CN105808762A - Resource sequencing method and device - Google Patents
Resource sequencing method and device Download PDFInfo
- Publication number
- CN105808762A CN105808762A CN201610158095.9A CN201610158095A CN105808762A CN 105808762 A CN105808762 A CN 105808762A CN 201610158095 A CN201610158095 A CN 201610158095A CN 105808762 A CN105808762 A CN 105808762A
- Authority
- CN
- China
- Prior art keywords
- model
- score
- resource
- candidate resource
- clicking rate
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a resource sequencing method and device. The resource sequencing method comprises the following steps: acquiring candidate resources; calculating first click through rate (CTR) scores of the candidate resources based on a logistic regression (LR) model; calculating second CTR scores and uncertainty scores of the candidate resources based on a linear confidence upper limit LinearUCB model; inputting the first CTR scores, the second CTR scores and the uncertainty scores serving as input features into a gradient boosting decision tree (GBDT) model, and calculating pre-estimated scores of the candidate resources based on the GBDT model; and sequencing the candidate resources according to the pre-estimated scores. Through adoption of the resource sequencing method, the accuracy of CTR pre-estimation can be increased effectively, so that sequencing optimization of the candidate resources is realized; candidate resources more preferred by a user are presented preferentially; and the user experience is improved.
Description
Technical field
The present invention relates to Internet technical field, particularly relate to a kind of resource ordering method and device.
Background technology
In fields such as internet search engine, product or advertisement promotion, it is frequently encountered candidate resource more, and limited candidate resource can only be represented every time.In order to candidate resource user being most interested in preferentially represents, currently mainly utilize machine learning to estimate the CTR (ClickThroughRate, clicking rate) of candidate resource, according to CTR, candidate resource is ranked up.Such as: calculate the CTR of a certain candidate resource, mainly the score of multiple CTR prediction models is carried out linear, additive.
But, said method cannot embody the non-linear correlation relation between each CTR prediction model, and when the order of magnitude of CTR prediction model is inconsistent, simply score is carried out linear, additive deleterious, causes that the degree of accuracy estimating the CTR of candidate resource declines.
Summary of the invention
It is contemplated that one of technical problem solved at least to a certain extent in correlation technique.For this, it is an object of the present invention to propose a kind of resource ordering method, it is possible to the sequence of candidate resource is optimized, preferentially represent the candidate resource that user is interested, promote user's experience.
Second purpose of the present invention is in that to propose a kind of resource collator.
To achieve these goals, first aspect present invention embodiment proposes a kind of resource ordering method, including: obtain candidate resource；Logic-based returns LR model and calculates the first clicking rate CTR score of described candidate resource；The second clicking rate CTR score of described candidate resource and uncertain score is calculated based on linear confidence upper limit LinearUCB model；Using described first clicking rate CTR score, described second clicking rate CTR score and described uncertain score as input feature vector, input to iteration decision tree GBDT model, and estimate score based on what described GBDT model calculated described candidate resource；And according to described score of estimating, described candidate resource is ranked up.
The resource ordering method of the embodiment of the present invention, by obtaining candidate resource, the first clicking rate CTR score of described candidate resource is calculated based on LR model, and the second clicking rate CTR score of described candidate resource and uncertain score is calculated based on LinearUCB model, again by described first clicking rate CTR score, described second clicking rate CTR score and described uncertain score are as input feature vector, input to iteration decision tree GBDT model, based on described GBDT model calculate described candidate resource estimate score, and according to described score of estimating, described candidate resource is ranked up, can effectively promote the CTR degree of accuracy estimated, thus realizing the sequence of candidate resource is optimized, preferentially represent the candidate resource that user is interested, promote user's experience.
Second aspect present invention embodiment proposes a kind of resource collator, including: acquisition module, it is used for obtaining candidate resource；First computing module, returns LR model for logic-based and calculates the first clicking rate CTR score of described candidate resource；Second computing module, for calculating the second clicking rate CTR score of described candidate resource and uncertain score based on linear confidence upper limit LinearUCB model；Estimate module, for using described first clicking rate CTR score, described second clicking rate CTR score and described uncertain score as input feature vector, input to iteration decision tree GBDT model, and estimate score based on what described GBDT model calculated described candidate resource；And order module, described in basis, estimate score described candidate resource is ranked up.
