CN110223108B - Click through rate prediction method, device and equipment - Google Patents
Click through rate prediction method, device and equipment Download PDFInfo
- Publication number
- CN110223108B CN110223108B CN201910444789.2A CN201910444789A CN110223108B CN 110223108 B CN110223108 B CN 110223108B CN 201910444789 A CN201910444789 A CN 201910444789A CN 110223108 B CN110223108 B CN 110223108B
- Authority
- CN
- China
- Prior art keywords
- feature data
- characteristic data
- data
- prediction model
- preset prediction
- 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
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/22—Indexing; Data structures therefor; Storage structures
-
- 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
Abstract
The method, the device and the equipment for predicting the click through rate provided by the embodiment of the invention are used for acquiring at least one characteristic data of the network advertisement to be delivered; aiming at each acquired feature data, searching a preset prediction model corresponding to the feature data from the corresponding relation between the preset feature data and the preset prediction model; the preset prediction model is a model obtained by utilizing sample characteristic data and a prediction result label of the sample characteristic data in advance for training, and the sample characteristic data and the characteristic data corresponding to the preset prediction model have the same type; and inputting the characteristic data into a corresponding preset prediction model aiming at each acquired characteristic data to obtain the predicted click through rate corresponding to the characteristic data. The method and the device can determine the characteristic data of the to-be-delivered network advertisement which influences the acquired predicted click through rate, so that the characteristics reflected by the characteristic data can be adjusted in a targeted manner, and convenience and efficiency of characteristic adjustment are improved.
Description
Technical Field
The invention relates to the technical field of network advertisements, in particular to a method, a device and equipment for predicting click through rate.
Background
In the network advertisement delivery, the CTR (Click-Through-Rate), i.e. the ratio of the actual number of clicks of the user on the network advertisement to the advertisement display amount, is an important index for measuring the advertisement delivery effect. In order to reasonably deliver the network advertisements, the click through rate of the network advertisements to be delivered can be predicted. After the predicted click through rate is obtained, adjusting the characteristics of the network advertisement to be delivered, such as delivery time, delivery position, user information and the like, according to the predicted click through rate, and improving the click through rate of the network advertisement through the adjusted characteristics, thereby optimizing the delivery effect of the network advertisement. For example, in the feature of the to-be-delivered web advertisement, the user information is a female aged 20 to 30 years, at this time, the predicted click through rate is relatively low, and after the user information is adjusted to a female aged 30 to 40 years, the predicted click through rate is improved.
In the related art, a preset click through rate prediction model is obtained by training in advance by using sample characteristic data and a prediction result label of the sample characteristic data. The characteristics reflected by the sample characteristic data are the same as those of the network advertisement to be predicted, so that the predicted click through rate output by the model can be obtained after the characteristic data of the network advertisement are input into the preset click through rate prediction model.
However, the basis for predicting the click through rate in the method is the characteristics of the network advertisement to be delivered, the characteristics of the network advertisement are various, and the obtained predicted click through rate is a prediction result obtained by integrating the overall influence of various characteristic data. According to the predicted click through rate obtained by integrating the overall influence of various feature data, which feature or features cause the click through rate to be relatively low cannot be indicated. Therefore, when adjusting the characteristics of the network advertisements to be delivered, operation and maintenance personnel are required to tentatively adjust the characteristics of the network advertisements to be delivered one by one, which results in that the characteristics of the network advertisements to be delivered are adjusted more complicatedly and the efficiency is lower.
Disclosure of Invention
The embodiment of the invention aims to provide a method, a device and equipment for predicting click through rate, so as to realize the effect of indicating the characteristics influencing the predicted click through rate in a network advertisement to be delivered and improving the convenience and efficiency of characteristic adjustment. The specific technical scheme is as follows:
in a first aspect, an embodiment of the present invention provides a method for predicting a click through rate, where the method includes:
acquiring at least one characteristic data of a network advertisement to be delivered; the type of the feature data is divided according to the features of the network advertisements reflected by the feature data;
aiming at each acquired feature data, searching a preset prediction model corresponding to the feature data from the corresponding relation between the preset feature data and the preset prediction model; the preset prediction model is obtained by utilizing sample characteristic data and a prediction result label of the sample characteristic data in advance for training, and the sample characteristic data is the same as the type of characteristic data corresponding to the preset prediction model;
and inputting the characteristic data into a corresponding preset prediction model aiming at each acquired characteristic data to obtain the predicted click through rate corresponding to the characteristic data.
Optionally, the obtaining at least one feature data of the network advertisement to be delivered includes:
acquiring the type of characteristic data of a network advertisement to be delivered and a plurality of characteristic data of the network advertisement to be delivered;
determining a construction rule of each kind of feature data according to the type of the acquired feature data;
constructing each kind of feature data by using the plurality of feature data according to the construction rule of each kind of feature data;
and taking the constructed at least one characteristic data as at least one characteristic data of the network advertisement to be delivered.
Optionally, after the feature data is constructed by using the plurality of feature data according to the construction rule of each kind of feature data, the method further includes:
inputting at least one constructed characteristic data into a preset hash algorithm to obtain a hash value corresponding to each characteristic data;
and using the obtained hash value as at least one characteristic data of the network advertisement to be delivered.
Optionally, each preset prediction model corresponds to one thread; the thread is used for loading a preset prediction model corresponding to the thread;
the step of inputting the acquired feature data into a corresponding preset prediction model for each type of feature data to obtain the predicted click through rate corresponding to the feature data includes:
detecting whether the thread corresponding to the characteristic data finishes running or not aiming at each acquired characteristic data;
if the operation is finished, the thread corresponding to the characteristic data is utilized, the preset prediction model corresponding to the characteristic data is loaded, the characteristic data is input into the corresponding preset prediction model, the predicted click through rate corresponding to the characteristic data is obtained, and otherwise, the detection of whether the operation of the thread corresponding to the characteristic data is finished or not is executed.
Optionally, the inputting, to each obtained feature data, the feature data into a corresponding preset prediction model to obtain a predicted click through rate corresponding to the feature data includes:
acquiring a second identifier of a preset prediction model corresponding to each acquired feature data;
and inputting the second identifier corresponding to the characteristic data into a preset prediction model interface function to obtain the predicted click through rate corresponding to the characteristic data.
