CN115409551A

CN115409551A - Event trigger probability prediction method, device and equipment

Info

Publication number: CN115409551A
Application number: CN202211037678.8A
Authority: CN
Inventors: 陈迪; 盛梦雪; 邵冬
Original assignee: Beijing Youzhuju Network Technology Co Ltd
Current assignee: Beijing Youzhuju Network Technology Co Ltd
Priority date: 2022-08-26
Filing date: 2022-08-26
Publication date: 2022-11-29

Abstract

The application discloses a method, a device and equipment for predicting event trigger probability, which are used for acquiring content characteristic data of target content, user characteristic data of a target user and characteristic data of a touch screen operator of the target user. And inputting the content characteristic data, the user characteristic data and the characteristic data of the touch screen operator into the event trigger probability prediction model to obtain the predicted event trigger probability of the target user on the target content. When the operators used by the target users for touching the screens are different, the event triggering probability of the target content is different. The operator used by the target user for touching the screen has a certain influence on the event triggering probability of the target content. Therefore, the event trigger probability prediction model mines the relation between the characteristic data of the touch screen operator of the target user and the content characteristic data of the target content, and the predicted event trigger probability of the target user on the target content, which is output by the model, is more accurate.

Description

Event trigger probability prediction method, device and equipment

Technical Field

The application relates to the technical field of internet, in particular to a method, a device and equipment for predicting event triggering probability.

Background

With the rapid development of online advertisement promotion technologies, advertisers are more and more concerned about the precise delivery of advertisements. The click-through rate of the advertisement is used to evaluate the user's preference for the advertisement. By predicting the click rate of an advertisement, accurate placement of the advertisement may be facilitated.

At present, a recommendation algorithm model can be used for predicting the click rate of a user on an advertisement, but the prediction accuracy of the existing recommendation algorithm model is low.

Disclosure of Invention

In view of this, embodiments of the present application provide a method, an apparatus, and a device for predicting event trigger probability, which can improve prediction accuracy.

In order to solve the above problem, the technical solution provided by the embodiment of the present application is as follows:

in a first aspect, an embodiment of the present application provides an event trigger probability prediction method, where the method includes:

acquiring content characteristic data of target content and user characteristic data of a target user;

acquiring characteristic data of a touch screen operator of the target user;

inputting the content characteristic data, the user characteristic data and the touch screen operator characteristic data into an event trigger probability prediction model, and acquiring the predicted event trigger probability of the target user on the target content; the event trigger probability prediction model is trained based on historical content feature data of historical content, historical user feature data of a historical user, historical touch screen operator feature data of the historical user and event trigger expected probability of the historical user on the historical content.

In a second aspect, an embodiment of the present application provides an event trigger probability prediction apparatus, where the apparatus includes:

a first acquisition unit configured to acquire content feature data of a target content and user feature data of a target user;

the second obtaining unit is used for obtaining characteristic data of a touch screen operator of the target user;

the input unit is used for inputting the content characteristic data, the user characteristic data and the touch screen operator characteristic data into an event trigger probability prediction model, and acquiring the predicted event trigger probability of the target user on the target content; the event trigger probability prediction model is trained based on historical content feature data of historical content, historical user feature data of a historical user, historical touch screen operator feature data of the historical user and event trigger expected probability of the historical user on the historical content.

In a third aspect, an embodiment of the present application provides an electronic device, including:

one or more processors;

a storage device having one or more programs stored thereon,

when executed by the one or more processors, cause the one or more processors to implement a method of event triggered probability prediction as described in any above.

In a fourth aspect, the present application provides a computer readable medium, on which a computer program is stored, wherein the program, when executed by a processor, implements the event trigger probability prediction method as described in any one of the above.

Therefore, the embodiment of the application has the following beneficial effects:

the embodiment of the application provides an event trigger probability prediction method, device and equipment, and content characteristic data of target content and user characteristic data of a target user are obtained. Meanwhile, the characteristic data of a touch screen operator of the target user are obtained. And inputting the content characteristic data, the user characteristic data and the characteristic data of the touch screen operator into the event trigger probability prediction model to obtain the predicted event trigger probability of the target user on the target content. The event triggering probability prediction model is obtained by training based on historical content characteristic data of historical content, historical user characteristic data of historical users, historical touch screen operator characteristic data of historical users and event triggering expectation probability of the historical users on the historical content. When the operators used by the target users for touching the screens are different, the event triggering probability of the target content is different. The operator used by the target user for touching the screen has a certain influence on the event triggering probability of the target content. Therefore, the event trigger probability prediction model provided by the embodiment of the application excavates the relationship between the characteristic data of the touch screen operator of the target user and the content characteristic data of the target content, and the predicted event trigger probability of the target user on the target content, which is output by the model, can be more accurate.

Drawings

Fig. 1 is a schematic diagram of a framework of an exemplary application scenario provided in an embodiment of the present application;

fig. 2 is a flowchart of a method for predicting event trigger probability according to an embodiment of the present disclosure;

FIG. 3 is a schematic diagram of an event triggered probability prediction model according to an embodiment of the present disclosure;

fig. 4 is a schematic structural diagram of an event trigger probability prediction apparatus according to an embodiment of the present disclosure;

fig. 5 is a schematic diagram of a basic structure of an electronic device according to an embodiment of the present application.

Detailed Description

In order to make the aforementioned objects, features and advantages of the present application more comprehensible, embodiments accompanying figures and detailed description thereof are described in further detail below.

In order to facilitate understanding and explaining the technical solutions provided by the embodiments of the present application, the following description will first describe the background art of the present application.

With the rapid development of online advertising promotion technology, the number of creative advertisements is in the order of billions, and the number of users reaches the level of billions. Advertisers are more and more concerned about the precise placement of advertisements, and hopefully recommend advertisements in which the advertisers are most interested for each user, so that the advertisement experience of the users is improved, and the conversion rate of the advertisements is improved. In practical applications, the click-through rate of the advertisement is used to evaluate the user's preference for the advertisement. By predicting the click rate of the advertisement, accurate placement of the advertisement may be facilitated.

Based on this, the embodiment of the application provides an event trigger probability prediction method, device and equipment, and content feature data of target content and user feature data of a target user are obtained. Meanwhile, the characteristic data of a touch screen operator of the target user are obtained. And inputting the content characteristic data, the user characteristic data and the characteristic data of the touch screen operator into the event trigger probability prediction model to obtain the predicted event trigger probability of the target user on the target content. The event triggering probability prediction model is obtained by training based on historical content characteristic data of historical content, historical user characteristic data of historical users, historical touch screen operator characteristic data of historical users and event triggering expectation probability of the historical users on the historical content. When the operators used by the target users for touching the screens are different, the event triggering probabilities of the target contents are different. The operator used by the target user for touching the screen has a certain influence on the event triggering probability of the target content. Therefore, the event trigger probability prediction model provided by the embodiment of the application excavates the relation between the characteristic data of the touch screen operator of the target user and the content characteristic data of the target content, and the predicted event trigger probability of the target user on the target content output by the model can be more accurate.

In order to facilitate understanding of the event trigger probability prediction method provided in the embodiment of the present application, the following description is made with reference to a scenario example shown in fig. 1. Referring to fig. 1, the drawing is a schematic diagram of a framework of an exemplary application scenario provided in an embodiment of the present application.

As shown in fig. 1, an implementation environment of the event trigger probability prediction method provided in the embodiment of the present application includes a terminal device 100, a network 200, and a server 300. The terminal device 100 is a user terminal or a client terminal, and the server 300 is a server terminal. The implementation environment may include any number of terminal devices 100, networks 200, and servers 300, as may be practical. The terminal device 100 includes, but is not limited to, a smart phone, a tablet computer, a notebook computer, a smart watch, and the like. The server 300 may be an independent physical server, a server cluster or a distributed system formed by a plurality of physical servers, or a cloud server providing basic cloud computing services such as cloud service, a cloud database, cloud computing, cloud storage, big data, an artificial intelligence platform, and the like.

The terminal device 100 and the server 300 can communicate with each other via the network 200. Network 200 may comprise various types of wired or wireless communication links. For example: the wired communication link comprises an optical fiber, a twisted pair wire or a coaxial cable, and the wireless communication link comprises a Bluetooth communication link, a wireless fidelity Wi-Fi communication link or a microwave communication link.

The terminal device 100 may interact with the server 300 through the network 200 to receive messages from the server 300 or to transmit messages and data to the server 300. Various communication client applications may be installed on the terminal device 100. For example: video playing applications, voice interaction applications, search-type applications, instant messaging tools, mailbox clients, social platform software, and the like.

It should be noted that, in some embodiments, the event triggering probability prediction method and the training method of the event triggering probability prediction model provided in the embodiments of the present application are generally performed by the server 300, and accordingly, the event triggering probability prediction model and the event triggering probability prediction apparatus are generally disposed in the server 300. In addition, in other embodiments, the terminal device 100 may also perform an event trigger probability prediction method and a training method of the event trigger probability prediction model.

In one or more embodiments, the server 300 obtains content characteristic data of the target content and user characteristic data of the target user transmitted by the terminal device 100, and obtains touch screen operator characteristic data of the target user from the terminal device 100. The server 300 inputs the content characteristic data, the user characteristic data and the touch screen operator characteristic data into the event triggering probability prediction model to obtain the predicted event triggering probability of the target user on the target content. The event trigger probability prediction model is trained by the server 300 based on historical content feature data of historical content, historical user feature data of historical users, historical touch screen operator feature data of the historical users and event trigger expected probability of the historical users on the historical content.

Those skilled in the art will appreciate that the block diagram shown in fig. 1 is only one example in which embodiments of the present application may be implemented. The scope of applicability of the embodiments of the present application is not limited in any way by this framework.

In order to facilitate understanding of the present application, a method for predicting event triggering probability provided by the embodiments of the present application is described below with reference to the accompanying drawings.

Referring to fig. 2, the figure is a flowchart of an event trigger probability prediction method according to an embodiment of the present application. The method is applicable to the server 300 of the above embodiment. As shown in fig. 2, the method may include S201-S203:

s201: and acquiring content characteristic data of the target content and user characteristic data of the target user.

The target content is content displayed in the terminal equipment. In some embodiments, the targeted content may be targeted advertisements. In other embodiments, the target content may also be multimedia content, such as non-advertising video, audio, etc., and the target content is not limited in this embodiment.

The content data corresponding to the target content is acquired first, and then the content characteristic data of the target content is acquired based on the content data of the target content. The content data is, for example, video data, audio data, advertisement data, or the like. The content characteristic data is a further representation of the content data on the basis of the content data. The content feature data may be represented by a content feature vector, a content feature array, or an integer value. The number of the target contents is not limited in the embodiment of the application, and when the target contents are the target advertisements, the number of the target advertisements may be one or more, but the contents of the advertisements are different.

The target user is a user who views the target content in the terminal device. The user characteristic data of the target user is obtained based on the user data of the target user after the user data of the target user is obtained. The user data is data related to a target user. The user characteristic data is a further representation of the user data on the basis of the user data. The user characteristic data can be represented by a user characteristic vector, a user characteristic array or an integer value. The number of the target users is not limited in the embodiment of the application, and the number of the target users can be one or more, when the target users are different.

In this step, the obtained content feature data of the target content and the user feature data of the target user may be used as input data of a subsequent event trigger probability prediction model.

In embodiments of the present application, the target user may implement an event trigger on the target content. Event triggers may be understood exemplarily as actions that implement certain specifics, such as clicking, forwarding, and/or likes. When the target user realizes a specific behavior aiming at the target content, the target user is indicated to implement event trigger on the target content. That is, if the target content is the target video and the target user has forwarded the target video, it is determined that the target user has implemented an event trigger of a forwarding behavior type for the target video. And if the target content is the target advertisement and the target user clicks the target advertisement, determining that the target user implements event triggering of the clicking behavior type on the target advertisement. It is to be understood that the embodiments of the present application do not limit the category and the number of event triggers.

It should be noted that, in the embodiment of the present application, the content data corresponding to the target content, the content feature data of the target content, the user data of the target user, and the user feature data of the target user do not relate to sensitive information of the user, and the content data corresponding to the target content, the content feature data of the target content, the user data of the target user, and the user feature data of the target user are obtained and used after being authorized by the user. In one example, before the content data corresponding to the target content, the content feature data of the target content, the user data of the target user and the user feature data of the target user are obtained, the corresponding interface displays prompt information related to the authorization of the use of the obtained data, and the user determines whether to approve the authorization based on the prompt information.

It is understood that, before the technical solutions disclosed in the embodiments of the present disclosure are used, the user should be informed of the type, the use range, the use scene, etc. of the personal information related to the present disclosure in a proper manner according to the relevant laws and regulations and obtain the authorization of the user.

For example, when user characteristic data of a target user is acquired, prompt information is sent to the user to explicitly prompt the user that the operation requested to be performed will require acquisition and use of the user characteristic data to the user. Thus, the user can autonomously select whether to provide personal user data to software or hardware such as an electronic device, an application program, a server, or a storage medium that performs the operations of the disclosed technical solution, according to the prompt information.

As an optional but non-limiting implementation manner, when the user characteristic data of the target user is obtained, the manner of sending the prompt information to the user may be, for example, a pop-up window manner, and the prompt information may be presented in a text manner in the pop-up window. In addition, a selection control for providing user data to the electronic device by the user selecting "agree" or "disagree" can be carried in the popup.

It is understood that the above notification and user authorization process is only illustrative and not limiting, and other ways of satisfying relevant laws and regulations may be applied to the implementation of the present disclosure.

S202: and acquiring characteristic data of a touch screen operator of a target user.

In practical application, when a target user touches a screen by using different operators, different event trigger probabilities are provided for target content. For example, some targeted content facilitates left-handed implementation of event triggers, and other targeted content facilitates right-handed implementation of event triggers. On the basis, the event trigger probability prediction model constructed in the embodiment of the application can mine the relation between the characteristic data of the touch screen operator of the user and the content characteristic data.

Firstly, acquiring characteristic data of a touch screen operator of a target user, wherein the characteristic data of the touch screen operator of the target user is input data of an event trigger probability prediction model.

Specifically, the terminal device in the embodiment of the present application is a touch screen terminal device. The target user can realize touch screen input through a touch screen of the terminal equipment. The touch screen manipulator of the target user may be exemplarily understood as a manipulator for realizing touch screen input by the target user. For example, when the target user implements the touch screen input with only the left hand, the operator of the target user is the left hand. When the target user only adopts the right hand to realize the touch screen input, the operating hand of the target user is the right hand. When the target user adopts the left hand to realize the touch screen input and adopts the right hand to realize the touch screen input, the operating hand of the target user is the left hand and the right hand.

The terminal equipment acquires the characteristic data of the touch screen operator of the target user and sends the characteristic data of the touch screen operator of the target user to the server. The characteristic data of the touch screen operator of the target user is data related to the touch screen operator used by the target user, and the specific content of the characteristic data of the touch screen operator is not limited in the embodiment of the application.

For example, when the left hand is represented by the symbol 1, the right hand is represented by the symbol 2, and the left and right hands are represented by the symbol 3, and it is recognized that the hand used by the target user is the left hand, the touch screen hand feature data is feature data obtained based on the symbols. The characteristic data is a re-representation result of the identification. The re-representation may be a vector representation, an array representation or an integer value representation, whereby the obtained feature data may be a feature vector, a feature array or an integer value. Therefore, the obtained feature data are represented to facilitate subsequent data processing. In another example, the use probabilities of the operators when the target user touches the screen can be obtained, and the use probabilities of the operators are used as the characteristic data of the touch-screen operators of the target user. For example, data processing may be performed based on the usage probability of each manipulator, and data obtained after the data processing may be used as the touch screen manipulator feature data of the target user. It can be known that the feature data in the above three examples are three types of feature data of different data contents, which can be used as the feature data of the touch screen operator of the target user.

It will be appreciated that the data content of the historical user's historical touch screen manipulator feature data required to satisfy the target user's touch screen manipulator feature data and the training event trigger probability prediction model is of the same type. The type of the data content of the characteristic data of the touch screen operator of the target user can be determined according to the type of the data content of the characteristic data of the historical touch screen operator of the historical user, so that the input data of the input event trigger probability prediction model are ensured to be the data of the same data content type.

It can be further understood that, in the embodiment of the present application, the content characteristic data of the target content and the user characteristic data of the target user are non-real-time data, and the characteristic data of the touch screen operator of the target user is real-time data.

Based on the above, in a possible implementation manner, the embodiment of the present application provides a specific implementation manner for obtaining characteristic data of a touch screen operator of a target user, and the specific implementation manner is referred to as the following A1-A3.

For example, when acquiring the touch screen operator characteristic data of the target user, prompt information is sent to the user to explicitly prompt the user that the operation requested to be performed by the user will need to acquire and use the touch screen operator characteristic data of the target user. Therefore, the user can autonomously select whether to provide the data related to the touch screen operator for software or hardware such as an electronic device, an application program, a server or a storage medium which executes the operation of the technical scheme of the disclosure according to the prompt information.

As an optional but non-limiting implementation manner, when obtaining the characteristic data of the touch screen operator of the target user, the manner of sending the prompt information to the user may be, for example, a pop-up window manner, where the prompt information may be presented in a text manner. In addition, a selection control for providing user data to the electronic device by the user selecting "agree" or "disagree" can be carried in the popup.

S203: inputting content characteristic data, user characteristic data and touch screen operator characteristic data into an event trigger probability prediction model, and acquiring the predicted event trigger probability of a target user on target content; the event trigger probability prediction model is trained based on historical content feature data of historical content, historical user feature data of historical users, historical touch screen operator feature data of the historical users and event trigger expected probability of the historical users on the historical content.

After the content characteristic data, the user characteristic data and the touch screen operator characteristic data are obtained, input data consisting of the content characteristic data, the user characteristic data and the touch screen operator characteristic data are input into a trained event trigger probability prediction model, and the predicted event trigger probability of the target user on the target content, which is output by the event trigger probability prediction model, is obtained. It can be known that, when the number of the target users is multiple and the number of the target contents is also multiple, the event triggering probability prediction model can output the predicted event triggering probability of each user for each target content.

The predicted event triggering probability is a predicted value of the event triggering probability of the target user to the target content. When the event trigger is click, the predicted event trigger probability is specifically predicted click probability (referred to as predicted click rate). The predicted click probability is a predicted value of the click probability and is used for predicting the probability of the target user clicking the target content. When the event trigger is praise, the predicted event trigger probability is specifically the predicted praise probability. The prediction praise probability is a prediction value of the praise probability and is used for predicting the probability that the target user praise the target content. When the event trigger is forwarding, the predicted event trigger probability is specifically a predicted forwarding probability. The predicted forwarding probability is a predicted value of the forwarding probability and is used for predicting the probability of the target user for forwarding the target content.

In a possible implementation manner, the embodiment of the present application provides a specific implementation manner for inputting content feature data, user feature data, and touch-screen operator feature data into an event trigger probability prediction model to obtain a predicted event trigger probability of a target user for target content, which is specifically referred to as following B1 to B4.

It should be noted that the event trigger probability prediction model in this step is obtained by training based on the historical content feature data of the historical content, the historical user feature data of the historical user, the historical touch screen operator feature data of the historical user, and the event trigger expectation probability of the historical user for the historical content, which is specifically referred to as D1 to D5 below.

It can be understood that the event triggering probability prediction model mines the relationship between the user touch-screen operator characteristic data and the content characteristic data, so that the input data of the event triggering probability prediction model in the embodiment of the present application includes not only the content characteristic data of the target content and the user characteristic data of the target user, but also the touch-screen operator characteristic data of the target user. Because the operator used by the target user for touching the screen has a certain influence on the event triggering probability of the target content, the predicted event triggering probability of the target user for the target content, which is output by the event triggering probability prediction model, is more accurate on the basis of considering the characteristic data of the user's touch screen operator.

Based on the contents of S201 to S203, the embodiment of the present application provides an event trigger probability prediction method, which obtains content feature data of target content and user feature data of a target user. Meanwhile, the characteristic data of the touch screen operator of the target user is obtained. And inputting the content characteristic data, the user characteristic data and the characteristic data of the touch screen operator into an event trigger probability prediction model to obtain the predicted event trigger probability of the target user on the target content. The event triggering probability prediction model is trained based on historical content feature data of historical content, historical user feature data of a historical user, historical touch screen operator feature data of the historical user and event triggering expectation probability of the historical user on the historical content. When the operators used by the target users for touching the screens are different, the event triggering probabilities of the target contents are different. The operator used by the target user for touching the screen has a certain influence on the event triggering probability of the target content. Therefore, the event trigger probability prediction model provided by the embodiment of the application excavates the relationship between the characteristic data of the touch screen operator of the target user and the content characteristic data of the target content, and the predicted event trigger probability of the target user on the target content, which is output by the model, can be more accurate.

In a possible implementation manner, an embodiment of the present application provides a specific implementation manner of acquiring characteristic data of a touch screen operator of a target user in S202, including:

a1: and acquiring touch screen operation information of a target user.

When a target user uses an operator to perform touch screen input, a touch screen operation track is left on a touch screen of the terminal device. The touch screen operation track is composed of a plurality of touch screen points triggered on the touch screen by the target user, and each touch screen point has a corresponding coordinate on the touch screen. The terminal device can acquire the touch screen operation information of the target user and send the touch screen operation information of the target user to the server for subsequent data processing.

As an optional example, the terminal device may obtain touch screen operation information of the target user based on a certain collection frequency. For example, every 10s or 20 s. When the interval time is short enough, the terminal device is considered to collect the touch screen operation information of the target user in real time.

It should be noted that, in the embodiment of the present application, the touch screen operation information of the target user and the touch screen operator characteristic data of the target user do not relate to sensitive information of the user, and the touch screen operation information of the target user and the touch screen operator characteristic data of the target user are obtained and used after being authorized by the user. In one example, before acquiring the touch screen operation information of the target user and the touch screen operator characteristic data of the target user, the corresponding interface displays prompt information related to data use authorization, and the user determines whether to approve the authorization based on the prompt information.

A2: and inputting touch screen operation information of the target user into the trained left-hand and right-hand feature extraction network to obtain the left-hand operation probability, the right-hand operation probability and the left-hand and right-hand operation probability of the target user.

The left-hand and right-hand characteristic extraction network is used for inputting touch screen operation information of a target user and outputting left-hand operation probability, right-hand operation probability and left-hand and right-hand operation probability of the target user. For example, the target user has a left-hand operation probability of 0.3560, a right-hand operation probability of 0.4590, and a left-hand operation probability of 0.1850. From the probability results, it is found that the probability that the operator used by the target user is the right hand is the greatest. In the embodiment of the application, the left-hand and right-hand feature extraction network is a trained network model.

When the left-hand and right-hand feature extraction network is trained, the parameters of the left-hand and right-hand feature extraction network can be adjusted by adopting historical touch screen operation information of a historical user and actual operation hand labels of the historical user. And when the training stopping condition is reached, determining the parameters of the optimal left-hand and right-hand feature extraction network, and acquiring the left-hand and right-hand feature extraction network after training. It can be understood that, in the embodiment of the present application, the specific structure of the left-right hand feature extraction network and the specific content of the training stopping condition are not limited, and may be determined according to actual needs.

A3: and acquiring characteristic data of a touch screen operator of the target user according to the left-hand operation probability, the right-hand operation probability and the left-hand operation probability.

As an alternative example, the left-hand operation probability, the right-hand operation probability, and the left-hand operation probability may be directly used as the touch screen operator characteristic data of the target user. For example, the left hand operation probability, the right hand operation probability, and the left and right hand operation probability are expressed in the form of a vector, such as (0.3560, 0.4590, 0.1850), which is taken as the touch-screen operator hand characteristic data of the target user.

In another possible implementation manner, data processing may be performed based on the usage probability of each manipulator, and data obtained after the data processing is used as the touch screen manipulator feature data of the target user. Based on this, the present application provides another specific implementation manner of A3, including:

a301: and respectively converting the left hand operation probability, the right hand operation probability and the left hand operation probability into a first integer value, a second integer value and a third integer value.

After the left hand operation probability, the right hand operation probability and the left hand and right hand operation probability are obtained, a first integer value is obtained based on the left hand operation probability, a second integer value is obtained based on the right hand operation probability, and a third integer value is obtained based on the left hand and right hand operation probability.

The left hand operation probability is represented as p1, the right hand operation probability is represented as p2, and the left hand operation probability and the right hand operation probability are represented as p 3. p1, p2, p3 may be referred to as the original probabilities. As an optional example, the decimal point in the left-hand operation probability is shifted backward by a first preset position and then rounded to obtain a first integer value, the decimal point in the right-hand operation probability is shifted backward by a second preset position and then rounded to obtain a second integer value, and the decimal point in the left-hand operation probability is shifted backward by a third preset position and then rounded to obtain a third integer value. The specific number of bits of the first preset bit, the second preset bit and the third preset bit is not limited. It will be appreciated that the number of bits that are shifted back by a decimal point represents the number of bits that remain for the decimal point in the original probability.

For example, the first preset bit, the second preset bit, and the third preset bit are all 3 bits, which is equivalent to multiplying by 1000 on the basis of the original probability. And after rounding, the decimal place in the original probability is reserved with three bits. That is, the first, second, and third integer values are int (p 1 × 1000), int (p 2 × 1000), and int (p 3 × 1000), respectively. In a specific example, when p1 is 0.3561, p1 × 1000 is 356.1, and 356 is obtained after rounding, that is, decimal places "3", "5" and "6" in p1 are retained, and an integer value 356 is obtained.

It can be understood that, the operation of moving back and rounding the decimal point of the original probability can realize that the original probability can be changed into an integer value while the information of the original probability is maximally retained.

A302: and taking a target integer value obtained by calculation based on the first integer value, the second integer value and the third integer value as the characteristic data of the touch screen operator of the target user.

In one possible implementation, the first integer value is multiplied by the first digit value to obtain the second value. The second integer value is added to the second value to obtain a third value. And multiplying the third value by the second bit value to obtain a fourth value. The third integer value is added to the fourth value to obtain a fifth value. And taking the fifth value as the characteristic data of the touch screen operator of the target user.

As an alternative example, the calculation formula of the characteristic data of the touch screen operator of the target user is as follows:

feature_id＝(int(p1*1000)*1000+int(p2*1000))*1000+int(p3*1000)

wherein id represents a target user, feature _ id is touch screen operator characteristic data of the target user, and int is a rounding function. In this equation, the first and second bitvalues are both 1000. The second value is int (p 1 × 1000) × 1000, the third value is int (p 1 × 1000) × 1000+ int (p 2 × 1000), the fourth value is (int (p 1 × 1000) × 1000+ int (p 2 × 1000))) 1000, and the fifth value is feature _ id.

For example, when p1=0.3560, p2=0.4590, and p3=0.1850, the calculation result of the above formula is (356000 + 459) 1000+185=356459185. The target user's touch-screen operator profile data may be represented by 356459185, with the profile data being represented by only one 64-bit integer (i.e., 4 bytes).

It will be appreciated that the left hand, right hand, and left hand operational probabilities are typically floating point numbers, i.e., float type data. When data transmission is carried out, information is easily lost by floating point numbers, so that transmitted data is inaccurate, the number of bytes occupied by the floating point numbers is large, and transmission delay is high due to large bandwidth pressure during transmission. Compared with the first example in which the left-hand operation probability, the right-hand operation probability and the left-hand operation probability are directly used as the characteristic data of the touch screen operator of the target user, a specific method for representing the characteristic data of the touch screen operator is designed in a mode of A301-A302, the characteristic data of the touch screen operator can be represented in a compressed mode by using simple bytes, information loss is avoided, the number of bytes can be reduced as far as possible, and data can be transmitted on line conveniently. In addition, when the server model is estimated and transmitted, only integer and Boolean types can be processed due to the limitation of bandwidth and delay, and the characteristic data of the touch screen operator obtained in the A301-A302 mode is an integer, so that the type requirement of data processing can be met.

In practical application, the steps A1 to A3 may be executed by the terminal device, and after the characteristic data of the touch screen operator of the target user is obtained, the characteristic data of the touch screen operator of the target user is sent to the server.

As an alternative example, the event triggering probability prediction model comprises a multilayer perceptron module and a linear regression module, wherein the multilayer perceptron module is a multilayer neural network. As another alternative example, the event triggered probability prediction model may also employ the Wide & deep model or a related model thereof.

Referring to fig. 3, fig. 3 is a schematic diagram of an event triggered probability prediction model according to an embodiment of the present application. In a possible implementation manner, when the event trigger probability prediction model includes a multi-layer perceptron module and a linear regression module, and with reference to fig. 3, an embodiment of the present application provides a specific implementation manner for inputting content feature data, user feature data, and touch-screen operator feature data into the event trigger probability prediction model in S203 to obtain a predicted event trigger probability of a target user for a target content, including:

b1: and converting input feature data consisting of the content feature data, the user feature data and the touch screen operator feature data into splicing expression of the feature vector and the offset vector.

The acquired content characteristic data, user characteristic data and touch screen operator characteristic data may constitute input characteristic data. Further, the input feature data is subjected to vector representation, and the vector representation can be formed by splicing feature vectors and offset vectors. For example, the vector may be represented by an N-dimensional feature vector and an M-dimensional bias vector.

In specific implementation, the content feature data can be represented by using a vector to obtain a content feature vector, and then the content feature vector is spliced and represented by using a first feature vector and a first offset vector. And representing the user characteristic data by using a vector to obtain a user characteristic vector, and splicing and representing the user characteristic data by using a second characteristic vector and a second offset vector. And representing the characteristic data of the touch screen operator by using vectors to obtain characteristic vectors of the touch screen operator, and splicing and representing the characteristic vectors of the touch screen operator by using a third characteristic vector and a third offset vector. And finally, splicing the first eigenvector, the second eigenvector and the third eigenvector into eigenvectors, and splicing the first offset vector, the second offset vector and the third offset vector into offset vectors. For example, the first feature vector, the second feature vector, and the third feature vector are 16-dimensional vectors, and the first offset vector, the second offset vector, and the third offset vector are 1-dimensional vectors.

B2: and inputting the feature vector into a multilayer perceptron module, and acquiring a first prediction event triggering probability of the target user on the target content.

And inputting the obtained feature vector into the multilayer perceptron module, and acquiring a first prediction event triggering probability of the target user on the target content output by the multilayer perceptron module.

It will be appreciated that the multi-layered perceptron module may handle non-linear problems for handling prediction and classification tasks. In this step, the multi-layer perceptron module is configured to estimate an event trigger probability for the target content for the target user to obtain a predicted event trigger probability, referred to as a first predicted event trigger probability.

B3: and inputting the bias vector into a linear regression module, and acquiring a second prediction event triggering probability of the target user on the target content.

And inputting the obtained bias vector into a linear regression module, and obtaining a second prediction event trigger probability of the target user on the target content, which is output by the linear regression module.

It is understood that a linear regression module is used to handle the prediction and regression tasks. In this step, the linear regression module is configured to estimate an event trigger probability of the target user on the target content to obtain a predicted event trigger probability, which is referred to as a second predicted event trigger probability.

B4: and determining the average value of the first prediction event triggering probability and the second prediction event triggering probability as the prediction event triggering probability of the target user to the target content.

After the first predicted event triggering probability and the second predicted event triggering probability are obtained, the average value of the first predicted event triggering probability and the second predicted event triggering probability is determined as the final predicted event triggering probability of the target user to the target content.

Based on the contents of B1-B4, in this embodiment, not only the multi-layer perceptron module but also the linear regression module are used to predict the event triggering probability. After the first prediction event triggering probability and the second prediction event triggering probability are obtained, the average value of the first prediction event triggering probability and the second prediction event triggering probability is determined as the prediction event triggering probability of the target user to the target content. Therefore, probability prediction is carried out by utilizing the multiple models, the final predicted event triggering probability is determined based on the probabilities output by the multiple models, and the accuracy of the predicted event triggering probability can be improved.

In practical application, after the predicted event triggering probability of the target user on the target content is obtained, the predicted event triggering probability can be utilized for subsequent processing. For example, a single event-triggered final bid for the targeted content may be obtained according to the predicted event-triggered probability. The final bid price triggered by a single event can be understood as the price of the owner of the target content when the target content is triggered by a single event.

For example, when the target content is the target advertisement and the predicted event trigger probability is the predicted click rate, after obtaining the predicted click rate of the target user on the target advertisement, the advertiser may calculate the final bid for a single click on the target advertisement by using the predicted click rate. The single click final bid is the price that the advertiser has placed for the targeted advertisement when the targeted advertisement is clicked once. Wherein, the advertiser can be understood as a merchant for promoting the target advertisement and is the owner of the target advertisement.

In specific implementation, the predicted click rate is recorded as pctr, the single click bid provided by the advertiser is recorded as bid, and the calculation formula of the current single click final bid is ecpm = bid × pctr. That is, when bid is 10-tuple and pctr is 0.1, the single click final bid ecpm = bid × pctr = 1-tuple. In addition, considering the exposure times of the target advertisement, the total click bid ecpm' = bid pctr exposure times given by the advertiser for the target advertisement can be obtained.

It will be appreciated that when there are multiple targeted advertisements, a single click final bid can be placed on each targeted advertisement by the targeted user. The promotion sequence of a plurality of target advertisements can be obtained according to the final bid size of a single click. The larger the single click final bid, the more advanced the recommendation order of the targeted advertisement. For example, the first 30 targeted advertisements in the top order may be obtained.

It is also understood that the target content is not limited to the target advertisement, the event trigger probability is not limited to the predicted click-through rate, and the single event trigger final bid of the target content is not limited to the single click final bid provided for the target advertisement, and may be defined according to actual situations.

Therefore, the calculation mode of the single click final bid does not consider the operator of the target user, so that the accuracy of the calculated single click final bid is low. Based on this, in a possible implementation manner, the embodiment of the present application provides a specific implementation manner of obtaining a single event-triggered final bid for target content, including:

c1: acquiring a target manipulator corresponding to target content; the target manipulator is the manipulator corresponding to the event trigger statistic probability of the target content when the target manipulator is the event trigger statistic probability is maximum; the target manipulator is a left-hand, right-hand, left-hand or right-hand manipulator or other type manipulator.

And the target manipulator is the manipulator corresponding to the event trigger with the maximum statistical probability of the target content. In distinction from the predicted event trigger probability in S203, the event trigger statistical probability in this step is a probability obtained based on a statistical method. The target manipulator may be a left-handed, right-handed, left-handed (i.e., two-handed), or other type of manipulator. In the embodiment of the application, when the user does not authorize the terminal device to capture the touch screen operation information, the operation hands used when the event triggering is performed on the target content are classified into other types of operation hands.

In a possible implementation manner, an embodiment of the present application provides a specific implementation manner of C1, including:

c101: and acquiring the left hand operation event triggering times, the right hand operation event triggering times, the left hand operation event triggering times and the right hand operation event triggering times in the event triggering historical times aiming at the target content, and other types of operation hand operation event triggering times.

After the target content is uploaded to the online, the terminal equipment can acquire and store the event triggering times aiming at the target content in real time. And acquiring the trigger times of left-hand operation events, the trigger times of right-hand operation events, the trigger times of left-hand operation events and the trigger times of other types of operation events from the historical event trigger times aiming at the target content. The event triggering history times for the target content are the times triggered by the event from the time point of uploading the target content to the current time point.

The left-hand operation event triggering times are times of event triggering of target content by a left hand of a user, the right-hand operation event triggering times are similar to the left-hand operation event triggering times and the right-hand operation event triggering times, and the other types of operation event triggering times are times of event triggering of target content by the user who is not authorized to capture touch screen operation information. It is understood that, in this step, the users who perform the event trigger on the target content may be different users, and the number of the users may be one or more.

C102: and respectively determining the quotient of the trigger times of the left-hand operation events and the exposure times of the target content, the quotient of the trigger times of the right-hand operation events and the exposure times of the target content, the quotient of the trigger times of the left-hand operation events and the right-hand operation events and the exposure times of the target content, and the quotient of the trigger times of the other types of operation events and the exposure times of the target content as a first statistical probability, a second statistical probability, a third statistical probability and a fourth statistical probability corresponding to the target content.

The number of exposures of the targeted content is the number of times the targeted content is uploaded or sent on-line, after which the targeted content may or may not be triggered by a user-implemented event.

The first statistical probability is the trigger statistical probability of the left-hand operation event, and the corresponding operator is the left hand. The second statistical probability is the right-hand operation event trigger statistical probability, and the corresponding operator is the right hand. The third statistical probability is the left-hand and right-hand operation event trigger statistical probability, and the corresponding operators are left hands and right hands. The fourth statistical probability is the trigger statistical probability of the operation events of the other types of operators, and the corresponding operators are the operators of the other types.

C103: and determining the manipulator corresponding to the maximum value of the first statistical probability, the second statistical probability, the third statistical probability and the fourth statistical probability as a target manipulator corresponding to the target content.

After the first statistical probability, the second statistical probability, the third statistical probability and the fourth statistical probability are obtained, the manipulator corresponding to the maximum value of the four statistical probability values is determined as the target manipulator corresponding to the target content. For example, if the first statistical probability is the maximum, the target manipulator corresponding to the target content is the left hand.

After the target manipulator corresponding to the target content is determined, the type corresponding to the target content can be determined. The type corresponding to the target content is a left-hand type, a right-hand type, a left-hand type or a right-hand type or other types of manipulator types. For example, when the target manipulator corresponding to the target content is a left hand, the type corresponding to the target content is a left-hand type, and the remaining types are similar and will not be described herein again.

Based on the contents of C101-C103, the corresponding target manipulator can be determined when the event trigger statistical probability of the target content is maximum. According to the target manipulator corresponding to the target content, the type corresponding to the target content can be determined. In addition, when a plurality of target contents exist, target operators corresponding to different target contents can be acquired. According to the target operators corresponding to different target contents, the type of each target content can be determined.

C2: and acquiring the occurrence probability that the manipulator of the target user is the target manipulator, and calculating the manipulator influence coefficient corresponding to the target manipulator according to the occurrence probability.

As an alternative example, the occurrence probability that the manipulator of the target user is the target manipulator can be obtained according to the left-right hand feature extraction network in A1-A2. For example, when the target manipulator is a left hand, that is, the type corresponding to the target content is a left-hand type, the obtained occurrence probability that the manipulator of the target user is a left hand is 0.3560.

And after the occurrence probability that the manipulator of the target user is the target manipulator is obtained, calculating a manipulator influence coefficient corresponding to the target manipulator according to the occurrence probability. And the manipulator influence coefficient corresponding to the target manipulator is used for evaluating the influence of the target user on the single event trigger final bid of the target content when the target user uses the target manipulator.

As an alternative example, the calculation formula of the operator influence coefficient coef corresponding to the target operator is as follows:

coef＝(p_class_cid-0.5)*x+1

wherein p _ class _ cid is the occurrence probability that the manipulator of the target user is the target manipulator, x is an adjustable hyper-parameter, and in practical application, x is usually set to be 1. It will be appreciated that the larger x and coef, the greater the impact that the operator of the target user has on calculating the single event trigger final bid for the target content.

It is understood that if x is set to 1, when p _ class _ cid is less than 0.5, coef is calculated to be less than 1. When p _ class _ cid is greater than 0.5, coef is calculated to be greater than 1.

C3: and determining the product of the single event trigger bid of the target content, the predicted event trigger probability of the target user on the target content and the operator influence coefficient as the single event trigger final bid of the target content.

As an alternative example, the calculation formula for the single event trigger final bid ecpm "for the targeted content is:

ecpm”＝bid*pctr*coef

therefore, the operator influence coefficient coef can be used for adjusting the original ecpm. When p _ class _ cid is greater than 0.5, coef is made greater than 1, and then ecpm "is made greater than ecpm, which enables support of a single event trigger final bid for target content. When p _ class _ cid is smaller than 0.5, coef is smaller than 1, and ecpm is further smaller than ecpm, the purpose that the single event of the target content triggers the pressing of the final bid is achieved.

Based on the contents of C1-C3, after a target manipulator corresponding to the target content is determined, a bid perturbation method based on the target manipulator is designed. Specifically, an operator influence coefficient corresponding to the target operator is obtained, and the final bid is triggered by a single event of disturbing the target content through the operator influence coefficient corresponding to the target operator. Therefore, when the manipulator of the target user is the target manipulator, the manipulator of the target user is more suitable for implementing event triggering of the target content, so that the final bid price of single event triggering corresponding to the target content is higher. When the target content is the target advertisement, the target advertisement can obtain higher advertisement display opportunity, and the advertisement effect can be improved.

It should be noted that, when the target manipulator is another type of manipulator, the occurrence probability that the manipulator of the target user is another type of manipulator is set to be 0.5, that is, when the type corresponding to the target content is another type of manipulator, the influence coefficient corresponding to the target manipulator is 1. At this point ecpm "= ecpm, and a single event trigger to the target content results in neither a hold nor a hold.

In practical applications, when the target content is a target advertisement and the event trigger is a click, the steps C1 to C3 are specifically the steps C11 to C13, which will not be described in detail here:

c11: acquiring a target manipulator corresponding to target content; when the target user uses the target manipulator for operation, the statistical click rate of the target content is maximum; the target manipulator is a left-hand manipulator, a right-hand manipulator, a left-hand manipulator, a right-hand manipulator or other types of manipulators;

c12: acquiring the occurrence probability that the manipulator of the target user is the target manipulator, and calculating the manipulator influence coefficient corresponding to the target manipulator according to the occurrence probability;

c13: and determining the product of the target content single click bid, the target user predicted click rate on the target content and the operator influence coefficient as the target content single click final bid.

When the target content is a target advertisement and the event trigger is a click, the step C11 of obtaining the target manipulator corresponding to the target content specifically includes the following steps, which are not described in detail here:

acquiring left-hand click times, right-hand click times, left-hand click times and right-hand click times and other types of click times in the historical click times aiming at the target content;

respectively determining the quotient of the left-hand click frequency and the exposure frequency of the target content, the quotient of the right-hand click frequency and the exposure frequency of the target content, the quotient of the left-hand click frequency and the right-hand click frequency and the exposure frequency of the target content, and the quotient of the other types of click frequencies and the exposure frequency of the target content as a left-hand operation statistical click rate, a right-hand operation statistical click rate, a left-hand operation statistical click rate and a right-hand operation statistical click rate and other types of operator statistical click rates corresponding to the target content;

and determining the manipulator corresponding to the maximum value of the left-hand operation statistical click rate, the right-hand operation statistical click rate, the left-hand operation statistical click rate and the right-hand operation statistical click rate and the other types of manipulators as the target manipulator corresponding to the target content.

It is understood that when the target content is the target advertisement and the event trigger is click, if there are a plurality of target advertisements, the maximum ecpm "target content is the advertisement winning the bid, and the advertisement can be recommended to the user to complete the presentation.

In a possible implementation manner, an embodiment of the present application provides a specific training process of an event triggered probability prediction model, including:

d1: and acquiring historical content characteristic data of historical content and historical user characteristic data of historical users.

D2: and acquiring historical touch screen operator characteristic data of a historical user.

Before the event trigger probability prediction model is trained, acquiring historical content characteristic data, historical user characteristic data and historical touch screen operator characteristic data, wherein the historical content characteristic data, the historical user characteristic data and the historical touch screen operator characteristic data are training data of the event trigger probability prediction model.

D3: and inputting the historical content characteristic data, the historical user characteristic data and the historical touch screen operator characteristic data into an event trigger probability prediction model, and acquiring the predicted event trigger probability of the historical user on the historical content.

The predicted event triggering probability of the historical user on the historical content is a predicted value of the event triggering probability of the historical user on the historical content.

D4: and calculating a loss value according to the predicted event triggering probability of the historical user to the historical content and the expected event triggering probability of the historical user to the historical content.

The expected probability of event triggering of the historical content by the historical user is a label value, and can be calibrated in practical application. After the prediction event triggering probability of the historical user on the historical content is obtained, a loss value is calculated according to the prediction event triggering probability of the historical user on the historical content and the event triggering expected probability of the historical user on the historical content.

It can be understood that the method for calculating the loss value is not limited in the embodiments of the present application, and the determination may be performed according to actual requirements.

D5: and adjusting model parameters of the event trigger probability prediction model through the loss values, re-inputting historical content characteristic data, historical user characteristic data and historical touch screen operator characteristic data into the event trigger probability prediction model, and acquiring the predicted event trigger probability of the historical user on the historical content and subsequent steps until the preset conditions are reached.

As an alternative example, the event triggered probability prediction model may be trained based on a gradient descent method.

Based on the contents of D1-D5, the event triggering probability prediction model obtained through training in the embodiment of the application not only mines the relationship between the user characteristic data and the content characteristic data, but also mines the relationship between the characteristic data of the touch screen operator and the content characteristic data, so that the result output by the event triggering probability prediction model obtained through training is more accurate.

It should be noted that, in the embodiment of the present application, the historical user characteristic data of the historical user and the historical touch screen operator characteristic data of the historical user do not relate to sensitive information of the user, and the historical user characteristic data of the historical user and the historical touch screen operator characteristic data of the historical user are obtained and used after being authorized by the user. In one example, prior to obtaining historical user characteristic data of the historical user and historical touch screen operator characteristic data of the historical user, the corresponding interface displays prompt information related to obtaining data use authorization, and the user determines whether to approve the authorization based on the prompt information.

In a possible implementation manner, when the event trigger probability prediction model includes a multi-layer perceptron module and a linear regression module, the embodiment of the present application provides a specific implementation manner for inputting the historical content feature data, the historical user feature data, and the historical touch screen operator feature data into the event trigger probability prediction model in D3 to obtain the predicted event trigger probability of the historical user on the historical content, and the specific implementation manner includes:

d301: and converting historical input feature data consisting of historical content feature data, historical user feature data and historical touch screen operator feature data into a spliced representation of a historical feature vector and a historical offset vector.

D302: and inputting the historical feature vector into the multilayer perceptron module, and acquiring a third prediction event triggering probability of the historical user on the historical content.

D303: and inputting the historical bias vector into a linear regression module, and acquiring the fourth predicted event triggering probability of the historical user on the historical content.

D304: and determining the average value of the third predicted event triggering probability and the fourth predicted event triggering probability as the predicted event triggering probability of the historical user on the historical content.

It should be noted that the technical details of D301 to D304 can refer to B1 to B4 in the above embodiments, and are not described herein again.

Based on the method for predicting the event triggering probability provided by the embodiment of the method, the embodiment of the application also provides a device for predicting the event triggering probability, and the device for predicting the event triggering probability is described below with reference to the accompanying drawings. Because the principle of the device in the embodiment of the present disclosure for solving the problem is similar to the event trigger probability prediction method in the embodiment of the present disclosure, the implementation of the device may refer to the implementation of the method, and repeated details are not repeated.

Referring to fig. 4, this figure is a schematic structural diagram of an event trigger probability prediction apparatus according to an embodiment of the present application. As shown in fig. 4, the event trigger probability prediction apparatus includes:

a first acquisition unit 401 configured to acquire content feature data of target content and user feature data of a target user;

a second obtaining unit 402, configured to obtain characteristic data of a touch-screen operator of the target user;

an input unit 403, configured to input the content feature data, the user feature data, and the touch-screen operator feature data into an event trigger probability prediction model, and obtain a predicted event trigger probability of the target user for the target content; the event trigger probability prediction model is trained based on historical content feature data of historical content, historical user feature data of a historical user, historical touch screen operator feature data of the historical user and event trigger expected probability of the historical user on the historical content.

In a possible implementation manner, the second obtaining unit 402 includes:

the first acquiring subunit is configured to acquire touch screen operation information of the target user;

the first input subunit is configured to input the touch screen operation information of the target user into a left-hand and right-hand feature extraction network after training is completed, and obtain a left-hand operation probability, a right-hand operation probability, and a left-hand and right-hand operation probability of the target user;

and the second obtaining subunit is configured to obtain touch screen operator characteristic data of the target user according to the left-hand operation probability, the right-hand operation probability and the left-hand and right-hand operation probability.

In a possible implementation manner, the second obtaining subunit includes:

a first conversion subunit, configured to convert the left-hand operation probability, the right-hand operation probability, and the left-hand operation probability into a first integer value, a second integer value, and a third integer value, respectively;

a first determining subunit, configured to use a target integer value obtained through calculation based on the first integer value, the second integer value, and the third integer value as the characteristic data of the touch-screen operator of the target user.

In a possible implementation manner, the event-triggered probabilistic predictive model includes a multi-layer perceptron module and a linear regression module, and the input unit 403 includes:

the second conversion subunit is used for converting input feature data formed by the content feature data, the user feature data and the touch screen operator feature data into splicing expression of feature vectors and offset vectors;

the second input subunit is used for inputting the feature vector into the multilayer perceptron module and acquiring a first predicted event triggering probability of the target user on the target content;

the third input subunit is configured to input the bias vector into the linear regression module, and obtain a second predicted event trigger probability of the target user for the target content;

a second determining subunit, configured to determine an average value of the first predicted event trigger probability and the second predicted event trigger probability as the predicted event trigger probability of the target user for the target content.

In one possible implementation, the apparatus further includes a training unit, and the training unit includes:

a third obtaining subunit, configured to obtain historical content feature data of the historical content and historical user feature data of the historical user;

the fourth obtaining subunit is configured to obtain historical touch screen operator characteristic data of the historical user;

a fourth input subunit, configured to input the historical content feature data, the historical user feature data, and the historical touch screen operator feature data into the event trigger probability prediction model, and obtain a predicted event trigger probability of the historical user for the historical content;

the calculating subunit is used for calculating a loss value according to the predicted event triggering probability of the historical user on the historical content and the event triggering expected probability of the historical user on the historical content;

and the execution subunit is configured to adjust a model parameter of the event trigger probability prediction model according to the loss value, re-execute the step of inputting the historical content feature data, the historical user feature data, and the historical touch screen operator feature data into the event trigger probability prediction model, and acquire the predicted event trigger probability of the historical user on the historical content and subsequent steps until a preset condition is reached.

In one possible implementation, the apparatus further includes:

a fifth obtaining subunit, configured to obtain a target manipulator corresponding to the target content; the target manipulator is the manipulator corresponding to the event trigger statistic probability of the target content when the event trigger statistic probability is maximum; the target manipulator is a left-hand manipulator, a right-hand manipulator, a left-hand manipulator or a right-hand manipulator or other types of manipulators;

a sixth obtaining subunit, configured to obtain an occurrence probability that the manipulator of the target user is the target manipulator, and calculate a manipulator influence coefficient corresponding to the target manipulator according to the occurrence probability;

and the third determining subunit is used for determining the product of the single event trigger bid of the target content, the predicted event trigger probability of the target user on the target content and the operator influence coefficient as the single event trigger final bid of the target content.

In a possible implementation manner, the fifth obtaining subunit includes:

a seventh obtaining subunit, configured to obtain left-hand operation event trigger times, right-hand operation event trigger times, left-hand operation event trigger times, and other types of operation event trigger times in the event trigger history times for the target content;

a fourth determining subunit, configured to respectively determine a quotient of the left-hand operation event trigger count and the exposure count of the target content, a quotient of the right-hand operation event trigger count and the exposure count of the target content, a quotient of the left-hand operation event trigger count and the exposure count of the target content, and a quotient of the other category of operation event trigger counts and the exposure count of the target content as a first statistical probability, a second statistical probability, a third statistical probability, and a fourth statistical probability corresponding to the target content;

a fifth determining subunit, configured to determine, as the target manipulator corresponding to the target content, the manipulator corresponding to the maximum value of the first statistical probability, the second statistical probability, the third statistical probability, and the fourth statistical probability.

Based on the method for predicting the event trigger probability provided by the embodiment of the method, the application further provides an electronic device, which comprises the following steps: one or more processors; a storage device having one or more programs stored thereon, which when executed by the one or more processors, cause the one or more processors to implement the event trigger probability prediction method of any of the embodiments described above.

Referring now to FIG. 5, a schematic diagram of a structure of an electronic device 1300 suitable for implementing embodiments of the present application is shown. The terminal device in the embodiment of the present application may include, but is not limited to, a mobile terminal such as a mobile phone, a notebook computer, a Digital broadcast receiver, a PDA (Personal Digital Assistant), a PAD (Portable android device), a PMP (Portable multimedia Player), a car terminal (e.g., car navigation terminal), and the like, and a fixed terminal such as a Digital TV (television), a desktop computer, and the like. The electronic device shown in fig. 5 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiments of the present application.

As shown in fig. 5, electronic device 1300 may include a processing device (e.g., central processing unit, graphics processor, etc.) 1301 that may perform various appropriate actions and processes according to a program stored in a Read Only Memory (ROM) 1302 or a program loaded from a storage device 1306 into a Random Access Memory (RAM) 1303. In the RAM1303, various programs and data necessary for the operation of the electronic apparatus 1300 are also stored. The processing device 1301, the ROM1302, and the RAM1303 are connected to each other via a bus 1304. An input/output (I/O) interface 1305 is also connected to bus 1304.

Generally, the following devices may be connected to the I/O interface 1305: input devices 1306 including, for example, touch screens, touch pads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, and the like; an output device 1307 including, for example, a Liquid Crystal Display (LCD), speaker, vibrator, etc.; storage devices 1306 including, for example, magnetic tape, hard disk, and the like; and a communication device 1309. The communications device 1309 may allow the electronic device 1300 to communicate wirelessly or by wire with other devices to exchange data. While fig. 5 illustrates an electronic device 1300 having various means, it is to be understood that not all illustrated means are required to be implemented or provided. More or fewer devices may be alternatively implemented or provided.

In particular, according to embodiments of the application, the processes described above with reference to the flow diagrams may be implemented as computer software programs. For example, embodiments of the present application include a computer program product comprising a computer program carried on a non-transitory computer readable medium, the computer program containing program code for performing the method illustrated by the flow chart. In such embodiments, the computer program may be downloaded and installed from a network via the communication device 1309, or installed from the storage device 1306, or installed from the ROM 1302. The computer program, when executed by the processing apparatus 1301, performs the above-described functions defined in the methods of the embodiments of the present application.

The electronic device provided by the embodiment of the present application and the event trigger probability prediction method provided by the embodiment of the present application belong to the same inventive concept, and technical details that are not described in detail in the embodiment of the present application can be referred to the embodiment of the present application, and the embodiment of the present application have the same beneficial effects.

Based on the event triggering probability prediction method provided in the foregoing method embodiments, an embodiment of the present application provides a computer readable medium, on which a computer program is stored, where the program, when executed by a processor, implements the event triggering probability prediction method according to any of the foregoing embodiments.

It should be noted that the computer readable medium mentioned above in the present application may be a computer readable signal medium or a computer readable storage medium or any combination of the two. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples of the computer readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present application, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In this application, however, a computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: electrical wires, optical cables, RF (radio frequency), etc., or any suitable combination of the foregoing.

In some embodiments, the clients, servers may communicate using any currently known or future developed network Protocol, such as HTTP (HyperText Transfer Protocol), and may interconnect with any form or medium of digital data communication (e.g., a communications network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), the Internet (e.g., the Internet), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future developed network.

The computer readable medium may be embodied in the electronic device; or may exist separately without being assembled into the electronic device.

The computer readable medium carries one or more programs which, when executed by the electronic device, cause the electronic device to perform the event-triggered probability prediction method.

Computer program code for carrying out operations for aspects of the present application may be written in any combination of one or more programming languages, including but not limited to an object oriented programming language such as Java, smalltalk, C + +, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider).

The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems that perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

The units described in the embodiments of the present application may be implemented by software or hardware. Where the name of a unit/module does not in some cases constitute a limitation on the unit itself, for example, a voice data collection module may also be described as a "data collection module".

The functions described herein above may be performed, at least in part, by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), systems on a chip (SOCs), complex Programmable Logic Devices (CPLDs), and the like.

In the context of this application, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

According to one or more embodiments of the present application, there is provided an event trigger probability prediction method, the method comprising:

acquiring characteristic data of a touch screen operator of the target user;

inputting the content characteristic data, the user characteristic data and the touch screen operator characteristic data into an event trigger probability prediction model, and acquiring the predicted event trigger probability of the target user on the target content; the event trigger probability prediction model is trained based on historical content characteristic data of historical content, historical user characteristic data of a historical user, historical touch screen operator characteristic data of the historical user and event trigger expected probability of the historical user on the historical content.

According to one or more embodiments of the present application, an event trigger probability prediction method is provided, where the obtaining of the feature data of the touch screen operator of the target user includes:

acquiring touch screen operation information of the target user;

inputting the touch screen operation information of the target user into a left-hand and right-hand feature extraction network after training is completed, and acquiring the left-hand operation probability, the right-hand operation probability and the left-hand and right-hand operation probability of the target user;

and acquiring the characteristic data of the touch screen operator of the target user according to the left-hand operation probability, the right-hand operation probability and the left-hand operation probability.

According to one or more embodiments of the present application, an event trigger probability prediction method is provided, wherein the obtaining of the touch screen operator hand characteristic data of the target user according to the left-hand operation probability, the right-hand operation probability and the left-hand operation probability includes:

converting the left hand operation probability, the right hand operation probability, and the left hand operation probability into a first integer value, a second integer value, and a third integer value, respectively;

and taking a target integer value calculated based on the first integer value, the second integer value and the third integer value as touch screen operator characteristic data of the target user.

According to one or more embodiments of the present application, an event triggering probability prediction method is provided [ example four ], where the event triggering probability prediction model includes a multi-layer perceptron module and a linear regression module, and the inputting the content feature data, the user feature data and the touch screen operator feature data into an event triggering probability prediction model to obtain the predicted event triggering probability of the target user on the target content includes:

converting input feature data consisting of the content feature data, the user feature data and the touch screen operator feature data into spliced representation of feature vectors and offset vectors;

inputting the feature vector into the multilayer perceptron module, and acquiring a first prediction event triggering probability of the target user on the target content;

inputting the bias vector into the linear regression module, and acquiring a second prediction event triggering probability of the target user on the target content;

determining an average of the first predicted event-trigger probability and the second predicted event-trigger probability as a predicted event-trigger probability of the target user for the target content.

According to one or more embodiments of the present application, [ example five ] there is provided an event-triggered probability prediction method, wherein a training process of the event-triggered probability prediction model includes:

acquiring historical content characteristic data of the historical content and historical user characteristic data of the historical user;

acquiring the characteristic data of a historical touch screen operator of the historical user;

inputting the historical content characteristic data, the historical user characteristic data and the historical touch screen operator characteristic data into the event trigger probability prediction model to obtain the predicted event trigger probability of the historical user on the historical content;

calculating a loss value according to the predicted event triggering probability of the historical user on the historical content and the event triggering expected probability of the historical user on the historical content;

and adjusting model parameters of the event trigger probability prediction model according to the loss value, re-executing the steps of inputting the historical content characteristic data, the historical user characteristic data and the historical touch screen operator characteristic data into the event trigger probability prediction model, and acquiring the predicted event trigger probability of the historical user on the historical content and the subsequent steps until a preset condition is reached.

According to one or more embodiments of the present application, [ example six ] there is provided an event triggering probability prediction method, the method further comprising:

acquiring a target manipulator corresponding to the target content; the target manipulator is the manipulator corresponding to the event trigger statistic probability of the target content when the event trigger statistic probability is maximum; the target manipulator is a left-hand manipulator, a right-hand manipulator, a left-hand manipulator or a right-hand manipulator or other types of manipulators;

acquiring the occurrence probability of the target user operator as the target operator, and calculating an operator influence coefficient corresponding to the target operator according to the occurrence probability;

and determining the product of the single event trigger bid of the target content, the predicted event trigger probability of the target user on the target content and the operator influence coefficient as the single event trigger final bid of the target content.

According to one or more embodiments of the present application, in example seven, there is provided an event trigger probability prediction method, where the obtaining of a target manipulator corresponding to the target content includes:

acquiring left hand operation event triggering times, right hand operation event triggering times, left hand operation event triggering times and other types of operation hand operation event triggering times in the event triggering historical times aiming at the target content;

respectively determining the quotient of the left-hand operation event trigger times and the exposure times of the target contents, the quotient of the right-hand operation event trigger times and the exposure times of the target contents, the quotient of the left-hand operation event trigger times and the exposure times of the target contents, and the quotient of the other types of operation event trigger times and the exposure times of the target contents as a first statistical probability, a second statistical probability, a third statistical probability and a fourth statistical probability corresponding to the target contents;

determining the manipulator corresponding to the maximum value of the first statistical probability, the second statistical probability, the third statistical probability and the fourth statistical probability as the target manipulator corresponding to the target content.

According to one or more embodiments of the present application, [ example eight ] there is provided an event trigger probability prediction apparatus, the apparatus comprising:

the input unit is used for inputting the content characteristic data, the user characteristic data and the touch screen operator characteristic data into an event trigger probability prediction model, and acquiring the predicted event trigger probability of the target user on the target content; the event trigger probability prediction model is trained based on historical content feature data of historical content, historical user feature data of a historical user, historical touch screen operator feature data of the historical user and event trigger expected probability of the historical user on the historical content. The second acquisition unit includes:

the first obtaining subunit is configured to obtain touch screen operation information of the target user;

the first input subunit is used for inputting the touch screen operation information of the target user into a trained left-hand and right-hand feature extraction network, and acquiring the left-hand operation probability, the right-hand operation probability and the left-hand and right-hand operation probability of the target user;

According to one or more embodiments of the present application, there is provided [ example nine ] an event trigger probability prediction apparatus, the second obtaining subunit including:

a first conversion subunit, configured to convert the left hand operation probability, the right hand operation probability, and the left and right hand operation probability into a first integer value, a second integer value, and a third integer value, respectively;

According to one or more embodiments of the present application, [ example ten ] there is provided an event trigger probability prediction apparatus, the second obtaining subunit including:

a first determining subunit, configured to use a target integer value calculated based on the first integer value, the second integer value, and the third integer value as touch-screen operator characteristic data of the target user.

According to one or more embodiments of the present application, [ example eleven ] there is provided an event-triggered probability prediction apparatus, the event-triggered probability prediction model including a multi-layered perceptron module and a linear regression module, the input unit including:

the second input subunit is used for inputting the feature vector into the multilayer perceptron module and acquiring a first prediction event triggering probability of the target user on the target content;

a second determining subunit, configured to determine an average of the first predicted event trigger probability and the second predicted event trigger probability as the predicted event trigger probability of the target user for the target content.

According to one or more embodiments of the present application, [ example twelve ] there is provided an event trigger probability prediction apparatus, the apparatus further comprising a training unit including:

a fourth obtaining subunit, configured to obtain historical touch screen operator characteristic data of the historical user;

a fourth input subunit, configured to input the historical content feature data, the historical user feature data, and the historical touch-screen operator feature data into the event trigger probability prediction model, and obtain a predicted event trigger probability of the historical user for the historical content;

According to one or more embodiments of the present application, [ example thirteen ] there is provided an event trigger probability prediction apparatus, the apparatus further comprising:

a fifth acquiring subunit, configured to acquire a target manipulator corresponding to the target content; the target manipulator is the manipulator corresponding to the event trigger statistic probability of the target content when the event trigger statistic probability is maximum; the target manipulator is a left hand, a right hand, a left hand, a right hand or other types of manipulators;

According to one or more embodiments of the present application, an event triggering probability prediction apparatus is provided [ example fourteen ], wherein the fifth obtaining subunit includes:

a seventh acquiring subunit, configured to acquire left-hand operation event trigger times, right-hand operation event trigger times, left-hand operation event trigger times, and other types of operation event trigger times in the event trigger history times for the target content;

According to one or more embodiments of the present application, [ example fifteen ] there is provided an electronic device, comprising:

one or more processors;

a storage device having one or more programs stored thereon,

According to one or more embodiments of the present application [ example sixteen ] there is provided a computer readable medium having a computer program stored thereon, wherein the program when executed by a processor implements a method of event triggered probability prediction as described in any of the above.

It should be noted that, in the present specification, each embodiment is described in a progressive manner, and each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments may be referred to each other. For the system or the device disclosed by the embodiment, the description is simple because the system or the device corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.

It should be understood that in the present application, "at least one" means one or more, "a plurality" means two or more. "and/or" for describing an association relationship of associated objects, indicating that there may be three relationships, e.g., "a and/or B" may indicate: only A, only B and both A and B are present, wherein A and B may be singular or plural. The character "/" generally indicates that the former and latter associated objects are in an "or" relationship. "at least one of the following" or similar expressions refer to any combination of these items, including any combination of the singular or plural items. For example, at least one (one) of a, b, or c, may represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", wherein a, b and c may be single or plural.

It is further noted that, herein, 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 phrases "comprising a," "8230," "8230," or "comprising" does not exclude the presence of additional like elements in a process, method, article, or apparatus that comprises the element.

The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module may reside in Random Access Memory (RAM), memory, read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for event triggered probability prediction, the method comprising:

acquiring characteristic data of a touch screen operator of the target user;

2. The method of claim 1, wherein the obtaining touch-screen operator characteristic data of the target user comprises:

acquiring touch screen operation information of the target user;

inputting the touch screen operation information of the target user into a trained left-hand and right-hand feature extraction network to obtain the left-hand operation probability, the right-hand operation probability and the left-hand and right-hand operation probability of the target user;

3. The method of claim 2, wherein the obtaining touch screen operator characteristic data of the target user according to the left hand operation probability, the right hand operation probability and the left hand and right hand operation probability comprises:

and taking a target integer value obtained by calculation based on the first integer value, the second integer value and the third integer value as the characteristic data of the touch screen operator of the target user.

4. The method as claimed in claim 1, wherein the event-triggered probability prediction model comprises a multi-layer perceptron module and a linear regression module, and the inputting the content feature data, the user feature data and the touch-screen operator feature data into the event-triggered probability prediction model to obtain the predicted event-triggered probability of the target user for the target content comprises:

converting input feature data composed of the content feature data, the user feature data and the touch screen operator feature data into splicing expression of feature vectors and offset vectors;

inputting the bias vector into the linear regression module, and acquiring a second prediction event trigger probability of the target user on the target content;

determining an average of the first predicted event-triggering probability and the second predicted event-triggering probability as a predicted event-triggering probability of the target user for the target content.

5. The method of claim 1, wherein the event triggers a training process of a probabilistic predictive model, comprising:

calculating a loss value according to the predicted event triggering probability of the historical user to the historical content and the event triggering expected probability of the historical user to the historical content;

6. The method of claim 1, further comprising:

7. The method according to claim 6, wherein the obtaining of the target manipulator corresponding to the target content comprises:

respectively determining the quotient of the left-hand operation event trigger times and the exposure times of the target content, the quotient of the right-hand operation event trigger times and the exposure times of the target content, the quotient of the left-hand operation event trigger times and the exposure times of the target content, and the quotient of the other types of operation event trigger times and the exposure times of the target content as a first statistical probability, a second statistical probability, a third statistical probability and a fourth statistical probability corresponding to the target content;

8. An event trigger probability prediction apparatus, the apparatus comprising:

the input unit is used for inputting the content characteristic data, the user characteristic data and the touch screen operator characteristic data into an event trigger probability prediction model, and acquiring the predicted event trigger probability of the target user on the target content; the event trigger probability prediction model is trained based on historical content characteristic data of historical content, historical user characteristic data of a historical user, historical touch screen operator characteristic data of the historical user and event trigger expected probability of the historical user on the historical content.

9. An electronic device, comprising:

one or more processors;

a storage device having one or more programs stored thereon,

when executed by the one or more processors, cause the one or more processors to implement the event-triggered probability prediction method as recited in any of claims 1-7.

10. A computer-readable medium, on which a computer program is stored, wherein the program, when executed by a processor, implements the event trigger probability prediction method according to any one of claims 1-7.