CN107526810B - Method and device for establishing click rate estimation model and display method and device - Google Patents
Method and device for establishing click rate estimation model and display method and device Download PDFInfo
- Publication number
- CN107526810B CN107526810B CN201710729982.1A CN201710729982A CN107526810B CN 107526810 B CN107526810 B CN 107526810B CN 201710729982 A CN201710729982 A CN 201710729982A CN 107526810 B CN107526810 B CN 107526810B
- Authority
- CN
- China
- Prior art keywords
- event
- value
- click
- click rate
- feature
- 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.)
- Active
Links
Images
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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0242—Determining effectiveness of advertisements
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0263—Targeted advertisements based upon Internet or website rating
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0277—Online advertisement
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Finance (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Economics (AREA)
- Game Theory and Decision Science (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The embodiment of the application discloses a method and a device for establishing a click rate pre-estimation model, a display method and a device, a terminal and a storage medium, wherein the method comprises the following steps: establishing an error function based on the type of each display event and the click rate pre-estimated value of the display event, wherein the value of the type of the display event is determined according to the actual click condition of a user, and the click rate pre-estimated value of the display event is calculated based on a built click rate pre-estimated model; establishing an error loss function of the click rate estimation model based on the error function; solving the weight value of the feature of each feature set in the click rate estimation model based on the error loss function and the value of the type of each display event; and determining a click rate estimation model according to the value of each weight obtained by solving.
Description
Technical Field
The present application relates to the field of network technologies, and in particular, to a method and an apparatus for establishing a click rate pre-estimation model, and a display method and an apparatus.
Background
The network technology is developed to the present day, more and more events, such as news information, articles, music, pictures and the like, need to be displayed to users, and the user needs to be specifically found for different display events to perform personalized display when the good effect is obtained by the display, and the core technical difficulty of the personalized display is how to accurately determine the events to be displayed.
Disclosure of Invention
In view of the foregoing problems, an object of the present application is to provide a method and an apparatus for building a click rate estimation model, a display method and apparatus, a terminal and a storage medium, which improve the accuracy of displaying events to a user.
In one aspect, the present application provides a method for establishing a click rate estimation model, including:
establishing an error function based on the type of each display event and the click rate pre-estimated value of the display event, wherein the value of the type of the display event is determined according to the actual click condition of a user, and the click rate pre-estimated value of the display event is calculated based on a built click rate pre-estimated model;
establishing an error loss function of the click rate estimation model based on the error function;
solving the weight value of the feature of each feature set in the click rate estimation model based on the error loss function and the value of the type of each display event;
and determining a click rate estimation model according to the value of each weight obtained by solving.
Optionally, the click rate prediction model is constructed by the following steps:
collecting characteristics of a user who shows an event;
classifying the characteristics of the user, dividing each type of characteristics into a plurality of characteristic sets, counting the standardized value of the characteristics in each characteristic set to each display event, and setting weight for the characteristics of each characteristic set;
and constructing a click rate pre-estimation model of each display event based on the normalized value of the features of each feature set to each display event and the weight of the features of each feature set.
Optionally, the method further comprises:
and calculating a click rate pre-estimated value of the event to be displayed based on the click rate pre-estimated model.
Optionally, after calculating the click rate pre-estimated value of the event to be displayed based on the click rate pre-estimation model, the method further includes:
calculating the expected income of the event to be displayed based on the calculated click rate estimated value of the event to be displayed;
ordering expected profits of a plurality of events to be displayed;
and displaying the event to be displayed with the expected income in the preset area to the user.
Optionally, the calculating a click rate pre-estimation value of the event to be shown based on the click rate pre-estimation model includes:
and substituting the normalized values of the events to be displayed and the weight of the features of each feature set into the click rate estimation model by the features of the feature sets to calculate to obtain the click rate estimated value of the events to be displayed.
Optionally, solving the value of each weight based on the error loss function and the value of the type of each display event includes:
setting an initial value for each weight;
performing iterative computation on the error loss function by taking the loss minimum of the error loss function as a target;
and stopping the iterative computation when the change rate of the error loss function is smaller than a preset threshold value, and taking the value of each weight at the moment as the value of the weight.
Optionally, the click rate prediction model is:
the error loss function is:
where j represents a presentation event, i represents a feature, θiRepresents the weight of the feature i, n represents the total number of features, m represents the total number of presentation events, flagjRepresenting the type of the j display event, 0 being a negative display event, 1 being a positive display event, if the j display event is clicked by the user, the j display event is a positive display event, if the j display event is not clicked by the user, the j display event is a negative display event, and pi，jAnd expressing the normalized value of the ith feature at the jth display event, wherein the calculation formula is as follows:
wherein f isi，jThe value, click (f), representing the ith feature at the jth presentation eventi，j) Representing a feature fi，jNumber of clicks, show (f)i，j) Representing a feature fi，jDisplay number of (d), pctr (f)i，j) Representing a feature fi，jThe click rate of (c).
In another aspect, the present application further provides a display method, including:
obtaining a click rate pre-estimated value of an event to be displayed;
calculating expected income obtained by clicking the event to be displayed based on the click rate pre-estimated value;
ordering expected profits of a plurality of events to be displayed;
and displaying the event to be displayed with the expected income in the preset area to the user.
Optionally, the calculating the expected profit obtained by clicking on the event to be shown based on the click rate pre-estimated value includes:
and multiplying the click rate estimated value by a preset click unit price to obtain expected benefits.
On the other hand, the application also provides a device for establishing a click rate pre-estimation model, which comprises the following steps:
the acquisition module is used for establishing an error function based on the type of each display event and the click rate pre-estimated value of the display event, wherein the click rate pre-estimated value of the display event is obtained by calculation based on a constructed click rate pre-estimated model, and the value of the type of the display event is determined according to the actual click condition of a user;
the establishing module is used for establishing an error loss function of the click rate estimation model based on the error function;
the solving module is used for solving the value of the weight of the feature of each feature set in the click rate estimation model based on the error loss function and the value of the type of each display event;
and the determining module is used for determining the click rate estimation model according to the value of each weight obtained by solving.
Optionally, the click rate prediction model is constructed by the following modules:
the acquisition module is used for acquiring the characteristics of the user who displays the event;
the statistical module is used for classifying the characteristics of the user, dividing each type of characteristics into a plurality of characteristic sets, counting the standardized value of the characteristics in each characteristic set to each display event, and setting weight for the characteristics of each characteristic set;
and the construction module is used for constructing a click rate pre-estimation model of each display event based on the normalized value of the features of each feature set to each display event and the weight of the features of each feature set.
Optionally, the apparatus further comprises:
and the first calculation module is used for calculating the click rate estimated value of the event to be displayed based on the click rate estimation model.
Optionally, the first computing module further comprises:
the second calculation module is used for calculating the expected income of the event to be displayed based on the calculated click rate pre-estimated value of the event to be displayed;
the sequencing module is used for sequencing the expected income of a plurality of events to be displayed;
and the display module is used for displaying the event to be displayed with the expected income in the preset area to the user.
Optionally, the first calculation module is specifically configured to substitute the normalized value of the event to be displayed by the features of the feature set and the weight of the features of each feature set into the click rate estimation model to calculate a click rate estimated value of the event to be displayed.
Optionally, the solving module includes:
the initialization module is used for setting an initial value for each weight;
and the iterative calculation module is used for performing iterative calculation on the error loss function by taking the minimum loss of the error loss function as a target, stopping the iterative calculation when the change rate of the error loss function is smaller than a preset threshold value, and taking the value of each weight at the moment as the value of the weight.
Optionally, the click rate prediction model is:
the error loss function is:
where j represents a presentation event, i represents a feature, θiWeight of the feature i, n the total number of features, m the showTotal number of elements, flagjRepresenting the type of the j display event, 0 being a negative display event, 1 being a positive display event, if the j display event is clicked by the user, the j display event is a positive display event, if the j display event is not clicked by the user, the j display event is a negative display event, and pi，jAnd expressing the normalized value of the ith feature at the jth display event, wherein the calculation formula is as follows:
wherein f isi，jThe value, click (f), representing the ith feature at the jth presentation eventi，j) Representing a feature fi，jNumber of clicks, show (f)i，j) Representing a feature fi，jDisplay number of (d), pctr (f)i，j) Representing a feature fi，jThe click rate of (c).
In another aspect, the present application further provides a display device, including:
the device comprises a pre-estimated value acquisition module, a display module and a display module, wherein the pre-estimated value acquisition module is used for acquiring a click rate pre-estimated value of an event to be displayed;
the profit calculation module is used for calculating expected profits obtained by clicking the events to be displayed based on the click rate pre-estimated value;
the income sorting module is used for sorting the expected income of a plurality of events to be displayed;
and the event display module is used for displaying the event to be displayed with the expected income in the preset area to the user.
Optionally, the profit calculating module is specifically configured to multiply the click rate estimated value by a preset click unit price to obtain an expected profit.
In another aspect, the present application further provides a terminal, including: a processor and a memory storing computer instructions;
the processor reads the computer instructions and executes a method for establishing a click rate prediction model as described above.
In another aspect, the present application further provides a terminal, including: a processor and a memory storing computer instructions;
the processor reads the computer instructions and executes a presentation method as described above.
In another aspect, the present application further provides a storage medium storing computer instructions, which when executed, implement the method for establishing a click-through rate estimation model as described above.
In another aspect, the present application further provides a storage medium storing computer instructions, which when executed, implement the display method as described above.
The method and the device for establishing the click rate pre-estimation model, the display method and the device, the terminal and the storage medium provided by the embodiment of the application are used for further dividing the characteristics into a plurality of characteristic sets on the basis of the click behaviors of a plurality of users for displaying events, counting the standardized value of the characteristics of each characteristic set, configuring the weight for the characteristics in each characteristic set and establishing the click rate pre-estimation model by taking the weight of each characteristic as a parameter. After each weight is determined, the click rate estimation model is determined. The click rate estimated value of any display event can be calculated by using the determined click rate estimated model, so that the event with high click rate estimated value can be displayed to the user, and the accuracy of displaying the event to the user is improved.
Drawings
The above and other objects, features and advantages of the present application will become apparent from the following detailed description, which proceeds with reference to the accompanying drawings. In the drawings:
fig. 1 is a flowchart of a method for establishing a click rate estimation model according to an embodiment of the present application;
FIG. 2 is a flowchart illustrating a method for building a click through rate estimation model according to an embodiment of the present disclosure;
FIG. 3 is a flowchart illustrating a method for building a click through rate estimation model according to an embodiment of the present application;
FIG. 4 is a flowchart illustrating a method for building a click through rate estimation model according to an embodiment of the present application;
FIG. 5 is a scene diagram illustrating an advertisement application click-through rate estimation model according to an embodiment of the present application;
FIG. 6 is a flowchart of a display method according to an embodiment of the present application;
FIG. 7 is a schematic structural diagram of an apparatus for building a click rate estimation model according to an embodiment of the present application;
fig. 8 is a schematic structural diagram of a display device according to an embodiment of the present application.
Detailed Description
Various aspects of the present application are described below. The teachings herein may be embodied in a wide variety of forms and any specific structure, function, or both being disclosed herein is merely representative. Based on the teachings herein one skilled in the art should appreciate that an aspect disclosed herein may be implemented independently of any other aspects and that two or more of these aspects may be combined in various ways. For example, an apparatus may be implemented or a method may be practiced using any number of the aspects set forth herein. In addition, such an apparatus may be implemented or such a method may be practiced using other mechanisms, functions, or structures and functions in addition to or other than one or more of the aspects set forth herein. Furthermore, any aspect described herein may include at least one element of a claim.
The application provides a method and a device for establishing a click rate estimation model, a terminal and a storage medium. The following description of specific embodiments of the present application refers to the accompanying drawings.
Referring to fig. 1, a method for establishing a click through rate prediction model according to an embodiment of the present application includes steps 101 to 104.
Step 101: and establishing an error function based on the type of each display event and the click rate pre-estimated value of the display event, wherein the click rate pre-estimated value of the display event is obtained by calculation based on a constructed click rate pre-estimated model, and the value of the type of the display event is determined according to the actual click condition of a user.
In the embodiment of the present application, the presentation event may be various events that are presented to the user and that the user wants to perform a click operation, such as presenting an advertisement, presenting an article, presenting an application, presenting music or a picture, presenting a movie, and the like.
Obtaining the type of each display event, taking the display advertisement as an example, displaying the advertisement for a certain user, wherein if the user clicks and views the display advertisement, the type of the display advertisement is a positive sample and can be represented by a numeral 1; if the user does not click to view the presented advertisement, the type of the presented advertisement is a negative example, which may be represented by the numeral 0.
Referring to fig. 2, in an embodiment of the present application, the click rate estimation model is constructed through steps 1011 to 1013.
Step 1011: features of a user presenting an event are collected.
In the embodiment of the application, the characteristics of the user may include self attribute characteristics, such as age, academic calendar, city, occupation, income, or a mobile phone application type preferred by the user; and mobile phone application type characteristics such as the type of the mobile phone application, the click rate of the mobile phone application and the like can also be included.
Taking advertisement display as an example, a certain advertisement is displayed to a plurality of users, and the characteristics of the user who has displayed the advertisement can be obtained by collecting the characteristics of the plurality of users.
In the embodiment of the present application, each user may be regarded as a feature set, for example, the user a is a feature set a having features of 95 th, male, online game, etc., the user B is a feature set B having features of 95 th, female, online game, etc., the user C is a feature set C having features of 90 th, male, online game, etc., all features of the feature set a, the feature set B, and the feature set C are collected, and then classified according to age features, that is, the features having 95 th are classified into one set, the features having 90 th are classified into one set, the features having online game are classified into one set, or classified according to gender features, that is, the features having male are classified into one set, and the features having female are classified into one set.
Step 1012: classifying the characteristics of the user, dividing each type of characteristics into a plurality of characteristic sets, counting the standardized value of the characteristics in each characteristic set to each display event, and setting weight for the characteristics of each characteristic set.
In the embodiment of the present application, each type of feature of the user may be divided into a plurality of different feature sets in some manner. For example, the age may be classified into "children", "teenagers", "young adults", "middle-aged people" and "old people", and may also be classified into "70 back", "80 back", "90 back", "95 back" and "00 back".
The dividing mode of each type of features can be determined according to actual needs, and the application is not limited.
Step 1013: and constructing a click rate pre-estimation model of each display event based on the normalized value of the features of each feature set to each display event and the weight of the features of each feature set.
In the embodiment of the application, the normalized value of the feature of each feature set to each display event is obtained by calculating the actual number of clicks of the feature of each feature set to each display event and the actual number of displays of the feature of each feature set to the event in advance.
In the embodiment of the application, the weight of the feature of each feature set represents the credibility of the feature set to the display event.
In an embodiment of the application, the click rate estimation model may be:
where j represents a presentation event, i represents a feature, θiRepresenting the weight of the feature i, n representing the total number of features, pi，jA normalized value representing the value of the ith feature at the jth display event.
Through the click rate estimation model, the click rate of a certain display event is determined by the normalized value of the feature pair display event in each feature set and the weight of the feature in each feature set, and the normalized value of the feature pair display event in each feature set can be obtained by calculating in advance according to the actual number of clicks of the feature pair display event in each feature set and the actual number of displays of the event to the feature of each feature set.
Step 102: and establishing an error loss function of the click rate estimation model based on the error function.
Step 103: and solving the weight value of the feature of each feature set in the click rate pre-estimation model based on the error loss function and the value of the type of each display event.
In an embodiment of the present application, based on the error loss function and the value of the type of each display event, solving the value of each weight may include:
setting an initial value for each weight;
performing iterative computation on the error loss function by taking the loss minimum of the error loss function as a target;
and stopping the iterative computation when the change rate of the error loss function is smaller than a preset threshold value, and taking the value of each weight at the moment as the value of the weight.
In the embodiment of the application, the weight of each feature to the presentation event is determined according to the actual click condition and the click rate estimated value of the user to the presentation event.
In an embodiment of the present application, if the click rate estimation model is shown as equation 1, the corresponding error function is:
the error loss function is:
where j represents a presentation event, i represents a feature, θiRepresents the weight of the feature i, n represents the total number of features, m represents the total number of presentation events, pi，jNormalized value, flag, representing the value of the ith feature at the jth display eventjThe j-th display event is represented by the type, 0 is a negative display event, 1 is a positive display event, if the j-th display event is clicked by a user, the j-th display event is a positive display event, and if the j-th display event is not clicked by the user, the j-th display event is a negative display event.
If the user clicks the presentation event j, flagj1, otherwise flagj＝0。
Wherein p isi，jThe calculation formula of (a) is as follows:
wherein f isi，jThe value, click (f), representing the ith feature at the jth presentation eventi，j) Representing a feature fi，jNumber of clicks, show (f)i，j) Representing a feature fi，jDisplay number of (d), pctr (f)i，j) Representing a feature fi，jThe click rate of (c).
Still taking the example of a presentation advertisement, the presentation advertisement is presented 100 times to a user having a "after 95" feature, where the number of click actions generated is 10, then the normalized value of the "after 95" feature for the presentation advertisement is:
if the weight of the features in the feature set of the shown advertisement is determined, the click rate of the shown advertisement can be estimated according to the weight of each feature to the shown advertisement.
As can be seen from equation 3, the embodiment of the present application can calculate p in advance according toi，jAnd the actual value of the type of the display event is taken by the user, and the value of the weight corresponding to each characteristic is solved.
In the embodiment of the application, a gradient descent method can be adopted, and based on the actual click value of the user who shows the event, the { theta ] of formula 1 is solvediI is more than or equal to 0 and less than or equal to n, i belongs to the optimal value of Z, namely the weight of each characteristic pair to the display event is solved. The solving method may include the steps of:
the first step is as follows: randomly given a set of numbers between 0-1 thetaiI is more than or equal to 0 and less than or equal to n, i belongs to Z, and is set as theta(0)Initializing the iteration step number k to be 0;
the second step is that: iterative computation
Wherein alpha is the step length of iteration and is taken as 0.001;
the third step: determining whether the error loss function converges
Δg(θ(k+1))＝|g(θ(k+1))-g(θ(k))|
If | Δ g (θ)(k+1))-Δg(θ(k)If | is less than β, then θ is returned(k+1)，θ(k+1)I.e. the estimated model parameters, otherwise go back to the second step to continue the calculation, where β is a small value, and β may be 0.01 · α.
Step 104: and determining a click rate estimation model according to the value of each weight obtained by solving.
In the embodiment of the application, the weight value of each feature to the display event is determined through the calculation of the steps, so that a click rate estimation model of the display event is established.
In the embodiment of the application, the click rate pre-estimated value of the event to be displayed can be calculated based on the click rate pre-estimated model.
The method for establishing the click rate estimation model provided by the embodiment of the application is characterized in that based on click behaviors of a plurality of users for displaying events, the characteristics are further divided into a plurality of characteristic sets, the standardized value of the characteristics of each characteristic set is counted, the weight is configured for the characteristics in each characteristic set, and the click rate estimation model is established by taking the weight of each characteristic as a parameter. After each weight is determined, the click rate estimation model is determined. The click rate estimated value of any display event can be calculated by using the determined click rate estimated model, so that the display event with high click rate estimated value can be selected, and the accuracy of displaying the event for the user is improved.
Referring to fig. 3, in an embodiment of the present application, the actual application of the click-through rate prediction model based on the event to be presented includes steps 301 to 304.
Step 301: and calculating a click rate pre-estimated value of the event to be displayed based on the click rate pre-estimated model.
Step 302: and calculating the expected income of the event to be displayed based on the calculated click rate estimated value of the event to be displayed.
In the embodiment of the present application, the click rate pre-estimated value of the event to be shown, which is obtained based on the calculation, includes: and substituting the normalized values of the events to be displayed and the weight of the features of each feature set into the click rate estimation model by the features of the feature sets to calculate to obtain the click rate estimated value of the events to be displayed.
Step 303: the expected revenue of a plurality of events to be presented is ranked.
In the embodiment of the application, the events to be displayed are sequenced according to the expected income, the events to be displayed with high expected income can be placed at a preferential position, and the events to be displayed with low expected income can be placed at a later position.
Step 304: and displaying the event to be displayed with the expected income in the preset area to the user.
For example, an event to be shown with the expected profit of the top 90 can be placed in the preset area, the event to be shown with the expected profit ranking of the top 90 is placed in the preset area according to the expected profit, and then 90 events to be shown in the preset area are shown to the user.
In the embodiment of the application, the click rate pre-evaluation value of the event to be displayed is calculated based on the click rate pre-evaluation model, the event to be displayed is sorted according to the click rate pre-evaluation value, the event with the highest click rate pre-evaluation value is displayed for a user, the click viewing interest of the user on the displayed event can be greatly improved, and the user experience is improved.
Referring to fig. 4, in an embodiment of the present application, taking the presentation event a as an example of presenting an advertisement, a method for establishing a click through rate estimation model provided by the present application is described, which includes steps 401 to 407.
Step 401: characteristics of a user exhibiting an advertisement are collected.
The click rate estimation model of the displayed advertisement is established by collecting the characteristics of a plurality of users displaying the advertisement. The number of users who present advertisements may be plural, and the characteristics of each user may also be plural.
Step 402: classifying the characteristics of the user, dividing each type of characteristics into a plurality of characteristic sets, counting the standardized value of the characteristics in each characteristic set to the displayed advertisement, and setting weight for the characteristics of each characteristic set.
In an embodiment of the application, a feature set characterizing the user exhibiting the advertisement may be collected from multiple dimensions.
Dimension 1: the user's preference for the presented advertisement characterizes, for example, a user who likes shopping characterizes a "shopping fan".
Dimension 2: the region attribute of the user is characterized, such as Beijing, Tianjin and Shanghai.
Dimension 3: by user natural attributes such as age, gender, etc.
Dimension 4: the social attributes of the user are characterized, such as cultural level, occupation, territory, and the like.
In practical applications, the selected dimension may be different according to different objects. This is not a limitation of the present application.
For each type of feature, the feature can be further divided into different feature sets.
Step 403: and constructing the click rate estimation model based on the weight of the features of each feature set to the normalized value of the displayed advertisement and the features of each feature set.
In the embodiment of the application, the normalized value of the feature of each feature set to the displayed advertisement is calculated by the actual number of clicks of the feature of each feature set to the displayed advertisement and the actual number of displays of the feature of each feature set to the displayed advertisement.
In the embodiment of the present application, the weight of the feature of each feature set represents different reliability of the feature of each feature set on the displayed advertisement.
Taking age as an example, if the type of features of age further includes 3 feature sets of "90 th", "95 th" and "00 th", the features in each feature set have the same weight for the presentation event of the same presentation advertisement. For example, a 26-year-old user and a 24-year-old user both belong to the feature set of "after 90", and the age of the 26-year-old user and the 24-year-old user has the same value of the weight of the presentation event of the same presentation advertisement.
The click rate estimation model of the displayed advertisement established in the embodiment of the application is shown as the formula 5:
where A represents a presentation event of presenting an advertisement, i represents a feature, and θiWeight, θ, representing characteristic i0Representing a constant, n representing the total number of features, pi，AAnd expressing the normalized value of the ith characteristic in the value of the display advertisement, wherein the calculation formula is as follows:
wherein f isi，AA value, click (f), indicating that the ith feature is showing an advertisementi，A) Representing a feature fi，ANumber of clicks, show (f)i，A) Representing a feature fi，ADisplay number of (d), pctr (f)i，A) Representing a feature fi，AThe click rate of (c).
For example: the presentation advertisement is presented 100 times to a user having a "after 95" feature, where the number of click actions generated is 10, then the normalized value of the "after 95" feature for the presentation advertisement is:
step 404: and obtaining the type of the displayed advertisement and a click rate pre-estimated value of the displayed advertisement calculated based on the constructed click rate pre-estimated model, wherein the value of the displayed advertisement is determined according to the actual click rate of the user.
In the embodiment of the application, if the user clicks and views the display advertisement, the type of the display advertisement is a positive sample and can be represented by a numeral 1; if the user does not click to view the shown advertisement, the type of the shown advertisement is a negative example, which may be represented by the numeral 0.
Step 405: and establishing an error function of the type of the displayed advertisement and the click rate estimated value, and establishing an error loss function of the click rate estimation model based on the error function.
The error function in the embodiment of the present application is shown in equation 7.
The error loss function is shown in equation 8.
Where A represents a presentation event of presenting an advertisement, i represents a feature, and θiWeight representing characteristic i, n representing total number of characteristics, m representing total number of advertisements shown, pi，AA normalized value, flag, representing the value of the ith feature in the shown advertisementAAnd the type of the display advertisement is represented, wherein 0 is a negative display event, 1 is a positive display event, if the display advertisement is clicked by a user, the display event is a positive display event, and if the display advertisement is not clicked by the user, the display event is a negative display event.
If the user clicks the display advertisement, flagA1, otherwise flagA＝0。
Step 406: and solving the value of each weight based on the error loss function and the value of the type of the displayed advertisement.
According to equation 8, in the embodiment of the present application, the value of the weight of the feature of each feature set to the displayed advertisement is solved based on the error loss function and the value of the type of the displayed advertisement, that is, the actual click value.
In the embodiment of the application, a gradient descent method can be adopted, and the weight of each characteristic to the display event is solved based on the actual click value of the user displaying the advertisement. The solving method may include the steps of:
the first step is as follows: randomly given a set of numbers between 0-1 thetaiI is more than or equal to 0 and less than or equal to n, i belongs to Z, and is set as theta(0)Initializing the iteration step number k to be 0;
the second step is that: iterative computation
Wherein alpha is the step length of iteration and is taken as 0.001;
the third step: determining whether the error loss function converges
Δg(θ(k+1))＝|g(θ(k+1))-g(θ(k))|
If | Δ g (θ)(k+1))-Δg(θ(k)) If | is less than β, then θ is returned(k+1)，θ(k+1)I.e. the estimated model parameters, otherwise go back to the second step to continue the calculation, where β is a small value, and β may be 0.01 · α.
Step 407: and determining a click rate estimation model according to the value of each weight obtained by solving.
In the embodiment of the application, the values of the weights of the characteristics to the displayed advertisements are calculated and determined through the steps, so that the click rate estimation model of the displayed advertisements is established.
FIG. 5 is a schematic diagram of an application scenario of the display advertisement click-through rate estimation model. For a user I who has not shown the advertisement, the user I includes 3 features, I1、i2、i3(ii) a First collecting the characteristics of multiple users who display the advertisement, and thenClassifying the characteristics of the users, dividing each type of characteristics into a plurality of characteristic sets, and counting the normalized value of the characteristics in each characteristic set to the displayed advertisement, namely
Wherein i1、i2、i3The corresponding weights have values of θ1、θ2、θ3The said i1、i2、i3The values of the corresponding weights have been determined by the modeling process described above. The click rate pre-estimated value of the user I to the displayed advertisement is as follows: ctrA＝θ0+p1，Aθ1+p2，Aθ2+p3，Aθ3。
The click rate pre-estimation model of the displayed advertisement is established based on the weight of the characteristic of each characteristic set of the displayed advertisement to the standardized value of the displayed advertisement and the characteristic of each characteristic set, the click rate pre-estimation value of other displayed advertisements can be calculated by utilizing the click rate pre-estimation model, and then the advertisement with the high click rate pre-estimation value is selected and displayed to the user, so that the delivery accuracy of the displayed advertisement to the user can be greatly improved, the advertisement display with low click rate is avoided, the delivery cost of the displayed advertisement is saved, and the economic benefit of the displayed advertisement is improved.
Referring to fig. 6, a display method according to an embodiment of the present application includes steps 601 to 604.
Step 601: and obtaining a click rate estimated value of the event to be displayed.
In the embodiment of the application, according to the formula of the click rate estimation model in the formula 1, click rate estimation values of different events to be displayed of a user are calculated.
Step 602: and calculating expected income obtained by clicking the event to be shown based on the click rate estimated value.
In the embodiment of the application, the normalized value of the event to be shown and the weight of the feature of each feature set are substituted into the click rate estimation model to calculate to obtain the click rate estimated value of the event to be shown, and the expected yield of the event to be shown can be obtained by multiplying the click rate estimated value by the preset click unit price.
Step 603: the expected revenue of a plurality of events to be presented is ranked.
In the embodiment of the application, after the expected income of the events to be shown is calculated, the events to be shown can be sequenced from high to low in sequence according to the expected income, and the events to be shown can also be sequenced according to a specific sequence.
Step 604: and displaying the event to be displayed with the expected income in the preset area to the user.
For example, a preset resource library has a set of all events to be shown as a, and the above steps can predict the probability P that the user u clicks any one event b to be shown in the set au(xb) Let xbClick unit price of pricebThe value is provided by the provider of the event to be shown, then the expected profit of the event to be shown b to the user u is earnb＝ctrj·pricebAnd performing descending order on the set A according to the value, and displaying the events to be displayed with expected income in a preset area to the user, namely intercepting the events to be displayed of the top100 and displaying the events to be displayed to the user u.
In the embodiment of the application, the click rate pre-evaluation value of the event to be displayed is calculated based on the click rate pre-evaluation model, the event to be displayed is sequenced according to the click rate pre-evaluation value, the event to be displayed with high expected income is placed at a high-quality position and is preferentially displayed to a user, and the profit maximization can be realized.
The click rate estimation model provided by the application is applied to music display and comprises a first step to a fourth step.
The first step is as follows: and obtaining a click rate estimated value of the music to be displayed.
In the embodiment of the application, according to the formula of the click rate estimation model in formula 1, click rate estimation values of different music to be displayed of a user are calculated.
The second step is that: and calculating expected income obtained by clicking the music to be displayed based on the click rate pre-estimated value.
In the embodiment of the application, the standard value of the displayed music and the weight of the feature of each feature set are substituted into the click rate estimation model to calculate to obtain the click rate estimated value of the music to be displayed, and the expected income of the music to be displayed can be obtained by multiplying the click rate estimated value by the preset click unit price of the music to be displayed.
The third step: the expected revenue of a plurality of music to be presented is ranked.
In the embodiment of the application, after the expected profits of a plurality of pieces of music to be displayed are all calculated, the pieces of music to be displayed are sequentially ranked from high to low according to the expected profits.
The fourth step: and displaying the music to be displayed with the expected income in the preset area to the user.
For example: the music to be shown with the expected profit of top80 can be placed in the preset area, the music to be shown with the expected profit ranking of top80 is placed in the preset area according to the expected profit, and then 80 pieces of music to be shown in the preset area are shown to the user.
In the embodiment of the application, the click rate pre-evaluation value of the music to be displayed is calculated based on the click rate pre-evaluation model, the music to be displayed is sequenced according to the click rate pre-evaluation value, the music to be displayed with high expected income is placed at a high-quality position and is preferentially displayed for a user, in the process of displaying the music, the maximization of profit in limited resources is realized, and the use effect of the user is also improved to the greatest extent.
The method for establishing the click rate estimation model comprises the following steps: firstly, collecting sample data, extracting features, then carrying out ctr-based standardization on the features, finally establishing a ctr pre-estimation linear model of the standardized features, and solving model parameters through a gradient descent method. And a second stage: and calculating the click probability of the user to different display events according to the model established in the first stage, and displaying the events with high click probability to the user.
The click-through rate estimation model established by the application is applied to commercial application promotion (equivalent to a display event of displaying advertisements), and the commercial application promotion is a main income source of an application store. In the commercial application promotion process, the main goal is how to realize profit maximization in limited resources (including user resources and application showing resources). How to realize profit maximization is a core problem that the click probability of a user to a displayed business application (namely, a display event such as a display advertisement) needs to be predicted, the expected profit of the business application can be obtained by calculating the click probability and multiplying the click probability by the unit price of the business application, and the goal of profit maximization can be realized by placing the business application with high profit at a high-quality position for preferential display.
Fig. 7 is a schematic structural diagram of an apparatus for building a click rate prediction model according to an embodiment of the present application. Because the apparatus embodiments are substantially similar to the method embodiments, reference may be made to some of the descriptions of the method embodiments for related points. The device embodiments described below are merely illustrative.
The device for establishing the click rate pre-estimation model comprises the following steps:
an obtaining module 701, configured to establish an error function based on the obtained type of each display event and the click rate pre-estimated value of the display event, where a value of the type of the display event is determined according to an actual click condition of a user, and the click rate pre-estimated value of the display event is calculated based on a built click rate pre-estimated model;
an establishing module 702, configured to establish an error loss function of the click rate estimation model based on the error function;
a solving module 703, configured to solve a value of the weight of the feature of each feature set in the click rate estimation model based on the error loss function and the value of the type of each display event;
and a determining module 704, configured to determine a click rate prediction model according to the solved value of each weight.
In an embodiment of the application, the click rate estimation model is constructed by the following modules:
the acquisition module is used for acquiring the characteristics of the user who displays the event;
the statistical module is used for classifying the characteristics of the user, dividing each type of characteristics into a plurality of characteristic sets, counting the standardized value of the characteristics in each characteristic set to each display event, and setting weight for the characteristics of each characteristic set;
and the construction module is used for constructing a click rate pre-estimation model of each display event based on the normalized value of the features of each feature set to each display event and the weight of the features of each feature set.
In an embodiment of the present application, the apparatus further includes:
and the first calculation module is used for calculating the click rate estimated value of the event to be displayed based on the click rate estimation model.
In an embodiment of the present application, the first computing module further includes:
the second calculation module is used for calculating the expected income of the event to be displayed based on the calculated click rate pre-estimated value of the event to be displayed;
the sequencing module is used for sequencing the expected income of a plurality of events to be displayed;
and the display module is used for displaying the event to be displayed with the expected income in the preset area to the user.
In an embodiment of the application, the first calculation module is specifically configured to substitute the feature of the feature set into the click rate estimation model to calculate the normalized value of the event to be displayed and the weight of the feature of each feature set, so as to obtain the click rate estimated value of the event to be displayed.
In an embodiment of the present application, the solving module 703 includes:
the initialization module is used for setting an initial value for each weight;
and the iterative calculation module is used for performing iterative calculation on the error loss function by taking the minimum loss of the error loss function as a target, stopping the iterative calculation when the change rate of the error loss function is smaller than a preset threshold value, and taking the value of each weight at the moment as the value of the weight.
In an embodiment of the present application, the click rate estimation model is:
the error loss function is:
where j represents a presentation event, i represents a feature, θiRepresents the weight of the feature i, n represents the total number of features, m represents the total number of presentation events, flagjRepresenting the type of the j display event, 0 being a negative display event, 1 being a positive display event, if the j display event is clicked by the user, the j display event is a positive display event, if the j display event is not clicked by the user, the j display event is a negative display event, and pi，jAnd expressing the normalized value of the ith feature at the jth display event, wherein the calculation formula is as follows:
wherein f isi，jThe value, click (f), representing the ith feature at the jth presentation eventi，j) Representing a feature fi，jNumber of clicks, show (f)i，j) Representing a feature fi，jDisplay number of (d), pctr (f)i，j) Representing a feature fi，jThe click rate of (c).
The device for establishing the click rate estimation model provided by the embodiment of the application is characterized in that the click behaviors of a plurality of users for displaying events are taken as the basis, the features are further divided into a plurality of feature sets, the standardized value of the features of each feature set is counted, the weight is configured for the features in each feature set, and the click rate estimation model is established by taking the weight of each feature as a parameter. After each weight is determined, the click rate estimation model is determined. The click rate estimated value of any display event can be calculated by using the determined click rate estimated model, so that the event with high click rate estimated value can be displayed to the user, and the accuracy of displaying the event to the user is improved.
The present application further provides a terminal, including: a processor and a memory storing computer instructions;
the processor reads the computer instructions and executes a method for establishing a click rate prediction model as described above.
The present application further provides a storage medium storing computer instructions that, when executed, implement the method for establishing a click-through rate estimation model as described above.
Fig. 8 is a schematic structural diagram of a display device according to an embodiment of the present application. Because the apparatus embodiments are substantially similar to the method embodiments, reference may be made to some of the descriptions of the method embodiments for related points. The device embodiments described below are merely illustrative.
The application provides a display device includes:
a pre-estimated value obtaining module 801, configured to obtain a pre-estimated value of a click rate of an event to be displayed;
a profit calculation module 802, configured to calculate, based on the click-through rate pre-estimated value, an expected profit obtained by clicking on the event to be shown;
a revenue ranking module 803, configured to rank the expected revenue of the multiple events to be displayed;
and the event display module 804 is configured to display the event to be displayed, in which the expected revenue is located in the preset area, to the user.
In an embodiment of the present application, the profit calculating module 802 is specifically configured to multiply the click rate estimated value by a preset click unit price to obtain an expected profit.
In the embodiment of the application, the click rate pre-evaluation value of the event to be displayed is calculated based on the click rate pre-evaluation model, the event to be displayed is sequenced according to the click rate pre-evaluation value, the display event with the highest click rate pre-evaluation value is displayed for a user, the click viewing interest of the user on the display event can be greatly improved, and the user experience is improved.
The present application further provides a terminal, including: a processor and a memory storing computer instructions;
the processor reads the computer instructions and executes a presentation method as described above.
The present application also provides a storage medium storing computer instructions that, when executed, implement the presentation method as described above.
It should be noted that, the click rate estimation model building device, if implemented in the form of a software functional unit and sold or used as an independent product, may be stored in a computer readable storage medium. Based on such understanding, all or part of the processes in the methods of the embodiments described above may be implemented by a computer program, which may be stored in a computer-readable storage medium and used by a computer to implement the steps of the embodiments of the methods described above. Wherein the computer program comprises computer program code, which may be in the form of source code, presentation event code, an executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, usb disk, removable hard disk, magnetic disk, optical disk, computer Memory, Read-Only Memory (ROM), Random Access Memory (RAM), electrical carrier wave signals, telecommunications signals, software distribution medium, and the like. It should be noted that the computer readable medium may contain content that is subject to appropriate increase or decrease as required by legislation and patent practice in jurisdictions, for example, in some jurisdictions, computer readable media does not include electrical carrier signals and telecommunications signals as is required by legislation and patent practice.
It should be noted that, for the sake of simplicity, the above-mentioned method embodiments are described as a series of acts or combinations, but those skilled in the art should understand that the present application is not limited by the described order of acts, as some steps may be performed in other orders or simultaneously according to the present application. Further, those skilled in the art should also appreciate that the embodiments described in the specification are preferred embodiments and that the acts and modules referred to are not necessarily required in this application.
In the above embodiments, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments.
The preferred embodiments of the present application disclosed above are intended only to aid in the explanation of the application. Alternative embodiments are not exhaustive and do not limit the application to the precise embodiments described. Obviously, many modifications and variations are possible in light of the above teaching. The embodiments were chosen and described in order to best explain the principles of the application and the practical application, to thereby enable others skilled in the art to best understand and utilize the application. The application is limited only by the claims and their full scope and equivalents.
The preferred embodiments and examples of the present application have been described in detail with reference to the accompanying drawings, but the present application is not limited to the embodiments and examples described above, and various changes can be made within the knowledge of those skilled in the art without departing from the concept of the present application.
Claims (22)
1. A method for establishing a click-through rate prediction model is characterized by comprising the following steps:
establishing an error function based on the type of each display event and the click rate pre-estimated value of the display event, wherein the value of the type of the display event is determined according to the actual click condition of a user, and the click rate pre-estimated value of the display event is calculated based on a built click rate pre-estimated model; the click rate pre-estimation model comprises a normalized value of the feature of each feature set to the display event and a weight of the feature of each feature set, and the normalized value of the feature of each feature set to the display event in the click rate pre-estimation model is a known value; the normalized value of the feature pair display event of each feature set is obtained by calculating the actual number of clicks of the feature pair display event of each feature set and the actual number of displays of the event to the feature pair display event of each feature set in advance;
establishing an error loss function of the click rate estimation model based on the error function;
solving the weight value of the feature of each feature set in the click rate estimation model based on the error loss function and the value of the type of each display event;
and determining a click rate estimation model according to the value of each weight obtained by solving.
2. The method of claim 1, wherein the click through rate prediction model is constructed by:
collecting characteristics of a user who shows an event;
classifying the characteristics of the user, dividing each type of characteristics into a plurality of characteristic sets, counting the standardized value of the characteristics in each characteristic set to each display event, and setting weight for the characteristics of each characteristic set;
and constructing a click rate pre-estimation model of each display event based on the normalized value of the features of each feature set to each display event and the weight of the features of each feature set.
3. The method of claim 1, further comprising:
and calculating a click rate pre-estimated value of the event to be displayed based on the click rate pre-estimated model.
4. The method of claim 3, wherein after calculating the click-through rate estimate for the event to be presented based on the click-through rate estimate model, the method further comprises:
calculating the expected income of the event to be displayed based on the calculated click rate estimated value of the event to be displayed;
ordering expected profits of a plurality of events to be displayed;
and displaying the event to be displayed with the expected income in the preset area to the user.
5. The method of claim 3, wherein said calculating a click-through rate estimate for the event to be presented based on the click-through rate estimation model comprises:
and substituting the normalized values of the events to be displayed and the weight of the features of each feature set into the click rate estimation model by the features of the feature sets to calculate to obtain the click rate estimated value of the events to be displayed.
6. The method of claim 1, wherein solving the value of each weight based on the error loss function and the value of the type of each presentation event comprises:
setting an initial value for each weight;
performing iterative computation on the error loss function by taking the loss minimum of the error loss function as a target;
and stopping the iterative computation when the change rate of the error loss function is smaller than a preset threshold value, and taking the value of each weight at the moment as the value of the weight.
7. The method of claim 6, wherein the click-through rate prediction model is:
the error loss function is:
where j represents a presentation event, i represents a feature, θ0Represents a constant value, θiRepresents the weight of the feature i, n represents the total number of features, m represents the total number of presentation events, flagjRepresenting the j type of the display event, 0 being a negative display event, 1 being a positive display event, if the j isIf the display event is clicked by the user, the display event is a positive display event, and if the display event is not clicked by the user, the display event is a negative display event, pi,jAnd expressing the normalized value of the ith feature at the jth display event, wherein the calculation formula is as follows:
wherein f isi,jThe value, click (f), representing the ith feature at the jth presentation eventi,j) Denotes fi,jNumber of clicks, show (f)i,j) Denotes fi,jDisplay number of (d), pctr (f)i,j) Denotes fi,jThe click rate of (c).
8. A method of displaying, comprising:
obtaining a click rate pre-estimated value of an event to be displayed, wherein the click rate pre-estimated value is calculated based on a click rate pre-estimated model, and the click rate pre-estimated model is determined according to the method of any one of claims 1-7;
calculating expected income obtained by clicking the event to be displayed based on the click rate pre-estimated value;
ordering expected profits of a plurality of events to be displayed;
and displaying the event to be displayed with the expected income in the preset area to the user.
9. The method of claim 8, wherein said calculating an expected revenue from clicking on the event to be presented based on the click-through rate estimate comprises:
and multiplying the click rate estimated value by a preset click unit price to obtain expected benefits.
10. An apparatus for building a click-through rate prediction model, comprising:
the acquisition module is used for establishing an error function based on the type of each display event and the click rate pre-estimated value of the display event, wherein the click rate pre-estimated value of the display event is obtained by calculation based on a constructed click rate pre-estimated model, and the value of the type of the display event is determined according to the actual click condition of a user; the click rate pre-estimation model comprises a normalized value of the feature of each feature set to the display event and a weight of the feature of each feature set, and the normalized value of the feature of each feature set to the display event in the click rate pre-estimation model is a known value; the normalized value of the feature pair display event in each feature set is obtained by calculating the actual number of clicks of the feature pair display event of each feature set and the actual number of displays of the event to the feature pair display event of each feature set in advance;
the establishing module is used for establishing an error loss function of the click rate estimation model based on the error function;
the solving module is used for solving the value of the weight of the feature of each feature set in the click rate estimation model based on the error loss function and the value of the type of each display event;
and the determining module is used for determining the click rate estimation model according to the value of each weight obtained by solving.
11. The apparatus of claim 10, wherein the click-through rate prediction model is constructed by:
the acquisition module is used for acquiring the characteristics of the user who displays the event;
the statistical module is used for classifying the characteristics of the user, dividing each type of characteristics into a plurality of characteristic sets, counting the standardized value of the characteristics in each characteristic set to each display event, and setting weight for the characteristics of each characteristic set;
and the construction module is used for constructing a click rate pre-estimation model of each display event based on the normalized value of the features of each feature set to each display event and the weight of the features of each feature set.
12. The apparatus of claim 10, further comprising:
and the first calculation module is used for calculating the click rate estimated value of the event to be displayed based on the click rate estimation model.
13. The apparatus of claim 12, wherein the first computing module further comprises:
the second calculation module is used for calculating the expected income of the event to be displayed based on the calculated click rate pre-estimated value of the event to be displayed;
the sequencing module is used for sequencing the expected income of a plurality of events to be displayed;
and the display module is used for displaying the event to be displayed with the expected income in the preset area to the user.
14. The apparatus according to claim 12, wherein the first calculating module is specifically configured to substitute the normalized value of the event to be shown and the weight of the feature of each feature set into the click-through rate estimation model to calculate the click-through rate estimated value of the event to be shown.
15. The apparatus of claim 10, wherein the solving module comprises:
the initialization module is used for setting an initial value for each weight;
and the iterative calculation module is used for performing iterative calculation on the error loss function by taking the minimum loss of the error loss function as a target, stopping the iterative calculation when the change rate of the error loss function is smaller than a preset threshold value, and taking the value of each weight at the moment as the value of the weight.
16. The apparatus of claim 15, wherein the click-through rate prediction model is:
the error loss function is:
where j represents a presentation event, i represents a feature, θ0Represents a constant value, θiRepresents the weight of the feature i, n represents the total number of features, m represents the total number of presentation events, flagjRepresenting the type of the j display event, 0 being a negative display event, 1 being a positive display event, if the j display event is clicked by the user, the j display event is a positive display event, if the j display event is not clicked by the user, the j display event is a negative display event, and pi,jAnd expressing the normalized value of the ith feature at the jth display event, wherein the calculation formula is as follows:
wherein f isi,jThe value, click (f), representing the ith feature at the jth presentation eventi,j) Denotes fi,jNumber of clicks, show (f)i,j) Denotes fi,jDisplay number of (d), pctr (f)i,j) Denotes fi,jThe click rate of (c).
17. A display device, comprising:
the system comprises a pre-evaluation value acquisition module, a display module and a display module, wherein the pre-evaluation value of the click rate of an event to be displayed is acquired by acquiring the pre-evaluation value of the click rate of the event to be displayed, wherein the pre-evaluation value of the click rate is calculated based on a click rate pre-evaluation model, and the click rate pre-evaluation model is determined according to the method of any one of claims 1-7;
the profit calculation module is used for calculating expected profits obtained by clicking the events to be displayed based on the click rate pre-estimated value;
the income sorting module is used for sorting the expected income of a plurality of events to be displayed;
and the event display module is used for displaying the event to be displayed with the expected income in the preset area to the user.
18. The apparatus of claim 17, wherein the profit computation module is specifically configured to multiply the click rate estimate by a predetermined click unit price to obtain an expected profit.
19. A terminal comprising a processor and a memory storing computer instructions;
the processor reads the computer instructions and executes a method of establishing a click through rate prediction model according to any one of claims 1-7.
20. A terminal comprising a processor and a memory storing computer instructions;
the processor reads the computer instructions and performs a presentation method as claimed in claim 8 or 9.
21. A storage medium having stored thereon computer instructions which, when executed, implement a method of creating a click through rate prediction model according to any one of claims 1-7.
22. A storage medium storing computer instructions which, when executed, implement a presentation method as claimed in claim 8 or 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710729982.1A CN107526810B (en) | 2017-08-23 | 2017-08-23 | Method and device for establishing click rate estimation model and display method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710729982.1A CN107526810B (en) | 2017-08-23 | 2017-08-23 | Method and device for establishing click rate estimation model and display method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107526810A CN107526810A (en) | 2017-12-29 |
| CN107526810B true CN107526810B (en) | 2021-01-26 |
Family
ID=60682193
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710729982.1A Active CN107526810B (en) | 2017-08-23 | 2017-08-23 | Method and device for establishing click rate estimation model and display method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107526810B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108427708B (en) * | 2018-01-25 | 2021-06-25 | 腾讯科技（深圳）有限公司 | Data processing method, data processing apparatus, storage medium, and electronic apparatus |
| CN110401855B (en) * | 2018-04-25 | 2022-03-04 | 腾讯科技（深圳）有限公司 | Information display method and device, processing platform, computing equipment and storage medium |
| CN110263959B (en) * | 2018-05-21 | 2024-01-09 | 腾讯科技（深圳）有限公司 | Click rate estimation method, click rate estimation device, machine equipment and computer readable storage medium |
| CN109389424B (en) * | 2018-09-20 | 2021-08-06 | 北京达佳互联信息技术有限公司 | Flow distribution method and device, electronic equipment and storage medium |
| CN111126649B (en) * | 2018-10-31 | 2023-08-11 | 北京字节跳动网络技术有限公司 | Method and device for generating information |
| CN112150182B (en) * | 2019-06-28 | 2023-08-29 | 腾讯科技（深圳）有限公司 | Multimedia file pushing method and device, storage medium and electronic device |
| CN112214665A (en) * | 2019-07-12 | 2021-01-12 | 北京字节跳动网络技术有限公司 | Content display method and device, electronic equipment and computer readable storage medium |
| CN112612951B (en) * | 2020-12-17 | 2022-07-01 | 上海交通大学 | Unbiased learning sorting method for income improvement |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104951965A (en) * | 2015-06-26 | 2015-09-30 | 深圳市腾讯计算机***有限公司 | Advertisement delivery method and device |
| CN105809487A (en) * | 2016-03-17 | 2016-07-27 | 广州筷子信息科技有限公司 | Advertisement element based advertisement release method and apparatus |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8732175B2 (en) * | 2005-04-21 | 2014-05-20 | Yahoo! Inc. | Interestingness ranking of media objects |
-
2017
- 2017-08-23 CN CN201710729982.1A patent/CN107526810B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104951965A (en) * | 2015-06-26 | 2015-09-30 | 深圳市腾讯计算机***有限公司 | Advertisement delivery method and device |
| CN105809487A (en) * | 2016-03-17 | 2016-07-27 | 广州筷子信息科技有限公司 | Advertisement element based advertisement release method and apparatus |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107526810A (en) | 2017-12-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107526810B (en) | Method and device for establishing click rate estimation model and display method and device | |
| US10405016B2 (en) | Recommending media items based on take rate signals | |
| JP6067713B2 (en) | Understand the effectiveness of communications propagated through social networking systems | |
| CN107301247B (en) | Method and device for establishing click rate estimation model, terminal and storage medium | |
| US9348898B2 (en) | Recommendation system with dual collaborative filter usage matrix | |
| US10032176B2 (en) | Real time statistics extraction from arbitrary advertising audiences | |
| JP6261547B2 (en) | Determination device, determination method, and determination program | |
| CN108805598B (en) | Similarity information determination method, server and computer-readable storage medium | |
| US8725559B1 (en) | Attribute based advertisement categorization | |
| US20130268479A1 (en) | System and method for presenting and managing social media | |
| CN105095431A (en) | Method and device for pushing videos based on behavior information of user | |
| CN109447688A (en) | Advertisement placement method, device and storage equipment, computer equipment | |
| CN106202393B (en) | Media information pushing method and device | |
| US8359238B1 (en) | Grouping user features based on performance measures | |
| US20210241314A1 (en) | Method and system for optimizing user grouping for advertisement | |
| CN108629608A (en) | User data processing method and processing device | |
| US20110251889A1 (en) | Inventory clustering | |
| US20140068515A1 (en) | System and method for classifying media | |
| CN110689402A (en) | Method and device for recommending merchants, electronic equipment and readable storage medium | |
| US10304081B1 (en) | Yielding content recommendations based on serving by probabilistic grade proportions | |
| CN111798280A (en) | Multimedia information recommendation method, device and equipment and storage medium | |
| CN111861605A (en) | Business object recommendation method | |
| Guan et al. | When images backfire: The effect of customer-generated images on product rating dynamics | |
| JPWO2020031232A1 (en) | Information processing system, information processing method, and program | |
| CN110796465B (en) | Resource display equipment determining method, advertisement display screen determining method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20200526 Address after: 310051 room 508, floor 5, building 4, No. 699, Wangshang Road, Changhe street, Binjiang District, Hangzhou City, Zhejiang Province Applicant after: Alibaba (China) Co.,Ltd. Address before: 510627 unit 02, floor 15, Tower B, Pingyun Plaza, radio and television, No. 163, xipingyun Road, Huangpu Avenue, Tianhe District, Guangzhou City, Guangdong Province (for office use only) Applicant before: GUANGZHOU UC NETWORK TECHNOLOGY Co.,Ltd. |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |