Method and apparatus for determining target user

Abstract

The application relates to the technical field of artificial intelligence, and particularly provides a target user determination method and device, which comprises the following steps: predicting a first probability corresponding to each user according to feature information of to-be-pushed content and user portrait features, wherein the first probability is used for indicating a click probability of the user for the to-be-pushed content; determining a candidate user set according to the first probability corresponding to each user; determining a target score corresponding to each candidate user according to the first probability corresponding to the candidate user and a second probability corresponding to the candidate user at a specified page display position, wherein the specified page display position is used for displaying the to-be-pushed content, and the second probability is used for indicating an exposure probability of the pushed content on the specified page display position relative to the candidate user; and determining a target user in the candidate user set according to the target score corresponding to each candidate user, so as to push the to-be-pushed content to the target user; the method can realize targeted pushing of content.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and more specifically, to a method and apparatus for determining a target user. Background Technology

[0002] More and more internet platforms offer content push functionality to proactively send content to users. However, current content push services lack targeting; the pushed content is either unnecessary for the user, or users have to spend a significant amount of time selecting the relevant content from a large pool of options. Therefore, improving the targeting of content pushes is a pressing technical problem that needs to be solved in current technology. Summary of the Invention

[0003] The embodiments of this application provide a method and apparatus for determining target users, so as to improve the targeting of content push.

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

[0005] According to one aspect of the embodiments of this application, a method for determining a target user is provided, comprising: predicting a first probability corresponding to each user based on feature information of content to be pushed and user profile features, wherein the first probability is used to indicate the probability of a user clicking on the content to be pushed; determining a set of candidate users based on the first probabilities corresponding to each user; determining a target score corresponding to each candidate user based on the first probabilities corresponding to the candidate users and a second probability corresponding to the candidate users in a designated page display position; wherein the designated page display position is used to display the content to be pushed; wherein the second probability is used to indicate the probability of the content being exposed relative to the candidate users in the designated page display position; and determining a target user from the set of candidate users based on the target scores corresponding to each candidate user, so as to push the content to be pushed to the target user.

[0006] According to one aspect of the embodiments of this application, a target user determination device is provided, comprising: a first probability determination module, configured to predict a first probability corresponding to each user based on feature information of content to be pushed and user profile features, wherein the first probability is used to indicate the click probability of a user on the content to be pushed; a candidate user set determination module, configured to determine a candidate user set based on the first probabilities corresponding to each user; a target score determination module, configured to determine a target score corresponding to each candidate user based on the first probability corresponding to the candidate user and a second probability corresponding to the candidate user in a designated page display position; wherein the designated page display position is used to display the content to be pushed; the second probability is used to indicate the exposure probability of the pushed content relative to the candidate user in the designated page display position; and a target user determination module, configured to determine a target user from the candidate user set based on the target score corresponding to each candidate user, so as to push the content to be pushed to the target user.

[0007] According to one aspect of the embodiments of this application, an electronic device is provided, including: a processor; and a memory storing computer-readable instructions, which, when executed by the processor, implement the method for determining a target user as described above.

[0008] According to one aspect of the embodiments of this application, a computer-readable storage medium is provided, on which computer-readable instructions are stored, which, when executed by a processor, implement the method for determining a target user as described above.

[0009] In this application's solution, firstly, based on the feature information of the content to be pushed and the user profile features, a first probability corresponding to a user is predicted, and users are initially screened based on the first probability to determine a candidate user set. Then, a target score for the candidate user is calculated based on the first probability and the second probability corresponding to the candidate user, and a second screening is performed on the candidate user set based on the target score to determine the target users of the content to be pushed, so as to push the content to the target users. Since the first probability corresponding to the candidate user is used to indicate the probability of the user clicking on the content to be pushed, and the second probability corresponding to the candidate user is used to indicate the probability of the pushed content being exposed relative to the candidate user in a specified page display position, it realizes the targeted determination of the target users corresponding to the content to be pushed by combining the probability of the pushed content being exposed relative to the user and the probability of the user clicking on the content to be pushed.

[0010] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0011] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. It is obvious that the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort. In the drawings:

[0012] Figure 1 A schematic diagram of an exemplary system architecture to which the technical solutions of the embodiments of this application can be applied is shown.

[0013] Figure 2 This is a flowchart illustrating a method for determining a target user according to an embodiment of this application.

[0014] Figure 3 This is a flowchart illustrating step 210 according to an embodiment of this application.

[0015] Figure 4 This is a flowchart illustrating step 230 according to an embodiment of this application.

[0016] Figure 5 This is a flowchart illustrating the steps preceding step 430 according to an embodiment of this application.

[0017] Figure 6 This is a flowchart illustrating step 420 according to an embodiment of this application.

[0018] Figure 7 This is a flowchart illustrating the steps for calculating the second probability corresponding to a candidate user according to an embodiment of this application.

[0019] Figure 8 This is a flowchart illustrating the steps for calculating the second probability corresponding to a candidate user according to another embodiment of this application.

[0020] Figure 9 This is a flowchart illustrating a method for determining a target user according to a specific embodiment.

[0021] Figure 10 This is a schematic diagram of the interface in the advertiser's client, according to a specific embodiment.

[0022] Figure 11 This is a schematic diagram of the interface in the client of the advertising placement party, according to a specific embodiment.

[0023] Figure 12 This is a block diagram of a target user determination device according to an embodiment of this application.

[0024] Figure 13A schematic diagram of the structure of a computer system suitable for implementing the electronic device of the present application is shown. Detailed Implementation

[0025] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided to make this application more comprehensive and complete, and to fully convey the concept of the exemplary embodiments to those skilled in the art.

[0026] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. Numerous specific details are provided in the following description to give a thorough understanding of embodiments of this application. However, those skilled in the art will recognize that the technical solutions of this application can be practiced without one or more of the specific details, or other methods, components, apparatuses, steps, etc., can be employed. In other instances, well-known methods, apparatuses, implementations, or operations are not shown or described in detail to avoid obscuring various aspects of this application.

[0027] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0028] The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be broken down, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0029] It should be noted that "multiple" in this article refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0030] Figure 1 A schematic diagram of an exemplary system architecture to which the technical solutions of the embodiments of this application can be applied is shown.

[0031] like Figure 1 As shown, the system architecture may include terminal devices (such as...) Figure 1The device shown includes one or more of the following: smartphone 101, tablet 102, portable computer 103, in-vehicle equipment, etc. (which could also be a desktop computer, etc.); network 104; and server 105. Network 104 is a medium used to provide a communication link between the terminal device and server 105. Network 104 can include various connection types, such as wired communication links, wireless communication links, etc.

[0032] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, there can be any number of terminal devices, networks, and servers. For example, server 105 could be a server cluster composed of multiple servers.