The resource collator of the embodiment of the present invention, by obtaining candidate resource, the first clicking rate CTR score of described candidate resource is calculated based on LR model, and the second clicking rate CTR score of described candidate resource and uncertain score is calculated based on LinearUCB model, again by described first clicking rate CTR score, described second clicking rate CTR score and described uncertain score are as input feature vector, input to iteration decision tree GBDT model, and based on described GBDT model calculate described candidate resource estimate score, and according to described score of estimating, described candidate resource is ranked up, can effectively promote the CTR degree of accuracy estimated, thus realizing the sequence of candidate resource is optimized, preferentially represent the candidate resource that user is interested, promote user's experience.
Accompanying drawing explanation
Fig. 1 is the flow chart of resource ordering method according to an embodiment of the invention.
Fig. 2 is the structural representation one of resource collator according to an embodiment of the invention.
Fig. 3 is the structural representation two of resource collator according to an embodiment of the invention.
Fig. 4 is the structural representation three of resource collator according to an embodiment of the invention.
Fig. 5 is the structural representation four of resource collator according to an embodiment of the invention.
Detailed description of the invention
Being 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 the element of same or like function from start to finish.The embodiment described below with reference to accompanying drawing is illustrative of, it is intended to is used for explaining the present invention, and is not considered as limiting the invention.
Below with reference to the accompanying drawings resource ordering method and the device of the embodiment of the present invention are described.
Fig. 1 is the flow chart of resource ordering method according to an embodiment of the invention.
As it is shown in figure 1, resource ordering method comprises the steps that
S1, acquisition candidate resource.
Wherein, candidate resource can be Search Results, it is also possible to be the product information etc. of businessman's issue.
In the present embodiment, candidate resource is group purchase information, information of discount etc..
S2, logic-based return LR model and calculate the first clicking rate CTR score of candidate resource.
Such as: a CTR score of a certain candidate resource can be calculated according to formula 1.
Formula 1:
Wherein, p_ctr is a CTR score, and X is characteristic vector, and θ is the parameter of LR model, the transposition of T representing matrix.Assume that resource characteristic includes three features, then its characteristic of correspondence vector X is represented by [x1, x2, x3].
S3, the second clicking rate CTR score calculating candidate resource based on linear confidence upper limit LinearUCB model and uncertain score.
In continuation, example is described, and can calculate the 2nd CTR score of this candidate resource according to formula 2.
Formula 2:p_ctr'=θ 'TX
Wherein, p_ctr ' is the 2nd CTR score, and X is characteristic vector, and θ ' is the parameter of LinearUCB model, the transposition of T representing matrix.
The uncertain score of this candidate resource is calculated according to formula 3.
Formula 3:p_unc=XTA-1X
Wherein, p_unc is uncertain score, and X is characteristic vector, and A is another parameter of LinearUCB model, the transposition of T representing matrix.
S4, using the first clicking rate CTR score, the second clicking rate CTR score and uncertain score as input feature vector, input to iteration decision tree GBDT model, and estimate score based on what GBDT model calculated candidate resource.
In continuation, example is described, assume that a CTR must be divided into 0.022905,2nd CTR must be divided into 0.042237, uncertain must be divided into 0.085515, then can by a CTR score, the 2nd CTR score, uncertain score inputs to GBDT (GradientBoostingDecisionTree, iteration decision tree) model, what finally give this candidate resource estimates score.
S5, basis are estimated score and candidate resource are ranked up.
After score is estimated in acquisition, according to estimating score order from big to small, candidate resource can be ranked up, so that the sequence estimating the high candidate resource of score is preferential.
In addition, return LR model at logic-based and calculate the first clicking rate CTR score of candidate resource, the second clicking rate CTR score of candidate resource and uncertain score is calculated, it is necessary to training in advance LR model and LinearUCB model based on linear confidence upper limit LinearUCB model.
