Advertisement effect attribution evaluation method and device, computer equipment and storage medium

Abstract

The invention relates to a multi-channel advertisement effect attribution evaluation method and apparatus, a computer device and a storage medium. The method comprises the steps of obtaining historical behavior data of each first user of a target advertisement in a plurality of advertisement channels and corresponding user device information; identifying a first user of each piece of historical behavior data based on each piece of user equipment information; obtaining a deep learning model based on the advertisement behavior complexity of the plurality of historical behavior data of the first user; obtaining an advertisement effect attribution value of each advertisement channel of the first user; training a deep learning model through the plurality of historical behavior data of the first user and each advertisement effect attribution value; and when a trained deep learning model is determined based on the advertisement behavior complexity of the second user for the multiple pieces of current behavior data of the target advertisement, inputting the multiple pieces of current behavior data into the trained deep learning model to obtain an advertisement effect attribution value of the second user in each advertisement channel. According to the method, the evaluation accuracy of the advertisement effect attribution is improved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a multi-channel advertising effect attribution evaluation method, device, computer equipment and storage medium. Background Art

[0002] With the rise of digital marketing and the popularity of Internet advertising, advertising has become an important means for companies to acquire customers and increase brand exposure. However, with the increase in advertising channels and the complexity of user behavior, how to accurately evaluate advertising effectiveness has become one of the challenges facing advertisers.

[0003] In the past, the evaluation of advertising effectiveness mainly relied on rules or intuition. For example, the advertising effectiveness of each advertising channel was evaluated by "last click" or "average distribution". However, these methods ignored the multiple interactions between users and advertisements and the long-term impact of advertisements on user behavior, resulting in inaccurate and incomplete evaluation of advertising effectiveness. Summary of the invention

[0004] Based on this, it is necessary to provide a multi-channel advertising effect attribution evaluation method, device, computer equipment and storage medium to address the above technical problems, which can comprehensively evaluate the attribution of advertising effects and improve the accuracy of the evaluation of advertising effect attribution.

[0005] A multi-channel advertising effect attribution evaluation method includes: obtaining historical behavior data generated by each first user of a target advertisement in multiple advertising channels and user device information corresponding to each historical behavior data; identifying the first user corresponding to each historical behavior data based on the device information of each user; obtaining multiple historical behavior data of each advertising channel of any first user and obtaining a deep learning model based on the advertising behavior complexity of the multiple historical behavior data; obtaining the advertising effect attribution value of the target advertisement of any first user in each advertising channel; training the deep learning model through the multiple historical behavior data of any first user and the advertising effect attribution value of the target advertisement in each advertising channel to obtain a trained deep learning model; when obtaining multiple current behavior data of any second user for the target advertisement and determining the trained deep learning model based on the advertising behavior complexity of the multiple current behavior data, the multiple current behavior data are input into the trained deep learning model to obtain the advertising effect attribution value of the target advertisement of any second user in each advertising channel, and any advertising effect attribution value is used to evaluate the advertising effect.

[0006] In one embodiment, the multiple advertising channels include any number of short video channels, application market channels, search engine channels, and mini program channels, and the multiple historical behavior data include any number of historical click behavior data, historical exposure behavior data, historical download behavior data, historical registration behavior data, and historical credit behavior data; the user device information corresponding to each historical behavior data is obtained by collecting device information in the user device when each historical behavior is identified.

[0007] In one of the embodiments, a multi-channel advertising effectiveness attribution evaluation method also includes: obtaining the data volume of multiple historical behavior data of any first user and the time interval between each historical behavior data; determining the advertising behavior complexity of the multiple historical behavior data of any first user based on the data volume and the time interval.

[0008] In one embodiment, a deep learning model is obtained based on the advertising behavior complexity of multiple historical behavior data, including: if the advertising behavior complexity of multiple historical behavior data of any first user is a first complexity, the acquired deep learning model is an RNN model; if the advertising behavior complexity of multiple historical behavior data of any first user is a second complexity, the acquired deep learning model is an LSTM model; if the advertising behavior complexity of multiple historical behavior data of any first user is a third complexity, the acquired deep learning model is a Transformer model; wherein the first complexity is less than the second complexity, and the second complexity is less than the third complexity.

[0009] In one of the embodiments, the advertising behavior complexity of multiple historical behavior data of a target user is determined based on data volume and time interval, including: obtaining a first data volume threshold and a first time interval threshold corresponding to a first complexity, a second data volume threshold and a second time interval threshold corresponding to a second complexity, and a third data volume threshold and a third time interval threshold corresponding to a third complexity, wherein the first data volume threshold is less than the second data volume threshold, the second data volume threshold is less than the third data volume threshold, the first time interval threshold is less than the second time interval threshold, and the second time interval threshold is less than the third time interval threshold; when the data volume of multiple historical behavior data of any first user is less than or equal to the first data volume threshold, and / or the time interval for generating each historical behavior data is less than or equal to the first time interval threshold, determining The advertising behavior complexity of multiple historical behavior data of any first user is determined to be a first complexity; when the data volume of multiple historical behavior data of any first user is greater than a first data volume threshold and less than or equal to a second data volume threshold, and / or the time interval for generating each historical behavior data is greater than the first time interval threshold and less than or equal to the second time interval threshold, the advertising behavior complexity of multiple historical behavior data of any first user is determined to be a second complexity; when the data volume of multiple historical behavior data of any first user is greater than a second data volume threshold and less than or equal to a third data volume threshold, and / or the time interval for generating each historical behavior data is greater than the second time interval threshold and less than or equal to the third time interval threshold, the advertising behavior complexity of multiple historical behavior data of the target user is determined to be a third complexity.

[0010] In one embodiment, after the step of obtaining the trained deep learning model, the method also includes: obtaining the target behavior data generated by the target advertisement in each advertising channel and the target user device information corresponding to each target behavior data after a preset time period; identifying the third user corresponding to each target behavior data based on the device information of each target user; obtaining multiple target behavior data of each third user in each advertising channel; screening out multiple target third users based on the advertising behavior complexity of the multiple target behavior data of each third user; inputting the multiple target behavior data of each target third user into the trained deep learning model to obtain the advertising effect attribution value of the target advertisement of each target third user in each advertising channel; obtaining the advertising delivery strategy of the target advertisement and obtaining the delivery cost of each advertising channel from the advertising delivery strategy; calculating the customer acquisition unit price of each advertising channel according to the delivery cost of each advertising channel and the advertising effect attribution value corresponding to each target third user; adjusting the model parameters of the trained deep learning model based on the customer acquisition unit price of each advertising channel, and the deep learning model after adjusting the model parameters is used to evaluate the advertising effect attribution value of the user's target advertisement in each advertising channel.

[0011] In one of the embodiments, after the step of obtaining the advertising effectiveness attribution value of the target advertisement of each target third user in each advertising channel, it also includes: adjusting the advertising delivery strategy of the target advertisement based on the advertising effectiveness attribution value of the target advertisement of each target third user in each advertising channel.

[0012] A multi-channel advertising effect attribution evaluation device comprises: a first acquisition module, used to acquire historical behavior data generated by each first user of a target advertisement in multiple advertising channels and user device information corresponding to each historical behavior data; an identification module, used to identify the first user corresponding to each historical behavior data based on the device information of each user; a second acquisition module, used to acquire multiple historical behavior data of each advertising channel of any first user and acquire a deep learning model based on the advertising behavior complexity of the multiple historical behavior data; a third acquisition module, used to acquire the advertising effect attribution value of the target advertisement of any first user in each advertising channel; a model training module, used to train the deep learning model through the multiple historical behavior data of any first user and the advertising effect attribution value of the target advertisement in each advertising channel to obtain a trained deep learning model; an evaluation module, used to input multiple current behavior data into the trained deep learning model when multiple current behavior data of any second user on the target advertisement is acquired and the trained deep learning model is determined based on the advertising behavior complexity of the multiple current behavior data, to obtain the advertising effect attribution value of the target advertisement of any second user in each advertising channel, and any advertising effect attribution value is used to evaluate the advertising effect.

[0013] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of any of the above-mentioned embodiments are implemented.

[0014] A computer-readable storage medium stores a computer program, which implements the steps of any of the above-mentioned method embodiments when the computer program is executed by a processor.

[0015] The above-mentioned multi-channel advertising effect attribution evaluation method, device, computer equipment and storage medium select the corresponding deep learning model based on the advertising behavior complexity of the user's historical behavior data in each advertising channel, and train the selected deep learning model. The trained deep learning model can learn the current behavior data of any user in each advertising channel in subsequent applications and determine the advertising effect attribution value of each advertising channel. Therefore, it can more comprehensively analyze the interaction between users and advertisements, not only considering the advertising channel of the last click, but also comprehensively considering the interaction between users and advertisements through multiple advertising channels. The trained deep learning model can learn the complex relationship between different advertising channels from massive data, so as to more accurately evaluate the contribution of each advertising channel to advertising conversion. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A diagram of an application environment of a multi-channel advertising effect attribution evaluation method in an embodiment;

[0017] Figure 2 A schematic diagram of a flow chart of a multi-channel advertising effect attribution evaluation method in one embodiment;

[0018] Figure 3 It is a flowchart of a multi-channel advertising effect attribution evaluation method in an embodiment in a specific application scenario;

[0019] Figure 4 for Figure 3 Schematic diagram of the processing flow of deep learning models in application scenarios;

[0020] Figure 5 It is a structural block diagram of a multi-channel advertising effect attribution evaluation device in one embodiment;

[0021] Figure 6 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0022] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0023] This application provides a multi-channel advertising effect attribution evaluation method, which is applied to Figure 1 In the application environment shown. Figure 1As shown, the server 100 is used to implement a multi-channel advertising effect attribution evaluation method of the present application. Specifically, there can be multiple terminals 200. The server 100 obtains the historical behavior data generated by each first user in multiple advertising channels for the target advertisement and the user device information corresponding to each historical behavior data from each terminal 200. Based on the device information of each user, the first user corresponding to each historical behavior data is identified, multiple historical behavior data of each advertising channel of any first user is obtained, and a deep learning model is obtained based on the advertising behavior complexity of the multiple historical behavior data, and the advertising effect attribution value of the target advertisement of any first user in each advertising channel is obtained. The deep learning model is trained by the multiple historical behavior data of any first user and the advertising effect attribution value of the target advertisement in each advertising channel to obtain a trained deep learning model. When multiple current behavior data of any second user for the target advertisement are obtained and the trained deep learning model is determined based on the advertising behavior complexity of the multiple current behavior data, the multiple current behavior data are input into the trained deep learning model to obtain the advertising effect attribution value of the target advertisement of any second user in each advertising channel, and any advertising effect attribution value is used to evaluate the advertising effect.

[0024] In one embodiment, Figure 2 As shown in the figure, a multi-channel advertising effect attribution evaluation method is provided. Figure 1 The server 100 in FIG. 1 is taken as an example to illustrate the method, which includes the following steps:

[0025] S202, obtaining historical behavior data generated by each first user for the target advertisement in multiple advertising channels and user device information corresponding to each historical behavior data.

[0026] In one embodiment, the multiple advertising channels include any number of short video channels, application market channels, search engine channels, and mini program channels, and the multiple historical behavior data include any number of historical click behavior data, historical exposure behavior data, historical download behavior data, historical registration behavior data, and historical credit behavior data; the user device information corresponding to each historical behavior data is obtained by collecting device information in the user device when each historical behavior is identified.

[0027] In this embodiment, historical behavior data generated by each first user's operation on the target advertisement and user device information corresponding to each historical behavior data are collected in any of the short video channel, the application market channel, the search engine channel, and the mini program channel according to the advertisement delivery strategy of the target advertisement. Each historical behavior data includes data generated by the user's behavior of clicking on the target advertisement, the behavior of being exposed to the target advertisement, the behavior of downloading the application corresponding to the target advertisement, the behavior of completing the corresponding registration operation through the target advertisement, and the behavior of completing the login operation.

[0028] Specifically, the contact data of the user's click behavior and exposure behavior on the Douyin channel, the application market channel, the Baidu search channel and the WeChat mini program. The contact data of the download behavior in the mobile application market. After all the contact data are collected, they are serialized according to time and advertising channels. When the user's behavior towards the target advertisement is identified, the device information of the user's device is collected. When storing data and information, the user's device information, advertising channel information, advertising delivery information and its delivery time information are retained. For example, the user registered for the first time on the WeChat mini program, and the user's device information can be obtained at this time. The advertising channel information, registration and first login behavior data, and user device information of the WeChat mini program are collected and stored.

[0029] S204: Identify the first user corresponding to each historical behavior data based on the device information of each user.

[0030] In this embodiment, in step S202, the historical behavior data and user device information generated by the first user in each advertising channel are collected. The historical behavior data of different advertising channels may correspond to the same user. Here, the user device information is used as a user identifier to identify each first user.

[0031] S206, obtaining multiple historical behavior data of each advertising channel of any first user and obtaining a deep learning model based on the advertising behavior complexity of the multiple historical behavior data.

[0032] In this embodiment, there are multiple first users, and any first user is selected as the object of data processing. After determining any first user, one or more historical behavior data of any first user in each advertising channel can be screened out through the user device information of any first user. Any first user may have historical behaviors in all advertising channels, or may have historical behaviors in some advertising channels. Then, determine the complexity of the advertising behavior of any first user. The complexity of advertising behavior is determined based on the complexity of its multiple historical behavior data. Then select a deep learning model based on the complexity of its advertising behavior. For example, according to the user device information, search for the behavior data of the target advertisement, obtain the advertising behavior of the user in each advertising channel in the Douyin channel, application market, Baidu search and WeChat applets, obtain the cleaned historical behavior data, and select a suitable AI model based on the complexity of its advertising behavior.

[0033] In one embodiment, before step S206, a step of calculating the advertising complexity of any first user is also included, that is, the following steps are also included: obtaining the data volume of multiple historical behavior data of any first user and the time interval for each historical behavior data; determining the advertising behavior complexity of multiple historical behavior data of any first user based on the data volume and the time interval.

[0034] In one example, a deep learning model is obtained based on the advertising behavior complexity of multiple historical behavior data, including: if the advertising behavior complexity of multiple historical behavior data of any first user is a first complexity, the acquired deep learning model is an RNN model; if the advertising behavior complexity of multiple historical behavior data of any first user is a second complexity, the acquired deep learning model is an LSTM model; if the advertising behavior complexity of multiple historical behavior data of any first user is a third complexity, the acquired deep learning model is a Transformer model; wherein the first complexity is less than the second complexity, and the second complexity is less than the third complexity.

[0035] Wherein, determining the advertising behavior complexity of multiple historical behavior data of the target user according to the data volume and time interval includes: obtaining a first data volume threshold and a first time interval threshold corresponding to a first complexity, a second data volume threshold and a second time interval threshold corresponding to a second complexity, and a third data volume threshold and a third time interval threshold corresponding to a third complexity, wherein the first data volume threshold is less than the second data volume threshold, the second data volume threshold is less than the third data volume threshold, the first time interval threshold is less than the second time interval threshold, and the second time interval threshold is less than the third time interval threshold; when the data volume of multiple historical behavior data of any first user is less than or equal to the first data volume threshold, and / or the time interval for generating each historical behavior data is less than or equal to the first time interval threshold, determining any The advertising behavior complexity of multiple historical behavior data of the first user is a first complexity; when the data volume of multiple historical behavior data of any first user is greater than the first data volume threshold and less than or equal to the second data volume threshold, and / or the time interval for generating each historical behavior data is greater than the first time interval threshold and less than or equal to the second time interval threshold, the advertising behavior complexity of multiple historical behavior data of any first user is determined to be a second complexity; when the data volume of multiple historical behavior data of any first user is greater than the second data volume threshold and less than or equal to the third data volume threshold, and / or the time interval for generating each historical behavior data is greater than the second time interval threshold and less than or equal to the third time interval threshold, the advertising behavior complexity of multiple historical behavior data of the target user is determined to be a third complexity.

[0036] In this embodiment, the complexity of advertising behavior is determined based on the amount of user's behavior data and the interval time of each behavior data. The larger the data amount, the greater the corresponding advertising behavior complexity. The longer the interval time, the greater the corresponding advertising behavior complexity. The first data amount, the second data amount threshold and the third data amount threshold are set for the amount of user's behavior data. The first time interval threshold, the second time interval threshold and the third time interval threshold are set for the time interval of the behavior data. The first data amount threshold is less than the second data amount threshold, the second data amount threshold is less than the third data amount threshold, the first time interval threshold is less than the second time interval threshold, and the second time interval threshold is less than the third time interval threshold. Further, based on this, the advertising behavior complexity of the user's behavior data is divided into the first complexity, the second complexity and the third complexity. The deep learning model corresponding to the first complexity is the RNN model, the deep learning model corresponding to the second complexity is the LSTM model, and the deep learning model corresponding to the third complexity is the Transformer model. Therefore, it is possible to configure a deep learning model adapted to the advertising behavior complexity of the user's behavior data based on the user's behavior data, so as to learn and understand the user's more complex user behavior pattern for the target advertisement through the adapted deep learning model, thereby obtaining a more accurate attribution result.

[0037] S208: Obtain an advertising effect attribution value of a target advertisement of any first user in each advertising channel.

[0038] In this embodiment, any first user performs advertising behavior on the target advertisement in each advertising channel, obtains multiple historical behavior data, and finally converts them into advertising effects on the target advertisement. Therefore, the advertising effect attribution value of the target advertisement of any first user in each advertising channel can be calculated.

[0039] S210, training a deep learning model using multiple historical behavior data of any first user and advertising effect attribution values ​​of target advertisements in various advertising channels to obtain a trained deep learning model.

[0040] In this embodiment, the deep learning model is trained by using multiple historical behavior data of any first user and the advertising effect attribution value of the target advertisement in each advertising channel. For example, assuming that the attribution result of a first user is as follows:

[0041] Short video channel: 20% of conversions come from this channel.

[0042] App Marketplace Channel: 35% of conversions come from this channel.

[0043] Search engine channel: 20% of conversions come from this channel.

[0044] Mini Program channel: 25% of conversions come from this channel.

[0045] Collect the historical behavior data of the first user in various advertising channels such as short video channels, application market channels, search engine channels and mini program channels, and input the historical behavior data of each advertising channel and the corresponding advertising attribution value into the deep learning model to perform model training on it.

[0046] In one embodiment, after the above-mentioned step of obtaining the trained deep learning model, it also includes: obtaining the target behavior data generated by the target advertisement in each advertising channel and the target user device information corresponding to each target behavior data after a preset period of time; identifying the third user corresponding to each target behavior data based on the device information of each target user; obtaining multiple target behavior data of each third user in each advertising channel; screening out multiple target third users based on the advertising behavior complexity of the multiple target behavior data of each third user; inputting the multiple target behavior data of each target third user into the trained deep learning model to obtain the advertising effect attribution value of the target advertisement of each target third user in each advertising channel; obtaining the advertising delivery strategy of the target advertisement and obtaining the delivery cost of each advertising channel from the advertising delivery strategy; calculating the customer acquisition unit price of each advertising channel according to the delivery cost of each advertising channel and the advertising effect attribution value corresponding to each target third user; adjusting the model parameters of the trained deep learning model based on the customer acquisition unit price of each advertising channel, and the deep learning model after adjusting the model parameters is used to evaluate the advertising effect attribution value of the user's target advertisement in each advertising channel.

[0047] In this embodiment, after the deep learning model is trained based on the multiple historical behavior data of the first user and the advertising effect attribution values ​​of the target advertisements in each advertising channel, in the subsequent model application stage, the model parameters need to be adjusted based on the advertising delivery strategy.

[0048] Specifically, the target behavior data generated in each advertising channel of the target advertisement and the target user device information corresponding to each target behavior data are collected after a preset time period, and the third user corresponding to each target behavior data is identified based on the device information of each target user. The third user here is different from the first user and the second user mentioned above. The third user is a user who operates the target advertisement to generate advertising behavior within the preset time period. Then, multiple target third users whose advertising behavior complexity is the same as the advertising behavior complexity of any second user mentioned above are selected from multiple third users. Multiple target behavior data of each target third user are input into the trained deep learning model to obtain the advertising effect attribution value of the target advertisement of each target third user in each advertising channel. Then, the delivery cost of each advertising channel is obtained from the advertising delivery strategy, and the customer acquisition unit price of each advertising channel is calculated based on the delivery cost of each advertising channel and the advertising effect attribution value of the target advertisement in each advertising channel, and the model parameters of the deep learning model are adjusted according to the customer acquisition unit price of each advertising channel.

[0049] For example, the delivery cost of the short video channel in the advertising delivery strategy of the target advertisement is 1,000 yuan per hour, and the advertising effect attribution value of the above short video channel is 20%. Assuming that 10 people are calculated from the short video channel within 1 hour according to the 20% advertising effect attribution value, the customer acquisition unit price of the short video channel is 100 yuan / user. Get the customer acquisition unit price threshold of the pre-set short video channel, for example, 80. If the calculated customer acquisition unit price is less than the customer acquisition unit price threshold, adjust the model parameters of the deep learning model and iterate the model again. Alternatively, the customer acquisition unit price threshold is 101. If the calculated customer acquisition unit price is less than 101, adjust the model parameters of the deep learning model and iterate the model again.

[0050] In one embodiment, after the above step of obtaining the advertising effectiveness attribution value of the target advertisement of each target third user in each advertising channel, it also includes: adjusting the advertising delivery strategy of the target advertisement based on the advertising effectiveness attribution value of the target advertisement of each target third user in each advertising channel.

[0051] Specifically, the advertising delivery strategy is optimized based on the advertising effect attribution value generated by the degree learning model. For example, in the above example, the advertising effect attribution value of the application market channel is the highest, indicating that it has the highest market share and the best delivery effect. Therefore, the delivery cost of the application market channel in the advertising delivery strategy can be increased.

[0052] S212, when multiple current behavior data of any second user for the target advertisement are obtained and a trained deep learning model is determined based on the advertising behavior complexity of the multiple current behavior data, the multiple current behavior data are input into the trained deep learning model to obtain the advertising effect attribution value of the target advertisement of any second user in each advertising channel, and any advertising effect attribution value is used to evaluate the advertising effect.

[0053] In this embodiment, the second user and the first user may be different users. The first user refers to a user who has generated historical behavior data for the target advertisement, that is, has generated advertising behavior for the target advertisement in various advertising channels and has been statistically evaluated for advertising effect attribution. The second user refers to the current user whose advertising effect attribution has not been statistically evaluated.

[0054] When any second user generates advertising behavior for the target advertisement in each advertising channel to obtain multiple current behavior data, if the complexity of its advertising behavior is analyzed to determine the trained deep learning model, the multiple current behavior data of the second user are input into the trained deep learning model to obtain the advertising effect attribution value of the second user in each advertising channel. Among them, the advertising effect attribution values ​​mentioned in this application are all used to evaluate the corresponding advertising effect.

[0055] The above-mentioned multi-channel advertising effect attribution evaluation method selects the corresponding deep learning model based on the advertising behavior complexity of the user's historical behavior data in each advertising channel, and trains the selected deep learning model. The trained deep learning model can learn the current behavior data of any user in each advertising channel in subsequent applications and determine the advertising effect attribution value of each advertising channel. Therefore, it can more comprehensively analyze the interaction between users and advertisements, not only considering the advertising channel of the last click, but also comprehensively considering the interaction between users and advertisements through multiple advertising channels. The trained deep learning model can learn the complex relationship between different advertising channels from massive data, so as to more accurately evaluate the contribution of each advertising channel to advertising conversion.

[0056] The following is a specific example of the above multi-channel advertising effect attribution evaluation method:

[0057] refer to Figure 3As shown: After the advertisements are placed on various advertising channels based on the advertisement delivery strategy, the advertisement data is collected. The advertisement data includes contact data such as clicks and exposures. At the same time, the user behavior data is collected. The user behavior data includes registration and first-time registration completion behavior data. In addition, the user device information on the user who has taken action on the advertisement is also collected to find all the corresponding advertisement data through the user device on which the action has taken place. Then, the corresponding deep learning model is selected based on the complexity of all the behavior data in the advertisement data of the user device. The attribution results of the advertisements in various channels are output through the model. Finally, the model is experimentally verified. If the experimental settings are met, the trained deep learning model is directly released. Otherwise, the model parameters of the deep learning model are adjusted. For the selection of deep learning models and experimental verification, adjustment of model parameters, model optimization and other processes, see Figure 4 shown.

[0058] Compared with the traditional advertising effect attribution algorithm, the multi-channel advertising effect attribution evaluation method of the present application has the following obvious advantages:

[0059] 1. Comprehensive consideration of multiple advertising channels: AI models, namely deep learning models, can more comprehensively analyze the interaction between users and ads, not only considering the advertising channel of the last click, but also comprehensively considering the interaction between users and ads through multiple advertising channels. Through deep learning and other technologies, AI models can learn the complex relationship between different advertising channels from massive data, so as to more accurately evaluate the contribution of each advertising channel to conversion.

[0060] 2. Consider the time effect: AI models can fully consider the impact of the time interval between user and ad exposure on conversion, and use methods such as time decay to better reflect the changes in advertising effects over time. By modeling time series data with a deep learning model, AI models can capture the dynamic changes in advertising effects over time, improving the accuracy and practicality of attribution results.

[0061] 3. Ability to process big data: AI models are highly efficient and scalable when processing large amounts of advertising data. Using technologies such as deep learning, AI models can automatically extract effective features from advertising data and perform efficient model training and reasoning, greatly improving the efficiency and accuracy of attribution analysis.

[0062] 4. Flexibility and adaptability: AI models have certain flexibility and adaptability, and can automatically adjust and optimize according to different advertising data and business needs. Through continuous learning and iteration, AI models can adapt to different advertising scenarios and user behavior patterns, providing advertisers with more personalized and accurate advertising delivery strategies and resource allocation suggestions.

[0063] 5. Comprehensive consideration of multiple factors: AI models can comprehensively consider the impact of multiple factors on advertising effectiveness, such as advertising content, user attributes, time factors, etc. Through technologies such as deep learning, AI models can learn the complex relationships between different factors from massive data, thereby more comprehensively evaluating advertising effectiveness and providing advertisers with more effective decision support.

[0064] It should be understood that, although the steps in the flowchart are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.

[0065] The present application also provides a multi-channel advertising effect attribution evaluation device. Figure 5 As shown, a multi-channel advertising effect attribution evaluation device includes a first acquisition module 502 , an identification module 504 , a second acquisition module 506 , a third acquisition module 508 , a model training module 510 and an evaluation module 512 . The first acquisition module 502 is used to obtain the historical behavior data generated by each first user of the target advertisement in multiple advertising channels and the user device information corresponding to each historical behavior data; the identification module 504 is used to identify the first user corresponding to each historical behavior data based on the device information of each user; the second acquisition module 506 is used to obtain multiple historical behavior data of each advertising channel of any first user and obtain a deep learning model based on the advertising behavior complexity of the multiple historical behavior data; the third acquisition module 508 is used to obtain the advertising effect attribution value of the target advertisement of any first user in each advertising channel; the model training module 510 is used to train the deep learning model through the multiple historical behavior data of any first user and the advertising effect attribution value of the target advertisement in each advertising channel to obtain a trained deep learning model; the evaluation module 512 is used to input the multiple current behavior data into the trained deep learning model when obtaining multiple current behavior data of any second user for the target advertisement and determining the trained deep learning model based on the advertising behavior complexity of the multiple current behavior data, and obtain the advertising effect attribution value of the target advertisement of any second user in each advertising channel, and any advertising effect attribution value is used to evaluate the advertising effect.

[0066] In one embodiment, the multiple advertising channels include any number of short video channels, application market channels, search engine channels, and mini program channels, and the multiple historical behavior data include any number of historical click behavior data, historical exposure behavior data, historical download behavior data, historical registration behavior data, and historical credit behavior data; the user device information corresponding to each historical behavior data is obtained by collecting device information in the user device when each historical behavior is identified.

[0067] In one of the embodiments, a multi-channel advertising effectiveness attribution evaluation device also includes a determination module for obtaining the data volume of multiple historical behavior data of any first user and the time interval between each historical behavior data; and determining the advertising behavior complexity of multiple historical behavior data of any first user based on the data volume and the time interval.

[0068] In one embodiment, a deep learning model is obtained based on the advertising behavior complexity of multiple historical behavior data, including: if the advertising behavior complexity of multiple historical behavior data of any first user is a first complexity, the acquired deep learning model is an RNN model; if the advertising behavior complexity of multiple historical behavior data of any first user is a second complexity, the acquired deep learning model is an LSTM model; if the advertising behavior complexity of multiple historical behavior data of any first user is a third complexity, the acquired deep learning model is a Transformer model; wherein the first complexity is less than the second complexity, and the second complexity is less than the third complexity.

[0069] In one of the embodiments, the advertising behavior complexity of multiple historical behavior data of a target user is determined based on data volume and time interval, including: obtaining a first data volume threshold and a first time interval threshold corresponding to a first complexity, a second data volume threshold and a second time interval threshold corresponding to a second complexity, and a third data volume threshold and a third time interval threshold corresponding to a third complexity, wherein the first data volume threshold is less than the second data volume threshold, the second data volume threshold is less than the third data volume threshold, the first time interval threshold is less than the second time interval threshold, and the second time interval threshold is less than the third time interval threshold; when the data volume of multiple historical behavior data of any first user is less than or equal to the first data volume threshold, and / or the time interval for generating each historical behavior data is less than or equal to the first time interval threshold, determining The advertising behavior complexity of multiple historical behavior data of any first user is determined to be a first complexity; when the data volume of multiple historical behavior data of any first user is greater than a first data volume threshold and less than or equal to a second data volume threshold, and / or the time interval for generating each historical behavior data is greater than the first time interval threshold and less than or equal to the second time interval threshold, the advertising behavior complexity of multiple historical behavior data of any first user is determined to be a second complexity; when the data volume of multiple historical behavior data of any first user is greater than a second data volume threshold and less than or equal to a third data volume threshold, and / or the time interval for generating each historical behavior data is greater than the second time interval threshold and less than or equal to the third time interval threshold, the advertising behavior complexity of multiple historical behavior data of the target user is determined to be a third complexity.

[0070] In one of the embodiments, after obtaining the trained deep learning model, it also includes: obtaining the target behavior data generated by the target advertisement in each advertising channel and the target user device information corresponding to each target behavior data after a preset period of time; identifying the third user corresponding to each target behavior data based on the device information of each target user; obtaining multiple target behavior data of each third user in each advertising channel; screening out multiple target third users based on the advertising behavior complexity of the multiple target behavior data of each third user; inputting the multiple target behavior data of each target third user into the trained deep learning model to obtain the advertising effect attribution value of the target advertisement of each target third user in each advertising channel; obtaining the advertising delivery strategy of the target advertisement and obtaining the delivery cost of each advertising channel from the advertising delivery strategy; calculating the customer acquisition unit price of each advertising channel according to the delivery cost of each advertising channel and the advertising effect attribution value corresponding to each target third user; adjusting the model parameters of the trained deep learning model based on the customer acquisition unit price of each advertising channel, and the deep learning model after adjusting the model parameters is used to evaluate the advertising effect attribution value of the user's target advertisement in each advertising channel.

[0071] In one of the embodiments, after obtaining the advertising effect attribution value of the target advertisement of each target third user in each advertising channel, it also includes: adjusting the advertising delivery strategy of the target advertisement based on the advertising effect attribution value of the target advertisement of each target third user in each advertising channel.

[0072] For the specific definition of a multi-channel advertising effect attribution evaluation device, please refer to the definition of a multi-channel advertising effect attribution evaluation method above, which will not be repeated here. Each module in the above-mentioned multi-channel advertising effect attribution evaluation device can be implemented in whole or in part through software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0073] In one embodiment, a computer device is provided. The computer device may be a server supporting the operation of the server 100. The internal structure diagram thereof may be as shown in FIG. Figure 6 As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to perform data interaction with a client, etc. When the computer program is executed by the processor, the above-mentioned multi-channel advertising effect attribution evaluation method is implemented.

[0074] Those skilled in the art will understand that Figure 6 The structure shown in the figure is only a block diagram of a part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0075] In one embodiment, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program: obtaining historical behavior data generated by each first user of a target advertisement in multiple advertising channels and user device information corresponding to each historical behavior data; identifying the first user corresponding to each historical behavior data based on the device information of each user; obtaining multiple historical behavior data of each advertising channel of any first user and obtaining a deep learning model based on the advertising behavior complexity of the multiple historical behavior data; obtaining an advertising effect attribution value of a target advertisement of any first user in each advertising channel; training the deep learning model through the multiple historical behavior data of any first user and the advertising effect attribution value of the target advertisement in each advertising channel to obtain a trained deep learning model; when obtaining multiple current behavior data of any second user for the target advertisement and determining a trained deep learning model based on the advertising behavior complexity of the multiple current behavior data, inputting the multiple current behavior data into the trained deep learning model to obtain an advertising effect attribution value of the target advertisement of any second user in each advertising channel, and any advertising effect attribution value is used to evaluate the advertising effect.

[0076] In one embodiment, the multiple advertising channels include any number of short video channels, application market channels, search engine channels, and mini program channels, and the multiple historical behavior data include any number of historical click behavior data, historical exposure behavior data, historical download behavior data, historical registration behavior data, and historical credit behavior data; the user device information corresponding to each historical behavior data is obtained by collecting device information in the user device when each historical behavior is identified.

[0077] In one embodiment, when the processor executes the computer program, the following steps are implemented: obtaining the data volume of multiple historical behavior data of any first user and the time interval between each historical behavior data; determining the advertising behavior complexity of multiple historical behavior data of any first user based on the data volume and the time interval.

[0078] In one of the embodiments, when the processor executes a computer program to implement the above-mentioned step of obtaining a deep learning model based on the advertising behavior complexity of multiple historical behavior data, the following steps are specifically implemented: if the advertising behavior complexity of multiple historical behavior data of any first user is a first complexity, the acquired deep learning model is an RNN model; if the advertising behavior complexity of multiple historical behavior data of any first user is a second complexity, the acquired deep learning model is an LSTM model; if the advertising behavior complexity of multiple historical behavior data of any first user is a third complexity, the acquired deep learning model is a Transformer model; wherein the first complexity is less than the second complexity, and the second complexity is less than the third complexity.

[0079] In one of the embodiments, when the processor executes a computer program to implement the above-mentioned step of determining the advertising behavior complexity of multiple historical behavior data of the target user based on the data volume and time interval, the following steps are specifically implemented: obtaining a first data volume threshold and a first time interval threshold corresponding to a first complexity, a second data volume threshold and a second time interval threshold corresponding to a second complexity, and a third data volume threshold and a third time interval threshold corresponding to a third complexity, wherein the first data volume threshold is less than the second data volume threshold, the second data volume threshold is less than the third data volume threshold, the first time interval threshold is less than the second time interval threshold, and the second time interval threshold is less than the third time interval threshold; when the data volume of multiple historical behavior data of any first user is less than or equal to the first data volume threshold, and / or the time interval for generating each historical behavior data is less than or equal to the first data volume threshold, When the data volume of multiple historical behavior data of any first user is greater than the first data volume threshold and less than or equal to the second data volume threshold, and / or the time interval for generating each historical behavior data is greater than the first time interval threshold and less than or equal to the second time interval threshold, the advertising behavior complexity of multiple historical behavior data of any first user is determined to be the second complexity; when the data volume of multiple historical behavior data of any first user is greater than the second data volume threshold and less than or equal to the third data volume threshold, and / or the time interval for generating each historical behavior data is greater than the second time interval threshold and less than or equal to the third time interval threshold, the advertising behavior complexity of multiple historical behavior data of the target user is determined to be the third complexity.

[0080] In one of the embodiments, the processor implements the following steps when executing the computer program: obtaining the target behavior data generated by the target advertisement in each advertising channel and the target user device information corresponding to each target behavior data after a preset time period; identifying the third user corresponding to each target behavior data based on the device information of each target user; obtaining multiple target behavior data of each third user in each advertising channel; screening out multiple target third users based on the advertising behavior complexity of the multiple target behavior data of each third user; inputting the multiple target behavior data of each target third user into the trained deep learning model to obtain the advertising effect attribution value of the target advertisement of each target third user in each advertising channel; obtaining the advertising delivery strategy of the target advertisement and obtaining the delivery cost of each advertising channel from the advertising delivery strategy; calculating the customer acquisition unit price of each advertising channel based on the delivery cost of each advertising channel and the advertising effect attribution value corresponding to each target third user; adjusting the model parameters of the trained deep learning model based on the customer acquisition unit price of each advertising channel, and the deep learning model after adjusting the model parameters is used to evaluate the advertising effect attribution value of the user's target advertisement in each advertising channel.

[0081] In one of the embodiments, when the processor executes the computer program, the following steps are implemented: adjusting the advertising delivery strategy of the target advertisement based on the advertising effect attribution value of the target advertisement of each target third user in each advertising channel.

[0082] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented: obtaining historical behavior data generated by each first user of a target advertisement in multiple advertising channels and user device information corresponding to each historical behavior data; identifying the first user corresponding to each historical behavior data based on the device information of each user; obtaining multiple historical behavior data of each advertising channel of any first user and obtaining a deep learning model based on the advertising behavior complexity of the multiple historical behavior data; obtaining an advertising effect attribution value of the target advertisement of any first user in each advertising channel; training the deep learning model through the multiple historical behavior data of any first user and the advertising effect attribution value of the target advertisement in each advertising channel to obtain a trained deep learning model; when obtaining multiple current behavior data of any second user for the target advertisement and determining a trained deep learning model based on the advertising behavior complexity of the multiple current behavior data, the multiple current behavior data are input into the trained deep learning model to obtain the advertising effect attribution value of the target advertisement of any second user in each advertising channel, and any advertising effect attribution value is used to evaluate the advertising effect.

[0083] In one embodiment, the multiple advertising channels include any number of short video channels, application market channels, search engine channels, and mini program channels, and the multiple historical behavior data include any number of historical click behavior data, historical exposure behavior data, historical download behavior data, historical registration behavior data, and historical credit behavior data; the user device information corresponding to each historical behavior data is obtained by collecting device information in the user device when each historical behavior is identified.

[0084] In one embodiment, when the computer program is executed by a processor, the following steps are implemented: obtaining the data volume of multiple historical behavior data of any first user and the time interval between each historical behavior data; determining the advertising behavior complexity of multiple historical behavior data of any first user based on the data volume and the time interval.

[0085] In one embodiment, when the computer program is executed by the processor to implement the above-mentioned steps of obtaining a deep learning model based on the advertising behavior complexity of multiple historical behavior data, the following steps are specifically implemented: if the advertising behavior complexity of multiple historical behavior data of any first user is a first complexity, the acquired deep learning model is an RNN model; if the advertising behavior complexity of multiple historical behavior data of any first user is a second complexity, the acquired deep learning model is an LSTM model; if the advertising behavior complexity of multiple historical behavior data of any first user is a third complexity, the acquired deep learning model is a Transformer model; wherein the first complexity is less than the second complexity, and the second complexity is less than the third complexity.

[0086] In one embodiment, when a computer program is executed by a processor to implement the above-mentioned step of determining the advertising behavior complexity of multiple historical behavior data of a target user based on the data volume and time interval, the following steps are specifically implemented: obtaining a first data volume threshold and a first time interval threshold corresponding to a first complexity, a second data volume threshold and a second time interval threshold corresponding to a second complexity, and a third data volume threshold and a third time interval threshold corresponding to a third complexity, wherein the first data volume threshold is less than the second data volume threshold, the second data volume threshold is less than the third data volume threshold, the first time interval threshold is less than the second time interval threshold, and the second time interval threshold is less than the third time interval threshold; when the data volume of multiple historical behavior data of any first user is less than or equal to the first data volume threshold, and / or the time interval for generating each historical behavior data is less than or equal to the first data volume threshold, When the data volume of multiple historical behavior data of any first user is greater than the first data volume threshold and less than or equal to the second data volume threshold, and / or the time interval for generating each historical behavior data is greater than the first time interval threshold and less than or equal to the second time interval threshold, the advertising behavior complexity of multiple historical behavior data of any first user is determined to be the second complexity; when the data volume of multiple historical behavior data of any first user is greater than the second data volume threshold and less than or equal to the third data volume threshold, and / or the time interval for generating each historical behavior data is greater than the second time interval threshold and less than or equal to the third time interval threshold, the advertising behavior complexity of multiple historical behavior data of the target user is determined to be the third complexity.

[0087] In one embodiment, the computer program implements the following steps when executed by the processor: obtaining the target behavior data generated by the target advertisement in each advertising channel and the target user device information corresponding to each target behavior data after a preset time period; identifying the third user corresponding to each target behavior data based on the device information of each target user; obtaining multiple target behavior data of each third user in each advertising channel; screening out multiple target third users based on the advertising behavior complexity of the multiple target behavior data of each third user; inputting the multiple target behavior data of each target third user into the trained deep learning model to obtain the advertising effect attribution value of the target advertisement of each target third user in each advertising channel; obtaining the advertising delivery strategy of the target advertisement and obtaining the delivery cost of each advertising channel from the advertising delivery strategy; calculating the customer acquisition unit price of each advertising channel based on the delivery cost of each advertising channel and the advertising effect attribution value corresponding to each target third user; adjusting the model parameters of the trained deep learning model based on the customer acquisition unit price of each advertising channel, and the deep learning model after adjusting the model parameters is used to evaluate the advertising effect attribution value of the user's target advertisement in each advertising channel.

[0088] In one of the embodiments, when the computer program is executed by a processor, the following steps are implemented: adjusting the advertising delivery strategy of the target advertisement based on the advertising effect attribution value of the target advertisement of each target third user in each advertising channel.

[0089] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0090] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0091] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A multi-channel advertising effectiveness attribution evaluation method. It is characterized in that The method comprises: Acquire historical behavior data generated by each first user in multiple advertising channels for the target advertisement and user device information corresponding to each historical behavior data; Identify the first user corresponding to each historical behavior data based on the device information of each user; Acquire multiple historical behavior data of each advertising channel of any first user and acquire a deep learning model based on the advertising behavior complexity of the multiple historical behavior data; Obtaining an advertising effect attribution value of a target advertisement of the arbitrary first user in each advertising channel; Performing model training on the deep learning model through the multiple historical behavior data of the arbitrary first user and the advertising effect attribution value of the target advertisement in each advertising channel to obtain a trained deep learning model; When multiple current behavior data of any second user for the target advertisement are obtained and the trained deep learning model is determined based on the advertising behavior complexity of the multiple current behavior data, the multiple current behavior data are input into the trained deep learning model to obtain the advertising effect attribution value of the target advertisement of any second user in each advertising channel, and any advertising effect attribution value is used to evaluate the advertising effect.

2. The method according to claim 1, It is characterized in that The multiple advertising channels include any multiple of short video channels, application market channels, search engine channels, and mini program channels, and the multiple historical behavior data include any multiple of historical click behavior data, historical exposure behavior data, historical download behavior data, historical registration behavior data, and historical credit behavior data; The user device information corresponding to each historical behavior data is obtained by collecting device information in the user device when each historical behavior is identified to have occurred.

3. The method according to claim 1, It is characterized in that The method further comprises: Obtaining the data volume of multiple historical behavior data of the arbitrary first user and the time interval between each historical behavior data; The advertising behavior complexity of the plurality of historical behavior data of the arbitrary first user is determined according to the data volume and the time interval.

4. The method according to claim 3, It is characterized in that The deep learning model for acquiring the complexity of advertising behavior based on multiple historical behavior data includes: If the advertising behavior complexity of the plurality of historical behavior data of any first user is a first complexity, the acquired deep learning model is an RNN model; If the advertising behavior complexity of the plurality of historical behavior data of any first user is the second complexity, the acquired deep learning model is an LSTM model; If the advertising behavior complexity of the plurality of historical behavior data of any first user is the third complexity, the acquired deep learning model is a Transformer model; The first complexity is smaller than the second complexity, and the second complexity is smaller than the third complexity.

5. The method according to claim 4, It is characterized in that The determining the advertising behavior complexity of the plurality of historical behavior data of the target user according to the data volume and the time interval includes: Obtain a first data volume threshold and a first time interval threshold corresponding to the first complexity, a second data volume threshold and a second time interval threshold corresponding to the second complexity, and a third data volume threshold and a third time interval threshold corresponding to the third complexity, wherein the first data volume threshold is less than the second data volume threshold, the second data volume threshold is less than the third data volume threshold, the first time interval threshold is less than the second time interval threshold, and the second time interval threshold is less than the third time interval threshold; When the data volume of the multiple historical behavior data of the arbitrary first user is less than or equal to the first data volume threshold, and / or the time interval for generating each historical behavior data is less than or equal to the first time interval threshold, determining that the advertising behavior complexity of the multiple historical behavior data of the arbitrary first user is the first complexity; When the data volume of the multiple historical behavior data of any first user is greater than the first data volume threshold and less than or equal to the second data volume threshold, and / or the time interval for generating each historical behavior data is greater than the first time interval threshold and less than or equal to the second time interval threshold, determining that the advertising behavior complexity of the multiple historical behavior data of any first user is the second complexity; When the data volume of multiple historical behavior data of any first user is greater than the second data volume threshold and less than or equal to the third data volume threshold, and / or the time interval for generating each historical behavior data is greater than the second time interval threshold and less than or equal to the third time interval threshold, the advertising behavior complexity of the multiple historical behavior data of the target user is determined to be the third complexity.

6. The method according to claim 1, It is characterized in that After the step of obtaining the trained deep learning model, the method further includes: After a preset time, the target behavior data generated by the target advertisement in each advertising channel and the target user device information corresponding to each target behavior data are obtained; Identify a third user corresponding to each target behavior data based on the device information of each target user; Acquire multiple target behavior data of each advertising channel of each third user; Screening out multiple target third users based on the advertising behavior complexity of multiple target behavior data of each third user; Inputting multiple target behavior data of each target third user into the trained deep learning model to obtain the advertising effect attribution value of the target advertisement of each target third user in each advertising channel; Obtaining an advertisement delivery strategy for the target advertisement and obtaining delivery costs of each advertisement channel from the advertisement delivery strategy; Calculate the customer acquisition unit price of each advertising channel according to the delivery cost of each advertising channel and the advertising effect attribution value corresponding to each target third user; The model parameters of the trained deep learning model are adjusted based on the customer acquisition unit price of each advertising channel. The deep learning model after adjusting the model parameters is used to evaluate the advertising effect attribution value of the user's target advertisement in each advertising channel.

7. The method according to claim 6, It is characterized in that After the step of obtaining the advertising effect attribution value of the target advertisement of each target third user in each advertising channel, the method further includes: The advertising delivery strategy of the target advertisement is adjusted based on the advertisement effect attribution value of the target advertisement of each target third user in each advertisement channel.

8. A multi-channel advertising effect attribution evaluation device, It is characterized in that The device comprises: A first acquisition module is used to acquire historical behavior data generated by each first user of the target advertisement in multiple advertising channels and user device information corresponding to each historical behavior data; An identification module, used to identify the first user corresponding to each historical behavior data based on each user device information; A second acquisition module is used to acquire a plurality of historical behavior data of each advertising channel of any first user and acquire a deep learning model based on the advertising behavior complexity of the plurality of historical behavior data; A third acquisition module is used to obtain the advertising effect attribution value of the target advertisement of the arbitrary first user in each advertising channel; A model training module, configured to perform model training on the deep learning model by using a plurality of historical behavior data of the arbitrary first user and the advertising effect attribution value of the target advertisement in each advertising channel to obtain a trained deep learning model; An evaluation module is used to input multiple current behavior data of any second user for the target advertisement into the trained deep learning model when the multiple current behavior data of the second user for the target advertisement are obtained and the advertising behavior complexity of the multiple current behavior data is determined based on the trained deep learning model, so as to obtain the advertising effect attribution value of the target advertisement of the second user in each advertising channel, and any advertising effect attribution value is used to evaluate the advertising effect.

9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, It is characterized in that When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images