[0033] In some embodiments of this application, the terminal device can run a client program, and the server 105 provides services to the client where the terminal device is located. The client in the terminal device can display a page, and all or part of the page displayed in the client can be used to display the content to be pushed in the scheme of this application. The area on the page used to display the pushed content is called the page display position. Therefore, it can also be understood as displaying the (to be) pushed content in the page display position on the page. In a specific embodiment, the page display position can be all or part of the homepage of the client, or an area in a subpage recorded on the homepage, etc.

[0034] The presentation format of the content to be pushed in this application can be one of text, image, video, or a combination of at least two. In specific embodiments, the content to be pushed can be articles (such as news, blogs, or text shared on content sharing platforms), videos, audio, advertisements, etc., without specific limitations.

[0035] In some embodiments of this application, the terminal device can also be used for users to upload push content. After the push content is uploaded to the server 105, the server 105 can determine the target user for the push content according to the method applied for, so as to push the push content to the target user accordingly.

[0036] In some embodiments of this application, the method for determining the target user can be executed by server 105. After determining the target user, the server pushes the content to be pushed to the target user's terminal device (or client). Correspondingly, after the target user enters the corresponding page in the client, they can see the content to be pushed displayed in a designated page display position. The designated page display position for displaying the content to be pushed can be preset, or it can be set by the user through their terminal device. Therefore, after determining the target user for the content to be pushed, server 105 uses the target user as the push target and displays the content to be pushed in the designated page display position when the target user enters the page containing the content.

[0037] In some embodiments of this application, to prevent the content to be pushed from being tampered with, the content can be written into a blockchain, and retrieved from the blockchain when content needs to be pushed. In this application scenario, Figure 1 The terminal devices and servers shown can all serve as blockchain nodes in the blockchain system. Each blockchain node has the same blockchain, and the content to be pushed is stored in the blockchain by packaging it into blocks.

[0038] In some embodiments of this application, the method of this application can also be executed by a cloud server, which uses cloud technology to unify a series of resources such as hardware, software, and network within a wide area network or local area network to realize data computation, storage, processing, and sharing.

[0039] Cloud technology is a collective term for network technology, information technology, integration technology, management platform technology, and application technology applied to the cloud computing business model. It can form resource pools, providing flexible and convenient on-demand access. Cloud computing technology will become a crucial support. Backend services of technical network systems require substantial computing and storage resources, such as video websites, image websites, and many portal websites. With the rapid development and application of the internet industry, every item may have its own identification mark in the future, requiring transmission to backend systems for logical processing. Data at different levels will be processed separately, and various industry data will require robust system support, which can only be achieved through cloud computing. With the support of cloud computing and cloud storage, targeted push content stored in the database can be delivered to numerous users.

[0040] In the scheme of this application, a first neural network model and / or a second neural network model can be constructed in server 105 to determine the target user by means of artificial intelligence technology. The first neural network model is used to predict the second probability corresponding to the candidate user, and the second neural network model is used to predict the first probability corresponding to the candidate user.

[0041] Artificial intelligence (AI) is the theory, methods, technology, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a comprehensive technology within computer science that attempts to understand the essence of intelligence and produce a new kind of intelligent machine that can react in a way similar to human intelligence. AI studies the design principles and implementation methods of various intelligent machines, enabling them to possess the functions of perception, reasoning, and decision-making.

[0042] Artificial intelligence (AI) is a comprehensive discipline encompassing a wide range of fields, including both hardware and software technologies. Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies primarily include computer vision, speech processing, natural language processing, and machine learning / deep learning.

[0043] By using the first neural network model to predict the second probability corresponding to the candidate user and the second neural network model to predict the first probability corresponding to the candidate user, the prediction accuracy and prediction speed of the first and second probabilities can be improved, thereby ensuring the efficiency of target user identification and the matching degree between the content to be pushed and the target user, making the information push more targeted.

[0044] The implementation details of the technical solutions in the embodiments of this application are described in detail below:

[0045] Figure 2 A flowchart illustrating a method for determining a target user according to an embodiment of this application is shown. This method can be executed by a computer device with processing capabilities, such as a server, etc., without specific limitation herein. (Refer to...) Figure 2 As shown, the method includes at least steps 210 to 240, which are described in detail below:

[0046] Step 210: Based on the feature information of the content to be pushed and the user profile features, predict the first probability corresponding to each user. The first probability is used to indicate the probability of the user clicking on the content to be pushed.

[0047] The feature information of the content to be pushed is used to describe the characteristics of the content to be pushed, such as the author, topic tags, and name. As described above, the content to be pushed can be articles (e.g., news, blogs, etc.), videos, audio, advertisements, etc. If the content to be pushed is an article, its features can be the author, title, topic tags, and type; if the content to be pushed is a video, its features can be the author, title, tags, and description; if the content to be pushed is an advertisement, its features can include the advertiser's preset user tag information, the advertisement's tag information, and the advertisement's theme; the advertiser's preset user tag information includes, for example, the user's preferences, age group, occupation, and user tags.

[0048] User profile features are used to describe user characteristics, such as age, occupation, user tags, preferred content tags, and user address. In specific embodiments, user profile features can be obtained by combining multiple dimensions of user features.

[0049] For users, the higher the match between the push content and the user, the higher the probability that the push content is the content the user needs, and consequently, the higher the probability that the user will click on the push content.

[0050] Therefore, based on the feature information of the content to be pushed and the user's user profile features, the matching degree between the content to be pushed and a user can be mined, and then the first probability corresponding to the user can be predicted. In some embodiments of this application, the click probability of the user on the content to be pushed can be predicted first, and then the first probability corresponding to the predicted click probability can be determined according to the mapping relationship between the click probability and the first probability, thereby realizing the prediction of the first probability corresponding to the user.

[0051] In some embodiments of this application, deep learning may be used to predict the initial probability of a user's response.

[0052] In some embodiments of this application, the prediction of the first probability can be performed using deep learning. Specifically, such as... Figure 3 As shown, step 210 includes:

[0053] Step 310: Input the feature information of the content to be pushed and the user profile features into the second neural network model; the second neural network model is trained based on the feature information of the pushed content, the user profile features of the sample user, and the click trigger information of the sample user for the pushed content; wherein, the click trigger information is used to indicate whether the sample user clicks on the displayed pushed content.

[0054] Step 320: The second neural network model performs score prediction based on the feature information of the content to be pushed and the user profile features to obtain the first probability corresponding to each user.

[0055] The second neural network model can be a model constructed using recurrent neural networks, convolutional neural networks, etc.

[0056] To ensure the accuracy of the first probability prediction performed by the second neural network model, it needs to be trained first. In this embodiment, the training data for the second neural network model includes feature information of the pushed content, user profile features of the sample users, and click trigger information of the sample users for the pushed content.

[0057] During the training of the second neural network model, the feature information of the pushed content and the user profile features of the sample users are input into the second neural network model. The second neural network model outputs a click label, which indicates whether the sample user triggered a click action on the pushed content. If the output click label does not match the sample user's click trigger information for the pushed content, the parameters of the second neural network model are adjusted until the click label output by the adjusted second neural network model matches the sample user's click trigger information for the pushed content.

[0058] Through the training process described above, the second neural network can learn the correlation between user profile features and push content features and whether the user triggers a click operation on the push content, thereby achieving a first probability prediction.

[0059] Step 220: Determine the candidate user set based on the first probability corresponding to each user.

[0060] In some embodiments of this application, users can be sorted from high to low according to a first probability, and a first specified number of users in the sorted order can be selected as candidate users and added to a candidate user set. The users in the candidate user set are referred to as candidate users. This first specified number can be set according to actual needs and is not specifically limited here.

[0061] In some embodiments of this application, a first probability threshold may be set, and users with a first probability not lower than the first probability threshold may be identified as candidate users and added to the candidate user set; the first probability threshold may be set according to actual needs, and is not specifically limited here.

[0062] Step 230: Determine the target score for each candidate user based on the first probability corresponding to the candidate user and the second probability corresponding to the candidate user in the designated page display position; the designated page display position is used to display the content to be pushed; the second probability is used to indicate the exposure probability of the pushed content relative to the candidate user in the designated page display position.

[0063] To push content, a page display position is set for the pushed content by default. The pushed content is associated with the page display position, so that during the content push process, the pushed content associated with the page display position is displayed in the page display position.

[0064] In the solution of this application, the designated page display position can be a page display position set for the content to be pushed. That is, after the target user is determined, the pushed content is displayed in the designated page display position on the target user's client page. There can be one or more designated page display positions, which can be set according to actual needs.

[0065] For a user, the push content displayed on a specific page will only be perceived by the user when their client triggers access to that page's display position; that is, the push content is exposed to the user. Conversely, if the user's client does not trigger access to the page's display position, even if the server pushes content to the user, the content is not exposed to them. Therefore, from this perspective, the probability of push content being exposed to a candidate user on a specific page's display position refers to the probability that the user's client triggers access to that specific page's display position, given that content is pushed to that position. In other words, the probability of push content being exposed to a candidate user on a specific page's display position reflects the likelihood that the candidate user will trigger access to that specific page's display position.

[0066] Understandably, since the designated page display area serves as the area where the associated content to be pushed is shown, if a candidate user does not trigger an entry into that designated page display area on their client, the user will not be aware of the content to be pushed to them. Therefore, the probability of the pushed content being exposed on the designated page display area relative to the candidate user reflects the probability that the user's client will enter the designated page display area.

[0067] In some embodiments of this application, the product of the first probability corresponding to the candidate user and the second probability corresponding to the candidate user in a designated page display position can be used as the target score for the candidate user. Alternatively, when there are multiple designated page display positions, a reference score can be calculated based on the second probability corresponding to the candidate user in each of the designated page display positions, and then the product of the first probability corresponding to the candidate user and the reference score can be used as the target score for the candidate user.

[0068] In some embodiments of the application, the reference score corresponding to the candidate user can be a weighted probability obtained by weighting the second probabilities corresponding to the candidate user in each of the specified page display positions.

[0069] Step 240: Based on the target scores corresponding to each candidate user, determine the target user from the candidate user set, and push the content to be pushed to the target user.

[0070] In some embodiments of this application, candidate users in the candidate user set can be sorted from high to low according to the target score, and the top specified number of candidate users in the sorted list can be selected as target users. This specified number can be set according to actual needs and is not specifically limited here. It is understood that if the users in the candidate user set are selected based on a first specified number, then this first specified number is not less than a specified number.

[0071] In some embodiments of this application, in order to ensure the basis for screening target users and avoid selecting users with a lower probability as target users, the first specified number can be a multiple of N, where N is a positive number greater than 1.

[0072] In some embodiments of this application, a target score threshold can also be set to identify candidate users whose target scores are not lower than the target score threshold as target users; the target score threshold can be set according to actual needs and is not specifically limited here.

[0073] In some embodiments of this application, after determining the target user corresponding to the content to be pushed, the content is pushed to the target user and displayed in a designated page display position. Thus, after the target user's client enters the designated page display position, the user can quickly obtain the content to be pushed displayed there.

[0074] In some embodiments of this application, a push period for the content to be pushed can be further set, and the content to be pushed is pushed to the target user within the set push period.

[0075] In this application's solution, firstly, based on the feature information of the content to be pushed and the user profile features, a first probability corresponding to a user is predicted, and users are initially screened based on the first probability to determine a candidate user set. Then, a target score for the candidate user is calculated based on the first probability and the second probability corresponding to the candidate user, and a second screening is performed on the candidate user set based on the target score to determine the target users of the content to be pushed, so as to push the content to the target users. Since the first probability corresponding to the candidate user is used to indicate the probability of the user clicking on the content to be pushed, and the second probability corresponding to the candidate user is used to indicate the probability of the pushed content being exposed relative to the candidate user in a specified page display position, it realizes the targeted and automatic determination of the target users corresponding to the content to be pushed by combining the probability of the pushed content being exposed relative to the user and the probability of the user clicking on the content to be pushed.

[0076] Furthermore, since the first probability corresponding to a candidate user reflects the matching degree between the candidate user and the content to be pushed, and the second probability corresponding to a candidate user reflects the exposure probability of the content to be pushed to the candidate user, if target users with high exposure probability and high matching with the content to be pushed are selected, then after pushing the content to be pushed to the target users, due to the high exposure probability of the content to be pushed, the target users can quickly obtain content with a high degree of matching with themselves, without the need for users to spend time searching through a massive amount of content.

[0077] In some embodiments of this application, such as Figure 4 As shown, step 230 includes:

[0078] Step 410: Correct the first probability corresponding to the candidate user according to the adjustment factor to obtain the corrected first probability corresponding to the candidate user. The adjustment factor is a positive number not exceeding 1.

[0079] In some embodiments of this application, the adjustment factor can be used as the exponent of the first probability, and the first probability corresponding to the candidate user can be exponentially calculated to obtain the corrected first probability corresponding to the candidate user; specifically, it can be expressed by the following formula (1):

[0080] K i =(k i ) t ; (Formula 1)

[0081] Among them, K i k represents the corrected first probability corresponding to the i-th candidate user; i represents the first probability (i.e., the first probability before correction) corresponding to the i-th candidate user; t is the adjustment factor.

[0082] In some embodiments of this application, the product of the adjustment factor and the first probability corresponding to the candidate user can also be used as the modified first probability corresponding to the candidate user.

[0083] Step 420: Weight the second probability of the candidate user in all specified page display positions to obtain the weighted probability of the candidate user.

[0084] In the scheme of this embodiment, a weighting coefficient can be pre-set for each specified page display position. Then, according to the weighting coefficient corresponding to each specified page display position, the second probability of the candidate user in all specified page display positions is weighted. The weighted result is the weighted probability of the candidate user.

[0085] The weighted probability of a candidate user can reflect the exposure of all specified page display positions relative to the candidate user.

[0086] Step 430: Calculate the target score for each candidate user based on the corrected first probability and the weighted probability corresponding to the candidate user.

[0087] In some embodiments of this application, the product of the modified first probability corresponding to the candidate user and the weighted probability corresponding to the candidate user can be used as the target score corresponding to the candidate user.

[0088] In some embodiments of this application, the sum of the corrected first probability corresponding to the candidate user and the weighted probability corresponding to the candidate user can also be used as the target score corresponding to the candidate user. In this embodiment, the corrected first probability corresponding to the candidate user can be the product of the adjustment factor and the first probability corresponding to the candidate user. In this application scenario, multiple selectable adjustment factors can be preset, such as 0.1, 0.3, 0.5, 0.7, and 0.9, and then the optimal adjustment factor among these multiple selectable adjustment factors can be tested, for example, by A / B testing to select the optimal adjustment factor; the determined optimal adjustment factor is the initial adjustment factor, and subsequent adjustments are based on this optimal adjustment factor.

[0089] In this embodiment, the first probability corresponding to a candidate user is corrected by a positive number less than 1 (i.e., an adjustment factor), making the corrected first probability of the candidate user smaller than the original first probability. Since the first probability of a candidate user reflects the matching degree between the candidate user and the content to be pushed, calculating the target score of the candidate target user using the corrected first probability is equivalent to appropriately relaxing the matching degree requirement between the content to be pushed and the candidate user. This facilitates the selection of more target users who may need the content to be pushed, allowing the content to be pushed to more target users.

[0090] In some embodiments of this application, the number of target users is a specified number; such as Figure 5 As shown, after step 430, the method further includes:

[0091] Step 510: Obtain the first parameter value of the specified statistical parameter. The first parameter value is obtained by statistical analysis of the first probability corresponding to the specified number of candidate users with the highest first probability in the candidate user set.

[0092] The first parameter value of the specified statistical parameter is used to reflect the comprehensive distribution of the first probability corresponding to the specified number of candidate users with the highest first probability in the candidate user set. This specified statistical parameter can be one of the following: mean, median, mode, set percentile, or sum of scores (i.e., the sum of the first probabilities corresponding to the specified number of candidate users with the highest first probability), and no specific limitation is made here.

[0093] Step 520: Obtain the second parameter value of the specified statistical parameter. The second parameter value is obtained by statistical analysis of the target scores corresponding to the specified number of target users with the highest target scores in the candidate user set.

[0094] The second parameter value of the specified statistical parameter is used to reflect the overall distribution of the target scores of the specified number of target users with the highest target scores in the candidate user set.

[0095] It is worth mentioning that in steps 510 and 520, the first parameter value and the second parameter value are values ​​calculated for the same specified statistical parameter. That is, if step 510 calculates the average of a specified number of first probabilities, then step 520 calculates the average of a specified number of target scores.

[0096] Step 530: Calculate the difference between the first parameter value and the second parameter value of the specified statistical parameter.

[0097] Step 540: If the difference is not lower than a preset threshold, the adjustment factor is reduced so as to recalculate the target score corresponding to each candidate user through the adjusted adjustment factor, and the target user is re-determined through the recalculated target score.

[0098] If the difference is not lower than the preset threshold, reducing the adjustment factor will reduce the difference based on the target score and target user determined by the smaller adjustment factor, and thus gradually reduce the difference to below the preset threshold.

[0099] In some embodiments of this application, a preset mapping relationship between the difference and the adjustment factor can be established. Therefore, after step 530, the adjustment factor corresponding to the calculated difference is determined, and the adjustment is reduced according to the indication of this adjustment factor. It is understood that in step 540, this adjustment factor is the reduction magnitude of the adjustment factor.

[0100] In some embodiments of this application, the adjustment factor can be adjusted by the same adjustment range each time. For example, when the difference is not lower than a preset threshold, the adjustment factor is reduced by 0.1 each time. Of course, in other embodiments, the adjustment range of the adjustment factor each time can also be different, and no specific limitation is made here.

[0101] After reducing the adjustment factor, the target score for each candidate user is recalculated using the adjusted adjustment factor. Then, the process of steps 510-530 above is repeated until the difference obtained by the recalculation is lower than the preset threshold. The target user is then re-determined based on the target score for each candidate user obtained by the last calculation.

[0102] The initial value of the adjustment factor can be set according to actual needs. For example, if the initial value of the adjustment factor is set to 0.3, then in the initial calculation of the target score corresponding to the candidate user, the calculation will be performed according to the initial value of the adjustment factor, and the adjustment factor will be adjusted based on the initial value.

[0103] In this embodiment, an adjustment factor is introduced to regulate the contribution of the first probability and the second probability to the target score. It is understood that a higher first probability results in a higher click-through rate (CTR) of the content to be pushed to the target user, while a higher second probability results in a higher reach (exposure) rate. However, when both the second and first probabilities are high, a high exposure rate can lead to a decrease in the CTR. This means that users still need to spend time filtering the content they want to push. In this embodiment, an adjustment factor is introduced and adjusted based on the difference between the first and second parameter values ​​to achieve a trade-off between exposure rate and CTR, reducing the impact of high exposure rate on the CTR of the pushed content.

[0104] In some embodiments of this application, in Figure 5 In the corresponding embodiment, step 240 further includes: if the difference is lower than the preset threshold, then the specified number of candidate users with the highest target scores in the candidate user set are determined as target users.

[0105] In other words, if the calculated difference is lower than the preset threshold, there is no need to adjust the adjustment factor. The target user is directly determined by the target score of each candidate user initially calculated based on the adjustment factor.

[0106] In some embodiments of this application, such as Figure 6 As shown, step 420 further includes:

[0107] Step 610: Obtain the number of times the content is exposed for each specified page display position.

[0108] The number of times content is displayed for a specific page ad slot refers to the number of times the push content associated with that ad slot is displayed. For a user, the push content displayed on that ad slot is only exposed to that user's client when the user triggers access to that ad slot on their client; conversely, if the user's client does not trigger access to the ad slot, the push content displayed on that ad slot is not exposed to that user's client. Therefore, from this perspective, the number of times content is displayed for a specific page ad slot can be considered as the number of times that specific page ad slot is accessed on the client.

[0109] In some embodiments of this application, since the push content associated with different time periods in the same page display position may be different, in step 610, the number of times the content corresponding to each specified page display position is exposed can be counted according to the first preset time period. In other words, the number of times the client enters the specified page display position in the first preset time period is counted, without needing to pay attention to whether the push content pushed in the specified page display position in the first preset time period is the same push content.

[0110] In some embodiments of this application, to ensure that the number of times the content corresponding to each specified page display position is related to the content to be pushed, in step 610, the number of times the content corresponding to the specified page display position is exposed can also be the number of times the content corresponding to the target similar content associated with the specified page display position is exposed within a first preset time period. Here, the target similar content refers to already pushed content whose similarity to the content to be pushed is not lower than a set similarity threshold.

[0111] In some embodiments of this application, if the content to be pushed has already been pushed in the designated page display position in the past, in step 610, the number of times the content corresponding to the designated page display position is exposed can also be the historical number of times the content to be pushed is displayed in the designated page display position in the first preset time period.

[0112] The number of times the content corresponding to a specified page display position is exposed reflects the number of times the client enters that specified display position. In a specific embodiment, the number of times the content corresponding to the specified page display position is exposed can be calculated based on browsing logs generated from the user's page browsing history in the client or server.

[0113] In the solution of this application, the designated page display position can be a page display position set for the content to be pushed. That is, after the target user is determined, the pushed content is displayed in the designated page display position on the target user's client page. There can be one or more designated page display positions, and no specific limitation is made here.

[0114] Step 620: Determine the weighting coefficient corresponding to each designated page display position based on the number of times the content is exposed for each designated page display position, wherein the weighting coefficient is proportional to the number of times the content is exposed for the corresponding designated page display position.

[0115] In some embodiments of the application, the ratio of the number of times the content corresponding to a specified page display position is exposed to the total number of times the content is exposed can be used as the weighting coefficient for that specified page display position. Here, the total number of times the content is exposed refers to the sum of the number of times the content is exposed for all specified page display positions.

[0116] Step 630: According to the weighting coefficient corresponding to each specified page display position, the second probability of the candidate user in all specified page display positions is weighted to obtain the weighted probability of the candidate user.

[0117] Since the number of times the content corresponding to a specified page display position is exposed reflects the number of times the client enters that specified page display position, a weighting coefficient is determined based on the number of times the content is exposed for each specified page display position. The second probability of the candidate user corresponding to all specified page display positions is then weighted according to the determined weighting coefficient. This ensures that the calculated weighted probability can reflect the overall entry situation of the candidate user's client into all specified page display positions.

[0118] In some embodiments of this application, prior to step 230, the method further includes: calculating a second probability corresponding to the candidate user based on the content exposure information of the candidate user on the designated page display position for the pushed content; wherein the content exposure information is used to indicate whether the pushed content carried in the designated page display position is exposed to the client of the corresponding candidate user.

[0119] Content exposure information can also be understood as an indication of whether the client of the candidate user enters the specified page display position.

[0120] In some embodiments of this application, such as Figure 7As shown, the second probability corresponding to a candidate user can be calculated as follows:

[0121] Step 710: Input the content exposure information of the candidate user in the designated page display position for the pushed content and the user profile features of the candidate user into the first neural network model; the first neural network model is trained based on the user profile features of the sample user and the content exposure information of the sample user for the historical pushed content.

[0122] Step 720: The first neural network model performs score prediction based on the content exposure information of the candidate user in the designated page display position for the pushed content and the user profile characteristics of the candidate user, to obtain the second probability corresponding to the candidate user.

[0123] The first neural network model can be a model constructed using recurrent neural networks, convolutional neural networks, etc. In some embodiments of this application, the first neural network model can be an XGBoost model, an LR model (Logistic Regression Model), or a DNN model (Deep Neural Networks).

[0124] To ensure the accuracy of the first neural network model in making the second probability prediction, it needs to be trained first. In this embodiment, the training data for the first neural network model includes the user profile features of the sample users and the content exposure information of the sample users for historically pushed content.

[0125] In some embodiments of this application, to ensure the training effect of the first neural network model, the sample users can include positive sample users and negative sample users, where positive sample users refer to users who have viewed the pushed content, and negative sample users refer to users who have not viewed the pushed content. In a specific embodiment, to ensure the accuracy of the first neural network model in making the first probability prediction, training data related to positive sample users and negative sample users can be selected according to a set ratio, such as 1:1.

[0126] In some embodiments of this application, to ensure the effectiveness of the training of the first neural network model, content exposure information of sample users for historically pushed content can be collected over a period of time, thereby forming a sequence of exposure behaviors of the sample user for historically pushed content during that period. For example, the exposure behavior sequence can be used to indicate the content exposure information of sample users for historically pushed content within 14 days. This exposure behavior sequence can be set as a 14-dimensional vector, where the m-th dimension represents the number of times the sample user exposed pushed content in a designated page display position on the m-th day in the past.

[0127] During the training of the first neural network model, the content exposure information of the candidate user in the designated page display position for the pushed content and the user profile features of the candidate user are input into the first neural network model. The first neural network model predicts the exposure tag corresponding to the candidate user on the next day. The exposure tag is used to indicate whether the candidate user will expose the pushed content in the designated page display position on the next day. If the predicted exposure tag is inconsistent with the candidate user's exposure in the designated page display position on the next day, the parameters of the first neural network model are adjusted until the re-predicted exposure tag is consistent with the candidate user's exposure in the designated page display position on the next day.

[0128] Through the training process described above, the first neural network model can learn to predict the second probability of a candidate user in a future time period based on the user profile characteristics and the content exposure information of the candidate user in the specified page display position for the pushed content.

[0129] In some embodiments of this application, the training data of the first neural network model may further include click trigger information of sample users on the pushed content. Correspondingly, in the process of the first neural network model predicting the second probability, the input of the model includes click trigger information of candidate users on the pushed content.

[0130] In some embodiments of this application, the parameter selection evaluation metric for the first neural network model can be AUC (Area Under Curve, the area under the ROC curve and the coordinate axes). The ROC curve (Receiver Operating Characteristic Curve) is a curve plotted with the true positive rate on the ordinate and the false positive rate on the abscissa based on a series of different binary classification methods (cutoff values ​​or decision thresholds). AUC is a performance metric for measuring the quality of a classifier.

[0131] In other embodiments of this application, such as Figure 8 As shown, the second probability corresponding to a candidate user can be calculated as follows:

[0132] Step 810: Based on the content exposure information of the candidate users on the designated page display position for the pushed content, count the number of target unit statistical time periods within the set time period; the target unit statistical time period refers to the unit statistical time period corresponding to the display of the pushed content on the designated page display position.

[0133] Step 820: Calculate the second probability corresponding to the candidate user based on the total number of unit statistical time periods in the set time period and the number of target unit statistical time periods.

[0134] In some embodiments of this application, the ratio of the number of target unit statistical time periods in a set time period to the total number of unit statistical time periods in the set time period can be used as the second probability corresponding to the candidate user.

[0135] The time period and the unit statistical time period can be set according to actual needs. For example, the time period can be set to 14 days and the unit statistical time period can be set to 1 day. Then, in step 810, it is necessary to count the number of days (let's assume it's P) during which the candidate user displays the pushed content in the designated page display position within the 14-day time period. This is the number of target unit statistical time periods. In this example, the total number of unit statistical time periods in the time period is 14. Therefore, the second probability corresponding to the candidate user is P / 14.

[0136] In some embodiments of this application, the number of times the content is exposed in a set time period can be counted based on the content exposure information of the candidate user on the specified page display position for the pushed content. Then, the number of times the content is exposed in the set time period is normalized to the maximum or minimum, and the result of the normalization is used as the second probability corresponding to the candidate user.

[0137] In some embodiments of this application, if the sample size of average exposed users (i.e., positive sample users) and average unexposed users (i.e., negative sample users) for a specified page display position is large, then the corresponding amount of training data is also large, which can meet the training needs of the first neural network model. Therefore, in this case, the first neural network model can be trained based on a large amount of training data, and the second probability corresponding to the candidate user can be predicted based on the trained first neural network model.

[0138] In some embodiments of this application, if the number of average exposure sample users for a specified page display position is small, the corresponding number of samples that can be collected is also small. This small number of samples is insufficient to meet the training requirements of the first neural network model. In such scenarios, a different approach can be adopted. Figure 8 The method in the corresponding embodiment calculates the second probability corresponding to the candidate user.

[0139] Figure 9 This is a flowchart illustrating a target user determination method according to a specific embodiment, such as... Figure 9 As shown, it includes:

[0140] Step 910: Calculate the weighted probability corresponding to the candidate user. Specifically, first follow... Figure 7 or Figure 8 The illustrated embodiment calculates the second probability corresponding to the candidate user; then according to... Figure 6 The example shown calculates the weighted probability corresponding to the candidate user.

[0141] Step 920: Determine the candidate user set; specifically, users can be screened according to the first probability corresponding to the candidate users obtained in step 210, and the first specified number of users with the highest first probability can be selected as candidate users.

[0142] Step 930: Determine the target users. Specifically, calculate the target score as in step 230, and then perform a secondary screening from the candidate user set based on the target scores of the candidate users to initially determine the target users.

[0143] Step 940: Calculate the difference between the first parameter value and the second parameter value of the specified statistical parameter. The first parameter value can be calculated according to the procedure in step 510, and the second parameter value can be calculated according to the procedure in step 520. Then, calculate the difference between the first parameter value and the second parameter value accordingly.

[0144] Step 950: Determine if the difference is lower than a preset threshold; if yes, proceed to step 960. If no, proceed to step 951: Decrease the adjustment factor; then return to step 930.

[0145] Step 960: Output the target user's information. The target user's information includes the target user's user identifier, and then push the content to be pushed to the target user.

[0146] This solution can be applied to ad delivery. The following is an explanation of the terminology used in ad delivery:

[0147] Audience targeting: Before launching an ad campaign, advertisers need to define the target audience as a targeting criterion. Audience targeting is the first step in ad campaigning.

[0148] Audience targeting reach rate: After an advertiser uses the selected targeting to deliver ads, the proportion of the targeted audience that is exposed to the ads is called the audience targeting reach rate, which can also be called the exposure rate.

[0149] Ad placement: refers to the location provided by the advertiser to the advertiser for displaying advertising materials and copy. Examples include splash screen ads in different applications, ads in content interaction platforms, blog posts, WeChat official account articles, news interstitials, and video interstitials.

[0150] Intelligent audience targeting: refers to the function of automatically generating targeted audience packages based on the performance of the advertisement.

[0151] When the information to be pushed is an advertisement, the process of automatically determining the target users for the information to be pushed is equivalent to realizing intelligent audience targeting and automatically generating targeted audience packages (i.e., the set of target users).

[0152] In some embodiments of this application, the content to be pushed may be an advertisement. The push of the advertisement involves the advertiser, the advertising platform, and the user. In this application scenario, the advertiser can set the target user according to the method of this application, or the advertising platform can set the target user of the advertisement according to the method of this application.

[0153] If the advertiser sets up the method to determine the target users according to this application, the client provided to the advertiser provides a trigger entry point for the method of this application. Figure 10 This is a schematic diagram of the interface in the advertiser's client application, illustrating a specific embodiment. For example... Figure 10 As shown, if the "Model Prediction Correction" control is triggered, the adjustment factor and the final target user will be adjusted according to the method described in this application. The intelligent targeting module in the advertiser's client will then combine the advertiser's selected optimization goal (click-through rate or conversion rate) and the ad's performance to redetermine the ad's target user.

[0154] If the advertiser determines the target users of the advertisement according to the method of this application, the client provided to the advertiser provides a trigger entry for the method of this application. Figure 11 This is a schematic diagram of the interface in the client application of the advertising platform, according to a specific embodiment. Figure 10 As shown, if the advertiser allows the "Automatic Scalability" option to be enabled, it will be enabled on the ad placement side. Figure 11 The "Automatic Targeting" option will activate the intelligent targeting function the day after tomorrow, allowing advertisers to adjust adjustment factors according to the method described in this application and identify target users to improve the reach of the ads to be pushed.

[0155] The following describes an apparatus embodiment of this application, which can be used to perform the methods described in the above embodiments of this application. For details not disclosed in the apparatus embodiments of this application, please refer to the method embodiments described in the above embodiments of this application.

[0156] Figure 12 This is a block diagram of a target user determination device according to one embodiment, such as... Figure 12As shown, the target user determination device includes: a first probability determination module 1210, used to predict a first probability corresponding to each user based on the feature information of the content to be pushed and user profile features, wherein the first probability is used to indicate the probability of a user clicking on the content to be pushed; a candidate user set determination module 1220, used to determine a candidate user set based on the first probabilities corresponding to each user; a target score determination module 1230, used to determine a target score corresponding to each candidate user based on the first probability corresponding to the candidate user and a second probability corresponding to the candidate user in a designated page display position; the second probability is used to indicate the exposure probability of the pushed content relative to the candidate user in the designated page display position; the designated page display position is used to display the content to be pushed; and a target user determination module 1240, used to determine a target user from the candidate user set based on the target score corresponding to each candidate user, so as to push the content to be pushed to the target user.

[0157] In some embodiments of this application, the target score determination module 1230 includes: a correction unit, configured to correct the first probability corresponding to the candidate user according to an adjustment factor to obtain the corrected first probability corresponding to the candidate user, wherein the adjustment factor is a positive number not exceeding 1; a weighted probability determination unit, configured to weight the second probability corresponding to the candidate user in all specified page display positions to obtain the weighted probability corresponding to the candidate user; and a target score determination unit, configured to calculate the target score corresponding to each candidate user according to the corrected first probability corresponding to the candidate user and the weighted probability corresponding to the candidate user.

[0158] In some embodiments of this application, the number of target users is a specified number; the target user determination device further includes: a first parameter value acquisition module, used to acquire a first parameter value of a specified statistical parameter, the first parameter value being obtained by statistical analysis of the first probability corresponding to a specified number of candidate users with the highest first probability in the candidate user set; a second parameter value acquisition module, used to acquire a second parameter value of the specified statistical parameter, the second parameter value being obtained by statistical analysis of the target score corresponding to a specified number of target users with the highest target score in the candidate user set; a difference calculation module, used to calculate the difference between the first parameter value of the specified statistical parameter and the second parameter value of the specified statistical parameter; and an adjustment factor adjustment module, used to reduce the adjustment factor if the difference is not lower than a preset threshold, so as to recalculate the target score corresponding to each candidate user through the adjusted adjustment factor, and redetermine the target user through the recalculated target score.

[0159] In some embodiments of this application, the target user determination module 1240 is further configured to: if the difference is lower than the preset threshold, determine a specified number of candidate users with the highest corresponding target scores in the candidate user set as target users.

[0160] In some embodiments of this application, the weighted probability determination unit includes: a quantity acquisition unit, configured to acquire the number of times the content corresponding to each specified page display position is exposed; a weighting coefficient determination unit, configured to determine a weighting coefficient corresponding to each specified page display position based on the number of times the content corresponding to each specified page display position is exposed, wherein the weighting coefficient is proportional to the number of times the content corresponding to the specified page display position is exposed; and a weighting unit, configured to weight the second probability of the candidate user in all specified page display positions according to the weighting coefficient corresponding to each specified page display position, to obtain the weighted probability of the candidate user.

[0161] In some embodiments of this application, the target user determination device further includes: a second probability calculation module, used to calculate a second probability corresponding to the candidate user based on the content exposure information of the candidate user in the designated page display position for the pushed content; wherein, the content exposure information is used to indicate whether the pushed content carried in the designated page display position is exposed to the client of the corresponding candidate user.

[0162] In some embodiments of this application, the second probability calculation module includes: a first input unit, configured to input the content exposure information of the candidate user in a designated page display position for the pushed content and the user profile features of the candidate user into a first neural network model; the first neural network model is trained based on the user profile features of the sample user and the content exposure information of the sample user for historically pushed content; and a second prediction unit, configured to obtain the second probability corresponding to the candidate user by the first neural network model performing score prediction based on the content exposure information of the candidate user in a designated page display position for the pushed content and the user profile features of the candidate user.

[0163] In some other embodiments of this application, the second probability calculation module includes: a statistics unit, used to count the number of target unit statistical time periods in a set time period based on the content exposure information of the candidate user for the pushed content in the designated page display position; the target unit statistical time period refers to the unit statistical time period corresponding to the display of the pushed content in the designated page display position; the second probability calculation unit is used to calculate the second probability corresponding to the candidate user based on the total number of unit statistical time periods in the set time period and the number of target unit statistical time periods.

[0164] In some embodiments of this application, the first probability determination module 1210 includes: a second input unit, used to input the feature information of the content to be pushed and the user profile features into a second neural network model; the second neural network model is trained based on the feature information of the pushed content, the user profile features of the sample user, and the click trigger information of the sample user for the pushed content; wherein, the click trigger information is used to indicate whether the sample user clicks on the displayed pushed content; and a second prediction unit, used by the second neural network model to perform score prediction based on the feature information of the content to be pushed and the user profile features to obtain the first probability corresponding to each user.

[0165] Figure 13 A schematic diagram of the structure of a computer system suitable for implementing the electronic device of the present application is shown.

[0166] It should be noted that, Figure 13 The computer system 1300 of the electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0167] like Figure 13 As shown, the computer system 1300 includes a Central Processing Unit (CPU) 1301, which can perform various appropriate actions and processes, such as executing the methods described in the above embodiments, based on programs stored in Read-Only Memory (ROM) 1302 or programs loaded from storage portion 1308 into Random Access Memory (RAM) 1303. The RAM 1303 also stores various programs and data required for system operation. The CPU 1301, ROM 1302, and RAM 1303 are interconnected via a bus 1304. An Input / Output (I / O) interface 1305 is also connected to the bus 1304.

[0168] The following components are connected to I / O interface 1305: an input section 1306 including a keyboard, mouse, etc.; an output section 1307 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 1308 including a hard disk, etc.; and a communication section 1309 including a network interface card such as a LAN (Local Area Network) card, modem, etc. The communication section 1309 performs communication processing via a network such as the Internet. A drive 1310 is also connected to I / O interface 1305 as needed. Removable media 1311, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., are installed on drive 1310 as needed so that computer programs read from them can be installed into storage section 1308 as needed.

[0169] Specifically, according to embodiments of this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 1309, and / or installed from removable medium 1311. When the computer program is executed by central processing unit (CPU) 1301, it performs various functions defined in the system of this application.

[0170] It should be noted that the computer-readable medium shown in the embodiments of this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, portable compact disc read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such transmitted data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to wireless, wired, etc., or any suitable combination thereof.

[0171] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. Each block in a flowchart or block diagram may represent a module, segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0172] The units described in the embodiments of this application can be implemented in software or hardware, and the described units can also be located in a processor. The names of these units do not necessarily limit the specific unit itself.

[0173] In another aspect, this application also provides a computer-readable storage medium, which may be included in the electronic device described in the above embodiments; or it may exist independently and not assembled into the electronic device. The aforementioned computer-readable storage medium carries computer-readable instructions that, when executed by a processor, implement the methods in any of the above embodiments.

[0174] According to one aspect of this application, an electronic device is also provided, comprising: a processor; and a memory storing computer-readable instructions that, when executed by the processor, implement the methods of any of the above embodiments.

[0175] According to one aspect of the embodiments of this application, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, causing the computer device to perform the methods of any of the above embodiments.

[0176] It should be noted that although several modules or units for the device used to perform actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to the embodiments of this application, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0177] Through the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, touch terminal, or network device, etc.) to execute the method according to the embodiments of this application.

[0178] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the embodiments disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein.

[0179] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A method for determining target users, characterized in that, include: Based on the feature information of the content to be pushed and the user profile features, a first probability corresponding to each user is predicted. The first probability is used to indicate the probability of the user clicking on the content to be pushed. The candidate user set is determined based on the first probability corresponding to each user. The target score for each candidate user is determined based on the first probability corresponding to the candidate user and the second probability corresponding to the candidate user in the designated page display position; the designated page display position is used to display the content to be pushed; the second probability is used to indicate the exposure probability of the pushed content relative to the candidate user in the designated page display position; The step of determining the target score for each candidate user based on the first probability corresponding to the candidate user and the second probability corresponding to the candidate user in a designated page display position includes: adjusting the first probability corresponding to the candidate user according to an adjustment factor to obtain the adjusted first probability corresponding to the candidate user, wherein the adjustment factor is a positive number not exceeding 1; weighting the second probability corresponding to the candidate user in all designated page display positions to obtain the weighted probability corresponding to the candidate user; calculating the target score corresponding to each candidate user based on the adjusted first probability corresponding to the candidate user and the weighted probability corresponding to the candidate user; or, using the product of the first probability corresponding to the candidate user and the second probability corresponding to the candidate user in a designated page display position as the target score corresponding to the candidate user. Based on the target scores corresponding to each candidate user, a target user is determined from the candidate user set, and the content to be pushed is pushed to the target user.

2. The method according to claim 1, characterized in that, The number of target users is a specified number; After calculating the target score for each candidate user based on the corrected first probability and the weighted probability corresponding to the candidate user, the method further includes: Obtain the first parameter value of the specified statistical parameter. The first parameter value is obtained by statistical analysis of the first probability corresponding to the specified number of candidate users with the highest first probability in the candidate user set. Obtain the second parameter value of the specified statistical parameter, which is obtained by statistical analysis of the target scores corresponding to the specified number of target users with the highest target scores in the candidate user set; Calculate the difference between the first parameter value and the second parameter value of the specified statistical parameter; If the difference is not lower than a preset threshold, the adjustment factor is reduced so that the target score corresponding to each candidate user is recalculated using the adjusted adjustment factor, and the target user is re-determined using the recalculated target score.

3. The method according to claim 2, characterized in that, The step of determining the target user from the candidate user set based on the target scores corresponding to each candidate user includes: If the difference is lower than the preset threshold, then the specified number of candidate users with the highest target scores in the candidate user set are determined as target users.

4. The method according to claim 1, characterized in that, The step of weighting the second probability corresponding to the candidate user in all specified page display positions to obtain the weighted probability corresponding to the candidate user includes: Get the number of times the content corresponding to each specified page display position is exposed; Based on the number of times the content corresponding to each designated page display position is exposed, a weighting coefficient corresponding to each designated page display position is determined, wherein the weighting coefficient is proportional to the number of times the content corresponding to the designated page display position is exposed. The second probability of the candidate user in all designated page display positions is weighted according to the weighting coefficient corresponding to each designated page display position, so as to obtain the weighted probability of the candidate user.

5. The method according to claim 1, characterized in that, Before determining the target score for each candidate user based on the first probability, the second probability, and the adjustment factor, the method further includes: Based on the content exposure information of the candidate user in the designated page display position for the pushed content, a second probability corresponding to the candidate user is calculated; wherein, the content exposure information is used to indicate whether the pushed content carried in the designated page display position is exposed to the client of the corresponding candidate user.

6. The method according to claim 5, characterized in that, The step of calculating the second probability corresponding to the candidate user based on the content exposure information of the candidate user in the designated page display position for the pushed content includes: The content exposure information of the candidate user in the designated page display position for the pushed content and the user profile features of the candidate user are input into the first neural network model; the first neural network model is trained based on the user profile features of the sample user and the content exposure information of the sample user for the historical pushed content; The first neural network model predicts the second probability of the candidate user by performing a score prediction based on the content exposure information of the candidate user in the designated page display position for the pushed content and the user profile characteristics of the candidate user.

7. The method according to claim 5, characterized in that, The step of calculating the second probability corresponding to the candidate user based on the content exposure information of the candidate user in the designated page display position for the pushed content includes: Based on the content exposure information of the candidate users on the designated page display position for the pushed content, the number of target unit statistical time periods is counted within a set time period; the target unit statistical time period refers to the unit statistical time period corresponding to the display of the pushed content on the designated page display position; The second probability corresponding to the candidate user is calculated based on the total number of statistical time periods per unit within the set time period and the number of statistical time periods per target unit.

8. The method according to claim 1, characterized in that, The step of predicting the first probability for each user based on the feature information of the content to be pushed and the user profile features includes: The feature information of the content to be pushed and the user profile features are input into the second neural network model; the second neural network model is trained based on the feature information of the pushed content, the user profile features of the sample user, and the click trigger information of the sample user for the pushed content; wherein, the click trigger information is used to indicate whether the sample user clicks on the displayed pushed content; The second neural network model performs score prediction based on the feature information of the content to be pushed and the user profile features to obtain the first probability corresponding to each user.

9. A device for determining a target user, characterized in that, include: The first probability determination module is used to predict the first probability corresponding to each user based on the feature information of the content to be pushed and the user profile features. The first probability is used to indicate the click probability of the user for the content to be pushed. A candidate user set determination module is used to determine a candidate user set based on the first probability corresponding to each user. The target score determination module is used to determine the target score for each candidate user based on the first probability corresponding to the candidate user and the second probability corresponding to the candidate user in the designated page display position; the designated page display position is used to display the content to be pushed. The second probability is used to indicate the probability of the pushed content being exposed relative to the candidate user in the specified page display position; The target score determination module includes: a correction unit, configured to correct the first probability corresponding to the candidate user according to an adjustment factor to obtain the corrected first probability corresponding to the candidate user, wherein the adjustment factor is a positive number not exceeding 1; a weighted probability determination unit, configured to weight the second probability corresponding to the candidate user in all designated page display positions to obtain the weighted probability corresponding to the candidate user; and a target score determination unit, configured to calculate the target score corresponding to each candidate user based on the corrected first probability corresponding to the candidate user and the weighted probability corresponding to the candidate user; or, the target score determination module is further configured to: use the product of the first probability corresponding to the candidate user and the second probability corresponding to the candidate user in the designated page display position as the target score corresponding to the candidate user; The target user determination module is used to determine the target user from the set of candidate users based on the target scores corresponding to each candidate user, so as to push the content to be pushed to the target user.

10. An electronic device, characterized in that, include: processor; A memory storing computer-readable instructions, which, when executed by the processor, implement the method for determining a target user as described in any one of claims 1-8.

11. A computer-readable storage medium storing computer-readable instructions thereon, characterized in that, When the computer-readable instructions are executed by the processor, the method for determining the target user as described in any one of claims 1-8 is implemented.

Images