Specifically, first can obtain first and represent click logs, and extract the first resource characteristic representing in click logs.Wherein, first represents click logs for whether track record candidate resource produces click behavior, and recording-related information and the daily record data that produces.First resource characteristic representing click logs can include user characteristics, order feature and environmental characteristic.Such as user characteristics can include sex, age, the consuming capacity etc.；Order feature can include price, discount, sales volume etc.；Environmental characteristic can include the distance etc. of current time, user and publisher.
Then logistic regression LR model and LinearUCB model can be trained according to resource characteristic.
Wherein, LR model is based primarily upon loss function, uses gradient decline optimization, and training obtains the parameter θ of LR model.
LinearUCB model then can train the parameter θ of correspondence according to formula 4,5,6 ' and A.
Formula 4:
Formula 5:bat=bat+rtxt,at
Formula 6:
Wherein, A is matrix parameter, and b represents vector, and the row number of a representing matrix, t represents the t time repetitive exercise.R indicates whether to click, if clicked, then r value is 1；If do not clicked on, then r value is 0.
After training LinearUCB model, GBDT model can be trained.
Specifically, second can be obtained and represent click logs, and extract the second resource characteristic representing in click logs, then can represent resource characteristic in click logs, LR model and LinearUCB model according to second, train GBDT model.
For example, training LR model and use on January 14,1 day to 2016 January in 2016 during LinearUCB model represent click logs, after training completes, represent click logs in conjunction with on January 15th, 2016, GBDT model can be trained.Wherein, GBDT model is model known in this field, repeats no more in the present embodiment.
The resource ordering method of the embodiment of the present invention, by obtaining candidate resource, the first clicking rate CTR score of candidate resource is calculated based on LR model, and the second clicking rate CTR score of candidate resource and uncertain score is calculated based on LinearUCB model, again by the first clicking rate CTR score, second clicking rate CTR score and uncertain score are as input feature vector, input to iteration decision tree GBDT model, and based on GBDT model calculate candidate resource estimate score, and according to estimating score, candidate resource is ranked up, can effectively promote the CTR degree of accuracy estimated, thus realizing the sequence of candidate resource is optimized, preferentially represent the candidate resource that user is interested, promote user's experience.
For achieving the above object, the present invention also proposes a kind of resource collator.
Fig. 2 is the structural representation one of resource collator according to an embodiment of the invention.
As in figure 2 it is shown, resource collator comprises the steps that acquisition module the 110, first computing module the 120, second computing module 130, estimates module 140 and order module 150.
Acquisition module 110 is used for obtaining candidate resource.Wherein, candidate resource can be Search Results, it is also possible to be the product information etc. of businessman's issue.In the present embodiment, candidate resource is group purchase information, information of discount etc..
First computing module 120 returns LR model for logic-based and calculates the first clicking rate CTR score of candidate resource.Such as: a CTR score of a certain candidate resource can be calculated according to formula 1.
Formula 1:
Wherein, p_ctr is a CTR score, and X is characteristic vector, and θ is the parameter of LR model, the transposition of T representing matrix.Assume that resource characteristic includes three features, then its characteristic of correspondence vector X is represented by [x1, x2, x3].
Second computing module 130 for calculating the second clicking rate CTR score of candidate resource and uncertain score based on linear confidence upper limit LinearUCB model.In continuation, example is described, and can calculate the 2nd CTR score of this candidate resource according to formula 2.
Formula 2:p_ctr'=θ 'TX
Wherein, p_ctr ' is the 2nd CTR score, and X is characteristic vector, and θ ' is the parameter of LinearUCB model, the transposition of T representing matrix.
The uncertain score of this candidate resource is calculated according to formula 3.
Formula 3:p_unc=XTA-1X
Wherein, p_unc is uncertain score, and X is characteristic vector, and A is another parameter of LinearUCB model, the transposition of T representing matrix.
Estimate module 140 for using the first clicking rate CTR score, the second clicking rate CTR score and uncertain score as input feature vector, input to iteration decision tree GBDT model, and estimate score based on what GBDT model calculated candidate resource.In continuation, example is described, assume that a CTR must be divided into 0.022905,2nd CTR must be divided into 0.042237, uncertain must be divided into 0.085515, then can by a CTR score, the 2nd CTR score, uncertain score inputs to GBDT (GradientBoostingDecisionTree, iteration decision tree) model, what finally give this candidate resource estimates score.
Order module 150 is estimated score for basis and candidate resource is ranked up.After score is estimated in acquisition, candidate resource can be ranked up by order module 150 according to estimating score order from big to small, so that the sequence estimating the high candidate resource of score is preferential.
Additionally, as it is shown on figure 3, the resource collator of the present embodiment may also include the first training module 160.
First training module 160 is for, before returning, at logic-based, the first clicking rate CTR score that LR model calculates candidate resource, training LR model.Specifically, first can obtain first and represent click logs, and extract the first resource characteristic representing in click logs.Wherein, resource characteristic can include user characteristics, order feature and environmental characteristic.Such as user characteristics can include sex, age, the consuming capacity etc.；Order feature can include price, discount, sales volume etc.；Environmental characteristic can include the distance etc. of current time, user and publisher.Then logistic regression LR model can be trained according to resource characteristic.LR model is based primarily upon loss function, uses gradient decline optimization, and training obtains the parameter θ of LR model.
It addition, as shown in Figure 4, the resource collator of the present embodiment may also include the second training module 170.
Second training module 170 is for, before the second clicking rate CTR score calculating candidate resource based on linear confidence upper limit LinearUCB model and uncertain score, training LinearUCB model.Similarly, first can obtain first and represent click logs, and extract the first resource characteristic representing in click logs.Wherein, resource characteristic can include user characteristics, order feature and environmental characteristic.Such as user characteristics can include sex, age, the consuming capacity etc.；Order feature can include price, discount, sales volume etc.；Environmental characteristic can include the distance etc. of current time, user and publisher.Then LinearUCB model can be trained according to resource characteristic.LinearUCB model then can train the parameter θ of correspondence according to formula 4,5,6 ' and A.
Formula 4:
Formula 5:bat=bat+rtxt,at
Formula 6:
Wherein, A is matrix parameter, and b represents vector, and the row number of a representing matrix, t represents the t time repetitive exercise.R indicates whether to click, if clicked, then r value is 1；If do not clicked on, then r value is 0.
As it is shown in figure 5, the resource collator of the present embodiment may also include the 3rd training module 180.
3rd training module 180 for calculate based on GBDT model candidate resource estimate score before, train GBDT model.Specifically, second can be obtained and represent click logs, and extract the second resource characteristic representing in click logs, then can represent resource characteristic in click logs, LR model and LinearUCB model according to second, train GBDT model.For example, training LR model and use on January 14,1 day to 2016 January in 2016 during LinearUCB model represent click logs, after training completes, represent click logs in conjunction with on January 15th, 2016, GBDT model can be trained.Wherein, GBDT model is model known in this field, repeats no more in the present embodiment.
The resource collator of the embodiment of the present invention, by obtaining candidate resource, the first clicking rate CTR score of candidate resource is calculated based on LR model, and the second clicking rate CTR score of candidate resource and uncertain score is calculated based on LinearUCB model, again by the first clicking rate CTR score, second clicking rate CTR score and uncertain score are as input feature vector, input to iteration decision tree GBDT model, and based on GBDT model calculate candidate resource estimate score, and according to estimating score, candidate resource is ranked up, can effectively promote the CTR degree of accuracy estimated, thus realizing the sequence of candidate resource is optimized, preferentially represent the candidate resource that user is interested, promote user's experience.
In the description of this specification, specific features, structure, material or feature that the description of reference term " embodiment ", " some embodiments ", " example ", " concrete example " or " some examples " etc. means in conjunction with this embodiment or example describe are contained at least one embodiment or the example of the present invention.In this manual, the schematic representation of above-mentioned term is necessarily directed to identical embodiment or example.And, the specific features of description, structure, material or feature can combine in one or more embodiments in office or example in an appropriate manner.Additionally, when not conflicting, the feature of the different embodiments described in this specification or example and different embodiment or example can be carried out combining and combining by those skilled in the art.
Although above it has been shown and described that embodiments of the invention, it is understandable that, above-described embodiment is illustrative of, it is impossible to be interpreted as limitation of the present invention, and above-described embodiment can be changed, revises, replace and modification by those of ordinary skill in the art within the scope of the invention.
Claims (14)
1. a resource ordering method, it is characterised in that comprise the following steps:
Obtain candidate resource；
Logic-based returns LR model and calculates the first clicking rate CTR score of described candidate resource；
The second clicking rate CTR score of described candidate resource and uncertain score is calculated based on linear confidence upper limit LinearUCB model；
Using described first clicking rate CTR score, described second clicking rate CTR score and described uncertain score as input feature vector, input to iteration decision tree GBDT model, and estimate score based on what described GBDT model calculated described candidate resource；And
According to described score of estimating, described candidate resource is ranked up.
2. the method for claim 1, it is characterised in that before logic-based returns the first clicking rate CTR score that LR model calculates described candidate resource, also include:
Train described LR model.
3. method as claimed in claim 2, it is characterised in that train described LR model, including:
Obtaining first and represent click logs, and extract described first resource characteristic representing in click logs, described resource characteristic includes user characteristics, order feature and environmental characteristic；
Logistic regression LR model is trained according to described resource characteristic.
4. the method for claim 1, it is characterised in that before the second clicking rate CTR score calculating described candidate resource based on linear confidence upper limit LinearUCB model and uncertain score, also include:
Train described LinearUCB model.
5. method as claimed in claim 4, it is characterised in that train described LinearUCB model, including:
Obtaining first and represent click logs, and extract described first resource characteristic representing in click logs, described resource characteristic includes user characteristics, order feature and environmental characteristic；
Logistic regression LR model is trained according to described resource characteristic.
6. the method for claim 1, it is characterised in that based on described GBDT model calculate described candidate resource estimate score before, also include:
Train described GBDT model.
7. method as claimed in claim 6, it is characterised in that train described GBDT model, including:
Obtain second and represent click logs, and extract described second resource characteristic representing in click logs；
Represent the resource characteristic in click logs, described LR model and described LinearUCB model according to described second, train described GBDT model.
8. a resource collator, it is characterised in that including:
Acquisition module, is used for obtaining candidate resource；
First computing module, returns LR model for logic-based and calculates the first clicking rate CTR score of described candidate resource；
Second computing module, for calculating the second clicking rate CTR score of described candidate resource and uncertain score based on linear confidence upper limit LinearUCB model；
Estimate module, for using described first clicking rate CTR score, described second clicking rate CTR score and described uncertain score as input feature vector, input to iteration decision tree GBDT model, and estimate score based on what described GBDT model calculated described candidate resource；And
Order module, estimates score described in basis and described candidate resource is ranked up.
9. device as claimed in claim 8, it is characterised in that also include:
First training module, before returning, at logic-based, the first clicking rate CTR score that LR model calculates described candidate resource, trains described LR model.
10. device as claimed in claim 9, it is characterised in that described first training module, is used for:
Obtaining first and represent click logs, and extract described first resource characteristic representing in click logs, described resource characteristic includes user characteristics, order feature and environmental characteristic；
Logistic regression LR model is trained according to described resource characteristic.
11. device as claimed in claim 8, it is characterised in that also include:
Second training module, for, before the second clicking rate CTR score calculating described candidate resource based on linear confidence upper limit LinearUCB model and uncertain score, training described LinearUCB model.
12. device as claimed in claim 11, it is characterised in that described second training module, it is used for:
Obtaining first and represent click logs, and extract described first resource characteristic representing in click logs, described resource characteristic includes user characteristics, order feature and environmental characteristic；
Logistic regression LR model is trained according to described resource characteristic.
13. device as claimed in claim 8, it is characterised in that also include:
3rd training module, for based on described GBDT model calculate described candidate resource estimate score before, train described GBDT model.
14. device as claimed in claim 13, it is characterised in that described 3rd training module, it is used for:
Obtain second and represent click logs, and extract described second resource characteristic representing in click logs；
Represent the resource characteristic in click logs, described LR model and described LinearUCB model according to described second, train described GBDT model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610158095.9A CN105808762B (en) | 2016-03-18 | 2016-03-18 | Resource ordering method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610158095.9A CN105808762B (en) | 2016-03-18 | 2016-03-18 | Resource ordering method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105808762A true CN105808762A (en) | 2016-07-27 |
| CN105808762B CN105808762B (en) | 2019-04-02 |
Family
ID=56454454
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610158095.9A Active CN105808762B (en) | 2016-03-18 | 2016-03-18 | Resource ordering method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105808762B (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107301247A (en) * | 2017-07-14 | 2017-10-27 | 广州优视网络科技有限公司 | Set up the method and device, terminal, storage medium of clicking rate prediction model |
| CN107844867A (en) * | 2017-11-22 | 2018-03-27 | 广州优视网络科技有限公司 | Data processing method, device and equipment |
| CN108121814A (en) * | 2017-12-28 | 2018-06-05 | 北京百度网讯科技有限公司 | Search results ranking model generating method and device |
| CN108182597A (en) * | 2017-12-27 | 2018-06-19 | 银橙(上海)信息技术有限公司 | A kind of clicking rate predictor method based on decision tree and logistic regression |
| CN108509499A (en) * | 2018-02-27 | 2018-09-07 | 北京三快在线科技有限公司 | A kind of searching method and device, electronic equipment |
| CN110275779A (en) * | 2019-06-20 | 2019-09-24 | 北京百度网讯科技有限公司 | A kind of resource acquiring method, device, equipment and storage medium |
| CN110288433A (en) * | 2019-06-12 | 2019-09-27 | 达疆网络科技(上海)有限公司 | One kind being based on clicking rate model prediction O2O real time individual sort method |
| CN110400166A (en) * | 2019-06-24 | 2019-11-01 | 阿里巴巴集团控股有限公司 | The method and apparatus for selecting the official documents and correspondence pushed to target user |
| CN111598677A (en) * | 2020-07-24 | 2020-08-28 | 北京淇瑀信息科技有限公司 | Resource quota determining method and device and electronic equipment |
| CN111598116A (en) * | 2019-02-21 | 2020-08-28 | 杭州海康威视数字技术股份有限公司 | Data classification method and device, electronic equipment and readable storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070036441A1 (en) * | 2005-08-10 | 2007-02-15 | Xerox Corporation | Monotonic classifier |
| CN103246985A (en) * | 2013-04-26 | 2013-08-14 | 北京亿赞普网络技术有限公司 | Advertisement click rate predicting method and device |
| CN103761266A (en) * | 2014-01-02 | 2014-04-30 | 北京集奥聚合网络技术有限公司 | Click rate predicting method and system based on multistage logistic regression |
| CN103942279A (en) * | 2014-04-01 | 2014-07-23 | 百度(中国)有限公司 | Method and device for showing search result |
| 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 |
| CN104424291A (en) * | 2013-09-02 | 2015-03-18 | 阿里巴巴集团控股有限公司 | Method and device for sorting search results |
-
2016
- 2016-03-18 CN CN201610158095.9A patent/CN105808762B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070036441A1 (en) * | 2005-08-10 | 2007-02-15 | Xerox Corporation | Monotonic classifier |
| CN103246985A (en) * | 2013-04-26 | 2013-08-14 | 北京亿赞普网络技术有限公司 | Advertisement click rate predicting method and device |
| CN104424291A (en) * | 2013-09-02 | 2015-03-18 | 阿里巴巴集团控股有限公司 | Method and device for sorting search results |
| CN103761266A (en) * | 2014-01-02 | 2014-04-30 | 北京集奥聚合网络技术有限公司 | Click rate predicting method and system based on multistage logistic regression |
| CN103942279A (en) * | 2014-04-01 | 2014-07-23 | 百度(中国)有限公司 | Method and device for showing search result |
| 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 |
Non-Patent Citations (2)
| Title |
|---|
| LIHONG LI等: "A Contextual-Bandit Approach to Personalized News Article Recommendation", 《WWW’10 PROCEEDINGS OF THE 19TH INTERNATIONAL CONFERENCE ON WORLD WIDE WEB》 * |
| 纪文迪等: "广告点击率估算技术综述", 《华东师范大学学报(自然科学版)》 * |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107301247A (en) * | 2017-07-14 | 2017-10-27 | 广州优视网络科技有限公司 | Set up the method and device, terminal, storage medium of clicking rate prediction model |
| CN107301247B (en) * | 2017-07-14 | 2020-12-18 | 阿里巴巴（中国）有限公司 | Method and device for establishing click rate estimation model, terminal and storage medium |
| CN107844867A (en) * | 2017-11-22 | 2018-03-27 | 广州优视网络科技有限公司 | Data processing method, device and equipment |
| CN108182597A (en) * | 2017-12-27 | 2018-06-19 | 银橙(上海)信息技术有限公司 | A kind of clicking rate predictor method based on decision tree and logistic regression |
| CN108121814A (en) * | 2017-12-28 | 2018-06-05 | 北京百度网讯科技有限公司 | Search results ranking model generating method and device |
| CN108121814B (en) * | 2017-12-28 | 2022-04-22 | 北京百度网讯科技有限公司 | Search result ranking model generation method and device |
| CN108509499A (en) * | 2018-02-27 | 2018-09-07 | 北京三快在线科技有限公司 | A kind of searching method and device, electronic equipment |
| CN111598116A (en) * | 2019-02-21 | 2020-08-28 | 杭州海康威视数字技术股份有限公司 | Data classification method and device, electronic equipment and readable storage medium |
| CN111598116B (en) * | 2019-02-21 | 2024-01-23 | 杭州海康威视数字技术股份有限公司 | Data classification method, device, electronic equipment and readable storage medium |
| CN110288433A (en) * | 2019-06-12 | 2019-09-27 | 达疆网络科技(上海)有限公司 | One kind being based on clicking rate model prediction O2O real time individual sort method |
| CN110275779A (en) * | 2019-06-20 | 2019-09-24 | 北京百度网讯科技有限公司 | A kind of resource acquiring method, device, equipment and storage medium |
| CN110275779B (en) * | 2019-06-20 | 2022-07-08 | 北京百度网讯科技有限公司 | Resource acquisition method, device, equipment and storage medium |
| CN110400166A (en) * | 2019-06-24 | 2019-11-01 | 阿里巴巴集团控股有限公司 | The method and apparatus for selecting the official documents and correspondence pushed to target user |
| CN111598677A (en) * | 2020-07-24 | 2020-08-28 | 北京淇瑀信息科技有限公司 | Resource quota determining method and device and electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105808762B (en) | 2019-04-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105808762A (en) | Resource sequencing method and device | |
| CN110737783B (en) | Method and device for recommending multimedia content and computing equipment | |
| CN110879864B (en) | Context recommendation method based on graph neural network and attention mechanism | |
| CN105678587A (en) | Recommendation feature determination method and information recommendation method and device | |
| US8290945B2 (en) | Web searching | |
| CN105160545B (en) | Method and device for determining release information style | |
| US8732014B2 (en) | Automatic classification of display ads using ad images and landing pages | |
| CN108805598B (en) | Similarity information determination method, server and computer-readable storage medium | |
| CN108898429A (en) | Electronic device, preference tendency prediction technique and computer readable storage medium | |
| CN105183772A (en) | Release information click rate estimation method and apparatus | |
| CN107818105A (en) | The recommendation method and server of application program | |
| CN103942257B (en) | Video search method and device | |
| CN107239564B (en) | Text label recommendation method based on supervision topic model | |
| CN106445954B (en) | Business object display method and device | |
| CN102591915A (en) | Recommending method based on label migration learning | |
| CN105677857B (en) | method and device for accurately matching keywords with marketing landing pages | |
| CN111046188A (en) | User preference degree determining method and device, electronic equipment and readable storage medium | |
| CN105786983A (en) | Employee individualized-learning recommendation method based on learning map and collaborative filtering | |
| KR102559950B1 (en) | An AI-based optimal advertising recommendation system | |
| US20190050890A1 (en) | Video dotting placement analysis system, analysis method and storage medium | |
| CN110162609A (en) | For recommending the method and device asked questions to user | |
| CN106503267A (en) | A kind of personalized recommendation algorithm suitable for user preference dynamic evolution | |
| US20210191995A1 (en) | Generating and implementing keyword clusters | |
| CN111523315A (en) | Data processing method, text recognition device and computer equipment | |
| CN105701227A (en) | Cross-media similarity measure method and search method based on local association graph |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |