Advertisement user portrait generation method and device based on deep neural network, equipment and medium

Abstract

The invention discloses an advertisement user portrait generation method and device based on a deep neural network, equipment and a medium, and the method comprises the steps: obtaining user feature data of a target user, the user feature data comprises static feature data and dynamic feature data, the static feature data is obtained through a business server, and the dynamic feature data is obtained through a business server; the dynamic feature data is obtained through a media server; determining at least one prediction classification type corresponding to the target user based on the at least one conversion target of the target user; and inputting the user feature data into an improved deep neural network, performing user portrait prediction according to the prediction classification type to obtain a user portrait corresponding to the target user, and generating a user portrait label by using the improved deep neural network, thereby effectively improving accuracy and reliability of user portrait prediction.

Description

Technical Field

[0001] The present disclosure generally relates to the field of user portrait technology, and specifically relates to a method, device, equipment and medium for generating an advertisement user portrait based on a deep neural network. Background Art

[0002] User profiling is the process of classifying and describing the advertising audience based on the user's basic information, interests, hobbies, behavioral habits, etc. By analyzing user profiling, advertisers can better understand the needs and characteristics of target users, thereby formulating more accurate advertising strategies and improving advertising effectiveness and investment returns.

[0003] In related technologies, decision tree models are usually used to generate user portrait labels. However, user feature data in advertising scenarios has characteristics such as high dimensionality, nonlinearity, and imbalance, and decision tree models have certain limitations when processing such data, and are prone to overfitting, thus affecting the accuracy of user portraits. Summary of the invention

[0004] In view of the above-mentioned defects or deficiencies in the prior art, it is desired to provide a method, device, equipment and medium for generating advertising user portraits based on a deep neural network, using an improved deep neural network to generate user portrait labels, thereby effectively improving the accuracy and reliability of user portrait predictions.

[0005] In a first aspect, an embodiment of the present application provides a method for generating an advertisement user portrait based on a deep neural network, comprising:

[0006] Acquire user characteristic data of the target user, wherein the user characteristic data includes static characteristic data and dynamic characteristic data, wherein the static characteristic data is acquired through a service server, and the dynamic characteristic data is acquired through a media server;

[0007] Based on at least one conversion goal for the target user, determining at least one predicted classification type corresponding to the target user;

[0008] The user feature data is input into an improved deep neural network, and user portrait prediction is performed according to the prediction classification type to obtain a user portrait corresponding to the target user.

[0009] In some embodiments, determining at least one predicted classification type corresponding to the target user based on at least one conversion goal for the target user includes:

[0010] Determining a target stage corresponding to the target user based on the user characteristic data of the target user;

[0011] Based on the target stage, at least one predicted classification type corresponding to the target user is determined.

[0012] In some embodiments, the improved deep neural network includes an input layer, an embedding layer, a concatenation layer, at least one hidden layer and an output layer.

[0013] The splicing layer includes a first splicing unit and a second splicing unit.

[0014] The first splicing unit is used to perform addition splicing on the user feature vectors obtained by the embedding layer to obtain a first splicing vector, and the second splicing unit is used to perform product splicing on the user feature vectors obtained by the embedding layer to obtain a second splicing vector;

[0015] The concatenation layer is used to input the first concatenation vector and the second concatenation vector into the at least one hidden layer respectively.

[0016] In some embodiments, it also includes:

[0017] The at least one hidden layer uses historical user feature data to adjust parameters based on an error back propagation algorithm;

[0018] The historical user feature data are user features of historical users who already have stage user portraits.

[0019] In some embodiments, the error back propagation algorithm is an error back propagation algorithm based on a gradient descent strategy.

[0020] In some embodiments, the gradient descent strategy is a stochastic gradient descent strategy.

[0021] In a second aspect, an embodiment of the present application provides an advertising user portrait generation device based on a deep neural network, comprising:

[0022] An acquisition module, used to acquire user characteristic data of a target user, wherein the user characteristic data includes static characteristic data and dynamic characteristic data, wherein the static characteristic data is acquired through a service server, and the dynamic characteristic data is acquired through a media server;

[0023] a determination module, configured to determine at least one predicted classification type corresponding to the target user based on at least one conversion target for the target user;

[0024] The prediction module is used to input the user feature data into the improved deep neural network, perform user portrait prediction according to the prediction classification type, and obtain the user portrait corresponding to the target user.

[0025] In a third aspect, an embodiment of the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method described in the embodiment of the present application when executing the program.

[0026] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in the embodiment of the present application.

[0027] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program, characterized in that when the computer program is executed by a processor, it implements the method described in the embodiment of the present application.

[0028] The advertising user portrait generation method, device, equipment and medium based on deep neural network proposed in the embodiments of the present application can make full use of the static feature data provided by the business service provider and the dynamic feature data provided by the media service provider, and use the improved deep neural network to predict the user portrait for the conversion target, thereby effectively improving the reliability and accuracy of the user portrait prediction.

[0029] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings:

[0031] Figure 1 The following is a diagram showing the implementation environment architecture of the method for generating an advertisement user portrait based on a deep neural network provided in an embodiment of the present application;

[0032] Figure 2 A schematic diagram of a process of generating an advertisement user portrait based on a deep neural network provided in an embodiment of the present application is shown;

[0033] Figure 3 A schematic diagram of the structure of a deep neural network provided by an embodiment of the present application is shown;

[0034] Figure 4 A curve diagram of the activation function Sigmoid provided in an embodiment of the present application is shown;

[0035] Figure 5 A schematic diagram of the structure of a neural network structure provided by an embodiment of the present application is shown;

[0036] Figure 6A schematic diagram of the structure of an improved deep neural network provided by an embodiment of the present application is shown;

[0037] Figure 7 A schematic diagram of the structure of an embedding layer provided in an embodiment of the present application is shown;

[0038] Figure 8 A schematic diagram of the structure of an advertising user portrait generation device based on a deep neural network provided in an embodiment of the present application is shown;

[0039] Fig. 9 A schematic diagram of the structure of a computer system of an electronic device or server suitable for implementing an embodiment of the present application is shown. DETAILED DESCRIPTION

[0040] The present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the relevant invention, rather than to limit the invention. It is also necessary to explain that, for ease of description, only the parts related to the invention are shown in the accompanying drawings.

[0041] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0042] It should also be noted that the acquisition or use of data in the embodiments of the present application requires the user's consent. The relevant data can only be obtained after the user's authorization and permission, and the acquisition or use of the data complies with relevant legal provisions.

[0043] The specific implementation environment of the advertising user portrait generation method based on deep neural network proposed in this application can be found in Figure 1 . Figure 1 The following is a diagram showing the implementation environment architecture of the method for generating an advertising user portrait based on a deep neural network provided in an embodiment of the present application.

[0044] like Figure 1 As shown, the implementation environment architecture includes: a business server 101, a media server 102 and a server 103.

[0045] The business server 101 is an advertisement delivery server, and is used to store user attribute information and business status information static feature data.

[0046] The media server 102 is a media server where advertisements are placed, such as a server of a video platform, a server of a trading platform, etc. The advertiser places advertisements on the media platform, so that when users are active on the media platform, they can enter the business server 101 based on the advertisement operations on the media platform (such as ad browsing, ad clicking, etc.) to implement the expected conversion business of the advertiser. The media server 102 is used to store the user's behavior data on apps, pages, etc. that have relevant advertisements placed on the media platform.

[0047] The server 103 obtains the characteristic data from the business server 101 and the media server 102 respectively, and then performs user portrait prediction based on the user characteristic data of the target user. Optionally, the server 103 may be a part of the business server 101 that performs the prediction function, that is, a server that performs different functions for the advertisement delivery party.

[0048] Business server 101, media server 102 and server 103 can all be independent physical servers, or they can be a server cluster or distributed system composed of multiple physical servers. They can also be cloud servers that provide basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, as well as big data and artificial intelligence platforms.

[0049] The service server 101, the media server 102 and the server 103 are directly or indirectly connected by wired or wireless communication. Optionally, the wireless network or wired network uses standard communication technology and / or protocol. The network is usually the Internet, and can also be any network, including but not limited to a local area network (LAN), a metropolitan area network (MAN), a wide area network (WAN), a mobile, wired or wireless network, a private network or any combination of a virtual private network.

[0050] The advertising user portrait generation method based on a deep neural network proposed in the present application can be implemented by an advertising user portrait generation device based on a deep neural network, and the advertising user portrait generation device based on a deep neural network can be installed on a terminal device or a server.

[0051] In order to further illustrate the technical solution provided by the embodiment of the present application, this is described in detail below in conjunction with the accompanying drawings and specific implementation methods. Although the embodiment of the present application provides the method operation instruction steps shown in the following embodiments or drawings, more or less operation instruction steps may be included in the method based on routine or no creative labor. In the steps where there is no necessary causal relationship logically, the execution order of these steps is not limited to the execution order provided by the embodiment of the present application. The method may be executed in the order of the method shown in the embodiment or drawings or in parallel during the actual processing process or when the device is executed.

[0052] Please refer to Figure 2 , Figure 2 The flowchart of the method for generating an advertisement user portrait based on a deep neural network provided by an embodiment of the present application is shown. Figure 2 As shown, the method includes:

[0053] Step 201, obtaining user characteristic data of a target user.

[0054] The user characteristic data includes static characteristic data and dynamic characteristic data. The static characteristic data is obtained through the service server, and the dynamic characteristic data is obtained through the media server.

[0055] Exemplarily, user characteristics include user attribute characteristics, user behavior characteristics and business status characteristics. Among them, user attribute characteristics and user business status characteristics are static feature data obtained through the business server, that is, the stage status information of the user who has registered or completed in the business server of the advertising party. Optionally, user attribute characteristics include but are not limited to basic information such as user age, gender, education, occupation, region, etc., and business status characteristics include but are not limited to business status information such as user registration, login, quota application, and quota usage. User behavior characteristics are dynamic characteristics obtained through the media server, that is, dynamic information of the user's interactive behavior on the advertising media platform. Optionally, user behavior characteristics include ad views, ad clicks, APP login times, APP usage time, stay time on different pages, page click behavior, etc.

[0056] It can be seen that the user features provided in the embodiments of the present application are sparse, high-dimensional feature data from both the advertising party and the media party. Compared with the traditional method of using the feature data from the advertising party or the media party alone, the dimension of the feature data can be effectively improved, so that more dimensional user information can be referenced when performing user portrait prediction and classification later, reducing the number and scope of users in a single classification, and realizing the refined regulation of user data, that is, improving the accuracy of user portrait prediction through more dimensional limited user data.

[0057] Step 202: Determine at least one predicted classification type corresponding to the target user based on at least one conversion goal of the target user.

[0058] It should be noted that the embodiments of the present application can make targeted predictions for target users based on the conversion goals that the advertiser expects to optimize.

[0059] In a feasible embodiment, the target stage corresponding to the target user may be determined based on the user characteristic data of the target user, and at least one predicted classification type corresponding to the target user may be determined based on the target stage.

[0060] That is to say, in the embodiment of the present application, a user can be predicted in a refined manner in stages. Specifically, by analyzing the user characteristics of the target user, the target stage corresponding to the target user is determined. For example, if the user characteristics include registration information or login information, it means that the user is a registered user, and then a portrait prediction can be made for the subsequent business content, such as credit approval rate prediction, credit usage rate prediction, etc., that is, the user portrait in the subsequent stage is predicted. On the contrary, if the user characteristics do not include registration information or login information, it means that the user has not yet registered, and then a user portrait prediction can be made for the previous registration stage.

[0061] Therefore, the embodiment of the present application can perform phased and refined portrait prediction for target users, which not only improves the accuracy of user portrait prediction, but also helps to carry out targeted stage conversion for target users and improve user conversion efficiency. At the same time, there is no need to perform full-stage portrait prediction for users, which effectively reduces redundant prediction results and saves the cost of adjusting advertising delivery strategies.

[0062] Step 203: Input the user feature data into the improved deep neural network, perform user portrait prediction according to the prediction classification type, and obtain a user portrait corresponding to the target user.

[0063] It should be noted that the improved deep neural network is an improvement on the traditional deep neural network DNN to improve the accuracy and efficiency of user portrait prediction and classification.

[0064] In a feasible embodiment, the user portrait prediction of the target user is to obtain the user portrait model score. The model score refers to the evaluation and scoring of user quality in advertising delivery, such as the model score of the user based on the user click-through rate (CTR), conversion rate, exposure, cost per thousand impressions (CPM), etc. In the embodiment of the present application, the model score of the user portrait corresponding to the user is predicted by using an improved deep neural network to evaluate.

[0065] The following is an introduction to the traditional deep neural network DNN.

[0066] like Figure 3 As shown, a neuron receives input signals x from n other neurons. i (i=1,…,n), these input signals are passed through the weighted j The total input value Σ received by the neuron will be compared with the threshold θ of the neuron, and then processed by the activation function to generate the output of the neuron.

[0067] In the embodiment of the present application, the Sigmoid function is selected as the activation function. Figure 4 As shown in the figure, it is used to compress the output value of the neuron within the output range of (0,1), so it is also called the "squeezing function". Therefore, by selecting the Sigmoid function as the activation function, the eigenvalues ​​of sparse multidimensional data can be compressed within the same output range, which is convenient for subsequent feature calculation and analysis.

[0068] Furthermore, by connecting multiple neurons in a certain hierarchical structure, a neural network is formed. Figure 5 As shown, the input layer includes 3 neurons and the output layer includes two neurons.

[0069] From this, we can see that for a given data set (x i ,y k ), if the weight w is known j and threshold θ, the relationship between x and y can be obtained. In the embodiment of the present application, it represents the relationship between user features and user portrait prediction model scores. Therefore, the deep neural network DNN has the feasibility of predictive classification and can be used to predict the user portrait prediction model scores of target users.

[0070] In a feasible embodiment, the threshold can be transformed into a special node multiplied by the weight w n+1 So that the weight and threshold can be learned together through the process of weight learning.

[0071] It should also be noted that in traditional deep neural networks, perceptrons are used for classification. The perceptron only performs activation function processing on neurons in the output layer, that is, there is only one layer of functional neurons, and its learning ability is limited. Specifically, without considering the activation function, the relationship between x and y is as follows:

[0072]

[0073] Among them, n+1 is the weighted sum including the threshold θ.

[0074] It can be seen that the perceptron can only handle linearly separable problems. It can be proved that if the two types of patterns are linearly separable, that is, there is a linear hyperplane that can separate them, then the perceptron will converge in the learning process and obtain the appropriate weight vector w = (w1, w2, ..., w n+1 ).

[0075] Based on this, in order to solve the nonlinear separability problem of user feature data, the present application further sets up multiple hidden layers with functional neurons in the output layer, so that each hidden layer has learning ability and has an activation function. In the embodiment of the present application, multiple hidden layers are fully connected, which can also be called a fully connected layer.

[0076] Specifically, if Figure 6 As shown, in an embodiment of the present application, the improved deep neural network includes an input layer, an embedding layer, a concatenation layer, at least one hidden layer and an output layer.

[0077] Among them, the input layer is used for inputting user feature data, and the embedding layer is used to convert high-dimensional user feature data into low-dimensional user feature data, for example, into one-dimensional feature data, that is, a one-dimensional user feature vector. It should be understood that the user feature data is reduced in dimension by the embedding layer, so that the trained user feature data still maintains a similar geometric relationship in the low-dimensional space. The operation of mapping user feature data into user feature vectors can be regarded as adding a full-link linear transformation to the original feature vector of the input layer, which can be expressed in mathematical symbols as the multiplication operation of the feature vector x and the weight matrix W, such as Figure 7 shown.

[0078] Furthermore, in an embodiment of the present application, the splicing layer includes a first splicing unit and a second splicing unit.

[0079] The first splicing unit is used to perform wig splicing on the user feature vectors obtained by the embedding layer to obtain a first splicing vector, and the second splicing unit is used to perform product splicing on the user feature vectors obtained by the embedding layer to obtain a second splicing vector.

[0080] The concatenation layer inputs the first concatenation vector and the second concatenation vector into at least one hidden layer.

[0081] Specifically, the product splicing is to multiply the user feature vectors in pairs and then add them to obtain the second splicing vector.

[0082] It should be understood that compared with the traditional deep neural network that only uses additive splicing to splice user feature vectors, the multiplication splicing performed simultaneously with the additive splicing is added, so that the spliced ​​vector features have more expressive information, and can better integrate the correlation between different features into the model, thereby improving the final prediction accuracy of the model.

[0083] In a feasible embodiment, during the training process, at least one hidden layer uses historical user feature data to adjust parameters based on an error back propagation algorithm.

[0084] Among them, historical user characteristic data refers to the user characteristics of historical users who already have stage user portraits, such as users who have applied, users who have been granted credit, etc.

[0085] It should be noted that the error back propagation algorithm is also called the back propagation algorithm, and its goal is to minimize the cumulative error on the training set. Specifically, the layer-by-layer recursion is expressed in the form of a matrix:

[0086] x L =σ(z L )=σ(w L x L-1 +b L )

[0087] C0=Loss=(x L -y) 2

[0088] Among them, x L is the output of the Lth layer, σ is the activation function Sigmoid, w L is the weight matrix of the Lth layer, x L-1 is the output of the previous layer, b L is the bias of the Lth layer.

[0089] When L A small change of δw L When C0 changes slightly, a series of reactions will occur. L The partial derivative of , we get:

[0090]

[0091] Similarly, we can get:

[0092]

[0093] Extended to multiple neurons, the loss function is defined as follows:

[0094]

[0095] The gradient of the output layer can be expressed as:

[0096]

[0097] Let δ of the l+1th layer be l+1 If it is known, then for δ l The solution is as follows:

[0098]

[0099] That is to say, when we get δ of any layer l When , we can make any ∥ l and b l In other words, the weights of the hidden layers obtained through the back propagation algorithm training can make at least one hidden layer approximate a continuous function of arbitrary complexity with arbitrary precision, so that the deep neural network with parameters trained through the back propagation algorithm can achieve high-precision user portrait prediction for advertising users.

[0100] Furthermore, the error back propagation algorithm is an error back propagation algorithm based on a gradient descent strategy.

[0101] Specifically, the user characteristics of historical users who already have stage user portraits include the characteristics of historical users and the portraits of historical users. Therefore, when using the user characteristics of historical users for training, the predicted portrait obtained through training can be compared with the actual portrait of the historical user to calculate the mean square error, and then the parameters can be adjusted in the negative gradient direction of the target based on the gradient descent strategy.

[0102] Furthermore, in order to avoid falling into the dilemma of the local minimum of the mean square error, the stochastic gradient descent strategy is used. Unlike the standard gradient descent method, which accurately calculates the gradient, the stochastic gradient descent method adds a random factor when calculating the gradient. Even if it falls into a local minimum, the calculated gradient may still be non-zero, so there is a chance to jump out of the local minimum to continue searching.

[0103] Therefore, the advertising user portrait generation method based on deep neural network proposed in the embodiment of the present application can make full use of the static feature data provided by the business service provider and the dynamic feature data provided by the media service provider, and use the improved deep neural network to predict the user portrait for the conversion target, thereby effectively improving the reliability and accuracy of the user portrait prediction.

[0104] It should be noted that although the operations of the method of the present invention are described in a particular order in the drawings, this does not require or imply that the operations must be performed in this particular order or that all illustrated operations must be performed to achieve desired results.

[0105] Figure 8 A schematic diagram of the structure of an advertising user portrait generation device based on a deep neural network provided in one embodiment of the present application is shown.

[0106] like Figure 8 As shown, the advertising user portrait generation device 10 based on a deep neural network includes:

[0107] An acquisition module 11 is used to acquire user characteristic data of a target user, wherein the user characteristic data includes static characteristic data and dynamic characteristic data, wherein the static characteristic data is acquired through a service server, and the dynamic characteristic data is acquired through a media server;

[0108] A determination module 12, configured to determine at least one predicted classification type corresponding to the target user based on at least one conversion target for the target user;

[0109] The prediction module 13 is used to input the user feature data into the improved deep neural network, perform user portrait prediction according to the prediction classification type, and obtain the user portrait corresponding to the target user.

[0110] In some embodiments, the determination module 12 is further configured to:

[0111] Determining a target stage corresponding to the target user based on the user characteristic data of the target user;

[0112] Based on the target stage, at least one predicted classification type corresponding to the target user is determined.

[0113] In some embodiments, the improved deep neural network includes an input layer, an embedding layer, a concatenation layer, at least one hidden layer and an output layer.

[0114] The splicing layer includes a first splicing unit and a second splicing unit.

[0115] The first splicing unit is used to perform addition splicing on the user feature vectors obtained by the embedding layer to obtain a first splicing vector, and the second splicing unit is used to perform product splicing on the user feature vectors obtained by the embedding layer to obtain a second splicing vector;

[0116] The concatenation layer is used to input the first concatenation vector and the second concatenation vector into the at least one hidden layer respectively.

[0117] In some embodiments, it also includes:

[0118] The at least one hidden layer uses historical user feature data to adjust parameters based on an error back propagation algorithm;

[0119] The historical user feature data are user features of historical users who already have stage user portraits.

[0120] In some embodiments, the error back propagation algorithm is an error back propagation algorithm based on a gradient descent strategy.

[0121] In some embodiments, the gradient descent strategy is a stochastic gradient descent strategy.

[0122] It should be understood that the modules or modules recorded in the advertising user portrait generation device 10 based on deep neural network are similar to those in the reference Figure 2 The various steps in the described method correspond to each other. Therefore, the operations and features described above for the method are also applicable to the advertising user portrait generation device 10 based on a deep neural network and the modules contained therein, and will not be repeated here. The advertising user portrait generation device 10 based on a deep neural network can be pre-implemented in a browser or other security application of an electronic device, or can be loaded into the browser of an electronic device or its security application by downloading or the like. The corresponding modules in the advertising user portrait generation device 10 based on a deep neural network can cooperate with the modules in the electronic device to implement the solution of the embodiment of the present application.

[0123] For the several modules or units mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.

[0124] Reference below Fig. 9 , Fig. 9 A schematic diagram of the structure of a computer system of an electronic device or server suitable for implementing an embodiment of the present application is shown.

[0125] like Fig. 9 As shown, the computer system includes a central processing unit (CPU) 901, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 902 or the program loaded from the storage part 908 into the random access memory (RAM) 903. In the RAM 903, various programs and data required for the operation instructions of the system are also stored. The CPU 901, the ROM 902 and the RAM 903 are connected to each other through a bus 904. An input / output (I / O) interface 905 is also connected to the bus 904.

[0126] The following components are connected to the I / O interface 905: an input section 906 including a keyboard, a mouse, etc.; an output section 907 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 908 including a hard disk, etc.; and a communication section 909 including a network interface card such as a LAN card, a modem, etc. The communication section 909 performs communication processing via a network such as the Internet. A drive 910 is also connected to the I / O interface 905 as needed. A removable medium 911, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 910 as needed so that a computer program read therefrom is installed into the storage section 908 as needed.

[0127] In particular, according to an embodiment of the present application, the above reference flow chart Figure 2 The described process can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flow chart. In such an embodiment, the computer program includes a program code for executing the method shown in the flow chart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 909, and / or installed from the removable medium 911. When the computer program is executed by the central processing unit (CPU) 901, the above-mentioned functions defined in the system of the present application are executed.

[0128] It should be noted that the computer-readable medium shown in the present application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, 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 disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium such as a computer-readable storage medium that can send, propagate or transmit a program for use by or in conjunction with an instruction execution system, device or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0129] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, functions and operating instructions of the system, method and computer program product according to various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the aforementioned module, program segment or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the box can also occur in a sequence different from that marked in the accompanying drawings. For example, the boxes represented by two connections can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operating instruction, or can be implemented with a combination of dedicated hardware and computer instructions.

[0130] The units or modules involved in the embodiments of the present application may be implemented by software or by hardware. The units or modules described may also be arranged in a processor. For example, they may be described as: a processor includes an acquisition module, a determination module, and a prediction module. Among them, the names of these units or modules do not constitute a limitation on the units or modules themselves under certain circumstances. For example, the acquisition module may also be described as "obtaining user feature data of the target user, the user feature data including static feature data and dynamic feature data, the static feature data being obtained through a service server, and the dynamic feature data being obtained through a media server".

[0131] As another aspect, the present application further provides a computer-readable storage medium, which may be included in the electronic device described in the above embodiment, or may exist independently without being assembled into the electronic device. The above computer-readable storage medium stores one or more programs, and when the above programs are used by one or more processors to execute the advertising user portrait generation method based on a deep neural network described in the present application.

[0132] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the aforementioned disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in this application (but not limited to) by each other to form a technical solution.

Claims

1. A method for generating advertising user portraits based on deep neural networks, characterized in that: include: Acquire user characteristic data of the target user, wherein the user characteristic data includes static characteristic data and dynamic characteristic data, wherein the static characteristic data is acquired through a service server, and the dynamic characteristic data is acquired through a media server; Based on at least one conversion goal for the target user, determining at least one predicted classification type corresponding to the target user; The user feature data is input into an improved deep neural network, and user portrait prediction is performed according to the prediction classification type to obtain a user portrait corresponding to the target user.

2. The method for generating an advertisement user portrait based on a deep neural network according to claim 1, characterized in that: The determining, based on at least one conversion target for the target user, at least one predicted classification type corresponding to the target user includes: Determining a target stage corresponding to the target user based on the user characteristic data of the target user; Based on the target stage, at least one predicted classification type corresponding to the target user is determined.

3. The method for generating an advertisement user portrait based on a deep neural network according to claim 1, characterized in that: The improved deep neural network comprises an input layer, an embedding layer, a concatenation layer, at least one hidden layer and an output layer, The splicing layer includes a first splicing unit and a second splicing unit. The first splicing unit is used to perform addition splicing on the user feature vectors obtained by the embedding layer to obtain a first splicing vector, and the second splicing unit is used to perform product splicing on the user feature vectors obtained by the embedding layer to obtain a second splicing vector; The concatenation layer is used to input the first concatenation vector and the second concatenation vector into the at least one hidden layer respectively.

4. The method for generating an advertisement user portrait based on a deep neural network according to claim 3, characterized in that: Also includes: The at least one hidden layer uses historical user feature data to adjust parameters based on an error back propagation algorithm; The historical user feature data are user features of historical users who already have stage user portraits.

5. The method for generating an advertisement user portrait based on a deep neural network according to claim 4, characterized in that: The error back propagation algorithm is an error back propagation algorithm based on a gradient descent strategy.

6. The method for generating an advertisement user portrait based on a deep neural network according to claim 5, characterized in that: The gradient descent strategy is a stochastic gradient descent strategy.

7. An advertising user portrait generation device based on a deep neural network, characterized in that: include: An acquisition module, used to acquire user characteristic data of a target user, wherein the user characteristic data includes static characteristic data and dynamic characteristic data, wherein the static characteristic data is acquired through a service server, and the dynamic characteristic data is acquired through a media server; a determination module, configured to determine at least one predicted classification type corresponding to the target user based on at least one conversion target for the target user; The prediction module is used to input the user feature data into the improved deep neural network, perform user portrait prediction according to the prediction classification type, and obtain the user portrait corresponding to the target user.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, it implements the advertising user portrait generation method based on deep neural network as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method for generating an advertising user portrait based on a deep neural network as described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for generating an advertising user portrait based on a deep neural network as described in any one of claims 1 to 6 is implemented.

Images