In a second aspect, an embodiment of the present invention provides an apparatus for predicting a click through rate, where the apparatus includes:
the characteristic data acquisition module is used for acquiring at least one characteristic data of the network advertisement to be delivered; the type of the feature data is divided according to the features of the network advertisements reflected by the feature data;
the prediction model acquisition module is used for searching a preset prediction model corresponding to each acquired feature data from the corresponding relation between the preset feature data and the preset prediction model; the preset prediction model is obtained by utilizing sample characteristic data and a prediction result label of the sample characteristic data in advance for training, and the sample characteristic data is the same as the type of characteristic data corresponding to the preset prediction model;
and the prediction module is used for inputting the characteristic data into a corresponding preset prediction model aiming at each acquired characteristic data to obtain the predicted click through rate corresponding to the characteristic data.
Optionally, the characteristic data obtaining module is specifically configured to:
acquiring the type of characteristic data of a network advertisement to be delivered and a plurality of characteristic data of the network advertisement to be delivered;
determining a construction rule of each kind of feature data according to the type of the acquired feature data;
constructing each kind of feature data by using the plurality of feature data according to the construction rule of each kind of feature data;
and taking the constructed at least one characteristic data as at least one characteristic data of the network advertisement to be delivered.
Optionally, the feature data obtaining module is further configured to, after the feature data is constructed according to the construction rule of each feature data by using the plurality of feature data, input at least one constructed feature data into a preset hash algorithm, so as to obtain a hash value corresponding to each feature data;
and using the obtained hash value as at least one characteristic data of the network advertisement to be delivered.
Optionally, each preset prediction model corresponds to one thread; the thread is used for loading a preset prediction model corresponding to the thread;
the prediction module comprises a judgment submodule and a loading submodule;
the judging submodule is used for detecting whether the thread corresponding to the characteristic data finishes running or not according to the acquired characteristic data; if the operation is finished, triggering the loading sub-module to operate;
and the loading submodule is used for loading a preset prediction model corresponding to the characteristic data by using the thread corresponding to the characteristic data under the trigger of the judging submodule, inputting the characteristic data into the corresponding preset prediction model to obtain the predicted click through rate corresponding to the characteristic data, and otherwise, executing the detection of whether the thread corresponding to the characteristic data is finished running.
Optionally, the prediction module is specifically configured to:
acquiring a second identifier of a preset prediction model corresponding to each acquired feature data;
and inputting the second identifier corresponding to the characteristic data into a preset prediction model interface function to obtain the predicted click through rate corresponding to the characteristic data.
In a third aspect, an embodiment of the present invention provides an electronic device, including:
the system comprises a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory complete mutual communication through the bus; a memory for storing a computer program; and a processor, configured to execute the program stored in the memory, and implement the steps of the method for predicting the click through rate according to the first aspect.
In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, where a computer program is stored in the storage medium, and when the computer program is executed by a processor, the steps of the method for predicting a click through rate provided in the first aspect are implemented.
In the scheme provided by the embodiment of the invention, when the click through rate is predicted, the preset prediction model is a model obtained by utilizing sample characteristic data and a prediction result label of the sample characteristic data in advance, and the sample characteristic data is the same as the type of the characteristic data corresponding to the preset prediction model. Therefore, for each acquired feature data, after the feature data is input into a corresponding preset prediction model, the predicted click through rate corresponding to the feature data can be obtained. The predicted click through rate corresponding to the feature data is obtained aiming at different types of feature data of the network advertisements, and when the features of the network advertisements to be delivered are adjusted according to the predicted click through rate, the feature data of the network advertisements to be delivered, which affect the obtained predicted click through rate, can be determined, so that operation and maintenance personnel can adjust the features reflected by the feature data in a targeted manner. Compared with the method that operation and maintenance personnel need to tentatively adjust the characteristics of the network advertisements to be delivered one by one, the method saves the step of tentatively adjusting the characteristics which do not influence the predicted click through rate, and can improve the convenience and efficiency of the characteristic adjustment of the network advertisements to be delivered.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below.
Fig. 1 is a schematic flow chart illustrating a method for predicting click through rate according to an embodiment of the present invention;
FIG. 2 is a flowchart illustrating a method for predicting click through rate according to another embodiment of the present invention;
fig. 3 is a schematic structural diagram of a device for predicting click through rate according to an embodiment of the present invention;
FIG. 4 is a schematic structural diagram of a device for predicting click through rate according to another embodiment of the present invention;
fig. 5 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
In order to make those skilled in the art better understand the technical solutions in the embodiments of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
First, a method for predicting the click through rate according to an embodiment of the present invention will be described.
The method for predicting the click through rate provided by the embodiment of the present invention may be applied to an electronic device capable of performing data processing, where the device may include a desktop computer, a portable computer, an internet television, an intelligent mobile terminal, a wearable intelligent terminal, a server, and the like, and is not limited herein, and any electronic device that can implement the embodiment of the present invention belongs to the protection scope of the embodiment of the present invention.
In addition, since the placement of the network advertisement may be specifically performed on various applications related to the internet, in a specific application, the execution subject of the method for predicting the click through rate provided by the embodiment of the present invention may be a server of various applications. Illustratively, it may be a server of a live application, a social application, or a video application, among other applications.
As shown in fig. 1, a flow of a method for predicting a click through rate according to an embodiment of the present invention may include:
s101, acquiring at least one characteristic data of the network advertisement to be delivered; the types of the feature data are classified according to the features of the network advertisement reflected by the feature data.
In order to ensure that the predicted click through rate corresponding to a certain feature data obtained in step S103 can indicate a feature affecting the predicted click through rate, the feature data may be obtained according to the type of the feature data, and the type of the feature data needs to be divided according to the feature of the web advertisement reflected by the feature data. Since the feature division of the web advertisement may be various, the kinds of the feature data of the web advertisement may be various.
Illustratively, the category of the feature data may be divided by the source of the features of the web advertisement. For example, the feature data whose source is the user information is divided into the user information, and the category of the feature data such as the user gender, the user geographic location, the user age and the like is the user information; the characteristic data with the source of the advertisement information is divided into the advertisement information, and the types of the characteristic data such as the content form of the network advertisement, the putting mode of the network advertisement, the putting time of the network advertisement and the like are the advertisement information.
Or, for example, the category of the feature data may be divided according to the construction rule of the features of the network advertisement. For example, when the construction rule is "each feature data is a kind of feature data", each feature data may be divided into a kind of feature data, for example, the user age may be a first kind of feature data, and the user gender may be a second kind of feature data; when the construction rule is "the feature data obtained after splicing at least two specified feature data is used as one feature data", each spliced feature data can be divided into one feature data, for example, data obtained by splicing the age and gender of a user, data obtained by splicing the content form of the network advertisement, the delivery mode of the network advertisement and the delivery time of the network advertisement, and data obtained by splicing the content form of the network advertisement and the delivery time of the network advertisement can be used as the second feature data.
Any type of network advertisement can be used in the present invention, and this embodiment does not limit this. The form of the characteristic data may be varied in a particular application. Illustratively, the characteristic data may be the characteristics of the web advertisement itself, e.g., "image" and "female" and so on. Or, for example, the feature data may be feature values of features of the web advertisement, such as feature values of the feature "image" and feature values of the feature "female", and so on.
Corresponding to different kinds of division modes of the feature data, at least one kind of feature data of the network advertisement to be delivered can be obtained in various modes. Illustratively, when the types of the feature data are classified according to the sources of the features of the network advertisements, at least one type of feature data of the network advertisements to be delivered can be directly acquired. For example, at least one of advertisement information and user information is extracted from placement setting information of a network advertisement to be placed. Or, exemplarily, when the types of the feature data are divided according to the construction rules of the features of the network advertisement, at least one feature data and data type of the network advertisement to be delivered may be obtained, and then the corresponding construction rules are determined according to the data types, and according to the construction rules, different types of feature data are constructed and obtained by using the at least one feature data. For ease of understanding and reasonable layout, the following description specifically describes the case of dividing the types of feature data according to the construction rules in an alternative embodiment.
Any method for acquiring at least one feature data of the network advertisement to be delivered can be used in the present invention, and this embodiment does not limit this.
S102, aiming at each acquired feature data, searching a preset prediction model corresponding to the feature data from the corresponding relation between the preset feature data and the preset prediction model. The preset prediction model is obtained by training sample characteristic data and a prediction result label of the sample characteristic data in advance, and the sample characteristic data and the characteristic data corresponding to the preset prediction model are the same in type.
The preset feature data may correspond to a preset prediction model in various ways. For example, the correspondence relationship between the preset feature data and the preset prediction model may be a pointer indicating the preset prediction model to which the feature data corresponds. Or, for example, the preset feature data and the preset prediction model may be a correspondence table of the feature data and the preset prediction model, where a first identifier of the feature data and a second identifier of the preset prediction model are stored in the correspondence table.
In order to ensure that the predicted click through rate corresponding to certain feature data obtained in step S103 can indicate features affecting the predicted click through rate, the sample feature data used for obtaining the preset prediction model needs to be the same as the type of the feature data of the network advertisement to be delivered. For example, the acquired feature data includes first feature data of which the type is user information and second feature data of which the type is advertisement information, and then the type of the sample feature data used for training the preset prediction model for predicting the first feature data is user information, and the type of the sample feature data used for training the preset prediction model for predicting the second feature data is advertisement information.
And the sample feature data used for training is the same as the feature data corresponding to the preset prediction model in type, so that the preset prediction model obtained by training corresponds to the feature data in type one to one. Illustratively, the preset LR model M1 is used to calculate feature data reflecting advertisement information, and the preset LR model M2 is used to calculate feature data reflecting user information. Or, for example, the preset LR model M3 is used to calculate the first feature data constructed by the age and gender of the user, and the preset LR model M4 is used to calculate the second feature data constructed by the content format of the web advertisement, the delivery mode of the web advertisement, and the delivery time of the web advertisement. In addition, the predetermined predictive model may be various in specific applications. For example, the preset prediction model may be a preset LR model or a preset neural network model. Any prediction model which can be obtained by utilizing the sample feature data with the same type as the feature data of the network advertisement to be delivered and the prediction result label training of the sample feature data in advance can be used in the invention, and the embodiment does not limit the prediction model.
S103, inputting the acquired feature data into a corresponding preset prediction model according to each feature data to obtain a predicted click through rate corresponding to the feature data.
Illustratively, the acquired feature data includes feature data "gender and age of user", the category of the feature data is "category T1", the feature data "placement location and placement time of web advertisement", and the category of the feature data is "category T2". Inputting the first feature data into a preset prediction model M1 to obtain a predicted click through rate P1 corresponding to feature data 'user gender and user age'; and inputting the second feature data into a preset prediction model M2 to obtain a predicted click through rate P2 corresponding to feature data 'the placement position and the placement time of the network advertisement'. The type of the sample characteristic data used for training the preset prediction model M1 is 'type T1'; the type of the sample feature data used for training to obtain the preset prediction model M2 is "type T2".
In a specific application, for each acquired feature data, the specific manner of inputting the feature data into the corresponding preset prediction model may be various. Illustratively, the input may be serial input, specifically: selecting a preset prediction model corresponding to input characteristic data from the acquired various characteristic data to obtain a predicted click through rate corresponding to the characteristic data; and after the predicted click through rate corresponding to the feature data selected last time is obtained, selecting the next feature data from various feature data which are not obtained the predicted click through rate to obtain the click through rate, and sequentially circulating until the predicted click through rate corresponding to each obtained feature data is obtained. Or, for example, parallel input may be performed, at this time, each preset prediction model corresponds to one thread, the thread is used for loading the preset prediction model corresponding to the thread, different types of feature data may be simultaneously input by using a plurality of threads, the predicted click through rate of each type of feature data is obtained independently, it is not necessary to wait for obtaining the predicted click through rate corresponding to one type of feature data, and then obtain the predicted click through rate corresponding to another type of feature data, so that the obtaining efficiency of the predicted click through rate may be effectively improved.
In addition, when the types of the feature data are multiple, and multiple predicted click through rates are obtained, in order to facilitate the maintenance personnel to determine the feature data corresponding to each obtained predicted click through rate, a first identifier capable of indicating the corresponding feature data may be added to the obtained predicted click through rate. Alternatively, the obtained predicted click through rate may be stored in a storage location corresponding to the feature data. For example, the predicted click through rate and the feature data may be stored in a storage table in which each kind of feature data is stored in advance; the obtained predicted click through rate may be stored in a designated storage location corresponding to the feature data, and for example, the predicted click through rate P1 corresponding to the feature data of the type "type T1" may be stored in the designated storage location S1, and the predicted click through rate P2 corresponding to the feature data of the type "type T2" may be stored in the designated storage location S2.
In the scheme provided by the embodiment of the invention, when the click through rate is predicted, the preset prediction model is a model obtained by utilizing sample characteristic data and a prediction result label of the sample characteristic data in advance for training, the type of the sample characteristic data is the same as that of the characteristic data of the network advertisement to be delivered, and the type of the characteristic data is divided according to the characteristics of the network advertisement reflected by the characteristic data. Therefore, for each type of acquired feature data, after the feature data is input into a corresponding preset prediction model, the predicted click through rates corresponding to the feature data can be obtained. The predicted click through rate corresponding to the characteristic data is obtained aiming at different types of characteristic data of the network advertisement, and when the characteristic of the network advertisement to be delivered is adjusted according to the predicted click through rate, the characteristic data of the network advertisement to be delivered, which influences the obtained predicted click through rate, can be determined, so that the operation and maintenance personnel can adjust the characteristic reflected by the characteristic data in a targeted manner. Compared with the method that operation and maintenance personnel need to tentatively adjust the characteristics of the network advertisements to be delivered one by one, the method saves the step of tentatively adjusting the characteristics which do not influence the predicted click through rate, and can improve the convenience and efficiency of the characteristic adjustment of the network advertisements to be delivered.
Optionally, the obtaining of at least one feature data of the network advertisement to be delivered may specifically include the following steps:
acquiring the type of the characteristic data of the network advertisement to be delivered and a plurality of characteristic data of the network advertisement to be delivered;
determining a construction rule of each kind of feature data according to the type of the acquired feature data;
constructing each kind of feature data by utilizing a plurality of feature data according to the construction rule of each kind of feature data;
and taking the constructed at least one characteristic data as at least one characteristic data of the network advertisement to be delivered.
In a specific application, features of the to-be-delivered network advertisement allowed to be adjusted in different application scenarios may be different, for example, a delivery form of the network advertisement in a certain game application is only one of full-service broadcast, and when click through rate prediction is performed, prediction does not need to be performed with respect to the feature of the delivery form. Therefore, the type of the feature data of the network advertisement to be delivered can be obtained to obtain at least one type of feature data corresponding to the current application scene, so that redundant obtaining of feature data which cannot be adjusted in the current application scene when fixed type of feature data is obtained is avoided, and resource waste is avoided. The obtaining mode of the type of the feature data of the network advertisement to be delivered can be various. For example, the type of the characteristic data of the network advertisement to be delivered, which is input by the maintenance personnel, can be received. Or, for example, a configuration file corresponding to a to-be-delivered web advertisement may be read, and the category of the feature data recorded in the configuration file may be extracted. Any method capable of obtaining the type of the feature data of the network advertisement to be delivered can be used in the present invention, and this embodiment does not limit this.
In addition, the acquisition mode of a plurality of characteristic data of the network advertisement to be delivered can be various. For example, when the to-be-delivered network advertisement carries characteristic data, at least one characteristic data may be directly extracted from the to-be-delivered network advertisement. Or, for example, when the feature data is carried in the placement setting information or the attribute information of the network advertisement to be placed, at least one feature data may be extracted from the placement setting information or the attribute information of the network advertisement to be placed. For example, at least one of the feature data of the placement time, placement position, placement form, and the like of the network advertisement to be placed is extracted.
The construction rules for different kinds of feature data may be various. For example, the construction rule may include using one found feature data as a feature data, splicing at least two found feature data into one feature data, using a specified feature data as a feature data, splicing at least two specified feature data into one feature data, and the like. The splicing form may include that at least two pieces of feature data are spliced according to a preset sequence, or at least two pieces of feature data are spliced according to an arbitrary sequence, and the like. The specific setting of the construction rule may be performed according to an application scenario, and any construction rule of different types of feature data may be used in the present invention, which is not limited in this embodiment.
In addition, determining a construction rule of each type of feature data according to the type of the acquired feature data may specifically include: and aiming at the type of each acquired feature data, searching for a construction rule corresponding to the type of the feature data from the corresponding relation between the type of the preset feature data and the construction rule. Illustratively, the correspondence between the preset feature data type and the construction rule includes: "category T1" corresponds to construction rules "to splice user gender and user age", "category T2" corresponds to construction rules "to splice placement position and placement time of network advertisement", and "category T3" corresponds to construction rules "to use the content form of network advertisement as a kind of feature data". The acquired feature data types are "type T1" and "type T2", then the construction rule of the feature data of "type T1" is found to be "splicing user gender and user age" from the corresponding relationship between the preset feature data types and the construction rule, and the construction rule of the feature data of type T2 "is" splicing network advertisement placement position and placement time ".
After the construction rule of each feature data is determined, the feature data can be constructed by using a plurality of feature data according to the construction rule of each feature data, so that at least one constructed feature data is used as at least one feature data of the network advertisement to be delivered. Illustratively, according to the construction rule of the feature data of the category T1, the feature data "female, 26" of the category T1 is constructed by using the feature data "female gender of user" and "age of user 26"; feature data "advertisement spot of drama a, time at which drama a starts playing" of "category T2" is constructed by using feature data "placement spot of network advertisement is advertisement spot of drama a" and "time at which drama a starts playing" of "release time" according to the rule of constructing feature data of category T2 ".
For the optional embodiment, compared with the case that feature data is not constructed, but a plurality of feature data are all input into a preset prediction model, the construction of different types of feature data separates the function of screening the feature data from the preset prediction model, and content decoupling between feature data acquisition and click through rate prediction is realized. After content decoupling is realized, the functions of the preset prediction model and the function of characteristic data acquisition are mutually independent, when only one of the functions needs to be expanded, a maintainer can directly expand the content corresponding to the function, and the content corresponding to the other function is invisible to the maintainer. Compared with the case of no decoupling, when the function is expanded by the maintainer, the content corresponding to the function to be expanded does not need to be found out from a large amount of content, and the convenience and the efficiency of maintenance can be improved.
Optionally, after the feature data is constructed by using a plurality of feature data according to the construction rule of each feature data, the method for predicting the click through rate provided by the embodiment of the present invention may further include:
inputting at least one constructed characteristic data into a preset hash algorithm to obtain a hash value corresponding to each characteristic data;
and using the obtained hash value as at least one characteristic data of the network advertisement to be delivered.
In a specific application, since the at least one constructed feature data is obtained by splicing a plurality of feature data, a relatively large amount of storage space may be occupied compared to the un-spliced feature data. In order to reduce the occupation of the constructed feature data on the storage space, a preset Hash algorithm, such as a Hash algorithm, may be used to convert at least one constructed feature data into a Key-Value form for storage, where Value is the feature data, and then the obtained Hash Value is used as at least one feature data of the network advertisement to be delivered. The feature data of the preset prediction model input subsequently can be a Key for inputting the feature data. Illustratively, the characteristic data "mai, 26" of "category T1" is input into a preset hash algorithm to obtain Key "1" of the characteristic data; the feature data "the advertisement space of the drama a, the time when the drama a starts playing" of the category T2 "is used to obtain Key" 2 "of the feature data.
As shown in fig. 2, a flow of a method for predicting click through rate according to another embodiment of the present invention may include:
s201, at least one feature data of the network advertisement to be delivered is obtained.
S201 is the same as S101 in the embodiment of fig. 1, and is not repeated herein, for details, see the description of the embodiment of fig. 1.
S202, aiming at each acquired feature data, searching a preset prediction model corresponding to the feature data from the corresponding relation between the preset feature data and the preset prediction model. Each preset prediction model corresponds to one thread, and the threads are used for loading the preset prediction models corresponding to the threads.
S202 is a similar step to S102 in the embodiment of fig. 1, except that each preset prediction model in S202 corresponds to a thread, and the thread is used for loading the preset prediction model corresponding to the thread. For the same parts, detailed description is omitted here, and the description of the embodiment of fig. 1 of the present invention is given in detail.
In order to improve the prediction efficiency when the click through rate of various feature data is predicted, a plurality of used preset prediction models can be loaded in parallel, and therefore, a thread can be respectively arranged for each preset prediction model, and the thread is used for loading the preset prediction model corresponding to the thread. And, in order to ensure the normal operation of the task executed by each thread when multiple threads are concurrently active, step S203 needs to be executed before inputting the feature data into the corresponding preset prediction model.
S203, detecting whether the thread corresponding to the characteristic data finishes running or not according to the acquired characteristic data; if the process is finished, step S204 is executed, otherwise, the process returns to step S203.
The detection mode of whether the thread corresponding to each kind of feature data has finished running or not may be various. For example, after the preset duration of the thread corresponding to each type of feature data is found, whether the thread corresponding to the type of feature data returns the first status word or not is judged, and if the thread returns, the detection result is that the thread is finished running; where the first state word is a state word indicating that the thread has finished running. Or after the thread corresponding to each kind of feature data is found, directly judging whether the thread corresponding to the feature data returns the second status word, and if the thread returns, determining that the detection result is that the thread does not finish running; where the second state word is a state word indicating that the thread is active. Any detection method for detecting whether the thread has finished running can be used in the present invention, and this embodiment does not limit this.
When it is detected that the thread corresponding to each type of feature data has finished running, indicating that the thread does not execute any task, acquiring the predicted click through rate by using the thread does not cause thread activity abnormality or data acquisition abnormality, so step S204 may be executed. When detecting that the thread corresponding to the feature data does not finish running according to each obtained feature data, indicating that the thread still has a task to execute, acquiring the predicted click through rate by using the thread may cause thread activity abnormality or data acquisition abnormality, and therefore, the method may return to the execution step S203 to wait for the thread to finish running, and acquire the predicted click through rate by using the thread in time when running is finished.
And S204, loading a preset prediction model corresponding to the characteristic data by using the thread corresponding to the characteristic data, and inputting the characteristic data into the corresponding preset prediction model to obtain the predicted click through rate corresponding to the characteristic data.
S204 is a similar step to S103 in the embodiment of fig. 1 of the present invention, except that in S204, for each type of acquired feature data, a thread corresponding to the feature data is utilized, and a preset prediction model corresponding to the feature data is loaded, so that for each thread, a task of the thread is corresponding to the feature data, and thread execution confusion caused by data input abnormality can be avoided. For the same parts, detailed description is omitted here, and the description of the embodiment of fig. 1 of the present invention is given in detail.
In the embodiment of fig. 2, a plurality of threads are utilized, different types of feature data can be simultaneously input, the obtaining of the predicted click through rate of each type of feature data is independent, and compared with a serial prediction mode, the obtaining of the predicted click through rate corresponding to one type of feature data is not required to wait until the predicted click through rate corresponding to another type of feature data is obtained, so that the obtaining efficiency of the predicted click through rate can be effectively improved. For example, in the serial prediction mode, it takes about 14105813 microseconds to complete the prediction of the click through rate of a certain to-be-delivered network advertisement; in the multi-thread parallel prediction mode, the click through rate of a certain to-be-delivered network advertisement is predicted, which takes about 8988260 microseconds.
Optionally, the above inputting, for each obtained feature data, the feature data into a corresponding preset prediction model to obtain a predicted click through rate corresponding to the feature data may specifically include the following steps:
acquiring a second identifier of a preset prediction model corresponding to each acquired feature data;
and inputting the second identifier corresponding to the characteristic data into a preset prediction model interface function to obtain the predicted click through rate corresponding to the characteristic data.
In specific applications, a preset prediction model can be encapsulated to achieve content decoupling between feature data acquisition and click through rate prediction. Therefore, the preset prediction model is called through the prediction model interface function. After content decoupling is realized, the functions of the preset prediction model and the function of acquiring the characteristic data are mutually independent, when only one of the functions needs to be expanded, a maintainer can directly expand the content corresponding to the function, and the content corresponding to the other function is invisible to the maintainer. Compared with the case of no decoupling, when the function is expanded by the maintainer, the content corresponding to the function to be expanded does not need to be found out from a large amount of content, and the convenience and the efficiency of maintenance can be improved.
In order to implement the calling, the correspondence between the preset feature data and the preset prediction model may be a correspondence table established by using a first identifier of the feature data and a second identifier of the preset prediction model. Correspondingly, for each type of acquired feature data, a second identifier of the preset prediction model corresponding to the type of feature data is acquired, specifically, the second identifier corresponding to the acquired first identifier of the type of feature data is searched from a corresponding relationship between the preset feature data and the preset prediction model. Inputting the second identifier of the searched preset prediction model into the prediction model interface function, and calling the preset prediction model corresponding to the second identifier, so that the called preset prediction model calculates the predicted click through rate corresponding to the feature data.
Corresponding to the method embodiment, the embodiment of the invention also provides a device for predicting the click through rate.
As shown in fig. 3, a structure of a device for predicting a click through rate according to an embodiment of the present invention may include:
a feature data obtaining module 301, configured to obtain at least one feature data of a network advertisement to be delivered; the type of the feature data is divided according to the features of the network advertisements reflected by the feature data;
a prediction model obtaining module 302, configured to search, for each obtained feature data, a preset prediction model corresponding to the feature data from a corresponding relationship between preset feature data and a preset prediction model; the preset prediction model is obtained by utilizing sample characteristic data and a prediction result label of the sample characteristic data in advance for training, and the sample characteristic data is the same as the type of characteristic data corresponding to the preset prediction model;
the predicting module 303 is configured to, for each acquired feature data, input the feature data into a corresponding preset prediction model to obtain a predicted click through rate corresponding to the feature data.
In the scheme provided by the embodiment of the invention, when the click through rate is predicted, the preset prediction model is a model obtained by utilizing sample characteristic data and a prediction result label of the sample characteristic data in advance for training, the type of the sample characteristic data is the same as that of the characteristic data of the network advertisement to be delivered, and the type of the characteristic data is divided according to the characteristics of the network advertisement reflected by the characteristic data. Therefore, for each type of acquired feature data, after the feature data is input into a corresponding preset prediction model, the predicted click through rates corresponding to the feature data can be obtained. The predicted click through rate corresponding to the characteristic data is obtained aiming at different types of characteristic data of the network advertisement, and when the characteristic of the network advertisement to be delivered is adjusted according to the predicted click through rate, the characteristic data of the network advertisement to be delivered, which influences the obtained predicted click through rate, can be determined, so that the operation and maintenance personnel can adjust the characteristic reflected by the characteristic data in a targeted manner. Compared with the method that operation and maintenance personnel need to tentatively adjust the characteristics of the network advertisements to be delivered one by one, the method saves the step of tentatively adjusting the characteristics which do not influence the predicted click through rate, and can improve the convenience and efficiency of the characteristic adjustment of the network advertisements to be delivered.
Optionally, the feature data obtaining module 401 is specifically configured to:
acquiring the type of characteristic data of a network advertisement to be delivered and a plurality of characteristic data of the network advertisement to be delivered;
determining a construction rule of each kind of feature data according to the type of the acquired feature data;
constructing each kind of feature data by using the plurality of feature data according to the construction rule of each kind of feature data;
and taking the constructed at least one characteristic data as at least one characteristic data of the network advertisement to be delivered.
Optionally, the feature data obtaining module 301 is further configured to, after the feature data is constructed by using the multiple feature data according to the construction rule of each type of feature data, respectively input at least one constructed feature data into a preset hash algorithm to obtain a hash value corresponding to each type of feature data;
and using the obtained hash value as at least one characteristic data of the network advertisement to be delivered.
Optionally, the prediction module 403 is specifically configured to:
acquiring a second identifier of a preset prediction model corresponding to each acquired feature data;
and inputting the second identifier corresponding to the characteristic data into a preset prediction model interface function to obtain the predicted click through rate corresponding to the characteristic data.
As shown in fig. 4, a structure of a device for predicting a click through rate according to another embodiment of the present invention may include:
a feature data obtaining module 401, configured to obtain at least one feature data of a network advertisement to be delivered; the type of the feature data is divided according to the features of the network advertisements reflected by the feature data;
a prediction model obtaining module 402, configured to search, for each obtained feature data, a preset prediction model corresponding to the feature data from a correspondence between preset feature data and a preset prediction model; the preset prediction model is obtained by utilizing sample characteristic data and a prediction result label of the sample characteristic data in advance for training, and the sample characteristic data is the same as the type of characteristic data corresponding to the preset prediction model; each preset prediction model corresponds to one thread; the thread is used for loading a preset prediction model corresponding to the thread;
the predicting module 403 is configured to, for each obtained feature data, input the feature data into a corresponding preset prediction model to obtain a predicted click through rate corresponding to the feature data.
The prediction module 403 comprises a judgment sub-module 4031 and a loading sub-module 4032;
the determining submodule 4031 is configured to detect, for each acquired feature data, whether a thread corresponding to the feature data has finished running; if the operation is finished, triggering the loading sub-module 4032 to operate;
the loading sub-module 4032 is configured to load a preset prediction model corresponding to the feature data by using a thread corresponding to the feature data under the trigger of the determining sub-module 4031, and input the feature data into the corresponding preset prediction model to obtain a predicted click through rate corresponding to the feature data, otherwise, execute the detection on whether the thread corresponding to the feature data has finished running.
Corresponding to the above embodiment, an embodiment of the present invention further provides an electronic device, as shown in fig. 5, where the electronic device may include:
the system comprises a processor 501, a communication interface 502, a memory 503 and a communication bus 504, wherein the processor 501, the communication interface 502 and the memory complete mutual communication through the communication bus 504 through the 503;
a memory 503 for storing a computer program;
the processor 501 is configured to implement the steps of the method for predicting the click through rate in any of the embodiments when executing the computer program stored in the memory 503.
In the scheme provided by the embodiment of the invention, when the click through rate is predicted, the preset prediction model is a model obtained by utilizing sample characteristic data and a prediction result label of the sample characteristic data in advance for training, the type of the sample characteristic data is the same as that of the characteristic data of the network advertisement to be delivered, and the type of the characteristic data is divided according to the characteristics of the network advertisement reflected by the characteristic data. Therefore, for each type of acquired feature data, after the feature data is input into a corresponding preset prediction model, the predicted click through rates corresponding to the feature data can be obtained. The predicted click through rate corresponding to the characteristic data is obtained aiming at different types of characteristic data of the network advertisement, and when the characteristic of the network advertisement to be delivered is adjusted according to the predicted click through rate, the characteristic data of the network advertisement to be delivered, which influences the obtained predicted click through rate, can be determined, so that the operation and maintenance personnel can adjust the characteristic reflected by the characteristic data in a targeted manner. Compared with the method that operation and maintenance personnel need to tentatively adjust the characteristics of the network advertisements to be delivered one by one, the method saves the step of tentatively adjusting the characteristics which do not influence the predicted click through rate, and can improve the convenience and efficiency of the characteristic adjustment of the network advertisements to be delivered.
The Memory may include a RAM (Random Access Memory) or an NVM (Non-Volatile Memory), such as at least one disk Memory. Optionally, the memory may also be at least one memory device located remotely from the processor.
The Processor may be a general-purpose Processor, including a Central Processing Unit (CPU), a Network Processor (NP), and the like; but also a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array) or other Programmable logic device, discrete Gate or transistor logic device, discrete hardware component.
The computer-readable storage medium provided by an embodiment of the present invention is embodied in an electronic device, and a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, the steps of any one of the above-mentioned methods for predicting a click through rate are implemented.
