Method and device for predicting advertising conversion characteristics

Abstract

The embodiments of the present application provide a method and device for predicting advertising conversion characteristics. The method includes: obtaining first advertising sample data and second advertising sample data, the first advertising sample data including first attribute feature data and click feature data of the advertisement, and the second advertising sample data including second attribute feature data and conversion feature data of the advertisement; using the first attribute feature data as the first input data and the click feature data as the first label data, the initial network model is trained to obtain an advertising click feature prediction model; using the second attribute feature data as the second input data and the conversion feature data as the second label data, the advertising click feature prediction model is trained to obtain an advertising conversion feature prediction model; according to the target attribute feature data of the target advertisement, the conversion characteristics of the target advertisement are predicted by the advertising conversion feature prediction model. The technical solution of the embodiments of the present application can improve the prediction accuracy of advertising conversion characteristics.

Description

Technical Field

[0001] The present application relates to the field of computer and artificial intelligence technology, and more specifically, to a method and device for predicting advertising conversion characteristics. Background Art

[0002] In the scenario of predicting the conversion characteristics of advertisements, it is usually done by statistically analyzing the historical conversion characteristics of advertisements and then predicting the conversion characteristics of the current advertisement based on the statistical values, or by predicting the conversion characteristic values ​​of the current advertisement based on the attribute characteristics of the current advertisement using traditional neural network methods. However, the inventors of the present application found that the accuracy of the advertisement conversion characteristics predicted by the above two methods is relatively low. Based on this, how to improve the prediction accuracy of advertisement conversion characteristics is a technical problem that needs to be solved urgently. Summary of the Invention

[0003] The embodiments of the present application provide a method, apparatus, computer program product or computer program, computer-readable medium, and electronic device for predicting advertising conversion features, thereby improving the accuracy of predicting advertising conversion features, at least to a certain extent.

[0004] Other features and advantages of the present application will become apparent from the following detailed description, or may be learned in part by practice of the present application.

[0005] According to one aspect of an embodiment of the present application, a method for predicting advertising conversion characteristics is provided, the method comprising: obtaining first advertising sample data and second advertising sample data, the first advertising sample data comprising first attribute characteristic data and click characteristic data of the advertisement, and the second advertising sample data comprising second attribute characteristic data and conversion characteristic data of the advertisement; using the first attribute characteristic data as first input data and the click characteristic data as first label data to train an initial network model to obtain an advertising click characteristic prediction model; using the second attribute characteristic data as second input data and the conversion characteristic data as second label data to train the advertising click characteristic prediction model to obtain an advertising conversion characteristic prediction model; and predicting the conversion characteristics of the target advertisement through the advertising conversion characteristic prediction model based on the target attribute characteristic data of the target advertisement.

[0006] According to one aspect of an embodiment of the present application, a device for predicting advertising conversion characteristics is provided, the device comprising: a first acquisition unit, used to acquire first advertising sample data and second advertising sample data, the first advertising sample data comprising first attribute characteristic data and click characteristic data of the advertisement, and the second advertising sample data comprising second attribute characteristic data and conversion characteristic data of the advertisement; a first training unit, used to train an initial network model using the first attribute characteristic data as first input data and the click characteristic data as first label data to obtain an advertising click characteristic prediction model; a second training unit, used to train the advertising click characteristic prediction model using the second attribute characteristic data as second input data and the conversion characteristic data as second label data to obtain an advertising conversion characteristic prediction model; a prediction unit, used to predict target conversion characteristics of the target advertisement through the advertising conversion characteristic prediction model based on the target attribute characteristic data of the target advertisement.

[0007] In some embodiments of the present application, based on the aforementioned scheme, the first acquisition unit is configured to: obtain advertising sample data of advertisements that have been delivered in history, wherein each piece of advertising sample data also includes the amount of advertising click data and the number of advertising conversions of the corresponding delivered advertisement; determine the advertising sample data whose number of advertising clicks exceeds a first threshold as the first advertising sample data; and determine the advertising sample data whose number of advertising conversions exceeds a second threshold as the second advertising sample data.

[0008] In some embodiments of the present application, based on the aforementioned scheme, the initial network model includes a shared network layer and a first exclusive network layer, and the first training unit is configured to: input the first input data into the shared network layer to output a first intermediate feature value; input the first intermediate feature value into the first exclusive network layer to output a click feature prediction value for the advertisement; determine a click feature prediction error based on the first label data and the click feature prediction value; and reversely update the parameters in the first exclusive network layer and the parameters in the shared network layer based on the click feature prediction error to obtain an advertisement click feature prediction model.

[0009] In some embodiments of the present application, based on the aforementioned scheme, the second training unit is configured to: replace the first exclusive network layer in the advertising click feature prediction model with the second exclusive network layer, and retain the shared network layer in the advertising click feature prediction model to obtain an initial advertising conversion feature prediction model; use the second attribute feature data as the second input data and the conversion feature data as the second label data to train the initial advertising conversion feature prediction model to obtain an advertising conversion feature prediction model.

[0010] In some embodiments of the present application, based on the aforementioned scheme, the second training unit is configured to: input the second input data into the shared network layer of the initial advertising conversion feature prediction model to output a second intermediate feature value; input the second intermediate feature value into the second exclusive network layer to output a conversion feature prediction value for the advertisement; determine a conversion feature prediction error based on the second label data and the conversion feature prediction value; and reversely update the parameters in the second exclusive network layer and the parameters in the shared network layer based on the conversion feature prediction error to obtain an advertising conversion feature prediction model.

[0011] In some embodiments of the present application, based on the aforementioned scheme, the first attribute characteristic data or the second attribute characteristic data or the target attribute characteristic data includes at least one advertising attribute value; the click characteristic data includes the advertising click-through rate or the advertising click cost; and the conversion characteristic data includes the advertising conversion rate or the advertising conversion cost.

[0012] In some embodiments of the present application, based on the aforementioned scheme, the device also includes: a second acquisition unit, which is used to obtain the estimated click-through rate and estimated conversion rate for the target advertisement after predicting the target conversion characteristics of the target advertisement through the advertisement conversion characteristic prediction model; and a determination unit, which is used to determine the pricing strategy for the target advertisement based on the estimated click-through rate, estimated conversion rate, and target conversion characteristics of the target advertisement.

[0013] According to one aspect of an embodiment of the present application, a computer program product or computer program is provided, comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the method for predicting advertising conversion characteristics described in the above embodiments.

[0014] According to one aspect of an embodiment of the present application, a computer-readable medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the method for predicting advertisement conversion features as described in the above embodiments is implemented.

[0015] According to one aspect of an embodiment of the present application, an electronic device is provided, comprising: one or more processors; a storage device for storing one or more programs, which, when executed by the one or more processors, enables the one or more processors to implement the method for predicting advertising conversion features as described in the above embodiments.

[0016] In the technical solutions provided in some embodiments of the present application, by taking the first attribute feature data of the advertisement as the first input data and the click feature data of the advertisement as the first label data, and training the initial network model, the initial network model can learn the correlation between the first attribute feature data of the advertisement and the click feature data of the advertisement, and obtain an advertisement click feature prediction model. Further, by taking the second attribute feature data of the advertisement as the second input data and the conversion feature data of the advertisement as the second label data, and training the advertisement click feature prediction model, the advertisement click feature prediction model can learn the correlation between the second attribute feature data of the advertisement and the conversion feature data of the advertisement, and obtain an advertisement conversion feature prediction model. On the one hand, since the conversion data of the advertisement is very sparse, a large amount of exposure has only a few conversions, and it is impossible to directly learn the high-dimensional cross-information between the attribute features of the advertisement and the conversion features of the advertisement through the model. On the other hand, since the conversion of advertising is not a process from exposure to conversion directly, but a process of exposure, click, and then conversion, click and conversion are related processes. Therefore, this application can make up for the problem of too little advertising conversion data by progressively learning advertising conversion data through learning a model with advertising click data information, thereby improving the accuracy of the advertising conversion feature prediction model, and then improving the accuracy of the advertising conversion features predicted by the advertising conversion feature prediction model.

[0017] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings are incorporated into and constitute a part of the specification, illustrating embodiments consistent with the present application and, together with the specification, explaining the principles of the present application. Obviously, the drawings described below are only some embodiments of the present application, and those skilled in the art can derive other drawings based on these drawings without inventive effort. In the drawings:

[0019] Figure 1 A schematic diagram showing an exemplary system architecture to which the technical solutions of the embodiments of the present application can be applied;

[0020] Figure 2 A flowchart of a method for predicting advertisement conversion characteristics according to one embodiment of the present application is shown;

[0021] Figure 3 A detailed flowchart of obtaining first advertisement sample data and second advertisement sample data according to one embodiment of the present application is shown;

[0022] Figure 4 A detailed flowchart of training an initial network model according to an embodiment of the present application is shown;

[0023] Figure 5 A schematic diagram of a model for training an initial network model according to an embodiment of the present application is shown;

[0024] Figure 6 A detailed flowchart of training an advertisement click feature prediction model according to one embodiment of the present application is shown;

[0025] Figure 7 A detailed flowchart of training an initial advertisement conversion feature prediction model according to one embodiment of the present application is shown;

[0026] Figure 8 A schematic diagram of a model for training an initial advertisement conversion feature prediction model according to one embodiment of the present application is shown;

[0027] Figure 9 A flowchart of a method after predicting the conversion features of the target advertisement by the advertisement conversion feature prediction model according to an embodiment of the present application is shown;

[0028] Figure 10 A block diagram of a device for predicting advertisement conversion features according to an embodiment of the present application is shown;

[0029] Figure 11 A schematic diagram of the structure of a computer system suitable for implementing an electronic device according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0030] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this application will be thorough and complete and will fully convey the concepts of the example embodiments to those skilled in the art.

[0031] In addition, described feature, structure or characteristic can be combined in one or more embodiments in any suitable manner.In the following description, many specific details are provided so as to provide a full understanding of the embodiments of the present application. However, it will be appreciated by those skilled in the art that the technical scheme of the present application can be put into practice without one or more of the specific details, or other methods, components, devices, steps etc. can be adopted. In other cases, known methods, devices, implementations or operations are not shown or described in detail to avoid blurring the various aspects of the application.

[0032] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically separate entities. That is, these functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0033] The flowcharts shown in the accompanying drawings are for illustrative purposes only and do not necessarily include all contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may be decomposed, while others may be combined or partially combined. Therefore, the actual execution order may vary depending on the actual situation.

[0034] It should be noted that the term "plurality" used in this document refers to two or more. "And / or" describes a relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A alone, A and B together, or B alone. The character " / " generally indicates an "or" relationship between the associated objects.

[0035] It should be noted that the terms "first," "second," and the like in the specification and claims of this application and the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, such that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described.

[0036] The embodiments of this application involve technologies related to artificial intelligence. Artificial Intelligence (AI) is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to achieve optimal results. In other words, artificial intelligence is a comprehensive technology in computer science that attempts to understand the essence of intelligence and produce a new type of intelligent machine that can respond in a manner similar to human intelligence. Artificial intelligence is the study of the design principles and implementation methods of various intelligent machines, giving machines the functions of perception, reasoning, and decision-making.

[0037] Artificial intelligence (AI) technology is a comprehensive discipline encompassing a wide range of fields, encompassing both hardware and software technologies. Foundational AI technologies generally include sensors, specialized AI chips, cloud computing, distributed storage, big data processing, operating / interaction systems, and mechatronics. AI software technologies primarily encompass computer vision, speech processing, natural language processing, and machine learning / deep learning.

[0038] Machine learning (ML) is a multidisciplinary field that encompasses probability theory, statistics, approximation theory, convex analysis, and algorithmic complexity theory. It specifically studies how computers can simulate or implement human learning behaviors to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is at the core of artificial intelligence and the fundamental way to make computers intelligent. Its applications span all areas of AI. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and self-learning.

[0039] Figure 1 A schematic diagram shows an exemplary system architecture to which the technical solutions of the embodiments of the present application can be applied.

[0040] like Figure 1 As shown, the system architecture may include terminal devices (such as Figure 1 The device 100 includes one or more of the smartphone 101, tablet computer 102, and portable computer 103 shown in the figure (although other terminal devices with positioning capabilities, such as water meters, electricity meters, etc., may also be used), a network 104, and a server 105. The network 104 is a medium for providing a communication link between the terminal device and the server 105. The network 104 can include various connection types, such as wired communication links, wireless communication links, and the like.

[0041] In one embodiment of the present application, the server 105 may obtain first advertising sample data and second advertising sample data from a terminal device (or other environment), wherein the first advertising sample data includes first attribute feature data and click feature data of the advertisement, and the second advertising sample data includes second attribute feature data and conversion feature data of the advertisement. Then, the server 105 uses the first attribute feature data as the first input data and the click feature data as the first label data to train the initial network model to obtain an advertising click feature prediction model, and uses the second attribute feature data as the second input data and the conversion feature data as the second label data to train the advertising click feature prediction model to obtain an advertising conversion feature prediction model. Finally, the server 105 obtains the target attribute feature data of the target advertisement, and predicts the conversion feature of the target advertisement through the advertising conversion feature prediction model.

[0042] It should be noted that the method for predicting advertising conversion features provided in the embodiments of the present application can be executed by the server 105. Accordingly, the device for predicting advertising conversion features is generally provided in the server 105. However, in other embodiments of the present application, the terminal device may also have similar functions to the server, thereby executing the prediction scheme for advertising conversion features provided in the embodiments of the present application.

[0043] It should also be noted that Figure 1 The number of terminal devices, networks, and servers in the description is for illustrative purposes only. Depending on the implementation requirements, the server may be an independent physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server that provides 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, and big data and artificial intelligence platforms. The terminal may be a smartphone, tablet computer, laptop computer, desktop computer, smart speaker, smart watch, etc., but is not limited thereto and is not limited to this.

[0044] It's important to explain that cloud computing, as described above, is a computing model that distributes computing tasks across a resource pool consisting of a large number of computers, enabling various application systems to access computing power, storage space, and information services as needed. The network that provides these resources is called the "cloud." To users, the resources in the "cloud" appear infinitely scalable and can be accessed, used, and expanded at any time. By establishing a cloud computing resource pool (referred to as a cloud platform, often referred to as an IaaS (Infrastructure as a Service) platform), various types of virtual resources are deployed within the resource pool for external clients to choose from. The cloud computing resource pool primarily includes computing devices (virtualized machines, including operating systems), storage devices, and network devices.

[0045] The following is a detailed description of the implementation details of the technical solution of the embodiment of the present application:

[0046] Figure 2 A flowchart of a method for predicting advertisement conversion features according to an embodiment of the present application is shown. The method for predicting advertisement conversion features can be executed by a device having a computing and processing function, such as Figure 1 Refer to the server 105 shown in Figure 2 As shown, the method for predicting advertisement conversion characteristics includes at least steps 210 to 270, which are described in detail as follows:

[0047] In step 210, first advertisement sample data and second advertisement sample data are obtained, wherein the first advertisement sample data includes first attribute characteristic data and click characteristic data of the advertisement, and the second advertisement sample data includes second attribute characteristic data and conversion characteristic data of the advertisement.

[0048] In this application, a brief explanation of the life cycle of an advertisement is required: during the delivery phase, an advertisement may first be exposed on the Internet according to a predetermined delivery environment and a certain delivery speed. When netizens browse the advertisement, some may choose to ignore the advertisement, while others may choose to click on the advertisement (this is an exposure-click process). Among the netizens who choose to click on the advertisement, some may perform conversion operations such as filling out forms, downloading, activating, and purchasing on the advertisement landing page (that is, the page to which the advertisement jumps after being clicked) (this is an exposure-click-conversion process).

[0049] The inventors of this application discovered that in the advertising life cycle as described above, after an advertisement is exposed, the number of advertisements that are ultimately converted is very small, which makes the advertisement conversion data very sparse. If the advertisement conversion data is directly learned through the model, the model cannot be accurately trained. However, the number of advertisements clicked is much higher than the number of advertisements converted, which makes the advertisement click data much richer than its conversion data. Ad clicks and ad conversions are a related process, and advertisements with high click-through rates also have high interest rates. Therefore, based on this discovery, the invention of this application uses advertisement click data to train the model, so that before the model is trained with advertisement conversion data, the model already has certain learning knowledge, thereby making up for the problem of too little conversion data.

[0050] It should be noted that in step 210, one piece of advertisement sample data corresponds to one advertisement, and obtaining the first advertisement sample data and the second advertisement sample data may be obtaining multiple pieces of first advertisement sample data and multiple pieces of second advertisement sample data, that is, advertisement sample data corresponding to multiple advertisements.

[0051] Specifically, the first advertising sample data includes the first attribute characteristic data of the advertisement, and the second advertising sample data includes the second attribute characteristic data of the advertisement, wherein the attribute characteristic data includes at least one advertising attribute value. For example, the advertising attribute value may include the advertising delivery effect (such as the splash screen effect), the advertising conversion type (such as downloading an APP), the advertising creativity, the advertising delivery area (such as Shenzhen), the industry described in the advertisement (such as the gaming industry), the advertising number, etc.

[0052] Specifically, the first advertising sample data includes advertisement click feature data, and the second advertising sample data includes advertisement conversion feature data, wherein the click feature data includes advertisement click-through rate or advertisement click cost, and the conversion feature data includes advertisement conversion rate or advertisement conversion cost.

[0053] It should be noted that, in this application, the click-through rate (CTR) of an ad may refer to the ratio of the total number of clicks on an ad to the total number of impressions of the ad. The cost per click (CPC) of an ad may refer to the fee borne by the advertiser when the ad is clicked once or a predetermined number of times. The conversion rate of an ad may refer to the ratio of the total number of conversions of an ad to the total number of impressions of the ad. The cost per conversion (CPC) of an ad may refer to the fee borne by the advertiser when the ad is converted once or a predetermined number of times.

[0054] In one embodiment of the present application, the first advertisement sample data and the second advertisement sample data can be obtained as follows: Figure 3 Follow the steps shown.

[0055] See also Figure 3 , shows a detailed flow chart of obtaining first advertisement sample data and second advertisement sample data according to an embodiment of the present application. Specifically, it includes steps 211 to 213:

[0056] In step 211, advertisement sample data of advertisements that have been delivered in the past is obtained, wherein each piece of advertisement sample data also includes the advertisement click data volume and advertisement conversion number of the corresponding delivered advertisement.

[0057] In step 212, the advertisement sample data whose number of advertisement clicks exceeds a first threshold is determined as the first advertisement sample data.

[0058] Specifically, for example, advertisement sample data with an advertisement click count exceeding 30 is determined as the first advertisement sample data.

[0059] In step 213 , the advertisement sample data whose advertisement conversion quantity exceeds the second threshold is determined as the second advertisement sample data.

[0060] Specifically, for example, the advertisement sample data with the number of advertisement conversions exceeding 30 is determined as the second advertisement sample data.

[0061] It should be noted that the first threshold and the second threshold in this embodiment are set according to actual needs and are not limited to a fixed value. In this application, the first threshold and the second threshold are used to filter the first advertising sample data and the second advertising sample data, which can prevent the problem of data sparsity in the advertising sample data, thereby improving data quality.

[0062] Continue to refer to Figure 2 In step 230, the initial network model is trained using the first attribute feature data as the first input data and the click feature data as the first label data to obtain an advertisement click feature prediction model.

[0063] In one embodiment of the present application, the initial network model includes a shared network layer and a first exclusive network layer, wherein the shared network layer and the first exclusive network layer both include learning parameters for learning data features during the model training process.

[0064] In this embodiment, the first attribute feature data is used as the first input data, and the click feature data is used as the first label data to train the initial network model to obtain an advertisement click feature prediction model. Figure 4 Follow the steps shown.

[0065] See also Figure 4 , shows a detailed flow chart of training the initial network model according to one embodiment of the present application. Specifically, it includes steps 231 to 234:

[0066] In step 231 , the first input data is input to the shared network layer to output a first intermediate eigenvalue.

[0067] In step 232, the first intermediate feature value is input to the first exclusive network layer to output a click feature prediction value for the advertisement.

[0068] In step 233, a click feature prediction error is determined based on the first label data and the click feature prediction value.

[0069] In step 234 , the parameters in the first exclusive network layer and the parameters in the shared network layer are reversely updated based on the click feature prediction error to obtain an advertisement click feature prediction model.

[0070] It should be noted that the first intermediate feature value is the learning result of the first attribute feature data output by the shared network layer within the model. The click feature prediction value for the advertisement can refer to the click-through rate of the advertisement or the cost per click of the advertisement.

[0071] In step 233, the click feature prediction value can be compared with the corresponding first label data to determine the click feature prediction error. The click feature prediction error is used to indicate the reliability of the initial network model's prediction of the click feature value. The lower the click feature prediction error, the closer the click feature prediction value of the advertisement obtained by the initial network model is to the actual click feature value, indicating that the initial network model has a stronger predictive ability.

[0072] Specifically, the click feature prediction error is determined, for example, by determining the difference between the first label data and the click feature prediction value as the click feature prediction error.

[0073] Subsequently, in step 234, the parameters in the first exclusive network layer and the parameters in the shared network layer are reversely updated based on the click feature prediction error. The output click feature prediction error is back-propagated and the error is distributed to all units in each layer, thereby obtaining the error signal of the units in each layer, and then correcting the weights of each unit. In essence, this process is a process of weight adjustment.

[0074] In the present application, the parameters in the first exclusive network layer and the parameters in the shared network layer are reversely updated based on the click feature prediction error, so as to enhance the prediction capability of the initial network model.

[0075] It should also be noted that, in the actual process of training the initial network model, steps 231 to 234 may be repeatedly performed until the initial network model meets the first preset condition, thereby obtaining an advertisement click feature prediction model.

[0076] Furthermore, by repeatedly executing steps 231 to 234 until the initial network model meets the first preset condition, an advertisement click feature prediction model is obtained, wherein the first preset condition can be the number of repetitions of steps 231 to 234 or the model reaches convergence.

[0077] See also Figure 5 , shows a model schematic diagram of training an initial network model according to an embodiment of the present application.

[0078] The initial network model can be trained using the first advertisement sample data 510 , and the parameters in the model can be updated based on the click feature prediction value 530 output by the model, ultimately obtaining an advertisement click feature prediction model 520 .

[0079] Continue to refer to Figure 2 In step 250, the advertisement click feature prediction model is trained using the second attribute feature data as the second input data and the conversion feature data as the second label data to obtain an advertisement conversion feature prediction model.

[0080] In one embodiment of the present application, the second attribute feature data is used as the second input data, and the conversion feature data is used as the second label data to train the advertisement click feature prediction model to obtain the advertisement conversion feature prediction model. Figure 6 Follow the steps shown.

[0081] See also Figure 6 , shows a detailed flow chart of training an ad click feature prediction model according to an embodiment of the present application. Specifically, it includes steps 251 to 252:

[0082] In step 251, the first exclusive network layer in the advertisement click feature prediction model is replaced with the second exclusive network layer, and the shared network layer in the advertisement click feature prediction model is retained to obtain an initial advertisement conversion feature prediction model.

[0083] In step 252, the initial advertisement conversion feature prediction model is trained using the second attribute feature data as second input data and the conversion feature data as second label data to obtain an advertisement conversion feature prediction model.

[0084] In this application, the initial network model is first trained using the first attribute feature data and click feature data of the advertisement to obtain an advertisement click feature prediction model, and then the first exclusive network layer in the advertisement click feature prediction model is replaced with the second exclusive network layer, and the shared network layer in the advertisement click feature prediction model is retained to obtain an initial advertisement conversion feature prediction model, and the initial advertisement conversion feature prediction model is trained using the second attribute feature data and conversion feature data of the advertisement to obtain an advertisement conversion feature prediction model, which can effectively transfer the features in the first attribute feature data and click feature data of the advertisement to the advertisement conversion feature prediction model, and by retaining the shared network layer in the advertisement click feature prediction model, the learning results of the first training stage can be retained during the second model training, while greatly reducing the number of parameter adjustments and improving the training speed and training effect of the overall model.

[0085] In this embodiment, the initial advertisement conversion feature prediction model is trained with the second attribute feature data as the second input data and the conversion feature data as the second label data to obtain the advertisement conversion feature prediction model. Figure 7 Follow the steps shown.

[0086] See also Figure 7 , shows a detailed flow chart of training the initial advertising conversion feature prediction model according to one embodiment of the present application. Specifically, it includes steps 2521 to 2524:

[0087] In step 2521, the second input data is input into the shared network layer of the initial advertising conversion feature prediction model to output a second intermediate feature value.

[0088] In step 2522, the second intermediate feature value is input to the second exclusive network layer to output a conversion feature prediction value for the advertisement.

[0089] In step 2523, a conversion feature prediction error is determined based on the second label data and the conversion feature prediction value.

[0090] In step 2524, the parameters in the second exclusive network layer and the parameters in the shared network layer are reversely updated based on the conversion feature prediction error to obtain an advertising conversion feature prediction model.

[0091] In addition, in other specific implementations, step 2524 may also be to reversely update the parameters in the second exclusive network layer or the parameters in the shared network layer based on the conversion feature prediction error to obtain an advertising conversion feature prediction model.

[0092] It should be noted that the second intermediate feature value is the learning result of the second attribute feature data output by the shared network layer within the model. The conversion feature prediction value for the advertisement may refer to the conversion rate of the advertisement, or it may refer to the conversion cost of the advertisement. In step 2523, the conversion feature prediction value may be compared with the corresponding second label data to determine the conversion feature prediction error. The conversion feature prediction error is used to indicate the true credibility of the initial advertisement conversion feature prediction model in predicting the conversion feature value. The lower the conversion feature prediction error, the closer the conversion feature prediction value of the advertisement obtained by the initial advertisement conversion feature prediction model is to the true conversion feature value, and the stronger the prediction ability of the initial advertisement conversion feature prediction model.

[0093] In the present application, the parameters in the second exclusive network layer and the parameters in the shared network layer are reversely updated based on the conversion feature prediction error, or the parameters in the second exclusive network layer are reversely updated based on the conversion feature prediction error, so as to enhance the prediction ability of the initial advertising conversion feature prediction model.

[0094] It should also be noted that in the actual process of training the initial advertising conversion feature prediction model, steps 2521 to 2524 can be repeatedly executed until the initial advertising conversion feature prediction model meets the second preset condition to obtain the advertising conversion feature prediction model.

[0095] Furthermore, by repeatedly executing steps 2521 to 2524 until the initial advertising conversion feature prediction model meets the second preset condition, an advertising conversion feature prediction model is obtained, wherein the second preset condition can be the number of repetitions of steps 2521 to 2524, or the model reaches convergence.

[0096] See also Figure 8 , shows a model schematic diagram of training an initial advertising conversion feature prediction model according to an embodiment of the present application.

[0097] The initial advertisement conversion feature prediction model may be trained using the second advertisement sample data 810 , and the parameters in the model may be updated based on the conversion feature prediction value 830 output by the model, ultimately obtaining an advertisement conversion feature prediction model 820 .

[0098] Continue to refer to Figure 2 In step 270, the conversion characteristics of the target advertisement are predicted using the advertisement conversion characteristic prediction model according to the target attribute characteristic data of the target advertisement.

[0099] In the present application, the conversion characteristics of the target advertisement are predicted using the advertisement conversion characteristic prediction model based on the target attribute characteristic data of the target advertisement. This may be achieved by directly inputting the target attribute characteristic data of the target advertisement into the advertisement conversion characteristic prediction model and directly outputting the conversion characteristics of the target advertisement. The target attribute characteristic data includes at least one advertisement attribute value, and the conversion characteristics of the target advertisement include a conversion rate or a conversion cost of the target advertisement.

[0100] In one embodiment of the present application, after the conversion features of the target advertisement are predicted by the advertisement conversion feature prediction model, the following steps may be performed: Figure 9 Steps shown.

[0101] See also Figure 9 , shows a flow chart of a method after predicting the conversion features of the target advertisement using the advertisement conversion feature prediction model according to an embodiment of the present application. Specifically, it includes steps 281 to 282:

[0102] In step 281 , the estimated click rate and the estimated conversion rate for the target advertisement are obtained.

[0103] In step 282 , a pricing strategy for the target advertisement is determined based on the estimated click-through rate, estimated conversion rate, and conversion characteristics of the target advertisement.

[0104] Specifically, when the conversion feature is the advertisement conversion cost, the pricing strategy for the target advertisement can be determined by the following formula:

[0105] ECPM=OCPA×pctr×pcvr

[0106] or

[0107] CPM=ECPM / 1000=(OCPA×pctr×pcvr) / 1000

[0108] or

[0109] CPC=ECPM / click=(OCPA×pctr×pcvr) / click

[0110] Among them, ECPM represents the price per thousand impressions of the advertisement; OCPA represents the predicted conversion cost of the advertisement; CPM represents the price per single impression of the advertisement; CPC represents the price per single click of the advertisement; pctr represents the estimated click-through rate of the advertisement; pcvr represents the estimated conversion rate of the advertisement; click represents the actual number of clicks on the advertisement.

[0111] In the technical solutions of the above-mentioned embodiments of the present application, the inventors of the present application found that, on the one hand, the conversion data of advertisements is very sparse, with a large number of exposures resulting in only a few conversions, and it is impossible to directly learn the high-dimensional cross-information between advertisement attribute features and advertisement conversion features through the model. On the other hand, the conversion of advertisements is not a process from exposure directly to conversion, but a process of exposure, click, and then conversion. Click and conversion are related processes. Based on this, the present application progressively learns advertisement conversion data by learning a model with advertisement click data information, which can make up for the problem of too little advertisement conversion data, thereby improving the accuracy of the advertisement conversion feature prediction model, and further improving the accuracy of the advertisement conversion features predicted by the advertisement conversion feature prediction model.

[0112] The following describes an embodiment of the device of the present application, which can be used to implement the method for predicting advertising conversion features in the above-mentioned embodiment of the present application. For details not disclosed in the embodiment of the device of the present application, please refer to the embodiment of the method for predicting advertising conversion features in the above-mentioned embodiment of the present application.

[0113] Figure 10 A block diagram of a device for predicting advertisement conversion features according to an embodiment of the present application is shown.

[0114] Reference Figure 10 As shown, an advertisement conversion feature prediction device 1000 according to an embodiment of the present application includes: a first acquisition unit 1001 , a first training unit 1002 , a second training unit 1003 and a prediction unit 1004 .

[0115] Among them, the first acquisition unit 1001 is used to obtain first advertising sample data and second advertising sample data, the first advertising sample data includes first attribute feature data and click feature data of the advertisement, and the second advertising sample data includes second attribute feature data and conversion feature data of the advertisement; the first training unit 1002 is used to use the first attribute feature data as the first input data and the click feature data as the first label data to train the initial network model to obtain an advertising click feature prediction model; the second training unit 1003 is used to use the second attribute feature data as the second input data and the conversion feature data as the second label data to train the advertising click feature prediction model to obtain an advertising conversion feature prediction model; the prediction unit 1004 is used to predict the target conversion feature of the target advertisement through the advertising conversion feature prediction model based on the target attribute feature data of the target advertisement.

[0116] In some embodiments of the present application, based on the aforementioned scheme, the first acquisition unit 1001 is configured to: obtain advertising sample data of advertisements that have been delivered in history, wherein each piece of advertising sample data also includes the amount of advertising click data and the number of advertising conversions of the corresponding delivered advertisement; determine the advertising sample data whose number of advertising clicks exceeds a first threshold as the first advertising sample data; and determine the advertising sample data whose number of advertising conversions exceeds a second threshold as the second advertising sample data.

[0117] In some embodiments of the present application, based on the aforementioned scheme, the initial network model includes a shared network layer and a first exclusive network layer, and the first training unit 1002 is configured to: input the first input data into the shared network layer to output a first intermediate feature value; input the first intermediate feature value into the first exclusive network layer to output a click feature prediction value for the advertisement; determine a click feature prediction error based on the first label data and the click feature prediction value; and reversely update the parameters in the first exclusive network layer and the parameters in the shared network layer based on the click feature prediction error to obtain an advertisement click feature prediction model.

[0118] In some embodiments of the present application, based on the aforementioned scheme, the second training unit 1003 is configured to: replace the first exclusive network layer in the advertising click feature prediction model with the second exclusive network layer, and retain the shared network layer in the advertising click feature prediction model to obtain an initial advertising conversion feature prediction model; use the second attribute feature data as the second input data and the conversion feature data as the second label data to train the initial advertising conversion feature prediction model to obtain an advertising conversion feature prediction model.

[0119] In some embodiments of the present application, based on the aforementioned scheme, the second training unit 1003 is configured to: input the second input data into the shared network layer of the initial advertising conversion feature prediction model to output a second intermediate feature value; input the second intermediate feature value into the second exclusive network layer to output a conversion feature prediction value for the advertisement; determine a conversion feature prediction error based on the second label data and the conversion feature prediction value; and reversely update the parameters in the second exclusive network layer and the parameters in the shared network layer based on the conversion feature prediction error to obtain an advertising conversion feature prediction model.

[0120] In some embodiments of the present application, based on the aforementioned scheme, the first attribute characteristic data or the second attribute characteristic data or the target attribute characteristic data includes at least one advertising attribute value; the click characteristic data includes the advertising click-through rate or the advertising click cost; and the conversion characteristic data includes the advertising conversion rate or the advertising conversion cost.

[0121] In some embodiments of the present application, based on the aforementioned scheme, the device also includes: a second acquisition unit, which is used to obtain the estimated click-through rate and estimated conversion rate for the target advertisement after predicting the target conversion characteristics of the target advertisement through the advertisement conversion characteristic prediction model; and a determination unit, which is used to determine the pricing strategy for the target advertisement based on the estimated click-through rate, estimated conversion rate, and target conversion characteristics of the target advertisement.

[0122] Figure 11 A schematic diagram of the structure of a computer system suitable for implementing an electronic device according to an embodiment of the present application is shown.

[0123] It should be noted that Figure 11 The computer system 1100 of the electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0124] like Figure 11As shown, the computer system 1100 includes a central processing unit (CPU) 1101, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 1102 or the program loaded from the storage part 1108 into the random access memory (RAM) 1103, such as executing the method described in the above embodiment. Various programs and data required for system operation are also stored in the RAM 1103. The CPU 1101, ROM 1102 and RAM 1103 are connected to each other via a bus 1104. An input / output (I / O) interface 1105 is also connected to the bus 1104.

[0125] The following components are connected to the I / O interface 1105: an input section 1106 including a keyboard, a mouse, and the like; an output section 1107 including devices such as a cathode ray tube (CRT), a liquid crystal display (LCD), and a speaker; a storage section 1108 including a hard disk; and a communication section 1109 including a network interface card such as a LAN (Local Area Network) card or a modem. The communication section 1109 performs communication processing via a network such as the Internet. A drive 1110 is also connected to the I / O interface 1105 as needed. Removable media 1111, such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, is installed in the drive 1110 as needed, so that computer programs read from the removable media can be installed in the storage section 1108 as needed.

[0126] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart 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 program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 1109, and / or installed from a removable medium 1111. When the computer program is executed by the central processing unit (CPU) 1101, the various functions defined in the system of the present application are executed.

[0127] It should be noted that the computer-readable medium shown in the embodiments of the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium can 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 can 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), a 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 can 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 can include a data signal propagated in baseband or as part of a carrier wave, which carries a computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, 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 suitable medium, including but not limited to wireless, wired, or any suitable combination thereof.

[0128] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. Among them, each box in the flowchart or block diagram can represent a module, program segment, or part of the code, and the above-mentioned module, program segment, or part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession 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 or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0129] The units involved in the embodiments described in this application may be implemented by software or hardware, and the units described may also be set in a processor. In some cases, the names of these units do not constitute limitations on the units themselves.

[0130] As another aspect, the present application further provides a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the method for predicting advertising conversion characteristics described in the above embodiments.

[0131] As another aspect, the present application also provides a computer-readable medium, which may be included in the electronic device described in the above embodiments, or may exist independently and not be incorporated into the electronic device. The computer-readable medium carries one or more programs, and when executed by the electronic device, the electronic device implements the method for predicting advertising conversion characteristics described in the above embodiments.

[0132] It should be noted that, although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiment of the application, the features and functions of two or more modules or units described above can be concretized 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 concretized.

[0133] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solution according to the embodiments of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a touch terminal, or a network device, etc.) to execute the method according to the embodiments of the present application.

[0134] Those skilled in the art will readily conceive of other embodiments of the present application after considering the specification and practicing the embodiments disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present application that follow the general principles of this application and include common knowledge or customary techniques in the art that are not disclosed herein.

[0135] It should be understood that the present application is not limited to the exact structures described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. A method for predicting advertising conversion characteristics, characterized in that: The method comprises: Acquire first advertisement sample data and second advertisement sample data, wherein the first advertisement sample data includes first attribute characteristic data and click characteristic data of the advertisement, and the second advertisement sample data includes second attribute characteristic data and conversion characteristic data of the advertisement; Using the first attribute feature data as first input data and the click feature data as first label data, the initial network model is trained to obtain an advertisement click feature prediction model; the advertisement click feature prediction model includes a shared network layer and a first exclusive network layer; Replacing the first exclusive network layer in the advertisement click feature prediction model with a second exclusive network layer, and retaining the shared network layer in the advertisement click feature prediction model, to obtain an initial advertisement conversion feature prediction model; Using the second attribute feature data as second input data and the conversion feature data as second label data, the initial advertisement conversion feature prediction model is trained to obtain an advertisement conversion feature prediction model; According to the target attribute feature data of the target advertisement, the conversion feature of the target advertisement is predicted by the advertisement conversion feature prediction model.

2. The method according to claim 1, characterized in that The obtaining of the first advertisement sample data and the second advertisement sample data includes: Obtaining sample data of advertisements that have been placed historically, wherein each piece of sample data also includes the number of clicks and conversions for the corresponding advertisement; Determining the advertisement sample data whose number of advertisement clicks exceeds a first threshold as first advertisement sample data; The advertisement sample data whose advertisement conversion quantity exceeds the second threshold is determined as the second advertisement sample data.

3. The method according to claim 1, characterized in that The initial network model includes the shared network layer and the first exclusive network layer; the initial network model is trained using the first attribute feature data as first input data and the click feature data as first label data to obtain an advertisement click feature prediction model, including: Inputting the first input data into the shared network layer to output a first intermediate eigenvalue; Inputting the first intermediate feature value into the first exclusive network layer to output a click feature prediction value for the advertisement; Determining a click feature prediction error based on the first label data and the click feature prediction value; Based on the click feature prediction error, the parameters in the first exclusive network layer and the parameters in the shared network layer are reversely updated to obtain an advertisement click feature prediction model.

4. The method according to claim 1, wherein The training of the initial advertisement conversion feature prediction model using the second attribute feature data as second input data and the conversion feature data as second label data to obtain the advertisement conversion feature prediction model includes: Inputting the second input data into the shared network layer of the initial advertising conversion feature prediction model to output a second intermediate feature value; Inputting the second intermediate feature value into the second exclusive network layer to output a conversion feature prediction value for the advertisement; determining a conversion feature prediction error based on the second label data and the conversion feature prediction value; The parameters in the second exclusive network layer and the parameters in the shared network layer are reversely updated based on the conversion feature prediction error to obtain an advertisement click feature prediction model.

5. The method according to any one of claims 1 to 4, characterized in that At least one of the first attribute characteristic data, the second attribute characteristic data, and the target attribute characteristic data includes at least one advertisement attribute value.

6. The method according to any one of claims 1 to 4, characterized in that The click characteristic data includes advertisement click rate or advertisement click cost; The conversion feature data includes an advertisement conversion rate or an advertisement conversion cost.

7. The method according to claim 1, characterized in that After predicting the conversion feature of the target advertisement using the advertisement conversion feature prediction model, the method further includes: Get the estimated click-through rate and conversion rate for target ads; A pricing strategy for the target advertisement is determined based on the estimated click-through rate, estimated conversion rate, and conversion characteristics of the target advertisement.

8. A device for predicting advertising conversion characteristics, characterized in that: The device comprises: A first acquisition unit is used to acquire first advertisement sample data and second advertisement sample data, wherein the first advertisement sample data includes first attribute feature data and click feature data of the advertisement, and the second advertisement sample data includes second attribute feature data and conversion feature data of the advertisement; a first training unit, configured to train an initial network model using the first attribute feature data as first input data and the click feature data as first label data to obtain an advertisement click feature prediction model; the advertisement click feature prediction model includes a shared network layer and a first exclusive network layer; a second training unit, configured to replace the first exclusive network layer in the ad click feature prediction model with a second exclusive network layer, while retaining the shared network layer in the ad click feature prediction model, to obtain an initial ad conversion feature prediction model; and train the initial ad conversion feature prediction model using the second attribute feature data as second input data and the conversion feature data as second label data to obtain an ad conversion feature prediction model; The prediction unit is used to predict the conversion characteristics of the target advertisement through the advertisement conversion characteristic prediction model according to the target attribute characteristic data of the target advertisement.

9. The device according to claim 8, characterized in that The first acquiring unit is configured as follows: Obtaining sample data of advertisements that have been placed historically, wherein each piece of sample data also includes the number of clicks and conversions for the corresponding advertisement; Determining the advertisement sample data whose number of advertisement clicks exceeds a first threshold as first advertisement sample data; The advertisement sample data whose advertisement conversion quantity exceeds the second threshold is determined as the second advertisement sample data.

10. The device according to claim 8, characterized in that The initial network model includes the shared network layer and the first exclusive network layer, and the first training unit is configured as follows: Inputting the first input data into the shared network layer to output a first intermediate eigenvalue; Inputting the first intermediate feature value into the first exclusive network layer to output a click feature prediction value for the advertisement; Determining a click feature prediction error based on the first label data and the click feature prediction value; Based on the click feature prediction error, the parameters in the first exclusive network layer and the parameters in the shared network layer are reversely updated to obtain an advertisement click feature prediction model.

11. The device according to claim 8, characterized in that The second training unit is configured as follows: Inputting the second input data into the shared network layer of the initial advertising conversion feature prediction model to output a second intermediate feature value; Inputting the second intermediate feature value into the second exclusive network layer to output a conversion feature prediction value for the advertisement; determining a conversion feature prediction error based on the second label data and the conversion feature prediction value; The parameters in the second exclusive network layer and the parameters in the shared network layer are reversely updated based on the conversion feature prediction error to obtain an advertisement conversion feature prediction model.

12. The device according to any one of claims 8 to 11, characterized in that At least one of the first attribute characteristic data, the second attribute characteristic data, and the target attribute characteristic data includes at least one advertisement attribute value.

13. The device according to any one of claims 8 to 11, characterized in that The click characteristic data includes advertisement click rate or advertisement click cost; The conversion feature data includes an advertisement conversion rate or an advertisement conversion cost.

14. The device according to claim 8, characterized in that The device further comprises: a second acquisition unit, configured to acquire an estimated click-through rate and an estimated conversion rate for a target advertisement after predicting the target conversion feature of the target advertisement using the advertisement conversion feature prediction model; The determination unit is used to determine a pricing strategy for the target advertisement based on the estimated click rate, estimated conversion rate, and target conversion characteristics of the target advertisement.

15. A computer-readable storage medium, characterized in that The computer-readable storage medium stores at least one program code, and the at least one program code is loaded and executed by a processor to implement the method for predicting advertisement conversion characteristics according to any one of claims 1 to 7.

16. A computer device, characterized in that: The computer device includes one or more processors and one or more memories, wherein at least one program code is stored in the one or more memories, and the at least one program code is loaded and executed by the one or more processors to implement the method for predicting advertising conversion characteristics as described in any one of claims 1 to 7.

17. A computer program product, characterized in that The computer program product includes a computer program, which is stored in a computer-readable storage medium. The processor of a computer device reads and executes the computer program from the computer-readable storage medium, so that the computer device executes the method for predicting advertising conversion characteristics according to any one of claims 1 to 7.

Images