CN112446736A - Click through rate CTR prediction method and device - Google Patents
Click through rate CTR prediction method and device Download PDFInfo
- Publication number
- CN112446736A CN112446736A CN202011393094.5A CN202011393094A CN112446736A CN 112446736 A CN112446736 A CN 112446736A CN 202011393094 A CN202011393094 A CN 202011393094A CN 112446736 A CN112446736 A CN 112446736A
- Authority
- CN
- China
- Prior art keywords
- user
- equipment
- model
- advertisements
- model parameters
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- 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
- G06Q30/0255—Targeted advertisements based on user history
-
- 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
Abstract
The embodiment of the application provides a Click Through Rate (CTR) prediction method and device. The click through rate CTR prediction method comprises the following steps: the first equipment trains the first model based on user data to obtain the trained first model and first model parameters; the first equipment sends the first model parameters to the second equipment; the first equipment receives a second model parameter sent by the second equipment; the first equipment updates the first model parameter of the first model into a second model parameter and trains the first model parameter until the loss value of the second model parameter is converged to obtain a second model; the first device predicts a probability that the user clicks each of the different types of advertisements, respectively, based on the second model. According to the method and the device, click prediction can be performed on different types of candidate advertisements, the advertisement putting efficiency can be improved, the advertisement cost is saved, the user retention rate is improved, and the user can be prevented from being disturbed by disliked advertisements.
Description
Technical Field
The application relates to the technical field of internet, in particular to a Click Through Rate (CTR) prediction method and device.
Background
Click Through Rate (CTR) is the probability of clicking on an advertisement after a user views the advertisement, and can reflect the attention degree of the advertisement, and is generally used to evaluate the efficiency of advertisement delivery. In the current CTR prediction method, the number of user features and combinations of different features need to be considered, however, in most practical cases, different platforms often can only acquire user features of limited dimensions. In order to improve the advertisement putting efficiency of the platform, various factors of the user, such as the change of recent work place or home address, the browsing record of shopping websites, the online period of games and the like, need to be considered, which may come from different platforms, and due to business competition or related privacy protection regulations, the data of each party are isolated from each other, and joint modeling is difficult to realize.
Therefore, on the premise of ensuring data security, the user data of different platforms are used for optimizing advertisement delivery, so that the advertisement click rate is improved, and better experience is brought to users.
Disclosure of Invention
In view of the above, the present application is proposed to provide a training generation method and apparatus for predicting click through rate, which overcomes or at least partially solves the above problems.
In a first aspect, an embodiment of the present application provides a click through rate CTR prediction method, which is applied to a first device, and may include:
the method comprises the steps that a first device trains a first model based on user data to obtain the trained first model and first model parameters, the first device is any one of a plurality of first devices, the user data comprises click information of whether a user clicks a plurality of types of advertisements, and the first model is used for predicting the probability that the user clicks each type of advertisements in the plurality of types of advertisements;
the first equipment sends the first model parameter to second equipment;
the first equipment receives second model parameters sent by the second equipment, wherein the second model parameters are determined based on first model parameters respectively corresponding to the plurality of first equipment;
the first equipment updates the first model parameter of the first model into the second model parameter and trains the first model parameter until the loss value of the second model parameter is converged to obtain a second model;
the first device predicts, based on the second model, a probability that a user clicks each of the different types of advertisements, respectively.
By the method of the first aspect, in the embodiment of the application, the first device may first train the first model based on user data to obtain the trained first model and first model parameters; the first equipment sends the first model parameters to the second equipment; the first equipment receives a second model parameter sent by the second equipment; secondly, the first equipment updates the first model parameter of the first model into a second model parameter and trains the second model parameter until the loss value of the second model parameter is converged to obtain a second model; the first device predicts a probability that the user clicks each of the different types of advertisements, respectively, based on the second model. The first equipment is service equipment of different platforms, and the second equipment is a server capable of performing longitudinal federal modeling, so that each platform obtains a partial model of the platform and performs click prediction on candidate advertisements of different classes. The method provides possibility for cooperation of different platforms through a federal modeling method, so that training and learning can be performed jointly under the condition that platform data are mutually isolated, user data of different platforms can be better utilized, the delivery efficiency of platform advertisements is improved, advertisement expenses and social resources are saved, the user experience is improved, and loss of platform users is reduced.
In one possible implementation, the method further includes: the first device sends a user identifier set to the second device, wherein the user identifier set comprises a plurality of user identifiers, and each user identifier in the plurality of user identifiers is used for identifying a corresponding user and user data respectively; the first device receives a target user identification set sent by the second device, wherein the target user identification set comprises user identifications of target users, and the target users are users common among the plurality of first devices; and the first equipment acquires the user data of the target user according to the target user identification set.
In one possible implementation manner, the sending, by the first device, the set of user identifications to the second device includes: and the first equipment carries out hash encryption on the user identification included in the user identification set and sends the encrypted user identification set to the second equipment.
In one possible implementation manner, the sending, by the first device, the first model parameter to the second device includes: and the first equipment homomorphically encrypts the first model parameter and then sends the first model parameter to second equipment.
In one possible implementation, the method further includes: and when the probability that the user clicks the target type advertisement is larger than a preset threshold value, delivering the target type advertisement to the user in the first device, wherein the target type advertisement belongs to one of the multiple types of advertisements.
In a second aspect, an embodiment of the present application provides another click through rate CTR prediction method, which is applied to a second device, and includes:
the second equipment receives first model parameters which are respectively sent by a plurality of first equipment, the first model parameters are model parameters obtained by the first equipment through training a first model based on user data, the user data comprise click information of whether a user clicks a plurality of types of advertisements, and the first model is used for predicting the probability that the user clicks each type of advertisements in the plurality of types of advertisements;
the second equipment optimizes the first model parameters corresponding to the plurality of first equipment to generate second model parameters;
and the second equipment respectively sends the second model parameters to the plurality of first equipment.
In one possible implementation, the method further includes: the second device receives a user identifier set respectively sent by the first devices, wherein the user identifier set comprises a plurality of user identifiers, and each user identifier in the user identifiers is respectively used for identifying a corresponding user and user data; the second equipment screens out the user identification of the common user among the plurality of first equipment from the user identification sets corresponding to the plurality of first equipment to obtain a target user identification set; and the second equipment respectively sends the target user identification sets to the plurality of first equipment.
In a third aspect, an embodiment of the present application provides a click through rate CTR prediction apparatus, which is applied to a first device, and includes:
the system comprises a first training unit, a second training unit and a third training unit, wherein the first training unit is used for training a first model based on user data to obtain the trained first model and first model parameters, the first equipment is any one of a plurality of first equipment, the user data comprises click information of whether a user clicks a plurality of types of advertisements, and the first model is used for predicting the probability that the user clicks each type of advertisements in the plurality of types of advertisements;
the first sending unit is used for sending the first model parameter to second equipment;
a first receiving unit, configured to receive a second model parameter sent by the second device, where the second model parameter is determined based on first model parameters respectively corresponding to the multiple first devices;
a second training unit, configured to update the first model parameter of the first model to the second model parameter and train the first model parameter until a loss value of the second model parameter converges to obtain a second model;
and the prediction unit is used for predicting the probability that the user clicks each type of advertisement in the different types of advertisements respectively based on the second model.
In one possible implementation, the apparatus further includes: a second sending unit, configured to send a user identifier set to the second device, where the user identifier set includes multiple user identifiers, and each user identifier in the multiple user identifiers is used to identify a corresponding user and user data; a second receiving unit, configured to receive a target user identifier set sent by the second device, where the target user identifier set includes a user identifier of a target user, and the target user is a user common among the multiple first devices; and the first equipment acquires the user data of the target user according to the target user identification set.
In a possible implementation manner, the second sending unit is specifically configured to hash and encrypt the user identifiers included in the user identifier set, and send the encrypted user identifier set to the second device.
In a possible implementation manner, the first sending unit is specifically configured to send, to the second device, the first device after homomorphically encrypting the first model parameter.
In one possible implementation, the apparatus further includes: and the delivery unit is used for delivering the target type advertisement to the user in the first equipment when the probability that the user clicks the target type advertisement is greater than a preset threshold value, wherein the target type advertisement belongs to one of the multiple types of advertisements.
In a fourth aspect, an embodiment of the present application provides another click through rate CTR prediction apparatus, which is applied to a second device, and includes:
a third receiving unit, configured to receive first model parameters sent by a plurality of first devices, where the first model parameters are model parameters obtained by the first devices through training a first model based on user data, the user data includes click information of whether a user clicks on multiple types of advertisements, and the first model is used to predict a probability that the user clicks each type of advertisement among the multiple types of advertisements;
the optimization unit is used for optimizing the first model parameters corresponding to the plurality of first devices to generate second model parameters;
and a third sending unit, configured to send the second model parameters to the multiple first devices, respectively.
In one possible implementation, the apparatus further includes: a fourth receiving unit, configured to receive a user identifier set sent by each of the multiple first devices, where the user identifier set includes multiple user identifiers, and each of the multiple user identifiers is used to identify a corresponding user and user data; the screening unit is used for screening out the user identifiers of the common users among the plurality of first devices from the user identifier sets corresponding to the plurality of first devices to obtain a target user identifier set; a fourth sending unit, configured to send the target user identifier sets to the multiple first devices respectively.
In a fifth aspect, the embodiment of the present application provides another click through rate CTR prediction apparatus, including a storage component, a processing component and a communication component, where the storage component is used for storing a computer program, and the communication component is used for performing information interaction with an external device; the processing component is configured to invoke the computer program to execute the method according to the first aspect, which is not described herein again
In a sixth aspect, the embodiment of the present application provides another click through rate CTR prediction apparatus, including a storage component, a processing component and a communication component, where the storage component is used for storing a computer program, and the communication component is used for performing information interaction with an external device; the processing component is configured to invoke the computer program to execute the method according to the second aspect, which is not described herein again.
In a seventh aspect, this application embodiment provides a computer-readable storage medium, where the computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement the method of the first aspect.
In an eighth aspect, the present embodiment provides a computer-readable storage medium, where the computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement the method of the second aspect.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments or the background art of the present application, the drawings required to be used in the embodiments or the background art of the present application will be described below.
Fig. 1 is a schematic diagram of an architecture of a click through rate CTR prediction system according to an embodiment of the present disclosure.
Fig. 2 is a schematic diagram of a flow of a click through rate CTR prediction method provided in an embodiment of the present application.
Fig. 3 is a schematic diagram of user data of multiple first devices according to an embodiment of the present application.
Fig. 4 is a schematic interaction diagram between a plurality of first devices and a second device provided in an embodiment of the present application.
Fig. 5 is a schematic structural diagram of a click through rate CTR prediction apparatus according to an embodiment of the present application.
Fig. 6 is a schematic structural diagram of another click through rate CTR prediction apparatus according to an embodiment of the present application.
Fig. 7 is a schematic structural diagram of another click through rate CTR prediction apparatus according to an embodiment of the present application.
Fig. 8 is a schematic structural diagram of another click through rate CTR prediction apparatus according to an embodiment of the present application.
Detailed Description
The embodiments of the present application will be described below with reference to the drawings.
The terms "first," "second," and "third," etc. in the description and claims of this application and in the accompanying drawings are used for distinguishing between different objects and not for describing a particular order. Furthermore, "include" and "have" and any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not limited to only those steps or elements listed, but may alternatively include other steps or elements not listed, or inherent to such process, method, article, or apparatus.
Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. It is explicitly and implicitly understood by one skilled in the art that the embodiments described herein can be combined with other embodiments.
As used in this application, the terms "second device," "unit," "system," and the like are intended to refer to a computer-related entity, either hardware, firmware, a combination of hardware and software, or software in execution. For example, the second device may be, but is not limited to, a processor, a data processing platform, a computing device, a computer, two or more computers, and the like.
First, some terms in the present application are explained so as to be easily understood by those skilled in the art.
(1) Federal machine learning is also known as Federal learning, Joint learning, and Union learning. Federal machine learning is a machine learning framework, and can effectively help a plurality of organizations to perform data use and machine learning modeling under the condition of meeting the requirements of user privacy protection, data safety and government regulations.
(2) Cisco's Internet Operating System (IOS), an operating system optimized for internetworking, is a software architecture separated from hardware, and with the continuous development of network technology, can be dynamically upgraded to adapt to the changing technology (hardware and software), and has modularity, flexibility, scalability and controllability.
(3) Windows Phone (abbreviated as WP) is a mobile Phone operating system formally released by Microsoft in 2010 at 10/21.M, and has a series of defensive operation experiences such as desktop customization, icon dragging, sliding control and the like. Its home screen displays new emails, short messages, missed calls, calendar appointments, etc. by providing a dashboard-like experience. It also includes an enhanced touch screen interface for more convenient finger operation.
Next, a description is given of one of click through rate CTR prediction system architectures on which the embodiments of the present application are based. Referring to fig. 1, fig. 1 is a schematic diagram of an architecture of a click through rate CTR prediction system according to an embodiment of the present application, including: a plurality of first devices 101 and second devices 102. Wherein:
the first device 101 may include, but is not limited to, a backend server, a component server, a data processing server, etc., a program that provides local services to a client. The local services may include, but are not limited to: the first model can be trained based on user data, the trained first model and first model parameters are obtained, the user data comprise click information of whether a user clicks on multiple types of advertisements, and the first model is used for predicting the probability that the user clicks each type of advertisements in the multiple types of advertisements respectively; sending the first model parameters to a second device; receiving second model parameters sent by the second device, wherein the second model parameters are determined based on first model parameters respectively corresponding to the plurality of first devices; updating the first model parameter of the first model into the second model parameter and training until the loss value of the second model parameter is converged to obtain a second model; predicting a probability that a user clicks each of the different types of advertisements, respectively, based on the second model.
The second device 102 may install and run the associated application. The application refers to corresponding to a first device, and when the second device 102 is a server, the server may communicate with a plurality of first devices through the internet, and the server also needs to run a corresponding program to provide a corresponding model training service, and so on. For example, the server may receive first model parameters sent by a plurality of first devices, where the first model parameters are model parameters obtained by the first devices through training a first model based on user data, where the user data includes click information of whether a user clicks on a plurality of types of advertisements, and the first model is used to predict a probability that the user clicks on each type of advertisements in the plurality of types of advertisements; optimizing first model parameters corresponding to the plurality of first devices to generate second model parameters; and respectively sending the second model parameters to the plurality of first devices.
The first device 101 may send information (e.g., first model parameters) to the second device and receive information (e.g., second model parameters) sent by the second device along with other shared information, etc. First device and second device in this embodimentThe device may include, but is not limited to, any electronic product based on an intelligent operating system, which can interact with a user through an input device such as a keyboard, a virtual keyboard, a touch pad, a touch screen, and a voice control device, such as a tablet computer, a personal computer, and the like. Smart operating systems include, but are not limited to, any operating system that enriches device functionality by providing various mobile applications to a mobile device, such as: android (Android)TM)、iOSTM、Windows PhoneTMAnd so on.
It is further understood that the click through rate CTR prediction system architecture of fig. 1 is only a partial exemplary implementation manner in the embodiment of the present application, and the click through rate CTR prediction system architecture in the embodiment of the present application includes, but is not limited to, the above click through rate CTR prediction system architecture.
Referring to fig. 2, fig. 2 is a schematic diagram of a flow of a click through rate CTR prediction method provided in an embodiment of the present application. Applicable to the system of fig. 1 described above, the interaction between the first device 101 and the second device 102 will be described below in connection with fig. 2. Wherein, the method may comprise the following steps S201 to S206.
Step S201, the first device trains the first model based on the user data to obtain the trained first model and the first model parameter.
Specifically, the first device trains a first model based on user data to obtain the trained first model and first model parameters, the first device is any one of a plurality of first devices, the user data includes click information of whether a user clicks a plurality of types of advertisements, and the first model is used for predicting the probability that the user clicks each type of advertisements in the plurality of types of advertisements. The plurality of first devices respectively represent different user platforms, and the data of each first device contains user information capable of uniquely identifying each user and user behavior characteristic data specific to the platform. That is, each first device has its own unique user data, and the user data in this embodiment includes click information of whether a user clicks on multiple types of advertisements, for example: and taking whether the user clicks on different types of candidate advertisements or not as a data label.
Referring to fig. 3, fig. 3 is a schematic diagram of user data of a plurality of first devices according to an embodiment of the present application, it should be noted that each user platform corresponds to a first device, as shown in fig. 3, for a room renting platform, the user data may be whether a user has a house browsing or searching record in the near future; for the job hunting platform, the user data can be the record of whether the user has an interview recently and the post information of the interview; for the question-answer community, the user data may be browsing records, question-answer records, and the like of whether the user has some kind of articles recently.
Optionally, the first device sends a user identifier set to the second device, where the user identifier set includes multiple user identifiers, and each user identifier in the multiple user identifiers is used to identify a corresponding user and user data, respectively; the first device receives a target user identification set sent by the second device, wherein the target user identification set comprises user identifications of target users, and the target users are users common among the plurality of first devices; and the first equipment acquires the user data of the target user according to the target user identification set. It can be understood that, since the platforms performing the cooperation are often different platforms in different fields, most of the user characteristics are inconsistent, but there are more users in common among different user platforms, so that common users among a plurality of first devices can be screened in advance to improve the accuracy of federal learning.
Optionally, the sending, by the first device, the user identifier set to the second device includes: and the first equipment carries out hash encryption on the user identification included in the user identification set and sends the encrypted user identification set to the second equipment.
Step S202, the first device sends the first model parameter to the second device.
Specifically, the first device sends the first model parameter to the second device. It will be appreciated that the first model parameters may include gradient and loss values. Referring to fig. 4, fig. 4 is an interaction diagram between a plurality of first devices and a second device provided in the embodiment of the present application, and it should be noted that the plurality of first devices send trained first model parameters to the second device, so that the second device can optimize the first model parameters according to the plurality of first devices, and the optimized second model parameters are used for the plurality of first devices to update the model parameters, thereby improving accuracy of model prediction. It is understood that the system as referred to in fig. 4 is the second device as referred to in the embodiments of the present application, and the platform as referred to in fig. 4 corresponds to the first device as referred to in the embodiments of the present application.
In one possible implementation manner, the sending, by the first device, the first model parameter to the second device includes: and the first equipment homomorphically encrypts the first model parameter and then sends the first model parameter to second equipment. It will be appreciated that the gradient and loss value encryptions are calculated in the following manner:
Specifically, the second device optimizes first model parameters corresponding to the plurality of first devices to generate second model parameters, and the second model parameters are determined according to the first model parameters sent by the plurality of first devices. For example, an average calculation may be performed on a plurality of first model parameters to obtain second model parameters. The present application is not specifically limited.
And step S204, the second equipment respectively sends the second model parameters to the plurality of first equipment.
Specifically, the second device sends the second model parameters to the plurality of first devices, respectively. It can be understood that, when the second device sends the second model parameter, the second model parameter may be homomorphic encrypted and then sent, so as to ensure the security of information communication.
Step S205, the first device updates the first model parameter of the first model to the second model parameter and trains the first model parameter until the loss value of the second model parameter converges to obtain the second model.
Specifically, the first device updates the first model parameter of the first model to the second model parameter and trains the first model parameter until a loss value of the second model parameter converges to obtain a second model. It is understood that the second model is a model of the first model with updated parameters, and is used for predicting the probability that the user clicks each type of advertisement in the different types of advertisements respectively.
In step S206, the first device predicts the probability that the user clicks each of the different types of advertisements based on the second model.
Specifically, the first device predicts the probability that the user clicks each type of advertisement in the different types of advertisements respectively based on the second model.
Optionally, when the probability that the user clicks the target type advertisement is greater than a preset threshold, the target type advertisement is delivered to the user in the first device, and the target type advertisement belongs to one of the multiple types of advertisements. For example, for a user who has a moving behavior recently, a shopping platform can push living goods for the user, a question and answer platform can push articles for house improvement for the user, and a job hunting platform can recommend nearby employment opportunities for the user according to his occupation; for a user who just loses love, the shopping platform can push books in emotion, the knowledge sharing platform can push questions and answers about losing love, and the takeaway platform can push a luxury single person package which is that one person also needs to eat good points, and the like.
According to the embodiment of the application, the first equipment can firstly train the first model based on user data to obtain the trained first model and first model parameters; the first equipment sends the first model parameters to the second equipment; the first equipment receives a second model parameter sent by the second equipment; secondly, the first equipment updates the first model parameter of the first model into a second model parameter and trains the second model parameter until the loss value of the second model parameter is converged to obtain a second model; the first device predicts a probability that the user clicks each of the different types of advertisements, respectively, based on the second model. The first equipment is service equipment of different platforms, and the second equipment is a server capable of performing longitudinal federal modeling, so that each platform obtains a partial model of the platform and performs click prediction on candidate advertisements of different classes. Browsing records, purchasing records and the like of a user are personal data, and circulation among different affiliated parties cannot be allowed, however, the user portrait dimension from a single platform is often single, and the user cannot be stereoscopically depicted from various aspects, so that inaccurate positioning of advertisements and pushing is easily caused, the experience of the user is reduced, and the loss of platform users is caused. The method provides possibility for cooperation of different platforms through a federal modeling method, so that training and learning can be performed jointly under the condition that platform data are mutually isolated, user data of different platforms can be better utilized, the delivery efficiency of platform advertisements is improved, advertisement expenses and social resources are saved, the user experience is improved, and loss of platform users is reduced.
The method of the embodiment of the present application is explained in detail above, and the following provides a click through rate CTR prediction apparatus related to the embodiment of the present application, which is applied to the first device, where the click through rate CTR prediction apparatus 30 may be a service device that provides various conveniences for a third party to use based on interactive data by quickly acquiring, processing, analyzing and extracting valuable data. Referring to fig. 5, fig. 5 is a schematic structural diagram of a click through rate CTR prediction apparatus according to an embodiment of the present disclosure. The click through rate CTR prediction apparatus 30 may include a first training unit 301, a first transmitting unit 302, a first receiving unit 303, a second training unit 304, and a prediction unit 305, and may further include a second transmitting unit 306, a second receiving unit 307, and a delivery unit 308.
A first training unit 301, configured to train a first model based on user data to obtain the trained first model and first model parameters, where the first device is any one of multiple first devices, the user data includes click information of whether a user clicks on multiple types of advertisements, and the first model is used to predict a probability that the user clicks on each type of advertisements in the multiple types of advertisements;
a first sending unit 302, configured to send the first model parameter to a second device;
a first receiving unit 303, configured to receive a second model parameter sent by the second device, where the second model parameter is determined based on first model parameters respectively corresponding to the multiple first devices;
a second training unit 304, configured to update the first model parameter of the first model to the second model parameter and train the first model parameter until a loss value of the second model parameter converges to obtain a second model;
a predicting unit 305, configured to predict, based on the second model, a probability that the user clicks each of the different types of advertisements, respectively.
In one possible implementation, the apparatus further includes: a second sending unit 306, configured to send a user identifier set to the second device, where the user identifier set includes multiple user identifiers, and each user identifier in the multiple user identifiers is used to identify a corresponding user and user data; a second receiving unit 307, configured to receive a target user identifier set sent by the second device, where the target user identifier set includes a user identifier of a target user, and the target user is a user common among the multiple first devices; and the first equipment acquires the user data of the target user according to the target user identification set.
In a possible implementation manner, the second sending unit 306 is specifically configured to hash and encrypt the user identifiers included in the user identifier set, and send the encrypted user identifier set to the second device.
In a possible implementation manner, the first sending unit 302 is specifically configured to send, to the second device, the first device after homomorphically encrypting the first model parameter.
In one possible implementation, the apparatus further includes: and the delivering unit 308 is configured to deliver the target type advertisement to the user in the first device when the probability that the user clicks the target type advertisement is greater than a preset threshold, where the target type advertisement belongs to one of the multiple types of advertisements.
It should be noted that implementation of each operation may also correspond to corresponding description of the method embodiments shown in fig. 2 to fig. 4, and details are not described here again.
As shown in fig. 6, fig. 6 is a schematic structural diagram of another click through rate CTR prediction apparatus provided in this embodiment, and is applied to a second device, where the apparatus 40 includes: the third receiving unit 401, the optimizing unit 402, and the third transmitting unit 403 may further include: a fourth receiving unit 404, a filtering unit 405 and a fourth transmitting unit 406.
A third receiving unit 401, configured to receive first model parameters sent by a plurality of first devices, where the first model parameters are model parameters obtained by the first devices through training a first model based on user data, the user data includes click information of whether a user clicks a plurality of types of advertisements, and the first model is used to predict a probability that the user clicks each type of advertisement among the plurality of types of advertisements;
an optimizing unit 402, configured to optimize first model parameters corresponding to the multiple first devices, and generate second model parameters;
a third sending unit 403, configured to send the second model parameters to the multiple first devices, respectively.
In one possible implementation, the apparatus further includes: a fourth receiving unit 404, configured to receive a user identifier set sent by each of the multiple first devices, where the user identifier set includes multiple user identifiers, and each of the multiple user identifiers is used to identify a corresponding user and user data; a screening unit 405, configured to screen out, from a user identifier set corresponding to multiple first devices, a user identifier of a common user among the multiple first devices, so as to obtain a target user identifier set; a fourth sending unit 406, configured to send the target user identity sets to the multiple first devices respectively.
It should be noted that implementation of each operation may also correspond to corresponding description of the method embodiments shown in fig. 2 to fig. 4, and details are not described here again.
As shown in fig. 7, fig. 7 is a schematic structural diagram of another click through rate CTR prediction apparatus provided in this embodiment, where the apparatus 50 is applied to a first device, and includes at least one processor 501, at least one memory 502, and at least one communication interface 503. In addition, the device may also include common components such as an antenna, which will not be described in detail herein.
The processor 501 may be a general purpose Central Processing Unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of programs according to the above schemes.
The Memory 502 may be, but is not limited to, a Read-Only Memory (ROM) or other type of static storage device that can store static information and instructions, a Random Access Memory (RAM) or other type of dynamic storage device that can store information and instructions, an Electrically Erasable Programmable Read-Only Memory (EEPROM), a Compact Disc Read-Only Memory (CD-ROM) or other optical Disc storage, optical Disc storage (including Compact Disc, laser Disc, optical Disc, digital versatile Disc, blu-ray Disc, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer. The memory may be self-contained and coupled to the processor via a bus. The memory may also be integral to the processor.
The memory 502 is used for storing application program codes for executing the above scheme, and is controlled by the processor 501 for execution. The processor 501 is used to execute application program code stored in the memory 502.
The code stored in the memory 502 may perform the click through rate CTR prediction method provided in fig. 2, for example, when the device 50 is a click through rate CTR prediction device, the first model may be trained based on user data, the user data includes click information indicating whether a user clicks on a plurality of types of advertisements, and the trained first model and first model parameters are obtained, the first model is used to predict a probability that the user clicks on each type of advertisements in the plurality of types of advertisements; sending the first model parameters to a second device; receiving second model parameters sent by the second device, wherein the second model parameters are determined based on first model parameters respectively corresponding to the plurality of first devices; updating the first model parameter of the first model into the second model parameter and training until the loss value of the second model parameter is converged to obtain a second model; predicting a probability that a user clicks each of the different types of advertisements, respectively, based on the second model.
It should be noted that, the functions of each functional unit in the click through rate CTR prediction apparatus described in the embodiment of the present application may refer to corresponding descriptions of the method embodiments shown in fig. 2 to fig. 4, and are not described again here.
As shown in fig. 8, fig. 8 is a schematic structural diagram of another click through rate CTR prediction apparatus provided in this embodiment, where the apparatus 60 is applied to a second device, and includes at least one processor 601, at least one memory 602, and at least one communication interface 603. In addition, the device may also include common components such as an antenna, which will not be described in detail herein.
The processor 601 may be a general purpose Central Processing Unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of programs according to the above schemes.
The Memory 602 may be a Read-Only Memory (ROM) or other type of static storage device that can store static information and instructions, a Random Access Memory (RAM) or other type of dynamic storage device that can store information and instructions, an Electrically Erasable Programmable Read-Only Memory (EEPROM), a Compact Disc Read-Only Memory (CD-ROM) or other optical Disc storage, optical Disc storage (including Compact Disc, laser Disc, optical Disc, digital versatile Disc, blu-ray Disc, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer, but is not limited to these. The memory may be self-contained and coupled to the processor via a bus. The memory may also be integral to the processor.
The memory 602 is used for storing application program codes for executing the above scheme, and the processor 601 controls the execution. The processor 601 is used to execute application program code stored in the memory 602.
The code stored in the memory 602 may perform the click through rate CTR prediction method provided in fig. 2, for example, when the apparatus 60 is a click through rate CTR prediction apparatus, the first model parameter sent by each of a plurality of first devices may be received, where the first model parameter is a model parameter obtained by the first device through training a first model based on user data, where the user data includes click information of whether a user clicks on a plurality of types of advertisements, and the first model is used to predict a probability that the user clicks on each type of advertisements in the plurality of types of advertisements; optimizing first model parameters corresponding to the plurality of first devices to generate second model parameters; and respectively sending the second model parameters to the plurality of first devices.
It should be noted that, the functions of each functional unit in the click through rate CTR prediction apparatus described in the embodiment of the present application may refer to corresponding descriptions of the method embodiments shown in fig. 2 to fig. 4, and are not described again here.
In this application, the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiments of the present application.
In addition, functional components in the embodiments of the present application may be integrated into one component, or each component may exist alone physically, or two or more components may be integrated into one component. The integrated components can be realized in a form of hardware or a form of software functional units.
The integrated components, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be substantially implemented as part of or contributed to by the prior art, or all or part of the technical solution may be embodied in a software product stored in a storage medium, and including instructions for causing a computer device (which may be a personal computer, a second device, or a network device) to perform all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.
While the invention has been described with reference to specific embodiments, the scope of the invention is not limited thereto, and those skilled in the art can easily conceive various equivalent modifications or substitutions within the technical scope of the invention. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
It should be understood that, in the various embodiments of the present application, the sequence numbers of the above-mentioned processes do not mean the execution sequence, and the execution sequence of each process should be determined by its function and inherent logic, and should not constitute any limitation to the implementation process of the embodiments of the present application. While the present application has been described herein in conjunction with various embodiments, other variations to the disclosed embodiments may be understood and effected by those skilled in the art in practicing the present application as claimed herein.
Claims (10)
1. A Click Through Rate (CTR) prediction method is applied to a first device and comprises the following steps:
the method comprises the steps that a first device trains a first model based on user data to obtain the trained first model and first model parameters, the first device is any one of a plurality of first devices, the user data comprises click information of whether a user clicks a plurality of types of advertisements, and the first model is used for predicting the probability that the user clicks each type of advertisements in the plurality of types of advertisements;
the first equipment sends the first model parameter to second equipment;
the first equipment receives second model parameters sent by the second equipment, wherein the second model parameters are determined based on first model parameters respectively corresponding to the plurality of first equipment;
the first equipment updates the first model parameter of the first model into the second model parameter and trains the first model parameter until the loss value of the second model parameter is converged to obtain a second model;
the first device predicts, based on the second model, a probability that a user clicks each of the different types of advertisements, respectively.
2. The method of claim 1, further comprising:
the first device sends a user identifier set to the second device, wherein the user identifier set comprises a plurality of user identifiers, and each user identifier in the plurality of user identifiers is used for identifying a corresponding user and user data respectively;
the first device receives a target user identification set sent by the second device, wherein the target user identification set comprises user identifications of target users, and the target users are users common among the plurality of first devices;
and the first equipment acquires the user data of the target user according to the target user identification set.
3. The method of claim 1, wherein the first device sends a set of user identities to the second device, comprising:
and the first equipment carries out hash encryption on the user identification included in the user identification set and sends the encrypted user identification set to the second equipment.
4. The method of claim 1, wherein the first device sends the first model parameters to a second device, comprising:
and the first equipment homomorphically encrypts the first model parameter and then sends the first model parameter to second equipment.
5. The method of claim 1, further comprising:
and when the probability that the user clicks the target type advertisement is larger than a preset threshold value, delivering the target type advertisement to the user in the first device, wherein the target type advertisement belongs to one of the multiple types of advertisements.
6. A Click Through Rate (CTR) prediction method applied to a second device comprises the following steps:
the second equipment receives first model parameters which are respectively sent by a plurality of first equipment, the first model parameters are model parameters obtained by the first equipment through training a first model based on user data, the user data comprise click information of whether a user clicks a plurality of types of advertisements, and the first model is used for predicting the probability that the user clicks each type of advertisements in the plurality of types of advertisements;
the second equipment optimizes the first model parameters corresponding to the plurality of first equipment to generate second model parameters;
and the second equipment respectively sends the second model parameters to the plurality of first equipment.
7. The method of claim 6, further comprising:
the second device receives a user identifier set respectively sent by the first devices, wherein the user identifier set comprises a plurality of user identifiers, and each user identifier in the user identifiers is respectively used for identifying a corresponding user and user data;
the second equipment screens out the user identification of the common user among the plurality of first equipment from the user identification sets corresponding to the plurality of first equipment to obtain a target user identification set;
and the second equipment respectively sends the target user identification sets to the plurality of first equipment.
8. The computer equipment is characterized by comprising a processing component, a storage component and a communication module component, wherein the processing component, the storage component and the communication module component are connected with each other, the storage component is used for storing a computer program, and the communication component is used for carrying out information interaction with external equipment; the processing component is configured for invoking a computer program for performing the method of any of claims 1-5.
9. The computer equipment is characterized by comprising a processing component, a storage component and a communication module component, wherein the processing component, the storage component and the communication module component are connected with each other, the storage component is used for storing a computer program, and the communication component is used for carrying out information interaction with external equipment; the processing component is configured to invoke a computer program to perform the method of any of claims 6-7.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program which is executed by a processor to implement the method of any of claims 1-5 or 6-7.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011393094.5A CN112446736A (en) | 2020-12-02 | 2020-12-02 | Click through rate CTR prediction method and device |
| PCT/CN2021/083530 WO2022116431A1 (en) | 2020-12-02 | 2021-03-29 | Click through rate (ctr) prediction method and apparatus |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011393094.5A CN112446736A (en) | 2020-12-02 | 2020-12-02 | Click through rate CTR prediction method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112446736A true CN112446736A (en) | 2021-03-05 |
Family
ID=74739284
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011393094.5A Pending CN112446736A (en) | 2020-12-02 | 2020-12-02 | Click through rate CTR prediction method and device |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN112446736A (en) |
| WO (1) | WO2022116431A1 (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113269232A (en) * | 2021-04-25 | 2021-08-17 | 北京沃东天骏信息技术有限公司 | Model training method, vectorization recall method, related device and storage medium |
| WO2022116431A1 (en) * | 2020-12-02 | 2022-06-09 | 平安科技（深圳）有限公司 | Click through rate (ctr) prediction method and apparatus |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116012066A (en) * | 2023-03-28 | 2023-04-25 | 江西时刻互动科技股份有限公司 | Advertisement conversion rate prediction method, device and readable storage medium |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170186030A1 (en) * | 2015-04-21 | 2017-06-29 | Tencent Technology (Shenzhen) Company Limited | Advertisement click-through rate correction method and advertisement push server |
| CN109446431A (en) * | 2018-12-10 | 2019-03-08 | 网易传媒科技（北京）有限公司 | For the method, apparatus of information recommendation, medium and calculate equipment |
| CN109472632A (en) * | 2018-09-25 | 2019-03-15 | 平安科技（深圳）有限公司 | Evaluate method, apparatus, medium and the electronic equipment of advertisement competition power |
| CN110110229A (en) * | 2019-04-25 | 2019-08-09 | 深圳前海微众银行股份有限公司 | A kind of information recommendation method and device |
| CN111008709A (en) * | 2020-03-10 | 2020-04-14 | 支付宝(杭州)信息技术有限公司 | Federal learning and data risk assessment method, device and system |
| CN111046294A (en) * | 2019-12-27 | 2020-04-21 | 支付宝(杭州)信息技术有限公司 | Click rate prediction method, recommendation method, model, device and equipment |
| CN111460511A (en) * | 2020-04-17 | 2020-07-28 | 支付宝(杭州)信息技术有限公司 | Federal learning and virtual object distribution method and device based on privacy protection |
| CN111553745A (en) * | 2020-05-08 | 2020-08-18 | 深圳前海微众银行股份有限公司 | Federal-based model updating method, device, equipment and computer storage medium |
| CN111967910A (en) * | 2020-08-18 | 2020-11-20 | 中国银行股份有限公司 | User passenger group classification method and device |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112446736A (en) * | 2020-12-02 | 2021-03-05 | 平安科技（深圳）有限公司 | Click through rate CTR prediction method and device |
-
2020
- 2020-12-02 CN CN202011393094.5A patent/CN112446736A/en active Pending
-
2021
- 2021-03-29 WO PCT/CN2021/083530 patent/WO2022116431A1/en active Application Filing
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170186030A1 (en) * | 2015-04-21 | 2017-06-29 | Tencent Technology (Shenzhen) Company Limited | Advertisement click-through rate correction method and advertisement push server |
| CN109472632A (en) * | 2018-09-25 | 2019-03-15 | 平安科技（深圳）有限公司 | Evaluate method, apparatus, medium and the electronic equipment of advertisement competition power |
| CN109446431A (en) * | 2018-12-10 | 2019-03-08 | 网易传媒科技（北京）有限公司 | For the method, apparatus of information recommendation, medium and calculate equipment |
| CN110110229A (en) * | 2019-04-25 | 2019-08-09 | 深圳前海微众银行股份有限公司 | A kind of information recommendation method and device |
| CN111046294A (en) * | 2019-12-27 | 2020-04-21 | 支付宝(杭州)信息技术有限公司 | Click rate prediction method, recommendation method, model, device and equipment |
| CN111008709A (en) * | 2020-03-10 | 2020-04-14 | 支付宝(杭州)信息技术有限公司 | Federal learning and data risk assessment method, device and system |
| CN111460511A (en) * | 2020-04-17 | 2020-07-28 | 支付宝(杭州)信息技术有限公司 | Federal learning and virtual object distribution method and device based on privacy protection |
| CN111553745A (en) * | 2020-05-08 | 2020-08-18 | 深圳前海微众银行股份有限公司 | Federal-based model updating method, device, equipment and computer storage medium |
| CN111967910A (en) * | 2020-08-18 | 2020-11-20 | 中国银行股份有限公司 | User passenger group classification method and device |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2022116431A1 (en) * | 2020-12-02 | 2022-06-09 | 平安科技（深圳）有限公司 | Click through rate (ctr) prediction method and apparatus |
| CN113269232A (en) * | 2021-04-25 | 2021-08-17 | 北京沃东天骏信息技术有限公司 | Model training method, vectorization recall method, related device and storage medium |
| CN113269232B (en) * | 2021-04-25 | 2023-12-08 | 北京沃东天骏信息技术有限公司 | Model training method, vectorization recall method, related equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2022116431A1 (en) | 2022-06-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9760723B2 (en) | Techniques for in-app user data authorization | |
| CN108280115B (en) | Method and device for identifying user relationship | |
| US9152477B1 (en) | System and method for communication among mobile applications | |
| CN103138954B (en) | A kind of method for pushing of recommendation items, system and recommendation server | |
| US8291018B2 (en) | Methods, apparatuses, and computer program products for providing activity coordination services | |
| CN112446736A (en) | Click through rate CTR prediction method and device | |
| US10129197B2 (en) | Computerized system and method for modifying a message to apply security features to the message's content | |
| AU2008245773B2 (en) | Behavioral advertisement targeting and creation of ad-hoc microcommunities through user authentication | |
| JP5571145B2 (en) | Advertisement distribution apparatus and advertisement distribution method | |
| CN108605008A (en) | E-mail server is acted on behalf of for route messages | |
| CN105530175A (en) | Message processing method, device and system | |
| CN109154940A (en) | Learn new words | |
| KR20150004350A (en) | Method and/or system for user authentication with targeted electronic advertising content through personal communication devices | |
| CN104335607A (en) | Systems and methods for identifying and suggesting emoticons | |
| CN103870553B (en) | A kind of input resource supplying method and system | |
| CN110785970B (en) | Techniques to automate robotic creation of web pages | |
| CN107124349A (en) | Information transferring method and device | |
| CN114448921A (en) | Information display method and device based on social relation chain and server | |
| CN106487655B (en) | Message interaction method and device and processing server | |
| WO2015149321A1 (en) | Personal digital engine for user empowerment and method to operate the same | |
| CN107258071A (en) | The method and system and recording medium of the abundant menu of official's account are provided in instant Communications service | |
| CN106101358A (en) | A kind of method of contact person information updating and smart machine | |
| CN106790350A (en) | A kind of information push-delivery apparatus, server and method | |
| CN108401005B (en) | Expression recommendation method and device | |
| CN111557014A (en) | Method and system for providing multiple personal data |
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 |