CN105160548A - Method and apparatus for predicting advertisement click-through rate - Google Patents
Method and apparatus for predicting advertisement click-through rate Download PDFInfo
- Publication number
- CN105160548A CN105160548A CN201510515917.XA CN201510515917A CN105160548A CN 105160548 A CN105160548 A CN 105160548A CN 201510515917 A CN201510515917 A CN 201510515917A CN 105160548 A CN105160548 A CN 105160548A
- Authority
- CN
- China
- Prior art keywords
- advertisement
- cross feature
- click
- amount
- cross
- 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
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention provides a solution for predicting an advertisement click-through rate. The solution comprises: acquiring characteristic related information of multiple characteristic types related to multiple past delivered advertisements in a predetermined past time period; performing cross combination on characteristic related information of at least two characteristic types of each past delivered advertisement, to determine multiple cross characteristic sets, and calculating to determine cross characteristic identifiers separately corresponding to the multiple cross characteristic sets; extracting an advertisement display quantity and an advertisement click quantity corresponding to each cross characteristic set, and calculating to determine an advertisement click-through rate corresponding to each cross characteristic set, so as to use the advertisement click-through rate as a cross characteristic value; performing training on a logistic regression model based on the cross characteristic identifiers and the cross characteristic values separately corresponding to the multiple cross characteristic sets, and calculating to determine a model training parameter; and performing prediction calculation on an advertisement click-through rate of a to-be-predicted advertisement based on the model training parameter. According to the solution, more reliable training data is provided for prediction calculation of an advertisement, so that accuracy of a prediction calculation result of an advertisement click-through rate is ensured.
Description
Technical field
The present invention relates to field of computer technology, specifically, the present invention relates to method and device that ad click rate is predicted.
Background technology
Along with the rise of internet, the web advertisement becomes the main profit mode of Ge great portal website, search engine, social networks and each terminal called appliance applications.The market of current Internet advertising increases with surprising rapidity, and the effectiveness that Internet advertising plays more and more seems important.But the number due to advertisement position is limited, therefore need a kind of scheme accurately estimating ad click rate, for advertisement putting side and the side of representing provide strong reference frame reliably and decision information.The error estimated between scenario outcomes and actual value of prior art is comparatively large, can not estimate accurately the clicking rate of advertisement to be predicted, thus cannot provide good data reference frame for improving advertisement delivery effect further.
Summary of the invention
For overcoming above-mentioned technical matters or solving the problems of the technologies described above at least in part, the following technical scheme of special proposition:
Embodiments of the invention propose a kind of method predicted ad click rate, comprising:
Obtain the feature relevant information of throwing in the relevant various features type of advertisement within the order history time period to multiple history;
The feature relevant information of each history being thrown at least two kinds of characteristic types of advertisement carries out combined crosswise to determine multiple cross feature collection, and calculates the cross feature mark determining that multiple cross feature collection is corresponding respectively;
Be extracted in showing advertisement amount and ad click amount that each cross feature set pair in the described order history time period answers, and calculate and determine that ad click rate that each cross feature set pair described answers is using as cross feature value;
The cross feature mark corresponding respectively based on multiple cross feature collection and cross feature value, train Logic Regression Models, to calculate Confirming model training parameter;
Based on described model training parameter, multiple advertisement to be predicted is carried out to the prediction and calculation of ad click rate.
Wherein, described characteristic type comprises following at least any one:
Throw in user; Throw in context; History throws in advertisement.
Wherein, the feature relevant information of described input user comprises following at least any one:
Customer attribute information; The history interest information of user; The real-time interest information of user.
Wherein, described input contextual feature relevant information comprises following at least any one: terminal device relevant information; Network related information; Geographical location information; Showing advertisement relevant information.
Preferably, the method also comprises:
Judge whether the showing advertisement amount that each cross feature set pair described is answered and/or ad click amount meet reprocessing condition;
When the showing advertisement amount that cross feature set pair is answered and/or ad click amount meet reprocessing condition, described calculating determines that ad click rate that each cross feature set pair described answers is using the step as cross feature value, specifically comprises:
The showing advertisement amount of answer this cross feature set pair and ad click amount carry out the process of Bayes's smoothing computation, to determine that ad click rate that this cross feature set pair answers is using as cross feature value.
Wherein, described reprocessing condition comprises following arbitrary situation:
The showing advertisement amount that cross feature set pair is answered is less than the predetermined amount of representing threshold value;
The ad click amount that cross feature set pair is answered is less than predetermined click volume threshold value;
The ratio of the showing advertisement amount that cross feature set pair is answered and ad click amount is less than predetermined clicking rate threshold value.
Preferably, based on described model training parameter, advertisement to be predicted is carried out to the prediction and calculation of ad click rate, specifically comprises:
The cross feature mark of the cross feature collection that the feature relevant information obtaining at least two kinds of characteristic types when each advertisement to be predicted is thrown in real time is corresponding after carrying out combined crosswise;
Cross feature based on described model training parameter and described advertisement to be predicted identifies, and carries out prediction and calculation to determine the ad click rate of described advertisement to be predicted.
Preferably, the method also comprises:
According to results of prediction and calculation, screen described multiple advertisement to be predicted.
Another embodiment of the present invention also proposed a kind of results of prediction and calculation according to ad click rate and carries out applying the method for recommending, and it is characterized in that, comprising:
Obtain the feature relevant information relevant to targeted customer;
The feature relevant information of described targeted customer being correlated with and multiple advertisement to be recommended carry out combined crosswise respectively to determine multiple cross feature collection, and calculate the cross feature mark determining that the plurality of cross feature set pair is answered;
The cross feature obtaining model training parameter and described multiple advertisement to be recommended based on training in advance identifies, and carries out prediction and calculation to determine the ad click rate of described multiple advertisement to be recommended;
According to the ad click rate that prediction and calculation obtains, described multiple advertisement to be recommended is screened;
Described targeted customer is recommended in the advertisement one or more to be recommended filtered out.
Wherein, the described feature relevant information relevant to targeted customer, specifically comprises:
Belong to the feature relevant information of throwing in user;
Or belong to the contextual feature relevant information of input.
Another embodiment of the present invention proposes a kind of device predicted ad click rate, comprising:
Feature acquisition module, for obtaining the feature relevant information of throwing in the relevant various features type of advertisement within the order history time period to multiple history;
Signature identification computing module, the feature relevant information at least two kinds of characteristic types each history being thrown in advertisement carries out combined crosswise to determine multiple cross feature collection, and calculates the cross feature mark determining that multiple cross feature collection is corresponding respectively;
Characteristic value calculating module, for being extracted in the showing advertisement amount and ad click amount that in the described order history time period, each cross feature set pair is answered, and calculating and determines that ad click rate that each cross feature set pair described answers is using as cross feature value;
Training module, identifies and cross feature value for the cross feature corresponding respectively based on multiple cross feature collection, trains, to calculate Confirming model training parameter to Logic Regression Models;
First predictor computation module, for carrying out the prediction and calculation of ad click rate to multiple advertisement to be predicted based on described model training parameter.
Wherein, described characteristic type comprises following at least any one:
Throw in user; Throw in context; History throws in advertisement.
Wherein, the feature relevant information of described input user comprises following at least any one: customer attribute information; The history interest information of user; The real-time interest information of user.
Wherein, described input contextual feature relevant information comprises following at least any one: terminal device relevant information; Network related information; Geographical location information; Showing advertisement relevant information.
Preferably, this device also comprises:
Judge module, for judging whether the showing advertisement amount that each cross feature set pair described is answered and/or ad click amount meet reprocessing condition;
When the showing advertisement amount that cross feature set pair is answered and/or ad click amount meet reprocessing condition, described characteristic value calculating module carries out the process of Bayes's smoothing computation specifically for the showing advertisement amount of answering this cross feature set pair and ad click amount, to determine that ad click rate that this cross feature set pair answers is using as cross feature value.
Wherein, described reprocessing condition comprises following arbitrary situation:
The showing advertisement amount that cross feature set pair is answered is less than the predetermined amount of representing threshold value; The ad click amount that cross feature set pair is answered is less than predetermined click volume threshold value; The ratio of the showing advertisement amount that cross feature set pair is answered and ad click amount is less than predetermined clicking rate threshold value.
Preferably, described predictor computation module specifically comprises:
Real-time characteristic acquiring unit, the cross feature mark of the cross feature collection that the feature relevant information for obtaining at least two kinds of characteristic types when each advertisement to be predicted is thrown in real time is corresponding after carrying out combined crosswise;
Clicking rate predicting unit, for identifying based on the cross feature of described model training parameter and described advertisement to be predicted, carries out prediction and calculation to determine the ad click rate of described advertisement to be predicted.
Preferably, this device also comprises:
First screening module, for according to results of prediction and calculation, screens described multiple advertisement to be predicted.
Another embodiment of the present invention additionally provides a kind of results of prediction and calculation according to ad click rate and carries out applying the device recommended, and comprising:
Relevant information acquisition module, for obtaining the feature relevant information relevant to targeted customer;
Characteristic crossover module, carries out combined crosswise to determine multiple cross feature collection respectively for the feature relevant information of described targeted customer being correlated with and multiple advertisement to be recommended, and calculates the cross feature mark determining that the plurality of cross feature set pair is answered;
Second predictor computation module, identifies for the cross feature obtaining model training parameter and described multiple advertisement to be recommended based on training in advance, carries out prediction and calculation to determine the ad click rate of described multiple advertisement to be recommended;
Second screening module, screens described multiple advertisement to be recommended for the ad click rate obtained according to prediction and calculation;
Recommending module, for recommending described targeted customer by the advertisement one or more to be recommended filtered out.
Wherein, the described feature relevant information relevant to targeted customer, specifically comprises:
Belong to the feature relevant information of throwing in user;
Or belong to the contextual feature relevant information of input.
Another embodiment of the present invention additionally provides a kind of results of prediction and calculation according to ad click rate and carries out applying the terminal device recommended, and it is characterized in that, comprising:
For obtaining the device of the feature relevant information relevant to targeted customer;
Carry out combined crosswise respectively to determine multiple cross feature collection for the feature relevant information of described targeted customer being correlated with and multiple advertisement to be recommended, and calculate the device determining the cross feature mark that the plurality of cross feature set pair is answered;
Cross feature for obtaining model training parameter and described multiple advertisement to be recommended based on training in advance identifies, and carries out prediction and calculation to determine the device of the ad click rate of described multiple advertisement to be recommended;
For the device that the ad click rate obtained according to prediction and calculation screens described multiple advertisement to be recommended;
For the advertisement one or more to be recommended filtered out being recommended the device of described targeted customer.
In technical scheme of the present invention, propose a kind of scheme that ad click rate is predicted, in the Primary Stage Data set-up procedure that ad click rate is predicted, the various features information of history being thrown in advertisement is intersected, make the cross feature after combined crosswise more effectively abundant, the actual input situation of advertisement putting can be reacted more exactly, thus provide training data more effectively reliably for the prediction and calculation of advertisement, avoid the inaccuracy only using single feature in prior art; Meanwhile, the model training parameter that the training data higher according to accuracy obtains can ensure the accuracy of the results of prediction and calculation of ad click rate further, thus carries out screening according to ad click rate to advertisement provide reliable basis for follow-up.
The aspect that the present invention adds and advantage will part provide in the following description, and these will become obvious from the following description, or be recognized by practice of the present invention.
Accompanying drawing explanation
The present invention above-mentioned and/or additional aspect and advantage will become obvious and easy understand from the following description of the accompanying drawings of embodiments, wherein:
Fig. 1 is the process flow diagram to the method that ad click rate is predicted of an embodiment in the present invention;
Fig. 2 is the process flow diagram to the method that ad click rate is predicted of a preferred embodiment in the present invention;
Fig. 3 is the process flow diagram that the results of prediction and calculation according to ad click rate of an embodiment in the present invention carries out applying the method for recommending;
Fig. 4 is the structural representation to the device that ad click rate is predicted of another embodiment in the present invention;
Fig. 5 is the structural representation that the results of prediction and calculation according to ad click rate of another embodiment in the present invention carries out applying the device recommended.
Embodiment
Be described below in detail embodiments of the invention, the example of described embodiment is shown in the drawings, and wherein same or similar label represents same or similar element or has element that is identical or similar functions from start to finish.Being exemplary below by the embodiment be described with reference to the drawings, only for explaining the present invention, and can not limitation of the present invention being interpreted as.
Those skilled in the art of the present technique are appreciated that unless expressly stated, and singulative used herein " ", " one ", " described " and " being somebody's turn to do " also can comprise plural form.Should be further understood that, the wording used in instructions of the present invention " comprises " and refers to there is described feature, integer, step, operation, element and/or assembly, but does not get rid of and exist or add other features one or more, integer, step, operation, element, assembly and/or their group.Should be appreciated that, when we claim element to be " connected " or " coupling " to another element time, it can be directly connected or coupled to other elements, or also can there is intermediary element.In addition, " connection " used herein or " coupling " can comprise wireless connections or wirelessly to couple.Wording "and/or" used herein comprises one or more whole or arbitrary unit listing item be associated and all combinations.
Those skilled in the art of the present technique are appreciated that unless otherwise defined, and all terms used herein (comprising technical term and scientific terminology), have the meaning identical with the general understanding of the those of ordinary skill in field belonging to the present invention.It should also be understood that, those terms defined in such as general dictionary, should be understood to that there is the meaning consistent with the meaning in the context of prior art, unless and by specific definitions as here, otherwise can not explain by idealized or too formal implication.
Fig. 1 is the process flow diagram to the method that ad click rate is predicted of an embodiment in the present invention.
Step S110: obtain the feature relevant information of throwing in the relevant various features type of advertisement within the order history time period to multiple history; Step S120: the feature relevant information of each history being thrown at least two kinds of characteristic types of advertisement carries out combined crosswise to determine multiple cross feature collection, and calculate the cross feature mark determining that multiple cross feature collection is corresponding respectively; Step S130: be extracted in showing advertisement amount and ad click amount that each cross feature set pair in the order history time period answers, and calculate and determine that ad click rate that each cross feature set pair answers is using as cross feature value; Step S140: the cross feature mark corresponding respectively based on multiple cross feature collection and cross feature value, trains Logic Regression Models, to calculate Confirming model training parameter; Step S150: the prediction and calculation of based on model training parameter, multiple advertisement to be predicted being carried out to ad click rate.
Step S110: obtain the feature relevant information of throwing in the relevant various features type of advertisement within the order history time period to multiple history.
Wherein, history throws in the advertisement that advertisement refers to carry out throwing within the order history time period.
Wherein, characteristic type includes but not limited to: throw in user; Throw in context; History throws in advertisement.
Wherein, the feature relevant information of throwing in user includes but not limited to:
Customer attribute information, such as, sex, age, education degree, income, occupational information etc.;
The history interest information of user, such as, advertising message, the info web of user's historical viewings, the purchase information of user's history that user's history is clicked or browsed;
The real-time interest information of user, such as, the current info web browsed of user.
Wherein, throw in contextual feature relevant information to include but not limited to:
Terminal device relevant information, such as, the model of terminal device, OS Type, operating system version;
Network related information, such as, the network operator information of terminal device interconnection network, network type;
Geographical location information, such as, the geographical location information of terminal device;
Showing advertisement relevant information, such as, advertisement size information, ad placement.
Such as, throw in advertisement for multiple history, each history obtained in the past in three months is thrown in the feature relevant information of the customer attribute information of advertisement, is thrown in the advertisement and identifier of contextual feature relevant information and history input advertisement.Such as, throw in advertisement for a history, obtain user ID, user's sex of relative input user, throw in the operation system information of contextual terminal device, throw in the APP_ID of application.
Step S120: the feature relevant information of each history being thrown at least two kinds of characteristic types of advertisement carries out combined crosswise to determine multiple cross feature collection, and calculate the cross feature mark determining that multiple cross feature collection is corresponding respectively.
Particularly, two or more the feature relevant information of each history being thrown in the various features type of advertisement carries out combined crosswise to determine multiple cross feature collection, subsequently, by the such as mode such as hash algorithm, calculate the numerical value with uniqueness determining that each cross feature collection is corresponding respectively, this numerical value is identified as cross feature, as cross feature ID.
Such as, connect example, advertisement is thrown in for a history, by user ID, user's sex of relative input user, throw in the operation system information of contextual terminal device, throw in application APP_ID carry out combined crosswise, determine multiple cross feature collection, as shown in table 1 below, every a line represents a cross feature collection; Subsequently, each feature relevant information that cross feature is concentrated all is characterized by numerical expression mode, calculated the numerical value with uniqueness determining that each cross feature collection is corresponding respectively by hash algorithm, using this numerical value as cross feature ID.
Table 1
Advertisement | User ID | User's sex | Operation system information | APP_ID |
AD1 | cc5f70ccb7ea3d1cde36edb866ee6265 | male | Samsung | 839227 |
AD1 | 5c30e6b9d1d35ef3bfa71b1d686ccfa6 | female | MIUI | 839228 |
AD1 | …… | …… | …… | …… |
AD1 | 283ae80f4d124851170f4d2e73764afd | female | Meizu | 839200 |
Step S130: be extracted in showing advertisement amount and ad click amount that each cross feature set pair in the order history time period answers, and calculate and determine that ad click rate that each cross feature set pair answers is using as cross feature value.
