Advertisement putting analysis method and device, electronic equipment and storage medium
By generating scene weights and target feature vectors and combining with multi-task network to predict advertising delivery effects, the problem of inaccurate delivery effect prediction in the existing technology is solved, and the accuracy and efficiency of delivery effect are improved.
Patent Information
- Application Number
- CN202411042922.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-31
- Publication Date
- 2024-12-03
AI Technical Summary
The prior art is difficult to accurately predict the effectiveness of advertising through specific channels, resulting in poor delivery results.
By generating scene weights and target feature vectors based on the feature information of the target delivery scene and the advertising/user feature information, input predetermined variables in combination with the multi-task network, determine the effect parameters of the advertising delivery, and fuse the output values to improve prediction accuracy.
It improves the accuracy and efficiency of advertising delivery performance prediction, fully considers the delivery effect in different scenarios, and enhances analysis capabilities.
Smart Images

Figure CN119067727A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of Internet technology, and in particular to an advertisement delivery analysis method, device, electronic device and storage medium. Background Art
[0002] In order to gain the attention of the target group and increase influence, advertisers often place advertisements on Internet multimedia platforms. When placing advertisements through different channel scenarios, there are often significant differences in demand. Before placing advertisements, it is necessary to predict the revenue value of placing advertisements through each channel. Summary of the invention
[0003] The embodiments of the present application provide an advertising delivery analysis method, device, electronic device, and storage medium to solve the technical problem of poor prediction of the effect of advertising delivered through a specific channel.
[0004] According to one aspect of an embodiment of the present application, there is provided an advertising delivery analysis method, including: determining a scenario weight corresponding to a target delivery scenario based on first feature information and second feature information; the first feature information is feature information of the target delivery scenario, and the second feature information at least includes feature information of a target advertisement and / or a target delivery user; generating a target feature vector of the target delivery scenario based on the scenario weight and the second feature information; determining effect parameters of delivering a target advertisement in the target delivery scenario by inputting the target feature vector and predetermined variables into a multi-task network; the predetermined variables represent whether the target advertisement is delivered or not delivered in the target delivery scenario.
[0005] Further, determining the scene weight corresponding to the target delivery scene based on the first feature information and the second feature information includes: generating a first feature vector based on the first feature information, and generating a second feature vector based on the second feature information; merging the first feature vector and the second feature vector and inputting them into a gating network to obtain the scene weight corresponding to the target delivery scene.
[0006] Further, generating a target feature vector of the target delivery scene based on the scene weight and the second feature information includes: performing a predetermined matrix operation on the scene weight and the second feature vector to obtain the target feature vector of the target delivery scene.
[0007] Further, the predetermined variables include: a first variable characterizing the delivery of the target advertisement in the target delivery scenario; determining the effect parameters of delivering the target advertisement in the target delivery scenario by inputting the target feature vector and the predetermined variables into a multi-task network includes: inputting the target feature vector and the first variable into a multi-task network; obtaining a first output value set output by the multi-task network based on the target feature vector and the first variable; the first output value set includes at least one output value characterizing the delivery effect; and determining the effect parameters of delivering the target advertisement in the target delivery scenario based on the first output value set.
[0008] Furthermore, the predetermined variables also include: a second variable characterizing that the target advertisement is not delivered in the target delivery scenario; determining the effect parameters of delivering the target advertisement in the target delivery scenario based on the first output value set includes: inputting the target feature vector and the second variable into a multi-task network; obtaining a second output value set output by the multi-task network based on the target feature vector and the second variable; and determining the effect parameters of delivering the target advertisement in the target delivery scenario based on the first output value set and the second output value set.
[0009] Further, determining the effect parameters of delivering the target advertisement in the target delivery scenario based on the first output value set and the second output value set includes: fusing the first output value set to obtain a first fusion value; fusing the second output value set to obtain a second fusion value; and determining the difference between the first fusion value and the second fusion value as the effect parameters of delivering the target advertisement in the target delivery scenario.
[0010] Furthermore, the first output value set includes at least one of the following: an estimated conversion rate, an estimated click-through rate, an estimated number of clicks, an estimated playback time, and an estimated revenue value.
[0011] According to another aspect of an embodiment of the present application, an advertising delivery analysis device is also provided, including: a determination module, used to determine a scenario weight corresponding to a target delivery scenario based on first feature information and second feature information; the first feature information is feature information of the target delivery scenario, and the second feature information at least includes feature information of a target advertisement and / or a target delivery user; a generation module, used to generate a target feature vector of the target delivery scenario based on the scenario weight and the second feature information; a processing module, used to determine effect parameters of delivering a target advertisement in the target delivery scenario by inputting the target feature vector and predetermined variables into a multi-task network; the predetermined variables represent whether the target advertisement is delivered or not delivered in the target delivery scenario.
[0012] Furthermore, the determination module is specifically used to: generate a first feature vector based on the first feature information, and generate a second feature vector based on the second feature information; merge the first feature vector and the second feature vector and input them into the gated network to obtain the scene weight corresponding to the target delivery scene.
[0013] Furthermore, the generation module is specifically used to: perform a predetermined matrix operation on the scene weight and the second feature vector to obtain a target feature vector of the target delivery scene.
[0014] Further, the predetermined variables include: a first variable characterizing the delivery of the target advertisement in the target delivery scenario; the processing module is specifically used to: input the target feature vector and the first variable into a multi-task network; obtain a first output value set output by the multi-task network based on the target feature vector and the first variable; the first output value set includes at least one output value characterizing the delivery effect; and determine the effect parameters of delivering the target advertisement in the target delivery scenario based on the first output value set.
[0015] Furthermore, the predetermined variables also include: a second variable characterizing that the target advertisement is not delivered in the target delivery scenario; the processing module is specifically used to: input the target feature vector and the second variable into a multi-task network; obtain a second output value set output by the multi-task network based on the target feature vector and the second variable; and determine effect parameters of delivering the target advertisement in the target delivery scenario based on the first output value set and the second output value set.
[0016] Further, the processing module is specifically used to: fuse the first output value set to obtain a first fusion value; fuse the second output value set to obtain a second fusion value; determine the difference between the first fusion value and the second fusion value as an effect parameter of delivering the target advertisement in the target delivery scenario.
[0017] Furthermore, the first output value set includes at least one of the following: an estimated conversion rate, an estimated click-through rate, an estimated number of clicks, an estimated playback time, and an estimated revenue value.
[0018] According to another aspect of an embodiment of the present application, a storage medium is further provided, which includes a stored program, and the above steps are executed when the program is run.
[0019] According to another aspect of an embodiment of the present application, there is also provided an electronic device, including a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus; wherein: the memory is used to store computer programs; the processor is used to execute the steps in the above method by running the program stored in the memory.
[0020] The embodiment of the present application also provides a computer program product including instructions, which, when executed on a computer, enables the computer to execute the steps in the above method.
[0021] An advertising delivery analysis method, device, electronic device and storage medium proposed in the embodiments of the present application determine the scene weight corresponding to the target delivery scene based on first feature information and second feature information; the first feature information is feature information of the target delivery scene, and the second feature information at least includes feature information of the target advertisement and / or the target delivery user; a target feature vector of the target delivery scene is generated based on the scene weight and the second feature information; the effect parameters of delivering the target advertisement in the target delivery scene are determined by inputting the target feature vector and predetermined variables into a multi-task network; the predetermined variables represent whether the target advertisement is delivered or not delivered in the target delivery scene. In this way, by weighting and fusing the features of the delivery scene with the features of the advertisement or the user, the impact on the delivery effect in different scenarios can be fully considered, and then combined with the multi-task network and the variable input of whether to deliver the advertisement, the effect of delivering the advertisement in a specific scene can be more accurately predicted, thereby improving the accuracy and efficiency of the analysis. BRIEF DESCRIPTION OF THE DRAWINGS
[0022] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:
[0023] Figure 1 is a hardware structure block diagram of a computer according to an embodiment of the present application;
[0024] Figure 2 is a flow chart of an advertising delivery analysis method according to an embodiment of the present application;
[0025] Figure 3 is a flow chart of an advertising delivery analysis method according to an embodiment of the present application;
[0026] Figure 4 is a flow chart of an advertising delivery analysis method according to an embodiment of the present application;
[0027] Figure 5 is a structural block diagram of an advertising delivery analysis device according to an embodiment of the present application;
[0028] Figure 6 It is a structural diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION
[0029] In order to enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only embodiments of a part of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work should fall within the scope of protection of the present application. It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict.
[0030] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0031] The method embodiment provided in the first embodiment of the present application can be executed in a mobile phone, a computer, a tablet or a similar computing device. Taking running on a computer as an example, Figure 1 is a hardware structure block diagram of a computer in an embodiment of the present application. Figure 1 As shown, the computer may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data. Optionally, the computer may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above-mentioned computer. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.
[0032] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as a computer program corresponding to an advertising delivery analysis method in an embodiment of the present application. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the computer via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.
[0033] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the computer. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.
[0034] In this embodiment, an advertisement delivery analysis method is provided. Figure 2 is a flow chart of an advertising delivery analysis method according to an embodiment of the present application. Figure 2 As shown, the process includes the following steps:
[0035] S10: Determine a scenario weight corresponding to a target delivery scenario based on the first feature information and the second feature information; the first feature information is feature information of the target delivery scenario, and the second feature information includes at least feature information of the target advertisement and / or the target delivery user.
[0036] In this embodiment, the advertising delivery analysis method can be applied to a delivery model. The delivery scenario can be a delivery channel, a delivery platform or a delivery environment, for example, it can include various applications or web pages, etc. In this delivery scenario, advertisements can be delivered through the Real Time-Application Programming Interface (RTA). Target advertisements are advertisements delivered by advertisers through target delivery scenarios, for example, they can include one or more forms such as text, pictures, videos, and voice. Target delivery users are the target population for target advertising delivery, and target delivery users can receive target advertisements in target delivery scenarios.
[0037] In one embodiment, the first characteristic information may include at least one of the following: channel type, target number of advertisements delivered in the channel, channel price, statistical characteristics of channel delivery effects, and other characteristics related to the channel.
[0038] In one embodiment, the characteristic information of the target advertisement may include at least one of the following: advertisement type, total advertisement duration, advertisement size, advertisement content type, and the like.
[0039] In one embodiment, the characteristic information of the target delivery user may include at least one of the following: user gender, user age, user education, user location, user type, user preference, user online time, user browsing time, user views, and user clicks, etc.
[0040] In one embodiment, the first characteristic information can be obtained according to the target delivery scenario, for example, by calling or querying the first characteristic information corresponding to the target delivery scenario. The second characteristic information can be determined according to the target advertisement to determine the target delivery user, or directly according to the target advertisement. For example, the corresponding second characteristic information is determined according to the type of the target advertisement.
[0041] In one embodiment, the scene weight may represent the importance of the first feature information corresponding to the target delivery scene in the advertisement delivery. For example, step S10 may include: generating a first feature vector based on the first feature information, and generating a second feature vector based on the second feature information; merging the first feature vector and the second feature vector and inputting them into the gating network to obtain the scene weight corresponding to the target delivery scene.
[0042] Here, the gating network may be a deep neural network (DNN).
[0043] In one embodiment, in order to improve the prediction efficiency of the model, feature information corresponding to multiple target delivery scenarios can be input simultaneously to determine the corresponding effect parameters. Therefore, determining the scene weight corresponding to the target delivery scenario can include: determining the scene weight corresponding to one or more target delivery scenarios. For example, based on the first feature information and the second feature information, multiple scene weights corresponding to multiple target delivery scenarios can be determined respectively. Among them, the first feature information includes the feature information of each of the multiple target delivery scenarios. In this way, the feature information of multiple delivery scenarios can be processed and the effect predicted through one model, thereby improving the prediction efficiency and accuracy.
[0044] S20: Generate a target feature vector for the target delivery scenario based on the scenario weight and the second feature information.
[0045] In one embodiment, the scenario weight is applied to the second feature information to obtain a target feature vector, which can represent the characteristics of delivering the target advertisement in the target delivery scenario.
[0046] In one embodiment, when step S10 includes determining multiple scene weights corresponding to multiple target delivery scenarios based on the first feature information and the second feature information, step S20 may include: generating multiple target feature vectors for the multiple target delivery scenarios based on the multiple scene weights corresponding to the multiple target delivery scenarios and the second feature information.
[0047] In one embodiment, step S20 may include: performing a Hadamard product operation on the scene weight and the second eigenvector to obtain a target eigenvector of the target delivery scene.
[0048] In one embodiment, step S20 may include: performing Hadamard product operations on the plurality of scene weights and the second eigenvector respectively, to obtain a plurality of target eigenvectors of the plurality of target delivery scenes respectively.
[0049] S30: Determine effect parameters of delivering a target advertisement in a target delivery scenario by inputting a target feature vector and a predetermined variable into a multi-task network; the predetermined variable represents whether the target advertisement is delivered or not delivered in the target delivery scenario.
[0050] In one embodiment, the multi-task network is used to determine estimated values for multiple targets that characterize the effectiveness or value of advertising delivery in a target delivery scenario, for example, to determine an estimated value for at least one of click-through rate, click volume, conversion rate, pull-up rate, playback duration, revenue value, etc.
[0051] In one embodiment, the predetermined variable may refer to a processing variable input by the network, for example, the variable is to place the target advertisement in the target delivery scenario, or not to place the target advertisement in the target delivery scenario. Step S30 may include: determining the effect parameter of placing the target advertisement in the target delivery scenario by inputting the target feature vector and a first variable representing the placement of the target advertisement in the target delivery scenario into the multi-task network, and inputting the target feature vector and a second variable representing the non-placement of the target advertisement in the target delivery scenario into the multi-task network.
[0052] Here, the multi-task network may be a multi-gate mixture of experts (MMOE) network, or a personalized learning environment (PLE) network.
[0053] In one embodiment, the process of inputting predetermined variables to obtain effect parameters can be implemented through a causal inference network, such as S-Learner, DragonNet and other networks.
[0054] In one embodiment, step S30 may include: inputting the target feature vector and the first variable into a multi-task network; obtaining a first output value set output by the multi-task network based on the target feature vector and the first variable; the first output value set includes at least one output value representing the effect of delivery; and determining the effect parameters of delivering the target advertisement in the target delivery scenario based on the first output value set.
[0055] In one embodiment, step S30 may include: inputting the target feature vector and the first variable into a multi-task network; obtaining a first output value set output by the multi-task network based on the target feature vector and the first variable; inputting the target feature vector and the second variable into the multi-task network; obtaining a second output value set output by the multi-task network based on the target feature vector and the second variable; and determining effect parameters of delivering the target advertisement in the target delivery scenario based on the first output value set and the second output value set.
[0056] Here, the output values in the output value set may include at least one of the following: estimated conversion rate, estimated click-through rate, estimated click volume, estimated play time, and estimated revenue value. That is, the first output value set includes the output values corresponding to the target advertisement being delivered in the target delivery scenario, and the second output value set includes the output values corresponding to the target advertisement not being delivered in the target delivery scenario.
[0057] In one embodiment, determining the effect parameters of delivering the target advertisement in the target delivery scenario based on the first output value set and the second output value set may include: fusing the first output value set to obtain a first fused value; fusing the second output value set to obtain a second fused value; and determining the difference between the first fused value and the second fused value as the effect parameters of delivering the target advertisement in the target delivery scenario.
[0058] In one embodiment, step S30 may include: inputting multiple target feature vectors and predetermined variables into a multi-task network to respectively determine multiple effect parameters of delivering target advertisements in multiple target delivery scenarios. After each target feature vector and predetermined variable are input into the multi-task network, a corresponding effect parameter may be determined. Therefore, a model may be used to predict delivery effects for multiple delivery scenarios and multiple task targets (i.e., output values), thereby improving the efficiency of model prediction.
[0059] In this way, by weighting the characteristics of the delivery scenario with the characteristics of the advertisement or user, we can fully consider the impact of different scenarios on the delivery effect. Then, by combining the multi-task network and the variable input of whether to deliver advertisements, we can more accurately predict the effect of delivering advertisements in specific scenarios and improve the accuracy and efficiency of analysis.
[0060] like Figure 3 As shown, this embodiment provides an advertisement delivery analysis method, including:
[0061] S11: generating a first feature vector based on the first feature information, and generating a second feature vector based on the second feature information;
[0062] S21: The first feature vector and the second feature vector are combined and input into a gating network to obtain a scene weight corresponding to the target delivery scene;
[0063] S31: Generate a target feature vector of the target delivery scene based on the scene weight and the second feature information;
[0064] S41: Determine effect parameters of delivering a target advertisement in a target delivery scenario by inputting a target feature vector and a predetermined variable into a multi-task network; the predetermined variable represents whether the target advertisement is delivered or not delivered in the target delivery scenario.
[0065] In this embodiment, the specific implementation and related embodiments of steps S31 and S41 can refer to the specific implementation and related embodiments of the aforementioned steps S20 and S30, which will not be repeated here. The first characteristic information is characteristic information of the target delivery scene, and the second characteristic information at least includes characteristic information of the target advertisement and / or the target delivery user.
[0066] In one embodiment, generating a first feature vector based on the first feature information may be performed by forming the first feature vector with at least one feature included in the first feature information. Similarly, generating a second feature vector based on the second feature information may be performed by forming the second feature vector with at least one feature included in the second feature information.
[0067] In one embodiment, step S11 may include: generating a first feature vector based on a plurality of first feature information respectively; and generating a second feature vector based on a second feature information. Step S21 may include: merging the plurality of first feature vectors with the second feature vector respectively and inputting them into a gating network to obtain a plurality of scene weights corresponding to a plurality of target delivery scenes respectively.
[0068] Here, the gating network may be a deep neural network (DNN) or the like.
[0069] In this way, the obtained scenario weight can more accurately represent the importance of the scenario characteristics of the corresponding target delivery scenario in the delivery effect, so that the corresponding delivery effect can be predicted more accurately and the impact of the differences between different scenarios on the prediction effect can be fully reduced.
[0070] In some embodiments, step S20 or S31 may include: performing a predetermined matrix operation on the scene weight and the second eigenvector to obtain a target eigenvector of the target delivery scene. Here, the predetermined matrix operation may be a Hadamard product operation or other types of matrix operations.
[0071] In one embodiment, for multiple delivery scenarios, step S20 or S31 may include: performing a predetermined matrix operation on the multiple scenario weights and the second feature vector respectively to obtain multiple target feature vectors of the multiple target delivery scenarios.
[0072] In this way, the obtained target feature vector can accurately integrate the characteristics of the corresponding delivery scenario and other conventional features such as advertisements or users, thereby improving the accuracy of effect prediction.
[0073] In some embodiments, the predetermined variable includes: a first variable characterizing the delivery of the target advertisement in the target delivery scenario. Step S30 or S41 may include: inputting the target feature vector and the first variable into the multi-task network; obtaining a first output value set output by the multi-task network based on the target feature vector and the first variable; the first output value set includes at least one output value characterizing the delivery effect; and determining the effect parameter of delivering the target advertisement in the target delivery scenario based on the first output value set.
[0074] In one embodiment, for multiple delivery scenarios, step S30 or S41 may include: inputting multiple target feature vectors and the first variable into a multi-task network; obtaining multiple first output value sets output by the multi-task network based on the multiple target feature vectors and the first variable; and determining multiple effect parameters for delivering the target advertisements in multiple target delivery scenarios based on the multiple first output value sets.
[0075] Here, each target delivery scenario corresponds to a target feature vector, a first output value set, and an effect parameter. At least one output value in each first output value set can represent the expected delivery effect of the advertisement delivered in the delivery scenario corresponding to it.
[0076] In this way, based on the first variable, the obtained effect parameter can accurately express the effect of delivering the target advertisement in the target delivery scenario.
[0077] In one embodiment, determining the effect parameters of delivering a target advertisement in a target delivery scenario based on a first output value set may include: inputting a target feature vector and a second variable into a multi-task network; obtaining a second output value set output by the multi-task network based on the target feature vector and the second variable; and determining the effect parameters of delivering the target advertisement in the target delivery scenario based on the first output value set and the second output value set.
[0078] Here, the target feature vector is input into the multi-task network twice with the first variable and the second variable respectively, so that the output value set corresponding to placing the target advertisement in the target delivery scenario and the output value set corresponding to not placing the target advertisement in the target delivery scenario can be obtained respectively.
[0079] In one embodiment, for multiple delivery scenarios, determining effect parameters of delivering a target advertisement in the target delivery scenario based on a first output value set may include: inputting multiple target feature vectors and a second variable into a multi-task network; obtaining multiple second output value sets output by the multi-task network based on the multiple target feature vectors and the second variable; and determining multiple effect parameters of delivering the target advertisement in the multiple target delivery scenarios based on the multiple first output value sets and the multiple second output value sets.
[0080] Here, determining multiple effect parameters for delivering target advertisements in multiple target delivery scenarios based on multiple first output value sets and multiple second output value sets may refer to determining the corresponding effect parameters based on the first output value set and the second output value set corresponding to each target delivery scenario.
[0081] In one embodiment, the effect parameters of delivering a target advertisement in a target delivery scenario are determined based on the first output value set and the second output value set. The difference between the output values of the same type in the first output value set and the second output value set can be calculated, and the final difference set can be merged to obtain the effect parameters.
[0082] Here, the fused difference set may be a fusion layer that inputs the difference set into a network, and the obtained effect parameter may represent the effect gain of placing the advertisement in the scenario compared to not placing the advertisement.
[0083] In one embodiment, determining the effect parameters of delivering a target advertisement in a target delivery scenario based on a first output value set and a second output value set may include: fusing the first output value set to obtain a first fused value; fusing the second output value set to obtain a second fused value; and determining the difference between the first fused value and the second fused value as the effect parameters of delivering the target advertisement in the target delivery scenario.
[0084] Here, fusing the output value set may be inputting the output value set into a fusion layer of the network to obtain a corresponding fusion value.
[0085] In this way, by comparing the effect output values when advertising is placed and not placed in different scenarios, the gain brought by advertising can be more intuitively and accurately reflected through difference calculation, thereby being used to accurately judge the effect of advertising.
[0086] In one embodiment, for multiple delivery scenarios, determining the effect parameters of delivering the target advertisement in the target delivery scenario based on the first output value set and the second output value set may include: fusing multiple first output value sets respectively to obtain multiple first fusion values; fusing multiple second output value sets respectively to obtain multiple second fusion values; determining the difference between the first fusion value and the second fusion value corresponding to each target delivery scenario as the effect parameter of delivering the target advertisement in the target delivery scenario.
[0087] In this way, for multiple delivery scenarios, one model can be used to complete the delivery gain prediction of multiple scenarios and multi-task targets, which is conducive to accurately determining the delivery effect.
[0088] In some embodiments, the first output value set may include at least one of the following: estimated conversion rate, estimated click-through rate, estimated click volume, estimated play time, estimated pull-up rate, and estimated revenue value, etc. Similarly, the second output value set also includes at least one of the above.
[0089] In one embodiment, for the same target delivery scenario, the types of output values included in the corresponding first output value set and the second output value set may be the same, so as to facilitate the judgment of the gain situation through the fusion value.
[0090] In one embodiment, for different target delivery scenarios, the output value types included in the corresponding first output value sets may be the same or different, and the output value types included in the corresponding second output value sets may be the same or different.
[0091] As a possible implementation, the advertising delivery analysis method may include:
[0092] When advertisers place advertisements through the RTA system, they need to judge the value of users. In this process, we must first consider the issue of multiple scenarios. Different scenarios have different placement effects. Therefore, when estimating the value of users, we need to process the scene features separately to obtain the scene feature vector, and then fuse the scene features and conventional features to obtain the scene weight through the gating unit, and then apply the weight to the conventional features to obtain the scene-specific feature vector.
[0093] After obtaining this feature vector, it is necessary to make value judgments on the user from multiple angles. At this time, the feature vector needs to be input into a multi-task network, such as MMOE, PLE, etc., to obtain multiple value estimates, and then fuse them into a fused estimate.
[0094] Finally, advertising focuses on the gain value of users, that is, the gain value of advertising and not advertising. At this time, it is necessary to combine the above network structure with the network structure of causal inference, such as S-Learner, DragonNet, etc., to obtain the gain value of the fusion estimate of advertising and not advertising in a specific scenario. The specific process and network structure are as follows Figure 4 As shown, the steps may include:
[0095] 1. Obtain various features of each channel during RTA delivery, including delivery channels, target number of ads delivered by channels, channel bids, channel delivery effect statistical features, etc. These features constitute the first feature vector, namely the scenario feature vector.
[0096] 2. The scene feature vector is merged with other regular feature vectors (i.e., the second feature vector), and then the scene weight is obtained through a gating network, such as a DNN.
[0097] 3. Perform Hadamard product on the scene weight and the regular feature vector to obtain the scene-specific feature vector, i.e., the target feature vector, which is used as the input of the multi-task network.
[0098] 4. The scene-specific feature vector is input into the multi-task network, such as MMOE, PLE, to obtain the scene-specific multi-objective estimation value, that is, the first output value set and the second output value set.
[0099] 5. Fuse the multi-target estimated values to obtain a fused estimated value, namely, a first fused value and a second fused value.
[0100] 6. Combined with causal inference networks, such as S-Learner and DragonNet, we can obtain the scene-specific fusion estimates of advertising and non-advertising, and obtain the gain value of the fusion estimate, that is, the effect parameter.
[0101] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.
[0102] In this embodiment, an advertising delivery analysis device is also provided to implement the above-mentioned embodiments and preferred implementation modes, which have been described and will not be repeated. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.
[0103] Figure 5 is a structural block diagram of an advertising delivery analysis device according to an embodiment of the present application, such as Figure 5 As shown, the device includes: a determination module 100, a generation module 200, and a processing module 300, wherein:
[0104] The determination module 100 is used to determine the scene weight corresponding to the target delivery scene based on the first characteristic information and the second characteristic information; the first characteristic information is the characteristic information of the target delivery scene, and the second characteristic information at least includes the characteristic information of the target advertisement and / or the target delivery user;
[0105] A generating module 200, configured to generate a target feature vector of the target delivery scene based on the scene weight and the second feature information;
[0106] The processing module 300 is used to determine the effect parameters of delivering the target advertisement in the target delivery scenario by inputting the target feature vector and a predetermined variable into a multi-task network; the predetermined variable represents whether to deliver or not deliver the target advertisement in the target delivery scenario.
[0107] Optionally, the determination module 100 is specifically used to: generate a first feature vector based on the first feature information, and generate a second feature vector based on the second feature information; merge the first feature vector and the second feature vector and input them into a gating network to obtain a scene weight corresponding to the target delivery scene.
[0108] Optionally, the generation module 200 is specifically used to: perform a predetermined matrix operation on the scene weight and the second feature vector to obtain a target feature vector of the target delivery scene.
[0109] Optionally, the predetermined variables include: a first variable characterizing the delivery of the target advertisement in the target delivery scenario; the processing module 300 is specifically used to: input the target feature vector and the first variable into a multi-task network; obtain a first output value set output by the multi-task network based on the target feature vector and the first variable; the first output value set includes at least one output value characterizing the delivery effect; and determine the effect parameters of delivering the target advertisement in the target delivery scenario based on the first output value set.
[0110] Optionally, the predetermined variables also include: a second variable characterizing that the target advertisement is not delivered in the target delivery scenario; the processing module 300 is specifically used to: input the target feature vector and the second variable into a multi-task network; obtain a second output value set output by the multi-task network based on the target feature vector and the second variable; and determine effect parameters of delivering the target advertisement in the target delivery scenario based on the first output value set and the second output value set.
[0111] Optionally, the processing module 300 is specifically used to: fuse the first output value set to obtain a first fused value; fuse the second output value set to obtain a second fused value; determine the difference between the first fused value and the second fused value as an effect parameter of delivering the target advertisement in the target delivery scenario.
[0112] Optionally, the first output value set includes at least one of the following: estimated conversion rate, estimated click-through rate, estimated click volume, estimated playback time and estimated revenue value.
[0113] It should be noted that the above modules can be implemented by software or hardware. For the latter, it can be implemented in the following ways, but not limited to: the above modules are all located in the same processor; or the above modules are located in different processors in any combination.
[0114] The embodiment of the present application also provides an electronic device, Figure 6 is a structural diagram of an electronic device according to an embodiment of the present application, such as Figure 6As shown, it includes a processor 41, a communication interface 42, a memory 43 and a communication bus 44, wherein the processor 41, the communication interface 42 and the memory 43 communicate with each other through the communication bus 44, and the memory 43 is used to store computer programs; when the processor 41 is used to execute the program stored in the memory 43, the following steps are implemented: based on the first feature information and the second feature information, the scene weight corresponding to the target delivery scene is determined; the first feature information is the feature information of the target delivery scene, and the second feature information at least includes the feature information of the target advertisement and / or the target delivery user; based on the scene weight and the second feature information, a target feature vector of the target delivery scene is generated; by inputting the target feature vector and a predetermined variable into a multi-task network, the effect parameters of delivering the target advertisement in the target delivery scene are determined; the predetermined variable represents whether the target advertisement is delivered or not delivered in the target delivery scene.
[0115] Further, determining the scene weight corresponding to the target delivery scene based on the first feature information and the second feature information includes: generating a first feature vector based on the first feature information, and generating a second feature vector based on the second feature information; merging the first feature vector and the second feature vector and inputting them into a gating network to obtain the scene weight corresponding to the target delivery scene.
[0116] Further, generating a target feature vector of the target delivery scene based on the scene weight and the second feature information includes: performing a predetermined matrix operation on the scene weight and the second feature vector to obtain the target feature vector of the target delivery scene.
[0117] Furthermore, the method of determining the effect parameters of delivering the target advertisement in the target delivery scenario by inputting the target feature vector and the predetermined variable into the multi-task network includes: inputting the target feature vector and the first variable into the multi-task network; obtaining a first output value set output by the multi-task network based on the target feature vector and the first variable; the first output value set includes at least one output value representing the delivery effect; and determining the effect parameters of delivering the target advertisement in the target delivery scenario based on the first output value set.
[0118] Further, determining the effect parameters of delivering the target advertisement in the target delivery scenario based on the first output value set includes: inputting the target feature vector and the second variable into a multi-task network; obtaining a second output value set output by the multi-task network based on the target feature vector and the second variable; and determining the effect parameters of delivering the target advertisement in the target delivery scenario based on the first output value set and the second output value set.
[0119] Further, determining the effect parameters of delivering the target advertisement in the target delivery scenario based on the first output value set and the second output value set includes: fusing the first output value set to obtain a first fusion value; fusing the second output value set to obtain a second fusion value; and determining the difference between the first fusion value and the second fusion value as the effect parameters of delivering the target advertisement in the target delivery scenario.
[0120] Furthermore, the first output value set includes at least one of the following: an estimated conversion rate, an estimated click-through rate, an estimated number of clicks, an estimated playback time, and an estimated revenue value.
[0121] The communication bus mentioned in the above terminal can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.
[0122] The communication interface is used for communication between the above terminal and other devices.
[0123] The memory may include a random access memory (RAM) or a non-volatile memory, such as at least one disk memory. Optionally, the memory may also be at least one storage device located away from the aforementioned processor.
[0124] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.
[0125] In another embodiment provided in the present application, a computer-readable storage medium is provided, in which instructions are stored. When the computer-readable storage medium is executed on a computer, the computer executes the advertising delivery analysis method described in any one of the above embodiments.
[0126] In another embodiment provided by the present application, a computer program product including instructions is also provided, which, when executed on a computer, enables the computer to execute the advertising delivery analysis method described in any one of the above embodiments.
[0127] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from a website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium may be a magnetic medium, (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive Solid State Disk (SSD)), etc.
[0128] The above description is only a preferred embodiment of the present application and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application are included in the protection scope of the present application.
[0129] The above description is only a specific implementation of the present application, so that those skilled in the art can understand or implement the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest range consistent with the principles and novel features applied for herein.
Claims
1. An advertising delivery analysis method, characterized in that: include: Determine a scenario weight corresponding to a target delivery scenario based on first feature information and second feature information; the first feature information is feature information of the target delivery scenario, and the second feature information includes at least feature information of a target advertisement and / or a target delivery user; Generate a target feature vector for the target delivery scene based on the scene weight and the second feature information; The effect parameters of delivering the target advertisement in the target delivery scenario are determined by inputting the target feature vector and the predetermined variables into the multi-task network; the predetermined variables represent whether to deliver the target advertisement in the target delivery scenario or not.
2. The advertising delivery analysis method according to claim 1, characterized in that: The determining, based on the first feature information and the second feature information, a scene weight corresponding to the target delivery scene includes: generating a first feature vector based on the first feature information, and generating a second feature vector based on the second feature information; The first feature vector and the second feature vector are combined and input into the gating network to obtain the scene weight corresponding to the target delivery scene.
3. The advertising delivery analysis method according to claim 2, characterized in that: The step of generating a target feature vector for the target delivery scene based on the scene weight and the second feature information includes: A predetermined matrix operation is performed on the scene weight and the second feature vector to obtain a target feature vector of the target delivery scene.
4. The advertising delivery analysis method according to claim 1, characterized in that: The predetermined variables include: a first variable characterizing the delivery of the target advertisement in the target delivery scenario; The step of inputting the target feature vector and the predetermined variable into a multi-task network to determine the effect parameter of delivering the target advertisement in the target delivery scenario includes: Inputting the target feature vector and the first variable into a multi-task network; Obtaining a first output value set output by the multi-task network based on the target feature vector and the first variable; the first output value set includes at least one output value representing a delivery effect; An effect parameter of delivering a target advertisement in the target delivery scenario is determined based on the first output value set.
5. The advertising delivery analysis method according to claim 4, characterized in that: The predetermined variables also include: a second variable representing that the target advertisement is not delivered in the target delivery scenario; The determining, based on the first output value set, an effect parameter of delivering a target advertisement in the target delivery scenario includes: Inputting the target feature vector and the second variable into a multi-task network; Obtaining a second output value set output by the multi-task network based on the target feature vector and the second variable; The effect parameters of delivering the target advertisement in the target delivery scenario are determined based on the first output value set and the second output value set.
6. The advertising delivery analysis method according to claim 5, characterized in that: The determining, based on the first output value set and the second output value set, an effect parameter of delivering the target advertisement in the target delivery scenario includes: fusing the first output value set to obtain a first fused value; fusing the second output value set to obtain a second fused value; Determine the difference between the first fusion value and the second fusion value as an effect parameter of delivering the target advertisement in the target delivery scenario.
7. The advertising delivery analysis method according to claim 4, characterized in that: The first output value set includes at least one of the following: Estimated conversion rate, estimated click-through rate, estimated clicks, estimated playback time, and estimated revenue value.
8. An advertising delivery analysis device, characterized in that: include: A determination module, configured to determine a scenario weight corresponding to a target delivery scenario based on first feature information and second feature information; the first feature information is feature information of the target delivery scenario, and the second feature information at least includes feature information of a target advertisement and / or a target delivery user; A generating module, configured to generate a target feature vector of the target delivery scene based on the scene weight and the second feature information; The processing module is used to determine the effect parameters of delivering the target advertisement in the target delivery scenario by inputting the target feature vector and a predetermined variable into a multi-task network; the predetermined variable represents whether to deliver the target advertisement in the target delivery scenario or not.
9. An electronic device, characterized in that: The electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, the steps of the advertising delivery analysis method according to any one of claims 1 to 7 are implemented.
10. A storage medium, characterized in that: The storage medium stores a computer program, which, when executed by a processor, implements the steps of the advertising delivery analysis method according to any one of claims 1 to 7.
Citation Information
Cited By
Advertisement putting method and device, storage medium and electronic equipment
CN119991213A
