WO2017028728A1

WO2017028728A1 - Determining method and device for click through rate (ctr)

Info

Publication number: WO2017028728A1
Application number: PCT/CN2016/094448
Authority: WO
Inventors: 谢海军; 邓国; 邓琨; 贺露
Original assignee: 北京金山安全软件有限公司
Priority date: 2015-08-18
Filing date: 2016-08-10
Publication date: 2017-02-23
Also published as: CN105205098A; CN105205098B

Abstract

A determining method and device for a click through rate (CTR). The method comprises: when a display request aiming at applications is detected, determining a characteristic value sequence of each application (S101), wherein the characteristic value sequence consists of an application characteristic value sequence for describing application information, a user characteristic value sequence for describing user information, and a flow characteristic value sequence; respectively using each characteristic value sequence as an input of a preset CTR computational algorithm, and acquiring an output, that corresponds to each characteristic value sequence, of the CTR computational algorithm (S102); and determining the output, that corresponds to each characteristic value sequence, of the CTR computational algorithm as a CTR of an application corresponding to the characteristic value sequence (S103). The method can quickly determine the CTR of an application and has low resource consumption.

Description

Method and device for determining click arrival rate CTR

Cross-reference to related applications

The present application is based on a Chinese patent application No. 201510507737.7, filed on Aug. 18, 2015, and the priority of which is hereby incorporated by reference.

Technical field

The present invention relates to the field of Internet technologies, and in particular, to a method and an apparatus for determining a click arrival rate CTR.

Background technique

In the field of Internet technology, Internet product promoters can use the application wall and other promotional resources to promote the application to the developer of the application, that is, the promotion of the Internet product will be based on the user's current operating scenario and the determined click-through rate (CTR) of the application. Click Through Rate) recommends one or more premium apps with a higher CTR for the user. It can be seen that quickly determining the application's CTR is particularly important in the promotion of the application.

Currently, the common CTR determination method is a statistical-based determination method, that is, the CTR is the same in the same operation scenario, and the CTR in the different operation scenarios is found to be the same as or similar to the current operation scenario. The CTR under the operating scenario is used as the CTR applied in the current operating scenario. However, since the operation scenario is composed of a plurality of feature dimensions, the statistics-based determination method relies on a large amount of historical sample data, and multiple feature dimensions need to be considered when searching for an operation scenario that is the same as or similar to the current operation scenario. The CTR of the application is determined and the resource consumption is large.

Summary of the invention

The embodiment of the invention discloses a method and a device for determining the click arrival rate CTR, which can quickly determine the CTR of the application and consume small resources.

The first aspect of the present invention discloses a method for determining a click arrival rate CTR, and the method includes:

When a display request for an application is detected, a sequence of feature values for each application is determined, the sequence of feature values being a sequence of applied feature values for describing application information, a sequence of user feature values for describing user information, and a flow feature value Sequence composition

And respectively outputting each of the feature value sequences as an input of a preset CTR calculation algorithm, and acquiring an output of the CTR calculation algorithm corresponding to each of the feature value sequences;

The output of the CTR calculation algorithm corresponding to each of the sequence of feature values is determined as the CTR of the application corresponding to the sequence of feature values.

In at least one embodiment, the CTR calculation algorithm is a logistic regression model based algorithm, and the calculation formula of the logistic regression model based algorithm is:

Wherein the y ^CTR is an output of the calculation formula,

For the input of the calculation formula, the

It is a coefficient of the calculation formula calculated in advance.

In at least one embodiment, the method further comprises: before determining a sequence of feature values for each application when detecting a display request for an application, the method further comprising:

Reading a first quantity of first sample data from pre-stored sample data, the sample data being represented by a sequence of feature values, a presentation identifier for identifying whether the sample data is presented, and for identifying whether the sample data is clicked Click markup composition;

And combining the first sample data of the first quantity of the first sample data having the same sequence of feature values to obtain a second quantity of second sample data, wherein the second sample data is formed by the second sample data a sum of a sequence of feature values of a sample of data, a sum of presentation identifiers of first sample data forming the second sample data, and a click identifier of first sample data forming the second sample data;

Expressing each of the second sample data into a likelihood expression, and multiplying all of the likelihood expressions to obtain a product likelihood expression;

Calculating the value of the unknown parameter in the product likelihood expression when the product likelihood expression takes the maximum value by using the Newton iteration method and the initial iteration parameter to perform the iteration of the target number of times, and determining the value of the unknown parameter For the stated

Reading a first quantity of first sample data from pre-stored sample data, the first sample data being represented by a sequence of feature values, a presentation identifier for identifying whether the sample data is presented, and for identifying the sample data Whether it is composed of clicked click marks;

Dividing the second quantity of second sample data into a sample group each including a third quantity of second sample data;

Each second sample data included in each of the sample sets is expressed as a likelihood expression, respectively, and all of the likelihood expressions corresponding to each of the sample sets are multiplied to obtain a product of the sample set Expression

Performing an iteration by Newton iteration method and initial iteration parameters, respectively calculating a first value of an unknown parameter in the product likelihood expression when each of the product likelihood expression takes a maximum value, each of the first values And as an initial iteration parameter of the next iteration, performing the operation of performing an iteration by the Newton iteration method and the initial iteration parameter until the number of iterations reaches the target number of times;

And determining a sum of the second values of the unknown parameters in each of the product likelihood expressions calculated after the target number of iterations is determined as

In at least one embodiment, after the merging the first sample data having the same sequence of feature values in the first number of first sample data, dividing the second number of second sample data into each Before the sample group including the third quantity of the second sample data, the method further includes:

The second quantity of second sample data is stored in a memory space of consecutive addresses.

The second aspect of the present invention discloses a determining device for a click arrival rate CTR, the device comprising a first determining unit, a first obtaining unit, and a second determining unit, wherein:

The first determining unit is configured to determine a sequence of feature values of each application when the display request for the application is detected, where the sequence of feature values is used by the application feature value sequence for describing application information, and is used to describe user information. User characteristic value sequence and flow characteristic value sequence composition;

The first acquiring unit is configured to respectively obtain, according to an input of a preset CTR calculation algorithm, an output of the CTR calculation algorithm corresponding to each of the feature value sequences;

The second determining unit is configured to determine an output of the CTR calculation algorithm corresponding to each of the feature value sequences as a CTR corresponding to the application of the feature value sequence.

Wherein the y _CTR is an output of the calculation formula,

For the input of the calculation formula, the

It is a coefficient of the calculation formula calculated in advance.

In at least one embodiment, the apparatus further includes a first reading unit, a first merging unit, a second obtaining unit, a first calculating unit, and a third determining unit, wherein:

The first reading unit is configured to read a first quantity of first sample data from pre-stored sample data, where the sample data is a sequence of feature values, and a display identifier for identifying whether the sample data is presented And a click identifier for identifying whether the sample data is clicked;

The first merging unit is configured to merge the first number of eigenvalue sequences in the first quantity of the first sample data to be the same The data is obtained to obtain a second quantity of second sample data, the second sample data is formed by a sequence of feature values of the first sample data forming the second sample data, and the first sample data forming the second sample data And a sum of the click identifiers of the first sample data forming the second sample data;

The second obtaining unit is configured to express each of the second sample data into a likelihood expression, and multiply all the likelihood expressions to obtain a product likelihood expression;

The first calculating unit is configured to perform an iteration of a target number of times by a Newton iteration method and an initial iteration parameter, and calculate a value of an unknown parameter in the product likelihood expression when the product likelihood expression takes a maximum value;

The third determining unit is configured to determine a value of the unknown parameter as the

In at least one embodiment, the apparatus further includes a second reading unit, a second merging unit, a averaging unit, a third obtaining unit, a second calculating unit, and a fourth determining unit, wherein:

The second reading unit is configured to read a first quantity of first sample data from pre-stored sample data, where the first sample data is used by a sequence of feature values to identify whether the sample data is displayed a presentation identifier and a click identifier for identifying whether the sample data is clicked;

The second merging unit is configured to merge the first sample data of the first number of first sample data with the same sequence of feature values to obtain a second quantity of second sample data, the second sample data a sum of a sequence of feature values of the first sample data forming the second sample data, a sum of presentation identifiers of the first sample data forming the second sample data, and a click of the first sample data forming the second sample data The sum of the logos;

The equalizing unit is configured to divide the second quantity of second sample data into a sample group including a third quantity of second sample data;

The third obtaining unit is configured to respectively express each second sample data included in each of the sample groups into a likelihood expression, and compare all the likelihood expressions corresponding to each of the sample groups Multiply by to obtain a product likelihood expression for the sample set;

The second calculating unit is configured to perform an iteration by using a Newton iteration method and an initial iteration parameter, and respectively calculate a first value of an unknown parameter in the product likelihood expression when each of the product likelihood expressions takes a maximum value Performing, by using the sum of each of the first values as an initial iteration parameter of the next iteration, performing an iterative operation by the Newton iteration method and the initial iteration parameter until the number of iterations reaches the target number of times;

The fourth determining unit is configured to determine a sum of the second value of the unknown parameter in each of the product likelihood expressions calculated after the target number of times is iterated as

In at least one embodiment, the apparatus further includes a storage unit, wherein:

The storage unit is configured to store the second quantity of second sample data in a memory space with consecutive addresses.

A third aspect of the present invention discloses a server, where the server includes:

Processor;

a memory for storing instructions executable by the processor,

Wherein the processor is configured to read executable instructions stored in the memory to execute a program corresponding to the instructions for performing a determination method of a click arrival rate CTR according to an embodiment of the first aspect of the present invention .

A fourth aspect of the present invention discloses a computer readable storage medium having instructions stored therein, when the processor of the server executes the instructions, the server performs the method according to the first aspect of the present invention. Click the method of determining the arrival rate CTR.

A fifth aspect of the present invention discloses a computer application that, when executed on a processor of a server, performs a method of determining a click arrival rate CTR as in the first aspect of the present invention.

In the embodiment of the present invention, when a display request for an application is detected, a sequence of feature values of each application is determined, wherein the sequence of feature values is a sequence of application feature values for describing application information, and a user for describing user information. The eigenvalue sequence and the flow eigenvalue sequence are composed, and each eigenvalue sequence is respectively input as a preset CTR calculation algorithm, and an output of the CTR calculation algorithm corresponding to each eigenvalue sequence is obtained, and each eigenvalue sequence is obtained. The output of the corresponding CTR calculation algorithm is determined to correspond to the CTR of the applied sequence of feature values. The embodiment of the present invention can quickly calculate the CTR of each application according to the determined sequence of feature values of each application and the preset CTR calculation algorithm, and does not need to calculate the application in different operation scenarios according to the historical sample data. The CTR finds the CTR in the same or similar operation scenario as the current operation scenario, and the resource consumption is low.

DRAWINGS

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings to be used in the embodiments will be briefly described below. It is obvious that the drawings in the following description are only some embodiments of the present invention. Those skilled in the art can also obtain other drawings based on these drawings without paying any creative work.

1 is a schematic flow chart of a method for determining a click arrival rate CTR according to an embodiment of the present invention;

2 is a schematic flow chart of another method for determining a click arrival rate CTR according to an embodiment of the present invention;

3 is a schematic flow chart of still another method for determining a click arrival rate CTR according to an embodiment of the present invention;

4 is a schematic structural diagram of a device for determining a click arrival rate CTR according to an embodiment of the present invention;

FIG. 5 is a schematic structural diagram of another apparatus for determining a click arrival rate CTR according to an embodiment of the present invention; FIG.

6 is a schematic structural diagram of still another apparatus for determining a click arrival rate CTR according to an embodiment of the present invention; and

FIG. 7 is a schematic structural diagram of a server according to an embodiment of the present invention.

detailed description

The technical solutions in the embodiments of the present invention are clearly and completely described in the following with reference to the accompanying drawings in the embodiments of the present invention. It is obvious that the described embodiments are only a part of the embodiments of the present invention, but not all embodiments. Based on the invention All of the other embodiments obtained by those skilled in the art without creative efforts are within the scope of the present invention.

The embodiment of the invention discloses a method and a device for determining a click arrival rate CTR, which can quickly calculate the CTR of each application according to the determined sequence of feature values of each application and a preset CTR calculation algorithm, and resource consumption. low. The details are described below separately.

Referring to FIG. 1, FIG. 1 is a schematic flowchart diagram of a method for determining a click arrival rate CTR according to an embodiment of the present invention. Among them, the method shown in FIG. 1 can be applied to a server. As shown in FIG. 1, the method for determining the click arrival rate CTR may include the following steps:

S101. Determine a sequence of feature values of each application when detecting a display request for an application.

In the embodiment of the present invention, the display request for the application may be triggered by the user through the terminal device, or may be actively triggered by the terminal device, and the sequence of feature values of each application is used to describe the application information (such as the category of the application). And a sequence of application feature values of the related description information of the application, etc., a sequence of user feature values for describing user information (such as the gender of the user and the interest of the user, etc.) and information describing the user behavior (such as time, place, language, etc.) The sequence of traffic characteristic values is composed, and the sequence of feature values is a feature value vector composed of a plurality of 0s and 1s as components. For example, if the application information is the category of the application (shooting application and game application), the user information is the user's gender (male and female), and the behavior information is the location (Shanghai and Beijing), then for an application, its characteristics The value sequence A = [a1, a2, a3, a4, a5, a6], where a1 and a2 are used to describe the category of the application, a1 = 1 and a2 = 0 indicates that the application is a shooting application, a1 = 0 and a2 = 1 Indicates that the application is a game application, a3 and a4 are used to describe the gender of the user, a3=1 and a4=0 indicates that the user gender is male, a3=0 and a4=1 indicates that the user gender is female, and a5 and a6 are used to describe The user's behavior location, a5=1 and a6=0 means that the user is in Shanghai, a5=0 and a6=1 means that the user is in Beijing.

The terminal device in the embodiment of the present invention may include: a mobile phone, a smart phone, a notebook computer, a digital broadcast receiver, a personal digital assistant (PDA), a tablet computer (PAD), a portable multimedia player (PMP), a navigation device, and a desktop computer. And so on.

S102. Each of the feature value sequences is respectively input as a preset CTR calculation algorithm, and an output of the CTR calculation algorithm corresponding to each feature value sequence is acquired.

In the embodiment of the present invention, the preset CTR calculation algorithm is used to represent the relationship between the applied feature value sequence (input) and the CTR (output) of the application, that is, the eigenvalue sequence of each application is respectively used as the CTR calculation algorithm. The input, the output of the corresponding CTR calculation algorithm is the CTR of the application. In this way, as long as the CTR calculation algorithm is set in advance, the CTR of the application can be quickly determined, and there is no need to rely on a large amount of historical sample data, and the resource consumption is low.

The CTR calculation algorithm is an algorithm based on a logistic regression model, and the calculation formula of the algorithm based on the logistic regression model is:

Where y _CTR is the output of the calculation formula (CTR of the application),

For the input of this calculation formula (the sequence of eigenvalues applied),

Is the coefficient of the calculation formula calculated in advance. In the embodiment of the present invention, the calculation formula is calculated according to certain historical sample data.

The CTR of the application can be quickly determined by acquiring the sequence of feature values of the application in the subsequent CTR determination process.

S103. Determine an output of a CTR calculation algorithm corresponding to each feature value sequence as a CTR corresponding to the application of the feature value sequence.

In the embodiment of the present invention, when a display request for an application is detected, a sequence of feature values of each application is determined, wherein the sequence of feature values is a sequence of application feature values for describing application information, and a user for describing user information. The eigenvalue sequence and the flow eigenvalue sequence are composed, and each eigenvalue sequence is used as an input of a preset CTR calculation algorithm, and an output of a CTR calculation algorithm corresponding to each eigenvalue sequence is obtained, and each eigenvalue sequence is corresponding. The output of the CTR calculation algorithm is determined to correspond to the CTR of the applied sequence of feature values. The embodiment of the present invention can quickly calculate the CTR of each application according to the determined sequence of feature values of each application and the preset CTR calculation algorithm, and does not need to count from the historical sample data to be applied in different operation scenarios. The CTR finds the CTR in the same or similar operation scenario as the current operation scenario, and the resource consumption is low.

Referring to FIG. 2, FIG. 2 is a schematic flowchart diagram of another method for determining a click arrival rate CTR according to an embodiment of the present invention. Among them, the method shown in Figure 2 is applied to the server. As shown in FIG. 2, the method for determining the click arrival rate CTR may include the following steps:

S201. Read a first quantity of first sample data from pre-stored sample data.

In the embodiment of the present invention, each sample data is composed of a sequence of feature values, a presentation identifier pv for identifying whether the sample data is presented, and a click identifier click for identifying whether the sample data is clicked, and the sequence of feature values is used. a sequence of application feature values describing application information (such as categories of applications and related description information of applications, etc.), sequence of user feature values for describing user information (such as gender of the user and interest of the user, etc.) and for describing user behavior A sequence of traffic characteristic values of information (such as time, place, language, etc.), wherein pv is 1 and click is 0 indicates that the sample data is presentation sample data, pv is 0 and click 1 indicates that the sample data is click sample data. .

S202. The first sample data with the same sequence of feature values in the first quantity of the first sample data is merged to obtain a second quantity of second sample data.

In the embodiment of the present invention, the first sample data having the same sequence of feature values is merged into a second sample data by using the same sequence of feature values as the principle of merging, that is, the first sample data of the first quantity is merged into the first sample data. Second quantity of the second sample Data, and the second number is equal to the number of different values of the sequence of feature values in the first sample data of the first number. The second quantity of second sample data is formed by a sum of a sequence of feature values of the first sample data forming the second sample data, a pv of all first sample data forming the second sample data, and forming the second The sum of the clicks of all the first sample data of the sample data.

S203. Express each second sample data as a likelihood expression, and multiply all likelihood expressions to obtain a product likelihood expression.

In the embodiment of the present invention, each second sample data may be represented as a likelihood expression, wherein the likelihood expression is:

among them,

And n takes all integers greater than or equal to 1 and less than or equal to the second number,

a sequence of feature values of the nth second sample data, pv _n is the sum of pvs of all first sample data forming the nth second sample data, and click _n is all the numbers forming the nth second sample data The sum of clicks of a sample of data, then the product likelihood expression is:

S204. Perform an iteration of the target number of times by a Newton iteration method and an initial iteration parameter, calculate a value of the unknown parameter in the product likelihood expression when the product likelihood expression takes the maximum value, and determine the value of the unknown parameter as

In the embodiment of the present invention, the initial iteration parameter

For the preset iteration parameter, that is, the value of the unknown parameter in the product likelihood expression is calculated when the product likelihood expression takes the maximum value by performing an iteration by the Newton iteration method.

will

As the initial iteration parameter at the next iteration

And so on, until the number of iterations reaches the target number m, the value of the unknown parameter is obtained.

will

Determined as

The target number m may be a preset number of times, or may be based on

versus

The angle is the minimum or according to

The modulo is calculated as the minimum value, which is not limited in the embodiment of the present invention.

In the embodiment of the present invention, specifically, to calculate the maximum value of the product likelihood expression, the logarithm of the product likelihood expression may be first obtained, and then multiplied by -1 to obtain

Then iterate through the Newton iteration method to iterate the target times m times.

The value of the unknown parameter when taking the minimum value

S205. Determine a sequence of feature values for each application when a display request for the application is detected.

In the embodiment of the present invention, the display request for the application may be triggered by the user through the terminal device, or may be The terminal device is actively triggered, and the sequence of feature values of each application is used to describe application information (such as the category of the application and related description information of the application, etc.), and is used to describe user information (such as the gender of the user). a sequence of user feature values of a user's interest, etc., and a sequence of traffic feature values for describing user behavior information (such as time, place, language, etc.), and the sequence of feature values is a feature value of a plurality of 0s and 1s as components vector.

S206. Each of the feature value sequences is respectively input as a preset CTR calculation algorithm, and an output of the CTR calculation algorithm corresponding to each feature value sequence is obtained.

In the embodiment of the present invention, the CTR calculation algorithm is used to represent the relationship between the application's sequence of feature values (input) and the CTR (output) of the application, that is, the sequence of feature values of each application is used as the input of the CTR calculation algorithm, respectively. The output of the corresponding CTR calculation algorithm is the CTR of the application. And the CTR calculation algorithm is an algorithm based on a logistic regression model, and the calculation formula of the algorithm based on the logistic regression model is:

Where y _CTR is the output of the calculation formula (CTR of the application),

For the above

S207. Determine an output of the CTR calculation algorithm corresponding to each feature value sequence as a CTR corresponding to the application of the feature value sequence.

In the embodiment of the present invention, after determining the CTR of each application, the application with the top CTR ranking may be recommended to the user.

After performing step S202 and before performing step S203, the following operations may also be performed:

The second quantity of the second sample data is stored in a memory space of consecutive addresses.

In the embodiment of the present invention, the second quantity of the second sample data is stored in the contiguous memory space and the head and tail pointer array may be used to identify the start memory and the end memory of each second sample data, thereby speeding up the second Reading of sample data.

It can be seen that the implementation of the embodiment of the present invention can obtain a calculation formula for indicating the relationship between the application of the feature value sequence and the CTR of the application by using a learning process, and can quickly determine the CTR based on the determined feature value sequence and the calculation formula. Calculate the CTR of the application, and then quickly display the appropriate application in the promotion resource, improve the user experience, and merge the sample data when the calculation formula is obtained, and the resource consumption is low.

Referring to FIG. 3, FIG. 3 is a schematic flowchart diagram of still another method for determining a click arrival rate CTR according to an embodiment of the present invention. Among them, the method shown in FIG. 3 can be applied to a server. As shown in FIG. 3, the click arrival rate CTR can be determined as follows:

S301. Read a first quantity of first sample data from pre-stored sample data.

S302. The first sample data with the same sequence of feature values in the first quantity of the first sample data is merged to obtain a second quantity of second sample data.

In the embodiment of the present invention, the first sample data having the same sequence of feature values is merged into a second sample data by using the same sequence of feature values as the principle of merging, that is, the first sample data of the first quantity is merged into the first sample data. And a second quantity is equal to the number of different values of the sequence of feature values in the first sample data of the first quantity. The second quantity of second sample data is formed by a sum of a sequence of feature values of the first sample data forming the second sample data, a pv of all first sample data forming the second sample data, and forming the second The sum of the clicks of all the first sample data of the sample data.

S303. Divide the second quantity of the second sample data into the sample group including the third quantity of the second sample data.

In the embodiment of the present invention, the third quantity may be less than or equal to the number of CPU cores in the server, so that the manner of equally dividing the second quantity of second sample data can simultaneously perform the same processing for each third quantity of the second sample data. , speed up the processing speed.

S304. Express each second sample data included in each sample group into a likelihood expression, and multiply all likelihood expressions corresponding to each sample group to obtain a product likelihood expression of the sample group.

S305, performing an iteration by a Newton iteration method and initial iteration parameters.

S306. Calculate a first value of an unknown parameter in the product likelihood expression when each product likelihood expression takes a maximum value, and calculate a sum of all the first values.

In the embodiment of the present invention, each product likelihood expression is iterated to obtain a value of the second number of unknown parameters, and then the values of the second number of unknown parameters are summed as the initial iteration parameter at the next iteration.

S307. Determine whether the number of iterations reaches the target number of times.

In the embodiment of the present invention, when the determination result in step S307 is YES, step S309 is performed; when the determination result in step S307 is NO, step S308 is performed.

In the embodiment of the present invention, the target number m may be a preset number of times, or may be a sum of values of the second number of unknown parameters obtained after the m-1th iteration.

The sum of the values of the second number of unknown parameters obtained after the mth iteration

Minimum angle or basis

S308. The sum of all the first values described above is used as an initial iteration parameter of the next iteration.

In the embodiment of the present invention, after step S308 is performed, step S305 is performed.

S309. Determine a sum of a second value of an unknown parameter in each product likelihood expression calculated after the target number of iterations is determined as

S310. Determine a sequence of feature values for each application when a display request for the application is detected.

S311. Each of the feature value sequences is respectively input as a preset CTR calculation algorithm, and an output of the CTR calculation algorithm corresponding to each feature value sequence is acquired.

S312. Determine an output of the CTR calculation algorithm corresponding to each feature value sequence as a CTR corresponding to the application of the feature value sequence.

After performing step S302 and before performing step S303, the following operations may also be performed:

Embodiments of the present invention can quickly determine the CTR of an application and consume small resources.

Referring to FIG. 4, FIG. 4 is a schematic structural diagram of a device for determining a click arrival rate CTR according to an embodiment of the present invention. As shown in Figure 4, the device can be installed in a server. As shown in FIG. 4, the apparatus may include a first determining unit 401, a first obtaining unit 402, and a second determining unit 403, where:

The first determining unit 401 is configured to determine a sequence of feature values for each application when a display request for the application is detected.

In the embodiment of the present invention, the display request for the application may be triggered by the user through the terminal device, or may be actively triggered by the terminal device, and the sequence of feature values of each application is used to describe the application information (such as the category of the application). And a sequence of application feature values of the related description information of the application, etc., a sequence of user feature values for describing user information (such as the gender of the user and the interest of the user, etc.) and information describing the user behavior (such as time, place, language, etc.) The sequence of traffic characteristic values is composed, and the sequence of feature values is a feature value vector composed of a plurality of 0s and 1s as components.

The first obtaining unit 402 is configured to respectively acquire each feature value sequence as an input of a preset CTR calculation algorithm, and acquire an output of a CTR calculation algorithm corresponding to each feature value sequence.

The second determining unit 403 is configured to determine an output of the CTR calculation algorithm corresponding to each sequence of feature values as the feature value. The sequence corresponds to the CTR of the application.

Where y ^CTR is the output of the calculation formula (CTR of the application),

Based on the device structure shown in FIG. 4, the device may further include a first reading unit 404, a first merging unit 405, a second obtaining unit 406, a first calculating unit 407, and a third determining unit 408. The structure of the apparatus can be as shown in FIG. 5. FIG. 5 is a schematic structural diagram of another apparatus for determining the click arrival rate CTR disclosed in the embodiment of the present invention. among them:

The first reading unit 404 is configured to read the first quantity of first sample data from the pre-stored sample data.

In the embodiment of the present invention, each sample data is composed of a sequence of feature values, a presentation identifier pv for identifying whether the sample is presented, and a click identifier click for identifying whether the sample data is clicked, and the sequence of feature values is used for A sequence of application feature values describing application information (such as the category of the application and related description information of the application, etc.), a sequence of user feature values for describing user information (such as the gender of the user and the user's interest, etc.) and information describing the user behavior A sequence of traffic characteristic values (such as time, place, language, etc.), wherein pv is 1 and click is 0 indicates that the sample data is presentation sample data, pv is 0 and click 1 indicates that the sample data is click sample data.

The first merging unit 405 is configured to merge the first sample data having the same sequence of feature values in the first quantity of the first sample data to obtain the second quantity of second sample data.

The second obtaining unit 406 is configured to express each second sample data as a likelihood expression and multiply all likelihood expressions to obtain a product likelihood expression.

The first calculating unit 407 is configured to perform the iteration of the target times by the Newton iteration method and the initial iteration parameter, and calculate The value of the unknown parameter in the above product likelihood expression is obtained when the above product likelihood expression takes the maximum value.

The third determining unit 408 is configured to determine the value of the above unknown parameter as the above

Based on the device structure shown in FIG. 4, the device may further include a second reading unit 409, a second merging unit 410, a storage unit 411, a averaging unit 412, a third obtaining unit 413, a second calculating unit 414, and The fourth determining unit 415, at this time, the structure of the device can be as shown in FIG. 6. FIG. 6 is a schematic structural diagram of another determining device for the click arrival rate CTR disclosed in the embodiment of the present invention. among them:

The second reading unit 409 is configured to read the first quantity of the first sample data from the pre-stored sample data.

The second merging unit 410 is configured to merge the first sample data having the same sequence of feature values in the first quantity of the first sample data to obtain a second quantity of second sample data.

The storage unit 411 is configured to store the second quantity of the second sample data in the memory space with consecutive addresses.

The averaging unit 412 is configured to divide the second number of second sample data into a sample group including a third number of second sample data.

The third obtaining unit 413 is configured to respectively express each second sample data included in each sample group into a likelihood expression, and multiply all likelihood expressions corresponding to each sample group to obtain a product of the sample group. Likelihood expression.

The second calculating unit 414 is configured to perform an iteration by the Newton iteration method and the initial iterative parameter, and respectively calculate the first value of the unknown parameter in the product likelihood expression when each product likelihood expression takes the maximum value, and each will be The sum of the first values is used as the initial iteration parameter for the next iteration, and an iterative operation is performed by the Newton iteration method and the initial iteration parameters until the number of iterations reaches the target number of times.

The fourth determining unit 415 is configured to determine, as the above, the sum of the second values of the unknown parameters in each of the product likelihood expressions calculated after the target number of times is iterated

The embodiment of the invention can quickly determine the CTR of the application and the resource consumption is small.

In order to implement the above embodiment, the present invention also proposes a server including the determining means of the click arrival rate CTR of another embodiment of the present invention described in FIGS. 4, 5 and 6.

Referring to FIG. 7, FIG. 7 is a schematic structural diagram of a server for performing a method for determining a click arrival rate CTR according to an embodiment of the present invention. The server may include at least one processor 501, such as a CPU, at least one network interface 504 or other user interface 503, memory 505, at least one communication bus 502. Communication bus 502 is used to implement connection communication between these components. The user interface 503 can optionally include a USB interface and other standard interfaces and wired interfaces. Network interface 504 can optionally include a Wi-Fi interface as well as other wireless interfaces. The memory 505 may include a high speed RAM memory and may also include a non-volatile memory such as at least one disk memory. The memory 505 can optionally include at least one storage device located remotely from the aforementioned processor 501. As shown in FIG. 7, an operating system 5051 and an application 5052 may be included in the memory 505 as a computer storage medium.

In some implementations, the memory 505 stores the following elements, executable modules or data structures, or a subset thereof, or their extended set:

Operating system 5051, including various system programs for implementing various basic services and processing hardware-based tasks;

The application 5052 includes various setting programs of a data block parameter setting program of the target network data, a partitioning program of the target data block, a comparison program of the target data block and the data block in the database, and a deletion program of the target data block. To achieve a variety of application services.

Specifically, the processor 501 is configured to call a program stored in the memory 505, and performs the following operations:

Receiving client feature information collected by the client, where the client feature information is formed according to related information of the application in the client operating system and related information of the client user, where the related information of the application includes the application At least one of operational information, behavior information of the application, and context information related to the application.

For the specific execution process of the foregoing steps and the steps of the processor 501 to be further executed by the processor 501, reference may be made to the description of the foregoing method embodiments, and details are not described herein again.

The embodiment of the present invention further provides a computer readable storage medium having instructions stored therein, when the processor of the mobile terminal executes the instruction, the mobile terminal performs the determining method of the click arrival rate CTR.

The embodiment of the present invention further provides a computer application program that performs the above-described determination method of the click arrival rate CTR when it is executed on the processor of the mobile terminal.

It should be noted that, in the above embodiments, the descriptions of the various embodiments are different, and the parts that are not described in detail in a certain embodiment may be referred to the related descriptions of other embodiments. In addition, those skilled in the art should also understand that the embodiments described in the specification are all preferred embodiments, and the actions and units involved are not necessarily required by the present invention.

The method and device for determining the click arrival rate CTR provided by the embodiments of the present invention are described in detail. The principles and implementation manners of the present invention are described in the specific examples. The description of the above embodiments is only The method for understanding the present invention and its core idea; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in specific embodiments and application scopes. The description should not be construed as limiting the invention.

The steps in the method of the embodiment of the present invention may be sequentially adjusted, merged, and deleted according to actual needs.

The units in the apparatus of the embodiment of the present invention may be combined, divided, and deleted according to actual needs.

The unit in the embodiment of the present invention may be implemented by a general-purpose integrated circuit, such as a CPU (Central Processing Unit), or an ASIC (Application Specific Integrated Circuit).

A person skilled in the art can understand that all or part of the process of implementing the above embodiment method can be completed by a computer program to instruct related hardware, and the program can be stored in a computer readable storage medium. When executed, the flow of an embodiment of the methods as described above may be included. The storage medium may be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

It should be noted that, in this context, relational terms such as first and second are used merely to distinguish one entity or operation from another entity or operation, without necessarily requiring or implying between these entities or operations. There are any such actual relationships or sequences. Furthermore, the term "comprises" or "comprises" or "comprises" or any other variations thereof is intended to encompass a non-exclusive inclusion, such that a process, method, article, or device that comprises a plurality of elements includes not only those elements but also Other elements, or elements that are inherent to such a process, method, item, or device. An element that is defined by the phrase "comprising a ..." does not exclude the presence of additional equivalent elements in the process, method, item, or device that comprises the element.

The logic and/or steps represented in the flowchart or otherwise described herein, for example, may be considered as an ordered list of executable instructions for implementing logical functions, and may be embodied in any computer readable medium, Used in conjunction with, or in conjunction with, an instruction execution system, apparatus, or device (eg, a computer-based system, a system including a processor, or other system that can fetch instructions and execute instructions from an instruction execution system, apparatus, or device) Or use with equipment. For the purposes of this specification, a "computer-readable medium" can be any apparatus that can contain, store, communicate, propagate, or transport a program for use in an instruction execution system, apparatus, or device, or in conjunction with the instruction execution system, apparatus, or device. More specific examples (non-exhaustive list) of computer readable media include the following: electrical connections (electronic devices) having one or more wires, portable computer disk cartridges (magnetic devices), random access memory (RAM), Read only memory (ROM), erasable editable read only memory (EPROM or flash memory), fiber optic devices, and portable compact disk read only memory (CDROM). In addition, the computer readable medium may even be a paper or other suitable medium on which the program can be printed, as it may be optically scanned, for example by paper or other medium, followed by editing, interpretation or, if appropriate, other suitable The method is processed to obtain the program electronically and then stored in computer memory.

It should be understood that portions of the invention may be implemented in hardware, software, firmware or a combination thereof. In the above implementation In the manner, multiple steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented by any one or combination of the following techniques well known in the art: having logic gates for implementing logic functions on data signals. Discrete logic circuits, application specific integrated circuits with suitable combinational logic gates, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

In the present invention, the terms "installation", "connected", "connected", "fixed" and the like shall be understood broadly, and may be either a fixed connection or a detachable connection, unless explicitly stated and defined otherwise. , or integrated; can be mechanical or electrical connection; can be directly connected, or indirectly connected through an intermediate medium, can be the internal communication of two elements or the interaction of two elements, unless otherwise specified Limited. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood on a case-by-case basis.

In the description of the present specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" and the like means a specific feature described in connection with the embodiment or example. A structure, material or feature is included in at least one embodiment or example of the invention. In the present specification, the schematic representation of the above terms is not necessarily directed to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. In addition, various embodiments or examples described in the specification, as well as features of various embodiments or examples, may be combined and combined.

Although the embodiments of the present invention have been shown and described, it is understood that the above-described embodiments are illustrative and are not to be construed as limiting the scope of the invention. The embodiments are subject to variations, modifications, substitutions and variations.

Claims

A method for determining a click arrival rate CTR, the method comprising:

When a display request for an application is detected, a sequence of feature values for each application is determined, the sequence of feature values being a sequence of applied feature values for describing application information, a sequence of user feature values for describing user information, and a flow feature value Sequence composition

And respectively outputting each of the feature value sequences as an input of a preset CTR calculation algorithm, and acquiring an output of the CTR calculation algorithm corresponding to each of the feature value sequences;

The output of the CTR calculation algorithm corresponding to each of the feature value sequences is determined as the CTR of the application of the feature value sequence.
The method according to claim 1, wherein the CTR calculation algorithm is a logistic regression model based algorithm, and the calculation formula of the logistic regression model based algorithm is:

Wherein the yCTR is an output of the calculation formula,
For the input of the calculation formula, the
It is a coefficient of the calculation formula calculated in advance.
The method according to claim 1 or 2, wherein the method further comprises: before determining a sequence of feature values for each application when detecting a display request for an application, the method further comprising:

Reading a first quantity of first sample data from pre-stored sample data, the sample data being represented by a sequence of feature values, a presentation identifier for identifying whether the sample data is presented, and for identifying whether the sample data is The clicked click mark consists of;

And combining the first sample data of the first number of first sample data with the same sequence of feature values to obtain a second quantity of second sample data, wherein the second sample data is formed by the second sample data a sum of a sequence of feature values of the first sample data, a sum of presentation identifiers of the first sample data forming the second sample data, and a click identifier of the first sample data forming the second sample data;

Expressing each of the second sample data into a likelihood expression, and multiplying all of the likelihood expressions to obtain a product likelihood expression;

Calculating the value of the unknown parameter in the product likelihood expression when the product likelihood expression takes the maximum value by using the Newton iteration method and the initial iteration parameter to perform the iteration of the target number of times, and determining the value of the unknown parameter For the stated
The method according to claim 1 or 2, wherein the method further comprises: before determining a sequence of feature values for each application when detecting a display request for an application, the method further comprising:

Reading a first quantity of first sample data from pre-stored sample data, the first sample data being represented by a sequence of feature values, a presentation identifier for identifying whether the sample data is presented, and for identifying the Whether the sample data is composed of clicked click identifiers;

And combining the first sample data of the first number of first sample data with the same sequence of feature values to obtain a second quantity of second sample data, wherein the second sample data is formed by the second sample data a sum of a sequence of feature values of the first sample data, a sum of presentation identifiers of the first sample data forming the second sample data, and a click identifier of the first sample data forming the second sample data;

Dividing the second quantity of second sample data into a sample group each including a third quantity of second sample data;

Each second sample data included in each of the sample sets is expressed as a likelihood expression, and all of the likelihood expressions corresponding to each of the sample sets are multiplied to obtain a product of the sample set Likelihood expression

Performing an iteration by the Newton iteration method and the initial iteration parameter, respectively calculating a first value of the unknown parameter in the product likelihood expression when each of the product likelihood expression takes a maximum value, each of the first The sum of the values is used as the initial iteration parameter of the next iteration, and the operation of performing an iteration by the Newton iteration method and the initial iteration parameter is performed until the number of iterations reaches the target number of times;

And determining a sum of the second values of the unknown parameters in each of the product likelihood expressions calculated after the target number of iterations is determined as
.
The method according to claim 4, wherein after the first sample data having the same sequence of feature values in the first number of first sample data is merged, the second number of Before the two sample data are divided into each sample group including the third quantity of the second sample data, the method further includes:

The second quantity of second sample data is stored in a memory space of consecutive addresses.
A determining device for a click arrival rate CTR, characterized in that the device comprises a first determining unit, a first obtaining unit and a second determining unit, wherein:

The first determining unit is configured to determine a sequence of feature values of each application when the display request for the application is detected, where the sequence of feature values is used by the application feature value sequence for describing application information, and is used to describe user information. User characteristic value sequence and flow characteristic value sequence composition;

The first acquiring unit is configured to respectively obtain, according to an input of a preset CTR calculation algorithm, an output of the CTR calculation algorithm corresponding to each of the feature value sequences;

The second determining unit is configured to determine an output of the CTR calculation algorithm corresponding to each of the feature value sequences as a CTR corresponding to the application of the feature value sequence.
The apparatus according to claim 6, wherein the CTR calculation algorithm is a logistic regression model based algorithm, and the calculation formula of the logistic regression model based algorithm is:

Wherein the yCTR is an output of the calculation formula,
For the input of the calculation formula, the
It is a coefficient of the calculation formula calculated in advance.
The device according to claim 6 or 7, wherein the device further comprises a first reading unit, a first merging unit, a second obtaining unit, a first calculating unit and a third determining unit, wherein:

The first reading unit is configured to read a first quantity of first sample data from pre-stored sample data, the sample data being represented by a sequence of feature values for identifying whether the sample data is presented An identifier and a click identifier for identifying whether the sample data is clicked;

The first merging unit is configured to merge the first sample data of the first number of first sample data with the same sequence of feature values to obtain a second quantity of second sample data, the second sample data a sum of a feature value sequence of the first sample data forming the second sample data, a presentation identifier of the first sample data forming the second sample data, and a first sample forming the second sample data The sum of the click marks of the data;

The second obtaining unit is configured to express each of the second sample data into a likelihood expression, and multiply all the likelihood expressions to obtain a product likelihood expression;

The first calculating unit is configured to perform an iteration of a target number of times by a Newton iteration method and an initial iteration parameter, and calculate a value of an unknown parameter in the product likelihood expression when the product likelihood expression takes a maximum value;

The third determining unit is configured to determine a value of the unknown parameter as the
The apparatus according to claim 6 or 7, wherein the apparatus further comprises a second reading unit, a second merging unit, a averaging unit, a third obtaining unit, a second calculating unit, and a fourth determining unit, among them:

The second reading unit is configured to read a first quantity of first sample data from pre-stored sample data, where the first sample data is used by a sequence of feature values to identify whether the sample data is a presentation identifier of the presentation and a click identifier for identifying whether the sample data is clicked;

The second merging unit is configured to merge the same first eigenvalue sequence in the first quantity of the first sample data The data is obtained to obtain a second quantity of second sample data, the second sample data being the same as the sequence of feature values of the first sample data forming the second sample data, forming the second sample data a sum of a presentation identifier of the data and a click identifier of the first sample data forming the second sample data;

The equalizing unit is configured to divide the second quantity of second sample data into a sample group including a third quantity of second sample data;

The third obtaining unit is configured to respectively express each second sample data included in each of the sample groups into a likelihood expression, and compare all the likelihood expressions corresponding to each of the sample groups Multiplying by to obtain a product likelihood expression of the sample set;

The second calculating unit is configured to perform an iteration by using a Newton iteration method and an initial iteration parameter, and respectively calculate a first unknown parameter in the product likelihood expression when each of the product likelihood expressions takes a maximum value a value, using the sum of each of the first values as an initial iteration parameter of the next iteration, performing the operation of performing an iteration by the Newton iteration method and the initial iteration parameter until the number of iterations reaches the target number of times;

The fourth determining unit is configured to determine a sum of the second value of the unknown parameter in each of the product likelihood expressions calculated after the target number of times is iterated as
The apparatus of claim 9 wherein said apparatus further comprises a storage unit, wherein:

The storage unit is configured to store the second quantity of second sample data in a memory space with consecutive addresses.
A server, comprising:

Processor;

a memory for storing instructions executable by the processor,

Wherein the processor is configured to read an executable instruction stored in the memory to execute a program corresponding to the instruction for performing the click arrival rate CTR according to any one of claims 1 to 5. Determine the method.
A computer readable storage medium having instructions stored therein, the server performing the determination method of the click arrival rate CTR according to any one of claims 1 to 5 when the processor of the server executes the instruction .
A computer application that, when executed on a processor of a server, performs the determination method of the click arrival rate CTR according to any one of claims 1 to 5.