Particularly, advertisement is thrown in for a history, be extracted in corresponding showing advertisement amount and ad click amount when feature relevant information that each cross feature in the order history time period concentrates all occurs, subsequently, the ratio calculating showing advertisement amount and ad click amount determines that ad click rate that each cross feature set pair answers is using as cross feature value.
Such as, as shown in table 2 below, throw in advertisement AD1 for history, the showing advertisement amount that the cross feature set pair determined when user ID, user's sex, operation system information and APP_ID occur simultaneously in three months of being extracted in over is answered and ad click amount, subsequently, by formula 1):
CTR=PV/Click formula 1)
Calculate and determine cross feature value, wherein, CTR represents cross feature value, and PV represents showing advertisement amount, and Click represents ad click amount.
Table 2
Advertisement | User ID | User's sex | Operation system information | APP_ID | Cross feature value |
AD1 | cc5f70ccb7ea3d1cde36edb866ee6265 | male | Samsung | 839227 | PV/Click |
AD1 | 5c30e6b9d1d35ef3bfa71b1d686ccfa6 | female | MIUI | 839228 | PV/Click |
AD1 | …… | …… | …… | …… | …… |
AD1 | 283ae80f4d124851170f4d2e73764afd | female | Meizu | 839200 | PV/Click |
Preferably, the method also comprises step S160 (not shown); Step S160: judge whether the showing advertisement amount that each cross feature set pair is answered and/or ad click amount meet reprocessing condition; When the showing advertisement amount that cross feature set pair is answered and/or ad click amount meet reprocessing condition, step S130: the showing advertisement amount of answer this cross feature set pair and ad click amount carry out the process of Bayes's smoothing computation, to determine that ad click rate that this cross feature set pair answers is using as cross feature value.
Wherein, reprocessing condition comprises following arbitrary situation:
The showing advertisement amount that cross feature set pair is answered is less than the predetermined amount of representing threshold value;
The ad click amount that cross feature set pair is answered is less than predetermined click volume threshold value;
The ratio of the showing advertisement amount that cross feature set pair is answered and ad click amount is less than predetermined clicking rate threshold value.
Particularly, judge whether the showing advertisement amount that each cross feature set pair extracted is answered and/or ad click amount meet reprocessing condition, when the showing advertisement amount that cross feature set pair is answered is less than the predetermined amount of representing threshold value, or when the ad click amount that cross feature set pair is answered is less than predetermined click volume threshold value, or when the ratio of the showing advertisement amount that cross feature set pair is answered and ad click amount is less than predetermined clicking rate threshold value, by following formula 2):
CTR=(Click+ α)/(PV+ alpha+beta) formula 2)
Wherein, CTR represents cross feature value, and PV represents showing advertisement amount, and Click represents ad click amount, α and β is regulating parameter;
The showing advertisement amount of answer this cross feature set pair and ad click amount carry out the process of Bayes's smoothing computation, to determine that ad click rate that this cross feature set pair answers is using as cross feature value.
Such as, the showing advertisement amount of answering when cross feature set pair is 1 time, when being less than the predetermined amount of representing threshold value 500 times, and ad click amount is 1 time, when being less than predetermined click volume threshold value 10 times, the showing advertisement amount of answer this cross feature set pair and ad click amount are by formula 2) carry out the process of Bayes's smoothing computation, to determine that ad click rate that this cross feature set pair answers is using as cross feature value.
In this preferred embodiment, by adopting the mode of Bayes's smoothing computation, ad click rate is optimized, with the special case such as avoid the too low or clicking rate of the amount of representing in advertisement putting too low to the data inaccuracy problem calculating the ad click rate that obtains and bring, because the ad click rate after smoothing processing is more close to actual value, follow-up input training data more accurately when Logic Regression Models is trained can be ensured.
Step S140: the cross feature mark corresponding respectively based on multiple cross feature collection and cross feature value, trains Logic Regression Models, to calculate Confirming model training parameter.
Particularly, the cross feature mark corresponding respectively based on multiple cross feature collection and cross feature value, the i.e. data corresponding relation of cross feature ID-CTR, assembling formed for train and test logic regression model training examples and test sample, Logic Regression Models is trained, and analytical Calculation Confirming model training parameter.
Step S150: the prediction and calculation of based on model training parameter, multiple advertisement to be predicted being carried out to ad click rate.
As shown in Figure 2, step S150 specifically comprises step S151 and step S152; Step S151: the cross feature mark of the cross feature collection that the feature relevant information obtaining at least two kinds of characteristic types when each advertisement to be predicted is thrown in real time is corresponding after carrying out combined crosswise; Step S152: the cross feature based on model training parameter and advertisement to be predicted identifies, carries out prediction and calculation to determine the ad click rate of advertisement to be predicted.
Particularly, the cross feature collection that the feature relevant information obtaining at least two kinds of characteristic types when each advertisement to be predicted is thrown in real time is corresponding after carrying out combined crosswise, and calculate cross feature and corresponding cross feature mark, as cross feature ID; Subsequently, the product of the cross feature ID of computation model training parameter and advertisement to be predicted is using the ad click rate after the prediction and calculation as advertisement to be predicted.
In the preferred embodiment, the direct impact prediction result of calculation of model training parameter, ad click rate after the higher then prediction and calculation of model training parameter is higher, can think that the clicked possibility of advertisement to be predicted is higher, ad click rate after prediction and calculation can be used as the evaluation criterion to advertisement to be predicted, for the follow-up screening to advertisement provides reliable basis.
In a preferred embodiment, the method also comprises step S170 (not shown); Step S170: according to results of prediction and calculation, screens multiple advertisement to be predicted.
Particularly, the ad click rate of the advertisement multiple to be predicted obtained after prediction and calculation and predetermined prediction clicking rate threshold value are compared, when predicting that the ad click rate obtained is less than predetermined prediction clicking rate threshold value, the corresponding advertisement to be predicted of filtering; Or the ad click rate of the advertisement multiple to be predicted obtained after prediction and calculation is sorted, subsequently, filtering is sorted the advertisement to be predicted of predetermined number rearward.
In this preferred embodiment, the ad click rate obtained after prediction and calculation can react the input effect of advertisement to a great extent, carry out the screening and filtering of advertisement based on the ad click rate after prediction and calculation, can be advertisement putting side and provide reliable advertisement putting reference frame with the side of representing.
In technical scheme of the present invention, propose a kind of scheme that ad click rate is predicted, in the Primary Stage Data set-up procedure that ad click rate is predicted, the various features information of history being thrown in advertisement is intersected, make the cross feature after combined crosswise more effectively abundant, the actual input situation of advertisement putting can be reacted more exactly, thus provide training data more effectively reliably for the prediction and calculation of advertisement, avoid the inaccuracy only using single feature in prior art; Meanwhile, the model training parameter that the training data higher according to accuracy obtains can ensure the accuracy of the results of prediction and calculation of ad click rate further, thus carries out screening according to ad click rate to advertisement provide reliable basis for follow-up.
Fig. 3 is the process flow diagram that the results of prediction and calculation according to ad click rate of an embodiment in the present invention carries out applying the method for recommending.
Step S310: obtain the feature relevant information relevant to targeted customer; Step S320: the feature relevant information of targeted customer being correlated with and multiple advertisement to be recommended carry out combined crosswise respectively to determine multiple cross feature collection, and calculate the cross feature mark determining that the plurality of cross feature set pair is answered; Step S330: the cross feature obtaining model training parameter and multiple advertisement to be recommended based on training in advance identifies, carries out prediction and calculation to determine the ad click rate of multiple advertisement to be recommended; Step S340: multiple advertisement to be recommended is screened according to the ad click rate that prediction and calculation obtains; Step S350: targeted customer is recommended in the advertisement one or more to be recommended filtered out.
Wherein, the feature relevant information relevant to targeted customer, specifically comprises:
Belong to the feature relevant information of throwing in user;
Or belong to the contextual feature relevant information of input.
Fig. 4 is the structural framing schematic diagram to the device that ad click rate is predicted of an embodiment in the present invention.
Feature acquisition module S410 obtains the feature relevant information of throwing in the relevant various features type of advertisement within the order history time period to multiple history; The feature relevant information that each history is thrown at least two kinds of characteristic types of advertisement by signature identification computing module S420 carries out combined crosswise to determine multiple cross feature collection, and calculates the cross feature mark determining that multiple cross feature collection is corresponding respectively; Characteristic value calculating module S430 is extracted in the showing advertisement amount and ad click amount that each cross feature set pair in the order history time period answers, and calculates and determine that ad click rate that each cross feature set pair answers is using as cross feature value; Training module S440 identifies and cross feature value based on the cross feature that multiple cross feature collection is corresponding respectively, trains, to calculate Confirming model training parameter to Logic Regression Models; First predictor computation module S450 carries out the prediction and calculation of ad click rate to multiple advertisement to be predicted based on model training parameter.
Feature acquisition module S410 obtains the feature relevant information of throwing in the relevant various features type of advertisement within the order history time period to multiple history.
Wherein, history throws in the advertisement that advertisement refers to carry out throwing within the order history time period.
Wherein, characteristic type includes but not limited to: throw in user; Throw in context; History throws in advertisement.
Wherein, the feature relevant information of throwing in user includes but not limited to:
Customer attribute information, such as, sex, age, education degree, income, occupational information etc.;
The history interest information of user, such as, advertising message, the info web of user's historical viewings, the purchase information of user's history that user's history is clicked or browsed;
The real-time interest information of user, such as, the current info web browsed of user.
Wherein, throw in contextual feature relevant information to include but not limited to:
Terminal device relevant information, such as, the model of terminal device, OS Type, operating system version;
Network related information, such as, the network operator information of terminal device interconnection network, network type;
Geographical location information, such as, the geographical location information of terminal device;
Showing advertisement relevant information, such as, advertisement size information, ad placement.
Such as, throw in advertisement for multiple history, each history obtained in the past in three months is thrown in the feature relevant information of the customer attribute information of advertisement, is thrown in the advertisement and identifier of contextual feature relevant information and history input advertisement.Such as, throw in advertisement for a history, obtain user ID, user's sex of relative input user, throw in the operation system information of contextual terminal device, throw in the APP_ID of application.
The feature relevant information that each history is thrown at least two kinds of characteristic types of advertisement by signature identification computing module S420 carries out combined crosswise to determine multiple cross feature collection, and calculates the cross feature mark determining that multiple cross feature collection is corresponding respectively.
Particularly, two or more the feature relevant information of each history being thrown in the various features type of advertisement carries out combined crosswise to determine multiple cross feature collection, subsequently, by the such as mode such as hash algorithm, calculate the numerical value with uniqueness determining that each cross feature collection is corresponding respectively, this numerical value is identified as cross feature, as cross feature ID.
Such as, connect example, advertisement is thrown in for a history, by user ID, user's sex of relative input user, throw in the operation system information of contextual terminal device, throw in application APP_ID carry out combined crosswise, determine multiple cross feature collection, as shown in Table 1 above, every a line represents a cross feature collection; Subsequently, each feature relevant information that cross feature is concentrated all is characterized by numerical expression mode, calculated the numerical value with uniqueness determining that each cross feature collection is corresponding respectively by hash algorithm, using this numerical value as cross feature ID.
Characteristic value calculating module S430 is extracted in the showing advertisement amount and ad click amount that each cross feature set pair in the order history time period answers, and calculates and determine that ad click rate that each cross feature set pair answers is using as cross feature value.
Particularly, advertisement is thrown in for a history, be extracted in corresponding showing advertisement amount and ad click amount when feature relevant information that each cross feature in the order history time period concentrates all occurs, subsequently, the ratio calculating showing advertisement amount and ad click amount determines that ad click rate that each cross feature set pair answers is using as cross feature value.
Such as, as shown in Table 2 above, advertisement AD1 is thrown in for history, the showing advertisement amount that the cross feature set pair determined when user ID, user's sex, operation system information and APP_ID occur simultaneously in three months of being extracted in over is answered and ad click amount, subsequently, by above-mentioned formula 1) calculate and determine cross feature value.
Preferably, this device also comprises judge module; Judge module judges whether the showing advertisement amount that each cross feature set pair is answered and/or ad click amount meet reprocessing condition; When the showing advertisement amount that cross feature set pair is answered and/or ad click amount meet reprocessing condition, the showing advertisement amount that characteristic value calculating module S330 answers this cross feature set pair and ad click amount carry out the process of Bayes's smoothing computation, to determine that ad click rate that this cross feature set pair answers is using as cross feature value.
Wherein, reprocessing condition comprises following arbitrary situation:
The showing advertisement amount that cross feature set pair is answered is less than the predetermined amount of representing threshold value;
The ad click amount that cross feature set pair is answered is less than predetermined click volume threshold value;
The ratio of the showing advertisement amount that cross feature set pair is answered and ad click amount is less than predetermined clicking rate threshold value.
Particularly, judge whether the showing advertisement amount that each cross feature set pair extracted is answered and/or ad click amount meet reprocessing condition, when the showing advertisement amount that cross feature set pair is answered is less than the predetermined amount of representing threshold value, or when the ad click amount that cross feature set pair is answered is less than predetermined click volume threshold value, or when the ratio of the showing advertisement amount that cross feature set pair is answered and ad click amount is less than predetermined clicking rate threshold value, by above-mentioned formula 2)
The showing advertisement amount of answer this cross feature set pair and ad click amount carry out the process of Bayes's smoothing computation, to determine that ad click rate that this cross feature set pair answers is using as cross feature value.
Such as, the showing advertisement amount of answering when cross feature set pair is 1 time, when being less than the predetermined amount of representing threshold value 500 times, and ad click amount is 1 time, when being less than predetermined click volume threshold value 10 times, the showing advertisement amount of answer this cross feature set pair and ad click amount are by formula 2) carry out the process of Bayes's smoothing computation, to determine that ad click rate that this cross feature set pair answers is using as cross feature value.
In this preferred embodiment, by adopting the mode of Bayes's smoothing computation, ad click rate is optimized, with the special case such as avoid the too low or clicking rate of the amount of representing in advertisement putting too low to the data inaccuracy problem calculating the ad click rate that obtains and bring, because the ad click rate after smoothing processing is more close to actual value, follow-up input training data more accurately when Logic Regression Models is trained can be ensured.
Training module S440 identifies and cross feature value based on the cross feature that multiple cross feature collection is corresponding respectively, trains, to calculate Confirming model training parameter to Logic Regression Models.
Particularly, the cross feature mark corresponding respectively based on multiple cross feature collection and cross feature value, the i.e. data corresponding relation of cross feature ID-CTR, assembling formed for train and test logic regression model training examples and test sample, Logic Regression Models is trained, and analytical Calculation Confirming model training parameter.
First predictor computation module S450 carries out the prediction and calculation of ad click rate to multiple advertisement to be predicted based on model training parameter.
First predictor computation module S450 specifically comprises real-time characteristic acquiring unit (not shown) and clicking rate predicting unit (not shown); The cross feature mark of the cross feature collection that the feature relevant information of at least two kinds of characteristic types that real-time characteristic acquiring unit obtains when each advertisement to be predicted is thrown in real time is corresponding after carrying out combined crosswise; Clicking rate predicting unit identifies based on the cross feature of model training parameter and advertisement to be predicted, carries out prediction and calculation to determine the ad click rate of advertisement to be predicted.
Particularly, the cross feature collection that the feature relevant information obtaining at least two kinds of characteristic types when each advertisement to be predicted is thrown in real time is corresponding after carrying out combined crosswise, and calculate cross feature and corresponding cross feature mark, as cross feature ID; Subsequently, the product of the cross feature ID of computation model training parameter and advertisement to be predicted is using the ad click rate after the prediction and calculation as advertisement to be predicted.
In the preferred embodiment, the direct impact prediction result of calculation of model training parameter, ad click rate after the higher then prediction and calculation of model training parameter is higher, can think that the clicked possibility of advertisement to be predicted is higher, ad click rate after prediction and calculation can be used as the evaluation criterion to advertisement to be predicted, for the follow-up screening to advertisement provides reliable basis.
In technical scheme of the present invention, propose a kind of scheme that ad click rate is predicted, in the Primary Stage Data set-up procedure that ad click rate is predicted, the various features information of history being thrown in advertisement is intersected, make the cross feature after combined crosswise more effectively abundant, the actual input situation of advertisement putting can be reacted more exactly, thus provide training data more effectively reliably for the prediction and calculation of advertisement, avoid the inaccuracy only using single feature in prior art; Meanwhile, the model training parameter that the training data higher according to accuracy obtains can ensure the accuracy of the results of prediction and calculation of ad click rate further, thus carries out screening according to ad click rate to advertisement provide reliable basis for follow-up.
In a preferred embodiment, this device also comprises the first screening module (not shown); First screening module, according to results of prediction and calculation, screens multiple advertisement to be predicted.
Particularly, by after prediction and calculation to ad click rate and the predetermined prediction clicking rate threshold value of advertisement multiple to be predicted compare, when the ad click rate that prediction obtains is less than predetermined prediction clicking rate threshold value, the corresponding advertisement to be predicted of filtering; Or the ad click rate of the advertisement multiple to be predicted obtained after prediction and calculation is sorted, subsequently, filtering is sorted the advertisement to be predicted of predetermined number rearward.
In this preferred embodiment, the ad click rate obtained after prediction and calculation can react the input effect of advertisement to a great extent, carry out the screening and filtering of advertisement based on the ad click rate after prediction and calculation, can be advertisement putting side and provide reliable advertisement putting reference frame with the side of representing.
Fig. 5 is the structural representation that the results of prediction and calculation according to ad click rate of another embodiment in the present invention carries out applying the device recommended.
Relevant information acquisition module 510 obtains the feature relevant information relevant to targeted customer; The feature relevant information that targeted customer is correlated with by characteristic crossover module 520 and multiple advertisement to be recommended carry out combined crosswise respectively to determine multiple cross feature collection, and calculate the cross feature mark determining that the plurality of cross feature set pair is answered; The cross feature that second predictor computation module 530 obtains model training parameter and multiple advertisement to be recommended based on training in advance identifies, and carries out prediction and calculation to determine the ad click rate of multiple advertisement to be recommended; The ad click rate that second screening module 540 obtains according to prediction and calculation screens multiple advertisement to be recommended; The advertisement one or more to be recommended filtered out is recommended targeted customer by recommending module 550.
Wherein, the feature relevant information relevant to targeted customer, specifically comprises:
Belong to the feature relevant information of throwing in user;
Or belong to the contextual feature relevant information of input.
In a preferred embodiment, the invention provides a kind of results of prediction and calculation according to ad click rate and carry out applying the terminal device recommended, comprising:
For obtaining the device of the feature relevant information relevant to targeted customer; Carry out combined crosswise respectively to determine multiple cross feature collection for the feature relevant information of targeted customer being correlated with and multiple advertisement to be recommended, and calculate the device determining the cross feature mark that the plurality of cross feature set pair is answered; Cross feature for obtaining model training parameter and multiple advertisement to be recommended based on training in advance identifies, and carries out prediction and calculation to determine the device of the ad click rate of multiple advertisement to be recommended; For the device that the ad click rate obtained according to prediction and calculation screens multiple advertisement to be recommended; For the advertisement one or more to be recommended filtered out being recommended the device of targeted customer.
Those skilled in the art of the present technique are appreciated that the one or more equipment that the present invention includes and relate to for performing in operation described in the application.These equipment for required object and specialized designs and manufacture, or also can comprise the known device in multi-purpose computer.These equipment have storage computer program within it, and these computer programs optionally activate or reconstruct.Such computer program can be stored in equipment (such as, computing machine) in computer-readable recording medium or be stored in and be suitable for store electrons instruction and be coupled in the medium of any type of bus respectively, described computer-readable medium includes but not limited to that the dish of any type (comprises floppy disk, hard disk, CD, CD-ROM, and magneto-optic disk), ROM (Read-OnlyMemory, ROM (read-only memory)), RAM (RandomAccessMemory, storer immediately), EPROM (ErasableProgrammableRead-OnlyMemory, Erarable Programmable Read only Memory), EEPROM (ElectricallyErasableProgrammableRead-OnlyMemory, EEPROM (Electrically Erasable Programmable Read Only Memo)), flash memory, magnetic card or light card.Namely, computer-readable recording medium comprises and being stored or any medium of transmission information with the form that can read by equipment (such as, computing machine).
Those skilled in the art of the present technique are appreciated that the combination that can realize the frame in each frame in these structural drawing and/or block diagram and/or flow graph and these structural drawing and/or block diagram and/or flow graph with computer program instructions.Those skilled in the art of the present technique are appreciated that, the processor that these computer program instructions can be supplied to multi-purpose computer, special purpose computer or other programmable data disposal routes realizes, thus is performed the scheme of specifying in the frame of structural drawing disclosed by the invention and/or block diagram and/or flow graph or multiple frame by the processor of computing machine or other programmable data disposal routes.
Those skilled in the art of the present technique are appreciated that various operations, method, the step in flow process, measure, the scheme discussed in the present invention can be replaced, changes, combines or delete.Further, there is various operations, method, other steps in flow process, measure, the scheme discussed in the present invention also can be replaced, change, reset, decompose, combine or delete.Further, of the prior art have also can be replaced with the step in operation various disclosed in the present invention, method, flow process, measure, scheme, changed, reset, decomposed, combined or deleted.
The above is only some embodiments of the present invention; it should be pointed out that for those skilled in the art, under the premise without departing from the principles of the invention; can also make some improvements and modifications, these improvements and modifications also should be considered as protection scope of the present invention.
Claims (10)
1. to the method that ad click rate is predicted, it is characterized in that, comprising:
Obtain the feature relevant information of throwing in the relevant various features type of advertisement within the order history time period to multiple history;
The feature relevant information of each history being thrown at least two kinds of characteristic types of advertisement carries out combined crosswise to determine multiple cross feature collection, and calculates the cross feature mark determining that multiple cross feature collection is corresponding respectively;
Be extracted in showing advertisement amount and ad click amount that each cross feature set pair in the described order history time period answers, and calculate and determine that ad click rate that each cross feature set pair described answers is using as cross feature value;
The cross feature mark corresponding respectively based on multiple cross feature collection and cross feature value, train Logic Regression Models, to calculate Confirming model training parameter;
Based on described model training parameter, multiple advertisement to be predicted is carried out to the prediction and calculation of ad click rate.
2. the method that ad click rate is predicted according to claim 1, it is characterized in that, the method also comprises:
Judge whether the showing advertisement amount that each cross feature set pair described is answered and/or ad click amount meet reprocessing condition;
When the showing advertisement amount that cross feature set pair is answered and/or ad click amount meet reprocessing condition, described calculating determines that ad click rate that each cross feature set pair described answers is using the step as cross feature value, specifically comprises:
The showing advertisement amount of answer this cross feature set pair and ad click amount carry out the process of Bayes's smoothing computation, to determine that ad click rate that this cross feature set pair answers is using as cross feature value.
3. the method predicted ad click rate according to claim 2, is characterized in that, described reprocessing condition comprises following arbitrary situation:
The showing advertisement amount that cross feature set pair is answered is less than the predetermined amount of representing threshold value;
The ad click amount that cross feature set pair is answered is less than predetermined click volume threshold value;
The ratio of the showing advertisement amount that cross feature set pair is answered and ad click amount is less than predetermined clicking rate threshold value.
4. the method predicted ad click rate according to any one of claim 1-3, is characterized in that, based on described model training parameter, advertisement to be predicted is carried out to the prediction and calculation of ad click rate, specifically comprises:
The cross feature mark of the cross feature collection that the feature relevant information obtaining at least two kinds of characteristic types when each advertisement to be predicted is thrown in real time is corresponding after carrying out combined crosswise;
Cross feature based on described model training parameter and described advertisement to be predicted identifies, and carries out prediction and calculation to determine the ad click rate of described advertisement to be predicted.
5. carry out applying a method of recommending according to the results of prediction and calculation of ad click rate, it is characterized in that, comprising:
Obtain the feature relevant information relevant to targeted customer;
The feature relevant information of described targeted customer being correlated with and multiple advertisement to be recommended carry out combined crosswise respectively to determine multiple cross feature collection, and calculate the cross feature mark determining that the plurality of cross feature set pair is answered;
The cross feature obtaining model training parameter and described multiple advertisement to be recommended based on training in advance identifies, and carries out prediction and calculation to determine the ad click rate of described multiple advertisement to be recommended;
According to the ad click rate that prediction and calculation obtains, described multiple advertisement to be recommended is screened;
Described targeted customer is recommended in the advertisement one or more to be recommended filtered out.
6. to the device that ad click rate is predicted, it is characterized in that, comprising:
Feature acquisition module, for obtaining the feature relevant information of throwing in the relevant various features type of advertisement within the order history time period to multiple history;
Signature identification computing module, the feature relevant information at least two kinds of characteristic types each history being thrown in advertisement carries out combined crosswise to determine multiple cross feature collection, and calculates the cross feature mark determining that multiple cross feature collection is corresponding respectively;
Characteristic value calculating module, for being extracted in the showing advertisement amount and ad click amount that in the described order history time period, each cross feature set pair is answered, and calculating and determines that ad click rate that each cross feature set pair described answers is using as cross feature value;
Training module, identifies and cross feature value for the cross feature corresponding respectively based on multiple cross feature collection, trains, to calculate Confirming model training parameter to Logic Regression Models;
First predictor computation module, for carrying out the prediction and calculation of ad click rate to multiple advertisement to be predicted based on described model training parameter.
7. the device predicted ad click rate according to any one of claim 6, is characterized in that, this device also comprises:
Judge module, for judging whether the showing advertisement amount that each cross feature set pair described is answered and/or ad click amount meet reprocessing condition;
When the showing advertisement amount that cross feature set pair is answered and/or ad click amount meet reprocessing condition, described characteristic value calculating module carries out the process of Bayes's smoothing computation specifically for the showing advertisement amount of answering this cross feature set pair and ad click amount, to determine that ad click rate that this cross feature set pair answers is using as cross feature value.
8. the device predicted ad click rate according to claim 7, is characterized in that, described reprocessing condition comprises following arbitrary situation:
The showing advertisement amount that cross feature set pair is answered is less than the predetermined amount of representing threshold value;
The ad click amount that cross feature set pair is answered is less than predetermined click volume threshold value;
The ratio of the showing advertisement amount that cross feature set pair is answered and ad click amount is less than predetermined clicking rate threshold value.
9. carry out applying the device recommended according to the results of prediction and calculation of ad click rate, it is characterized in that, comprising:
Relevant information acquisition module, for obtaining the feature relevant information relevant to targeted customer;
Characteristic crossover module, carries out combined crosswise to determine multiple cross feature collection respectively for the feature relevant information of described targeted customer being correlated with and multiple advertisement to be recommended, and calculates the cross feature mark determining that the plurality of cross feature set pair is answered;
Second predictor computation module, identifies for the cross feature obtaining model training parameter and described multiple advertisement to be recommended based on training in advance, carries out prediction and calculation to determine the ad click rate of described multiple advertisement to be recommended;
Second screening module, screens described multiple advertisement to be recommended for the ad click rate obtained according to prediction and calculation;
Recommending module, for recommending described targeted customer by the advertisement one or more to be recommended filtered out.
10. carry out applying the terminal device recommended according to the results of prediction and calculation of ad click rate, it is characterized in that, comprising:
For obtaining the device of the feature relevant information relevant to targeted customer;
Carry out combined crosswise respectively to determine multiple cross feature collection for the feature relevant information of described targeted customer being correlated with and multiple advertisement to be recommended, and calculate the device determining the cross feature mark that the plurality of cross feature set pair is answered;
Cross feature for obtaining model training parameter and described multiple advertisement to be recommended based on training in advance identifies, and carries out prediction and calculation to determine the device of the ad click rate of described multiple advertisement to be recommended;
For the device that the ad click rate obtained according to prediction and calculation screens described multiple advertisement to be recommended;
For the advertisement one or more to be recommended filtered out being recommended the device of described targeted customer.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510515917.XA CN105160548A (en) | 2015-08-20 | 2015-08-20 | Method and apparatus for predicting advertisement click-through rate |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510515917.XA CN105160548A (en) | 2015-08-20 | 2015-08-20 | Method and apparatus for predicting advertisement click-through rate |
Publications (1)
Publication Number | Publication Date |
---|---|
CN105160548A true CN105160548A (en) | 2015-12-16 |
Family
ID=54801397
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510515917.XA Pending CN105160548A (en) | 2015-08-20 | 2015-08-20 | Method and apparatus for predicting advertisement click-through rate |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105160548A (en) |
Cited By (46)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105630996A (en) * | 2015-12-25 | 2016-06-01 | 腾讯科技(深圳)有限公司 | Information processing method and server |
CN105654200A (en) * | 2015-12-30 | 2016-06-08 | 上海珍岛信息技术有限公司 | Deep learning-based advertisement click-through rate prediction method and device |
CN105654345A (en) * | 2015-12-30 | 2016-06-08 | 上海珍岛信息技术有限公司 | Bayesian network-based advertisement click-through rate prediction method and device |
CN105678335A (en) * | 2016-01-08 | 2016-06-15 | 车智互联(北京)科技有限公司 | Click rate pre-estimation method, device and calculating equipment |
CN105701191A (en) * | 2016-01-08 | 2016-06-22 | 腾讯科技(深圳)有限公司 | Push information click rate estimation method and device |
CN106096629A (en) * | 2016-06-01 | 2016-11-09 | 南京邮电大学 | A kind of based on user between the ad click rate Forecasting Methodology of similarity relation |
CN106294743A (en) * | 2016-08-10 | 2017-01-04 | 北京奇虎科技有限公司 | The recommendation method and device of application function |
CN106886915A (en) * | 2017-01-17 | 2017-06-23 | 华南理工大学 | A kind of ad click predictor method based on time decay sampling |
WO2017118328A1 (en) * | 2016-01-04 | 2017-07-13 | 腾讯科技(深圳)有限公司 | Push information rough selection sorting method, device and computer storage medium |
CN106997549A (en) * | 2017-02-14 | 2017-08-01 | 火烈鸟网络(广州)股份有限公司 | The method for pushing and system of a kind of advertising message |
CN107527243A (en) * | 2017-08-29 | 2017-12-29 | 有米科技股份有限公司 | A kind of method that prediction APP rushes spending on ads needed for list |
CN107590689A (en) * | 2017-08-25 | 2018-01-16 | 北京奇艺世纪科技有限公司 | A kind of ad data recommends method and system |
CN107688582A (en) * | 2016-08-05 | 2018-02-13 | 腾讯科技(深圳)有限公司 | The acquisition methods and device of resource recommendation model |
CN107742221A (en) * | 2016-08-23 | 2018-02-27 | 腾讯科技(深圳)有限公司 | A kind of processing method of promotion message, device and system |
CN107767174A (en) * | 2017-10-19 | 2018-03-06 | 厦门美柚信息科技有限公司 | The Forecasting Methodology and device of a kind of ad click rate |
CN107944589A (en) * | 2016-10-12 | 2018-04-20 | 北京奇虎科技有限公司 | The Forecasting Methodology and prediction meanss of ad click rate |
CN107993078A (en) * | 2017-12-22 | 2018-05-04 | 北京三快在线科技有限公司 | For evaluation information bandwagon effect method and apparatus and computing device |
CN108038541A (en) * | 2017-12-07 | 2018-05-15 | 百度在线网络技术(北京)有限公司 | Method, apparatus, equipment and the computer-readable medium that CTR is estimated |
CN108122122A (en) * | 2016-11-29 | 2018-06-05 | 腾讯科技(深圳)有限公司 | Advertisement placement method and system |
CN108229994A (en) * | 2016-12-21 | 2018-06-29 | 北京金山云网络技术有限公司 | A kind of information-pushing method and device |
CN108875022A (en) * | 2018-06-20 | 2018-11-23 | 北京奇艺世纪科技有限公司 | A kind of video recommendation method and device |
CN109214841A (en) * | 2017-06-30 | 2019-01-15 | 北京金山安全软件有限公司 | Method and device for obtaining advertisement predicted value and terminal |
CN109274987A (en) * | 2018-08-30 | 2019-01-25 | 武汉斗鱼网络科技有限公司 | A kind of video collection sort method, server and readable storage medium storing program for executing |
CN109376381A (en) * | 2018-09-10 | 2019-02-22 | 平安科技(深圳)有限公司 | Method for detecting abnormality, device, computer equipment and storage medium are submitted an expense account in medical insurance |
CN109816412A (en) * | 2017-11-21 | 2019-05-28 | 腾讯科技(深圳)有限公司 | A kind of training pattern generation method, device, equipment and computer storage medium |
CN109858942A (en) * | 2018-11-06 | 2019-06-07 | 北京奇虎科技有限公司 | Promotion message methods of exhibiting, device, electronic equipment and readable storage medium storing program for executing |
CN109902849A (en) * | 2018-06-20 | 2019-06-18 | 华为技术有限公司 | User's behavior prediction method and device, behavior prediction model training method and device |
CN110019163A (en) * | 2017-12-05 | 2019-07-16 | 北京京东尚科信息技术有限公司 | Method, system, equipment and the storage medium of prediction, the recommendation of characteristics of objects |
CN110598845A (en) * | 2019-08-13 | 2019-12-20 | 中国平安人寿保险股份有限公司 | Data processing method, data processing device, computer equipment and storage medium |
CN110796465A (en) * | 2018-08-03 | 2020-02-14 | 阿里巴巴集团控股有限公司 | Resource display equipment determining method, advertisement display screen determining method and device |
CN110825966A (en) * | 2019-10-31 | 2020-02-21 | 广州市百果园信息技术有限公司 | Information recommendation method and device, recommendation server and storage medium |
CN110969460A (en) * | 2018-09-29 | 2020-04-07 | 北京国双科技有限公司 | Method and device for predicting delivery effect of information flow advertisement |
CN111177575A (en) * | 2020-04-07 | 2020-05-19 | 腾讯科技(深圳)有限公司 | Content recommendation method and device, electronic equipment and storage medium |
CN111242690A (en) * | 2020-01-14 | 2020-06-05 | 苏宁云计算有限公司 | Advertisement picture evaluation method and device, storage medium and computer equipment |
CN111566684A (en) * | 2020-04-13 | 2020-08-21 | 支付宝(杭州)信息技术有限公司 | Method and system for optimizing user grouping of advertisements |
CN111626805A (en) * | 2019-02-28 | 2020-09-04 | 北京沃东天骏信息技术有限公司 | Information display method and device |
CN112055038A (en) * | 2019-06-06 | 2020-12-08 | 阿里巴巴集团控股有限公司 | Method for generating click rate estimation model and method for predicting click probability |
CN112101993A (en) * | 2020-09-11 | 2020-12-18 | 厦门美图之家科技有限公司 | Offline anti-cheating method and device, electronic equipment and readable storage medium |
CN112508638A (en) * | 2020-11-30 | 2021-03-16 | 上海哔哩哔哩科技有限公司 | Data processing method and device and computer equipment |
CN113139827A (en) * | 2020-01-20 | 2021-07-20 | 上海哔哩哔哩科技有限公司 | User personalized feature mining method and system |
CN113435523A (en) * | 2021-06-29 | 2021-09-24 | 北京百度网讯科技有限公司 | Method and device for predicting content click rate, electronic equipment and storage medium |
CN113657926A (en) * | 2021-07-28 | 2021-11-16 | 上海明略人工智能(集团)有限公司 | Advertisement effect prediction method, system, electronic device and storage medium |
CN113724016A (en) * | 2021-09-09 | 2021-11-30 | 北京有竹居网络技术有限公司 | Method, device, medium and equipment for acquiring attention degree of multimedia resource |
CN113763000A (en) * | 2020-08-20 | 2021-12-07 | 北京沃东天骏信息技术有限公司 | Method and device for determining object click rate |
CN115545779A (en) * | 2022-10-11 | 2022-12-30 | 西窗科技(苏州)有限公司 | Big data-based advertisement delivery early warning management method and system |
WO2023226353A1 (en) * | 2022-05-26 | 2023-11-30 | 上海二三四五网络科技有限公司 | Method and apparatus for calculating ctr trend content based on click position factor improvement |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103310003A (en) * | 2013-06-28 | 2013-09-18 | 华东师范大学 | Method and system for predicting click rate of new advertisement based on click log |
CN103996088A (en) * | 2014-06-10 | 2014-08-20 | 苏州工业职业技术学院 | Advertisement click-through rate prediction method based on multi-dimensional feature combination logical regression |
CN104536983A (en) * | 2014-12-08 | 2015-04-22 | 北京掌阔技术有限公司 | Method and device for predicting advertisement click rate |
-
2015
- 2015-08-20 CN CN201510515917.XA patent/CN105160548A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103310003A (en) * | 2013-06-28 | 2013-09-18 | 华东师范大学 | Method and system for predicting click rate of new advertisement based on click log |
CN103996088A (en) * | 2014-06-10 | 2014-08-20 | 苏州工业职业技术学院 | Advertisement click-through rate prediction method based on multi-dimensional feature combination logical regression |
CN104536983A (en) * | 2014-12-08 | 2015-04-22 | 北京掌阔技术有限公司 | Method and device for predicting advertisement click rate |
Non-Patent Citations (1)
Title |
---|
FALAO_BEILIU: "广告点击率的贝叶斯平滑", 《HTTP://BLOG.CSDN.NET/MYTESTMY/ARTICLE/DETAILS/19088519》 * |
Cited By (68)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105630996A (en) * | 2015-12-25 | 2016-06-01 | 腾讯科技(深圳)有限公司 | Information processing method and server |
CN105654200A (en) * | 2015-12-30 | 2016-06-08 | 上海珍岛信息技术有限公司 | Deep learning-based advertisement click-through rate prediction method and device |
CN105654345A (en) * | 2015-12-30 | 2016-06-08 | 上海珍岛信息技术有限公司 | Bayesian network-based advertisement click-through rate prediction method and device |
WO2017118328A1 (en) * | 2016-01-04 | 2017-07-13 | 腾讯科技(深圳)有限公司 | Push information rough selection sorting method, device and computer storage medium |
CN105678335B (en) * | 2016-01-08 | 2019-07-02 | 车智互联(北京)科技有限公司 | It estimates the method, apparatus of clicking rate and calculates equipment |
CN105678335A (en) * | 2016-01-08 | 2016-06-15 | 车智互联(北京)科技有限公司 | Click rate pre-estimation method, device and calculating equipment |
CN105701191A (en) * | 2016-01-08 | 2016-06-22 | 腾讯科技(深圳)有限公司 | Push information click rate estimation method and device |
CN105701191B (en) * | 2016-01-08 | 2020-12-29 | 腾讯科技(深圳)有限公司 | Pushed information click rate estimation method and device |
CN106096629A (en) * | 2016-06-01 | 2016-11-09 | 南京邮电大学 | A kind of based on user between the ad click rate Forecasting Methodology of similarity relation |
CN106096629B (en) * | 2016-06-01 | 2019-07-16 | 南京邮电大学 | A kind of ad click rate prediction technique based on similarity relation between user |
CN107688582B (en) * | 2016-08-05 | 2020-07-07 | 腾讯科技(深圳)有限公司 | Resource recommendation model obtaining method and device |
CN107688582A (en) * | 2016-08-05 | 2018-02-13 | 腾讯科技(深圳)有限公司 | The acquisition methods and device of resource recommendation model |
CN106294743A (en) * | 2016-08-10 | 2017-01-04 | 北京奇虎科技有限公司 | The recommendation method and device of application function |
CN107742221A (en) * | 2016-08-23 | 2018-02-27 | 腾讯科技(深圳)有限公司 | A kind of processing method of promotion message, device and system |
CN107944589A (en) * | 2016-10-12 | 2018-04-20 | 北京奇虎科技有限公司 | The Forecasting Methodology and prediction meanss of ad click rate |
CN108122122A (en) * | 2016-11-29 | 2018-06-05 | 腾讯科技(深圳)有限公司 | Advertisement placement method and system |
CN108229994A (en) * | 2016-12-21 | 2018-06-29 | 北京金山云网络技术有限公司 | A kind of information-pushing method and device |
CN106886915B (en) * | 2017-01-17 | 2020-07-28 | 华南理工大学 | Advertisement click estimation method based on time attenuation sampling |
CN106886915A (en) * | 2017-01-17 | 2017-06-23 | 华南理工大学 | A kind of ad click predictor method based on time decay sampling |
CN106997549A (en) * | 2017-02-14 | 2017-08-01 | 火烈鸟网络(广州)股份有限公司 | The method for pushing and system of a kind of advertising message |
CN109214841A (en) * | 2017-06-30 | 2019-01-15 | 北京金山安全软件有限公司 | Method and device for obtaining advertisement predicted value and terminal |
CN107590689A (en) * | 2017-08-25 | 2018-01-16 | 北京奇艺世纪科技有限公司 | A kind of ad data recommends method and system |
CN107527243A (en) * | 2017-08-29 | 2017-12-29 | 有米科技股份有限公司 | A kind of method that prediction APP rushes spending on ads needed for list |
CN107767174A (en) * | 2017-10-19 | 2018-03-06 | 厦门美柚信息科技有限公司 | The Forecasting Methodology and device of a kind of ad click rate |
CN109816412A (en) * | 2017-11-21 | 2019-05-28 | 腾讯科技(深圳)有限公司 | A kind of training pattern generation method, device, equipment and computer storage medium |
CN110019163A (en) * | 2017-12-05 | 2019-07-16 | 北京京东尚科信息技术有限公司 | Method, system, equipment and the storage medium of prediction, the recommendation of characteristics of objects |
CN108038541B (en) * | 2017-12-07 | 2021-02-19 | 百度在线网络技术(北京)有限公司 | CTR (China train redundancy) estimation method, device, equipment and computer readable medium |
CN108038541A (en) * | 2017-12-07 | 2018-05-15 | 百度在线网络技术(北京)有限公司 | Method, apparatus, equipment and the computer-readable medium that CTR is estimated |
CN107993078A (en) * | 2017-12-22 | 2018-05-04 | 北京三快在线科技有限公司 | For evaluation information bandwagon effect method and apparatus and computing device |
CN107993078B (en) * | 2017-12-22 | 2020-08-14 | 北京三快在线科技有限公司 | Method and device for evaluating information display effect and computing equipment |
US11531867B2 (en) | 2018-06-20 | 2022-12-20 | Huawei Technologies Co., Ltd. | User behavior prediction method and apparatus, and behavior prediction model training method and apparatus |
CN109902849A (en) * | 2018-06-20 | 2019-06-18 | 华为技术有限公司 | User's behavior prediction method and device, behavior prediction model training method and device |
CN108875022A (en) * | 2018-06-20 | 2018-11-23 | 北京奇艺世纪科技有限公司 | A kind of video recommendation method and device |
CN108875022B (en) * | 2018-06-20 | 2021-03-02 | 北京奇艺世纪科技有限公司 | Video recommendation method and device |
CN110796465A (en) * | 2018-08-03 | 2020-02-14 | 阿里巴巴集团控股有限公司 | Resource display equipment determining method, advertisement display screen determining method and device |
CN109274987A (en) * | 2018-08-30 | 2019-01-25 | 武汉斗鱼网络科技有限公司 | A kind of video collection sort method, server and readable storage medium storing program for executing |
CN109376381A (en) * | 2018-09-10 | 2019-02-22 | 平安科技(深圳)有限公司 | Method for detecting abnormality, device, computer equipment and storage medium are submitted an expense account in medical insurance |
CN110969460A (en) * | 2018-09-29 | 2020-04-07 | 北京国双科技有限公司 | Method and device for predicting delivery effect of information flow advertisement |
CN110969460B (en) * | 2018-09-29 | 2023-10-31 | 北京国双科技有限公司 | Method and device for predicting putting effect of information flow advertisement |
CN109858942B (en) * | 2018-11-06 | 2023-12-15 | 三六零科技集团有限公司 | Popularization information display method and device, electronic equipment and readable storage medium |
CN109858942A (en) * | 2018-11-06 | 2019-06-07 | 北京奇虎科技有限公司 | Promotion message methods of exhibiting, device, electronic equipment and readable storage medium storing program for executing |
CN111626805B (en) * | 2019-02-28 | 2023-11-03 | 北京沃东天骏信息技术有限公司 | Information display method and device |
CN111626805A (en) * | 2019-02-28 | 2020-09-04 | 北京沃东天骏信息技术有限公司 | Information display method and device |
CN112055038A (en) * | 2019-06-06 | 2020-12-08 | 阿里巴巴集团控股有限公司 | Method for generating click rate estimation model and method for predicting click probability |
CN112055038B (en) * | 2019-06-06 | 2022-04-15 | 阿里巴巴集团控股有限公司 | Method for generating click rate estimation model and method for predicting click probability |
CN110598845B (en) * | 2019-08-13 | 2023-04-07 | 中国平安人寿保险股份有限公司 | Data processing method, data processing device, computer equipment and storage medium |
CN110598845A (en) * | 2019-08-13 | 2019-12-20 | 中国平安人寿保险股份有限公司 | Data processing method, data processing device, computer equipment and storage medium |
CN110825966A (en) * | 2019-10-31 | 2020-02-21 | 广州市百果园信息技术有限公司 | Information recommendation method and device, recommendation server and storage medium |
CN111242690A (en) * | 2020-01-14 | 2020-06-05 | 苏宁云计算有限公司 | Advertisement picture evaluation method and device, storage medium and computer equipment |
CN113139827A (en) * | 2020-01-20 | 2021-07-20 | 上海哔哩哔哩科技有限公司 | User personalized feature mining method and system |
US11893071B2 (en) | 2020-04-07 | 2024-02-06 | Tencent Technology (Shenzhen) Company Limited | Content recommendation method and apparatus, electronic device, and storage medium |
CN111177575A (en) * | 2020-04-07 | 2020-05-19 | 腾讯科技(深圳)有限公司 | Content recommendation method and device, electronic equipment and storage medium |
CN111177575B (en) * | 2020-04-07 | 2020-07-24 | 腾讯科技(深圳)有限公司 | Content recommendation method and device, electronic equipment and storage medium |
WO2020143849A3 (en) * | 2020-04-13 | 2021-02-11 | Alipay (Hangzhou) Information Technology Co., Ltd. | Method and system for optimizing user grouping for advertisement |
US11227309B2 (en) | 2020-04-13 | 2022-01-18 | Alipay (Hangzhou) Information Technology Co., Ltd. | Method and system for optimizing user grouping for advertisement |
CN111566684A (en) * | 2020-04-13 | 2020-08-21 | 支付宝(杭州)信息技术有限公司 | Method and system for optimizing user grouping of advertisements |
CN111566684B (en) * | 2020-04-13 | 2022-05-27 | 支付宝(杭州)信息技术有限公司 | Method and system for optimizing user grouping of advertisements |
CN113763000A (en) * | 2020-08-20 | 2021-12-07 | 北京沃东天骏信息技术有限公司 | Method and device for determining object click rate |
CN112101993B (en) * | 2020-09-11 | 2022-12-23 | 厦门美图之家科技有限公司 | Offline anti-cheating method and device, electronic equipment and readable storage medium |
CN112101993A (en) * | 2020-09-11 | 2020-12-18 | 厦门美图之家科技有限公司 | Offline anti-cheating method and device, electronic equipment and readable storage medium |
CN112508638A (en) * | 2020-11-30 | 2021-03-16 | 上海哔哩哔哩科技有限公司 | Data processing method and device and computer equipment |
CN113435523B (en) * | 2021-06-29 | 2023-09-26 | 北京百度网讯科技有限公司 | Method, device, electronic equipment and storage medium for predicting content click rate |
CN113435523A (en) * | 2021-06-29 | 2021-09-24 | 北京百度网讯科技有限公司 | Method and device for predicting content click rate, electronic equipment and storage medium |
CN113657926A (en) * | 2021-07-28 | 2021-11-16 | 上海明略人工智能(集团)有限公司 | Advertisement effect prediction method, system, electronic device and storage medium |
CN113724016A (en) * | 2021-09-09 | 2021-11-30 | 北京有竹居网络技术有限公司 | Method, device, medium and equipment for acquiring attention degree of multimedia resource |
WO2023226353A1 (en) * | 2022-05-26 | 2023-11-30 | 上海二三四五网络科技有限公司 | Method and apparatus for calculating ctr trend content based on click position factor improvement |
CN115545779A (en) * | 2022-10-11 | 2022-12-30 | 西窗科技(苏州)有限公司 | Big data-based advertisement delivery early warning management method and system |
CN115545779B (en) * | 2022-10-11 | 2023-09-01 | 西窗科技(苏州)有限公司 | Early warning management method and system for advertisement delivery based on big data |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105160548A (en) | Method and apparatus for predicting advertisement click-through rate | |
CN107871244B (en) | Method and device for detecting advertising effect | |
CN105550903B (en) | Target user determination method and device | |
US10354210B2 (en) | Quality prediction | |
CN108205766A (en) | Information-pushing method, apparatus and system | |
Queiruga et al. | Evolution of the electronic waste management system in Spain | |
EP3407285A1 (en) | Target user orientation method and device, and computer storage medium | |
CN108805598B (en) | Similarity information determination method, server and computer-readable storage medium | |
CN105160545B (en) | Method and device for determining release information style | |
CN103310003A (en) | Method and system for predicting click rate of new advertisement based on click log | |
CN110971659A (en) | Recommendation message pushing method and device and storage medium | |
CN115408586B (en) | Intelligent channel operation data analysis method, system, equipment and storage medium | |
CN112149352B (en) | Prediction method for marketing activity clicking by combining GBDT automatic characteristic engineering | |
US20190050890A1 (en) | Video dotting placement analysis system, analysis method and storage medium | |
Wagner et al. | Exploring 14 years of repair records–information retrieval, analysis potential and data gaps to improve reparability | |
CN105808641A (en) | Mining method and device of off-line resources | |
CN110633919A (en) | Method and device for evaluating business entity | |
CA3121421A1 (en) | A system and method of reconstructing browser interaction from session data having incomplete tracking data | |
CN111242713A (en) | Information pushing method and device | |
US20140188941A1 (en) | Using a Graph Database to Match Entities by Evaluating Boolean Expressions | |
Amatuni et al. | Quantifying total lifetimes of consumer products: Stochastic modelling accounting for second-hand use and establishing an open-collaborative database | |
CN106708835A (en) | Data table classification method and device | |
Bano et al. | Addressing the challenges of alignment of requirements and services: a vision for user-centered method | |
CN111445284B (en) | Determination method and device of orientation label, computing equipment and storage medium | |
CN114662005A (en) | Message pushing method, device, equipment and storage medium for user behavior track |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20151216 |