In the scheme provided by the embodiment of the invention, when the click through rate is predicted, the preset prediction model is a model obtained by utilizing sample characteristic data and a prediction result label of the sample characteristic data in advance for training, the type of the sample characteristic data is the same as that of the characteristic data of the network advertisement to be delivered, and the type of the characteristic data is divided according to the characteristics of the network advertisement reflected by the characteristic data. Therefore, for each type of acquired feature data, after the feature data is input into a corresponding preset prediction model, the predicted click through rates corresponding to the feature data can be obtained. The predicted click through rate corresponding to the feature data is obtained aiming at different types of feature data of the network advertisements, and when the features of the network advertisements to be delivered are adjusted according to the predicted click through rate, the feature data of the network advertisements to be delivered, which affect the obtained predicted click through rate, can be determined, so that operation and maintenance personnel can adjust the features reflected by the feature data in a targeted manner. Compared with the method that operation and maintenance personnel need to tentatively adjust the characteristics of the network advertisements to be delivered one by one, the method saves the step of tentatively adjusting the characteristics which do not influence the predicted click through rate, and can improve the convenience and efficiency of the characteristic adjustment of the network advertisements to be delivered.
In yet another embodiment of the present invention, there is also provided a computer program product containing instructions, which when run on a computer, causes the computer to execute the method for predicting click through rate as described in any of the above embodiments.
In the above embodiments, the implementation may be wholly or partially realized by software, hardware, firmware, or any combination thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When loaded and executed on a computer, cause the processes or functions described in accordance with the embodiments of the invention to occur, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device. The computer instructions may be stored in or transmitted from a computer-readable storage medium to another computer-readable storage medium, for example, from a website, computer, server, or data center, over a wired (e.g., coaxial cable, fiber optic, DSL (Digital Subscriber Line), or wireless (e.g., infrared, radio, microwave, etc.) network, to another website, computer, server, or data center, to any available medium that is accessible by a computer or that is a data storage device including one or more integrated servers, data centers, etc. the available medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD (Digital Versatile Disc, digital versatile disc)), or a semiconductor medium (e.g.: SSD (Solid State Disk)), etc.
In this document, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
All the embodiments in the present specification are described in a related manner, and the same and similar parts among the embodiments may be referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, as for the apparatus and electronic device embodiments, since they are substantially similar to the method embodiments, the description is relatively simple, and reference may be made to some descriptions of the method embodiments for relevant points.
The above description is only for the preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention shall fall within the protection scope of the present invention.
Claims (10)
1. A method for predicting click through rate, the method comprising:
acquiring at least one characteristic data of a network advertisement to be delivered; the type of the feature data is divided according to the features of the network advertisements reflected by the feature data;
aiming at each acquired feature data, searching a preset prediction model corresponding to the feature data from the corresponding relation between the preset feature data and the preset prediction model; the preset prediction model is obtained by utilizing sample characteristic data and a prediction result label of the sample characteristic data in advance for training, and the sample characteristic data is the same as the type of characteristic data corresponding to the preset prediction model;
inputting the characteristic data into a corresponding preset prediction model aiming at each acquired characteristic data to obtain a predicted click through rate corresponding to the characteristic data;
each preset prediction model corresponds to one thread; the thread is used for loading a preset prediction model corresponding to the thread;
the step of inputting the acquired feature data into a corresponding preset prediction model for each type of feature data to obtain the predicted click through rate corresponding to the feature data includes:
detecting whether the thread corresponding to the characteristic data finishes running or not aiming at each acquired characteristic data;
and if the operation is finished, loading a preset prediction model corresponding to the characteristic data by using the thread corresponding to the characteristic data, inputting the characteristic data into the corresponding preset prediction model to obtain the predicted click through rate corresponding to the characteristic data, and otherwise, executing the detection to determine whether the operation of the thread corresponding to the characteristic data is finished.
2. The method of claim 1, wherein the obtaining at least one feature data of the network advertisement to be delivered comprises:
acquiring the type of characteristic data of a network advertisement to be delivered and a plurality of characteristic data of the network advertisement to be delivered;
determining a construction rule of each type of feature data according to the type of the acquired feature data;
constructing each kind of feature data by using the plurality of feature data according to the construction rule of each kind of feature data;
and taking the constructed at least one characteristic data as at least one characteristic data of the network advertisement to be delivered.
3. The method according to claim 2, wherein after said constructing the feature data using the plurality of feature data according to the construction rule of each of the feature data, respectively, the method further comprises:
inputting at least one constructed characteristic data into a preset hash algorithm to obtain a hash value corresponding to each characteristic data;
and using the obtained hash value as at least one characteristic data of the network advertisement to be delivered.
4. The method according to claim 1, wherein the step of inputting each type of feature data into a corresponding preset prediction model for obtaining the predicted click through rate corresponding to the type of feature data comprises:
acquiring a second identifier of a preset prediction model corresponding to each acquired feature data;
and inputting the second identifier corresponding to the characteristic data into a preset prediction model interface function to obtain the predicted click through rate corresponding to the characteristic data.
5. An apparatus for predicting click through rate, the apparatus comprising:
the characteristic data acquisition module is used for acquiring at least one characteristic data of the network advertisement to be delivered; the type of the feature data is divided according to the features of the network advertisements reflected by the feature data;
the prediction model acquisition module is used for searching a preset prediction model corresponding to each acquired feature data from the corresponding relation between the preset feature data and the preset prediction model; the preset prediction model is obtained by utilizing sample characteristic data and a prediction result label of the sample characteristic data in advance for training, and the sample characteristic data is the same as the type of characteristic data corresponding to the preset prediction model;
the prediction module is used for inputting the characteristic data into a corresponding preset prediction model aiming at each acquired characteristic data to obtain a predicted click through rate corresponding to the characteristic data;
each preset prediction model corresponds to one thread; the thread is used for loading a preset prediction model corresponding to the thread;
the prediction module comprises a judgment submodule and a loading submodule;
the judging submodule is used for detecting whether the thread corresponding to the characteristic data finishes running or not according to the acquired characteristic data; if the operation is finished, triggering the loading sub-module to operate;
and the loading submodule is used for loading a preset prediction model corresponding to the characteristic data by using the thread corresponding to the characteristic data under the trigger of the judging submodule, inputting the characteristic data into the corresponding preset prediction model to obtain the predicted click through rate corresponding to the characteristic data, and otherwise, executing the detection of whether the thread corresponding to the characteristic data is finished running.
6. The apparatus of claim 5, wherein the feature data acquisition module is specifically configured to:
acquiring the type of characteristic data of a network advertisement to be delivered and a plurality of characteristic data of the network advertisement to be delivered;
determining a construction rule of each kind of feature data according to the type of the acquired feature data;
constructing each kind of feature data by using the plurality of feature data according to the construction rule of each kind of feature data;
and taking the constructed at least one characteristic data as at least one characteristic data of the network advertisement to be delivered.
7. The apparatus according to claim 6, wherein the feature data obtaining module is further configured to, after the feature data is constructed according to the construction rule of each type of feature data by using the plurality of feature data, input at least one constructed feature data into a preset hash algorithm to obtain a hash value corresponding to each type of feature data;
and using the obtained hash value as at least one characteristic data of the network advertisement to be delivered.
8. The apparatus of claim 5, wherein the prediction module is specifically configured to:
acquiring a second identifier of a preset prediction model corresponding to each acquired feature data;
and inputting the second identifier corresponding to the characteristic data into a preset prediction model interface function to obtain the predicted click through rate corresponding to the characteristic data.
9. An electronic device is characterized by comprising a processor, a communication interface, a memory and a communication bus, wherein the processor and the communication interface are used for realizing the communication between the processor and the memory through the bus; a memory for storing a computer program; a processor for executing a program stored in the memory to perform the method steps of any of claims 1 to 4.
10. A computer-readable storage medium, characterized in that a computer program is stored in the storage medium, which computer program, when being executed by a processor, carries out the method steps of any one of claims 1-4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910444789.2A CN110223108B (en) | 2019-05-27 | 2019-05-27 | Click through rate prediction method, device and equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910444789.2A CN110223108B (en) | 2019-05-27 | 2019-05-27 | Click through rate prediction method, device and equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110223108A CN110223108A (en) | 2019-09-10 |
| CN110223108B true CN110223108B (en) | 2022-05-24 |
Family
ID=67818015
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910444789.2A Active CN110223108B (en) | 2019-05-27 | 2019-05-27 | Click through rate prediction method, device and equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110223108B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112115365B (en) * | 2020-09-25 | 2021-09-14 | 贝壳找房(北京)科技有限公司 | Model collaborative optimization method, device, medium and electronic equipment |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105630996A (en) * | 2015-12-25 | 2016-06-01 | 腾讯科技（深圳）有限公司 | Information processing method and server |
| CN107657486A (en) * | 2017-10-19 | 2018-02-02 | 厦门美柚信息科技有限公司 | A kind of advertisement placement method and device |
| CN108629630A (en) * | 2018-05-08 | 2018-10-09 | 广州太平洋电脑信息咨询有限公司 | A kind of feature based intersects the advertisement recommendation method of joint deep neural network |
| CN109034880A (en) * | 2018-07-09 | 2018-12-18 | 麒麟合盛网络技术股份有限公司 | revenue prediction method and device |
| CN109740068A (en) * | 2019-01-29 | 2019-05-10 | 腾讯科技（北京）有限公司 | Media data recommended method, device and storage medium |
| CN110188942A (en) * | 2019-05-27 | 2019-08-30 | 北京金山安全软件有限公司 | Click through rate prediction method, device and equipment |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8924252B2 (en) * | 2012-05-14 | 2014-12-30 | Iqzone, Inc. | Systems and methods for providing timely advertising to portable devices |
| CN105227975B (en) * | 2015-09-29 | 2018-07-13 | 北京奇艺世纪科技有限公司 | Advertisement placement method and device |
| CN106126597A (en) * | 2016-06-20 | 2016-11-16 | 乐视控股（北京）有限公司 | User property Forecasting Methodology and device |
| CN107742221A (en) * | 2016-08-23 | 2018-02-27 | 腾讯科技（深圳）有限公司 | A kind of processing method of promotion message, device and system |
| CN106803190A (en) * | 2017-01-03 | 2017-06-06 | 北京掌阔移动传媒科技有限公司 | A kind of ad personalization supplying system and method |
-
2019
- 2019-05-27 CN CN201910444789.2A patent/CN110223108B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105630996A (en) * | 2015-12-25 | 2016-06-01 | 腾讯科技（深圳）有限公司 | Information processing method and server |
| CN107657486A (en) * | 2017-10-19 | 2018-02-02 | 厦门美柚信息科技有限公司 | A kind of advertisement placement method and device |
| CN108629630A (en) * | 2018-05-08 | 2018-10-09 | 广州太平洋电脑信息咨询有限公司 | A kind of feature based intersects the advertisement recommendation method of joint deep neural network |
| CN109034880A (en) * | 2018-07-09 | 2018-12-18 | 麒麟合盛网络技术股份有限公司 | revenue prediction method and device |
| CN109740068A (en) * | 2019-01-29 | 2019-05-10 | 腾讯科技（北京）有限公司 | Media data recommended method, device and storage medium |
| CN110188942A (en) * | 2019-05-27 | 2019-08-30 | 北京金山安全软件有限公司 | Click through rate prediction method, device and equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110223108A (en) | 2019-09-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20210042355A1 (en) | Method and device for searching for information in applications | |
| US9183072B1 (en) | Error troubleshooting using a correlated knowledge base | |
| US20130138674A1 (en) | System and method for recommending application by using keyword | |
| US11310066B2 (en) | Method and apparatus for pushing information | |
| CN109753601B (en) | Method and device for determining click rate of recommended information and electronic equipment | |
| US9645914B1 (en) | Apps store with integrated test support | |
| CN110704677B (en) | Program recommendation method and device, readable storage medium and terminal equipment | |
| KR20210038454A (en) | User grouping method, apparatus thereof, computer, computer-readable recording meduim and computer program | |
| CN109547807B (en) | Information processing method and device based on live broadcast and server | |
| CN109831532B (en) | Data sharing method, device, equipment and medium | |
| CN108153909B (en) | Keyword putting word-expanding method and device, electronic equipment and storage medium | |
| CN110262851B (en) | Interface configuration method and device | |
| CN108366012B (en) | Social relationship establishing method and device and electronic equipment | |
| CN111159563A (en) | Method, device and equipment for determining user interest point information and storage medium | |
| WO2015185020A1 (en) | Information category obtaining method and apparatus | |
| JP5264813B2 (en) | Evaluation apparatus, evaluation method, and evaluation program | |
| CN111382182A (en) | Data processing method and device, electronic equipment and storage medium | |
| CN110223108B (en) | Click through rate prediction method, device and equipment | |
| CN112346951A (en) | Service testing method and device | |
| CN114625407A (en) | Method, system, equipment and storage medium for implementing AB experiment | |
| CN109885729B (en) | Method, device and system for displaying data | |
| CN111553749A (en) | Activity push strategy configuration method and device | |
| CN107480269B (en) | Object display method and system, medium and computing equipment | |
| CN116225690A (en) | Memory multidimensional database calculation load balancing method and system based on docker | |
| CN110929207B (en) | Data processing method, device and computer readable storage medium |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |