object.
In one embodiment, the user's operation information that object has been displayed further include: the object that has been displayed exists
The time persistently exposed in default display area；Object is had been displayed into multiple history including the current display object
User's operation information inputs preset reward function respectively, to have obtained multiple history including the current display object
Show the reward value of object, comprising:
Each history in object has been displayed for multiple history including the current display object to have been displayed pair
As determining that the reward value of object has been displayed in the history according to following formula:
R=R1+aR2
Wherein, R indicates that the reward value of object, R has been displayed in the history1Indicate that object has been displayed whether by user's point in the history
It hits, a indicates default weight, R2Indicate that the history has been displayed object samples and shows that object persistently exposes in the default display area
The time of light.
Object has been displayed for each, determines that the process of the cumulative award desired value of object has been displayed in this are as follows: firstly, determining
The reward value of object has been displayed in this, and it is each to determine that object has been displayed in multiple history of the display order before object has been displayed in this
From reward value；Then, these reward values are calculated desired.Wherein it is determined that a reward value that object has been displayed has
And it is not limited to following two embodiment:
This: being had been displayed the user's operation information of object by the first embodiment, preset reward function is inputted, to be somebody's turn to do
The reward value of object has been displayed, wherein this have been displayed object user's operation information include at least this have been displayed object whether by
User clicks.
Second of embodiment: it includes that the user's operation of object has been displayed in this that the user's operation information that object has been displayed, which is removed,
Information includes at least this and has been displayed except whether object clicked by user, further includes: object has been displayed in default display area in this
The time inside persistently exposed.In the case, according to following formula, determine that the reward value of object has been displayed in this:
R=R1+aR2
Wherein, R indicates that the reward value of object, R has been displayed in this1Indicate that this has been displayed whether object is clicked by user, a is indicated
Default weight, R2Indicate that this has been displayed object samples and shows the time that object persistently exposes in the default display area.
In the present embodiment, it is contemplated that object has been displayed for each, whether user clicks this and object has been displayed, can not
Accurately embody the interest level that user this has been displayed object, it is possible to which the case where occurring is: user does not click on this and shown
Show object, but this to have been displayed the time that object persistently exposes in default display area longer, then user actually to this
Display object be also it is more interested, do not click on this only and object have been displayed.It is possible that occur another situation is that: user
It clicks this and has been displayed object, but this has been displayed that the time that object persistently exposes in default display area is shorter, then having can
It can be user misoperation, user actually has been displayed object to this and loses interest in.Therefore, this can whether have been clicked in conjunction with user
Display object and this time that object persistently exposes in default display area has been displayed, had been displayed pair to capture user to this
The interest level of elephant.Specifically, whether click that object has been displayed in this and object has been displayed in default display in this according to user
The time persistently exposed in region determines that user this has been displayed the journey interested of object in length range at a reasonable time
Degree.
In one embodiment, object can has been displayed with this in the interest level that object has been displayed to one in user
Reward value characterizes.The reward value that object has been displayed is obtained according to following formula:
R=R1+aR2
Wherein, R indicates that the reward value of object, R has been displayed in this1Indicate that this has been displayed whether object is clicked by user, a is indicated
Default weight, R2Indicate that this has been displayed object samples and shows the time that object persistently exposes in the default display area.
In the specific implementation process, a is pre-set, characterizes this and object samples display object has been displayed described default
The time persistently exposed in display area, the quantization table of significance level during determining that the reward value of object has been displayed in this
Show.
In the present embodiment, the time that object persistently exposes in default display area will have been displayed, as having calculated this
Show a parameter of the reward value of object, namely a Consideration as the point of interest for capturing user, it follows that
The reward value that object has been displayed more accurately reflects the point of interest of user, is obtained more based on the reward value that object has been displayed
Click Probabilistic Prediction Model after new carries out probabilistic forecasting using the updated click Probabilistic Prediction Model, improves probability
The accuracy of prediction.
Based on the same inventive concept, one embodiment of the application provides a kind of device of prediction click probability.With reference to Fig. 5, Fig. 5
It is the schematic diagram that the device of probability is clicked in the prediction that one embodiment of the application provides.As shown in figure 5, the device includes:
Module 501 is obtained, for obtaining the parameter information that object has been displayed, the parameter information that object has been displayed is at least
Have been displayed the type of object including this, the user's operation information that object has been displayed include at least this have been displayed object whether by
User clicks；
Prediction module 502, for according to the parameter information of object to be shown and clicking Probabilistic Prediction Model, prediction it is described to
Show the click probability of object；
Wherein, the operation of the historical user under the parameter information of object has been displayed with each history for the click Probabilistic Prediction Model
The desired value of information is reward function, is inputted using the parameter information that object has been displayed in each history as state, is respectively gone through with described
The click probability of the parameter information of object has been displayed as action output in history.
Optionally, described device further include:
Determining module, for determining that the cumulative award desired value of object has been displayed in each history respectively, wherein described each
The cumulative award desired value that object has been displayed in history is to be had been displayed according to multiple history including object currently has been displayed pair
What the user's operation information of elephant determined；
Object has been displayed for being greater than any history in the cumulative award desired value of the current display object in training module
Cumulative award desired value in the case where, with it is described it is current display object parameter information and user's operation information be training sample
This, is trained update to the click Probabilistic Prediction Model.
Optionally, the determining module includes:
Submodule is obtained, the user for object to be had been displayed in multiple history including the current display object grasps
Make information, input preset reward function respectively, with obtain include it is described it is current show object including multiple history have been displayed pair
The reward value of elephant；
First determines submodule, for object to have been displayed according to multiple history including the current display object
Reward value determines the cumulative award desired value of the current display object.
Optionally, the user's operation information that object has been displayed further include: the object that has been displayed is in default viewing area
The time persistently exposed in domain；
The acquisition submodule includes:
Second determines submodule, for having been displayed in object for multiple history including the current display object
Each history object has been displayed, according to following formula, determine that the reward value of object has been displayed in the history:
R=R1+aR2
Wherein, R indicates that the reward value of object, R has been displayed in the history1Indicate that object has been displayed whether by user's point in the history
It hits, a indicates default weight, R2Indicate that the history has been displayed object samples and shows that object persistently exposes in the default display area
The time of light.
In conjunction with above each embodiment, one embodiment of the application provides a kind of device of prediction click probability.With reference to Fig. 6,
Fig. 6 is the schematic diagram of the device for the transmission object to be shown that one embodiment of the application provides.As shown in fig. 6, the device is except including
It obtains except module 501 and prediction module 502, further includes:
Sending module 503 predicts for the click probability to multiple objects to be shown, and prediction is obtained
The object to be shown that probability is clicked greater than predetermined probabilities is sent to display terminal.
Optionally, the sending module includes:
Sending submodule, for the multiple object to be shown to be sent to the display eventually according to probability descending is clicked
End.
Based on the same inventive concept, another embodiment of the application provides a kind of computer readable storage medium, stores thereon
There is computer program, the step in the method as described in any of the above-described embodiment of the application is realized when which is executed by processor
Suddenly.
Based on the same inventive concept, one embodiment of the application provides a kind of electronic equipment.It is the application one with reference to Fig. 7, Fig. 7
The schematic diagram for the electronic equipment that embodiment proposes.As shown in fig. 7, electronic equipment 100 includes: memory 110 and processor 120,
It is connected between memory 110 and processor 120 by bus communication, is stored with computer program in memory 110, the computer
Program can be run on processor 120, and then realize the step in method described in any of the above-described embodiment of the application.
For device embodiment, since it is basically similar to the method embodiment, related so being described relatively simple
Place illustrates referring to the part of embodiment of the method.
All the embodiments in this specification are described in a progressive manner, the highlights of each of the examples are with
The difference of other embodiments, the same or similar parts between the embodiments can be referred to each other.
It should be understood by those skilled in the art that, the embodiments of the present application may be provided as method, apparatus or calculating
Machine program product.Therefore, the embodiment of the present application can be used complete hardware embodiment, complete software embodiment or combine software and
The form of the embodiment of hardware aspect.Moreover, the embodiment of the present application can be used one or more wherein include computer can
With in the computer-usable storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) of program code
The form of the computer program product of implementation.
The embodiment of the present application is referring to according to the method for the embodiment of the present application, terminal device (system) and computer program
The flowchart and/or the block diagram of product describes.It should be understood that flowchart and/or the block diagram can be realized by computer program instructions
In each flow and/or block and flowchart and/or the block diagram in process and/or box combination.It can provide these
Computer program instructions are set to general purpose computer, special purpose computer, Embedded Processor or other programmable data processing terminals
Standby processor is to generate a machine, so that being held by the processor of computer or other programmable data processing terminal devices
Capable instruction generates for realizing in one or more flows of the flowchart and/or one or more blocks of the block diagram
The device of specified function.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing terminal devices
In computer-readable memory operate in a specific manner, so that instruction stored in the computer readable memory generates packet
The manufacture of command device is included, which realizes in one side of one or more flows of the flowchart and/or block diagram
The function of being specified in frame or multiple boxes.
These computer program instructions can also be loaded into computer or other programmable data processing terminal devices, so that
Series of operation steps are executed on computer or other programmable terminal equipments to generate computer implemented processing, thus
The instruction executed on computer or other programmable terminal equipments is provided for realizing in one or more flows of the flowchart
And/or in one or more blocks of the block diagram specify function the step of.
Although preferred embodiments of the embodiments of the present application have been described, once a person skilled in the art knows bases
This creative concept, then additional changes and modifications can be made to these embodiments.So the following claims are intended to be interpreted as
Including preferred embodiment and all change and modification within the scope of the embodiments of the present application.
Finally, it is to be noted that, herein, relational terms such as first and second and the like be used merely to by
One entity or operation are distinguished with another entity or operation, without necessarily requiring or implying these entities or operation
Between there are any actual relationship or orders.Moreover, the terms "include", "comprise" or its any other variant meaning
Covering non-exclusive inclusion, so that process, method, article or terminal device including a series of elements not only wrap
Those elements are included, but also including other elements that are not explicitly listed, or further includes for this process, method, article
Or the element that terminal device is intrinsic.In the absence of more restrictions, being wanted by what sentence "including a ..." limited
Element, it is not excluded that there is also other identical elements in process, method, article or the terminal device for including the element.
Method, apparatus, storage medium and the electronic equipment for clicking probability to a kind of prediction provided herein above, into
It has gone and has been discussed in detail, specific examples are used herein to illustrate the principle and implementation manner of the present application, the above implementation
The explanation of example is merely used to help understand the present processes and its core concept；Meanwhile for the general technology people of this field
Member, according to the thought of the application, there will be changes in the specific implementation manner and application range, in conclusion this explanation
Book content should not be construed as the limitation to the application.
Claims (10)
1. a kind of method that probability is clicked in prediction, which is characterized in that the described method includes:
The parameter information that object has been displayed is obtained, the parameter information that object has been displayed includes at least this and the class of object has been displayed
Type, the user's operation information that object has been displayed include at least this and have been displayed whether object is clicked by user；
According to the parameter information of object to be shown and Probabilistic Prediction Model is clicked, predicts the click probability of the object to be shown；
Wherein, historical user's operation information under the parameter information of object has been displayed with each history for the click Probabilistic Prediction Model
Desired value be reward function, using each history have been displayed object parameter information be state input, with each history
The click probability for showing the parameter information of object is action output.
2. the method according to claim 1, wherein the method also includes:
Determine that the cumulative award desired value of object has been displayed in each history respectively, wherein object has been displayed in each history
Cumulative award desired value is that the user's operation information of object has been displayed according to multiple history including object currently has been displayed
Determining；
It is greater than the cumulative award desired value that object has been displayed in any history in the cumulative award desired value of the current display object
In the case where, using the parameter information of the current display object and user's operation information as training sample, to the click probability
Prediction model is trained update.
3. according to the method described in claim 2, it is characterized in that, determining the cumulative award expectation of the current display object
Value, comprising:
Multiple history including the current display object are had been displayed to the user's operation information of object, input is default respectively
Reward function, to obtain the reward value that object has been displayed in multiple history including the current display object；
The reward value that object has been displayed according to multiple history including the current display object, determines the current display
The cumulative award desired value of object.
4. according to the method described in claim 3, it is characterized in that, the user's operation information that object has been displayed further include:
The time that object has been displayed and is persistently exposed in default display area；
Multiple history including the current display object are had been displayed to the user's operation information of object, input is default respectively
Reward function, to obtain the reward value that object has been displayed in multiple history including the current display object, comprising:
Each history in object has been displayed for multiple history including the current display object, object has been displayed, presses
According to following formula, determine that the reward value of object has been displayed in the history:
R=R1+aR2
Wherein, R indicates that the reward value of object, R has been displayed in the history1Indicate that the history has been displayed whether object is clicked by user, a
Indicate default weight, R2Indicate that the history has been displayed object samples and shows what object persistently exposed in the default display area
Time.
5. method according to claim 1 to 4, which is characterized in that the method also includes:
The click probability predicted the click probability of multiple objects to be shown, and prediction is obtained is greater than predetermined probabilities
Object to be shown be sent to display terminal.
6. according to the method described in claim 5, it is characterized in that, by prediction obtain click probability be greater than predetermined probabilities to
Display object is sent to display terminal, comprising:
The multiple object to be shown is sent to the display terminal according to probability descending is clicked.
7. the device that probability is clicked in a kind of prediction, which is characterized in that described device includes:
Module is obtained, for obtaining the parameter information that object has been displayed, the parameter information that object has been displayed includes at least should
The type of object has been displayed, the user's operation information that object has been displayed includes at least this and object has been displayed whether by user's point
It hits；
Prediction module, for the parameter information and click Probabilistic Prediction Model according to object to be shown, it is described to be shown right to predict
The click probability of elephant；
Wherein, historical user's operation information under the parameter information of object has been displayed with each history for the click Probabilistic Prediction Model
Desired value be reward function, using each history have been displayed object parameter information be state input, with each history
The click probability for showing the parameter information of object is action output.
8. device according to claim 7, which is characterized in that described device further include:
Determining module, for determining that the cumulative award desired value of object has been displayed in each history respectively, wherein each history
The cumulative award desired value that object has been displayed is that object has been displayed according to multiple history including object currently has been displayed
What user's operation information determined；
The tired of object has been displayed for being greater than any history in the cumulative award desired value of the current display object in training module
It is right using the parameter information of the current display object and user's operation information as training sample in the case that bonuses distributed according to strict calculations encourages desired value
The click Probabilistic Prediction Model is trained update.
9. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the program is held by processor
The step in the method as described in claim 1-6 is any is realized when row.
10. a kind of electronic equipment including memory, processor and stores the calculating that can be run on a memory and on a processor
Machine program, which is characterized in that the step of method as described in claim 1-6 is any is realized when the processor executes.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910024668.2A CN109886729A (en) | 2019-01-10 | 2019-01-10 | Method, apparatus, electronic equipment and the readable storage medium storing program for executing of probability are clicked in prediction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910024668.2A CN109886729A (en) | 2019-01-10 | 2019-01-10 | Method, apparatus, electronic equipment and the readable storage medium storing program for executing of probability are clicked in prediction |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109886729A true CN109886729A (en) | 2019-06-14 |
Family
ID=66925872
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910024668.2A Pending CN109886729A (en) | 2019-01-10 | 2019-01-10 | Method, apparatus, electronic equipment and the readable storage medium storing program for executing of probability are clicked in prediction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109886729A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111260416A (en) * | 2020-02-13 | 2020-06-09 | 支付宝(杭州)信息技术有限公司 | Method and device for determining associated user of object |
| CN112214387A (en) * | 2020-10-13 | 2021-01-12 | 中国银行股份有限公司 | Knowledge graph-based user operation behavior prediction method and device |
| CN112784151A (en) * | 2019-11-08 | 2021-05-11 | 北京搜狗科技发展有限公司 | Method and related device for determining recommendation information |
| CN112801700A (en) * | 2021-01-29 | 2021-05-14 | 北京达佳互联信息技术有限公司 | Virtual object changing method and device, electronic device and storage medium |
-
2019
- 2019-01-10 CN CN201910024668.2A patent/CN109886729A/en active Pending
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112784151A (en) * | 2019-11-08 | 2021-05-11 | 北京搜狗科技发展有限公司 | Method and related device for determining recommendation information |
| CN112784151B (en) * | 2019-11-08 | 2024-02-06 | 北京搜狗科技发展有限公司 | Method and related device for determining recommended information |
| CN111260416A (en) * | 2020-02-13 | 2020-06-09 | 支付宝(杭州)信息技术有限公司 | Method and device for determining associated user of object |
| CN112214387A (en) * | 2020-10-13 | 2021-01-12 | 中国银行股份有限公司 | Knowledge graph-based user operation behavior prediction method and device |
| CN112801700A (en) * | 2021-01-29 | 2021-05-14 | 北京达佳互联信息技术有限公司 | Virtual object changing method and device, electronic device and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR101121280B1 (en) | Dynamic layout for a search engine results page based on implicit user feedback | |
| US20220292395A1 (en) | Method and system for creating a predictive model for targeting web-page to a surfer | |
| CN109886729A (en) | Method, apparatus, electronic equipment and the readable storage medium storing program for executing of probability are clicked in prediction | |
| US9396436B2 (en) | Method and system for providing targeted content to a surfer | |
| CN110060090A (en) | Method, apparatus, electronic equipment and the readable storage medium storing program for executing of Recommendations combination | |
| US20020062247A1 (en) | Selecting web site content to be displayed to a web site visitor based upon a probability distribution | |
| JP2013546086A (en) | Realization of predictive web browsing | |
| CN101432713A (en) | Method and system for providing content to users based on frequency of interaction | |
| CN105302903A (en) | Search method, apparatus and system and search result sequence adjustment basis determination method | |
| WO2012113791A1 (en) | Systems, methods, and media for executing and optimizing online marketing initiatives | |
| US20140274354A1 (en) | Intelligent merchandising of games | |
| CN110175264A (en) | Construction method, server and the computer readable storage medium of video user portrait | |
| CN109508218A (en) | The push of APP message shows control method, device, equipment and storage medium | |
| WO2014118524A1 (en) | Method and apparatus for accessing web sites | |
| US20110016161A1 (en) | System and method for the selection of context sensitive user profile fragments | |
| CN111371605A (en) | Network service scoring method and device | |
| CN108153891A (en) | Active time statistical method of surfing the Internet and device | |
| CN106648301A (en) | Dynamic adjustment method for classifying display of applications | |
| US20160107086A1 (en) | Game Retention Value Optimization System | |
| JP2009070156A (en) | Information retrieval system and information retrieval method | |
| JP2024502609A (en) | Providing ambient information and associated systems and devices based on learned user context and interactions | |
| US20230350969A1 (en) | Automatic display control method for web content | |
| Paireekreng et al. | Adaptive mobile content personalisation using the time-of-day | |
| WO2017220951A1 (en) | System and methods for customized internet searching and advertising |
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 |