A method, device, equipment and medium for determining a target population

Abstract

The embodiments of the present disclosure provide a method, device, equipment and medium for determining a target population, and relate to the field of computer technology. The method includes: generating multi-order population packages for multiple industries based on user characteristic data and advertisement characteristic data, each order population package in the multi-order population package contains a preset number of user populations; determining an initial order population package for the target industry corresponding to the advertisement to be delivered; obtaining delivery data of the advertisement to be delivered during the delivery process, and determining the order of the target order population package based on the delivery data; and expanding the delivery population of the advertisement to be delivered from the initial order population package to the target order population package based on the order of the target order population package. The embodiments of the present disclosure can deliver advertisements based on the industry multi-order population package during the cold start phase of the advertisement, thereby achieving intelligent promotion of advertisements while ensuring the user conversion effect.

Description

Background Art

[0002] This section is intended to provide a background or context to the embodiments of the disclosure that are recited in the claims. No description herein is admitted to be prior art by inclusion in this section.

[0003] With the development of network technology, information flow advertising has become an important part of the network field. Among them, information flow advertising refers to advertising in the dynamics of social media users' friends, or in the content flow of information media and audio-visual media. This kind of advertising interspersed in the content flow has a relatively good experience for users, and advertisers can use user tags for precise delivery. Therefore, especially after the advent of the mobile Internet era, it has ushered in explosive growth, and almost all Internet media have launched information flow advertising platforms.

[0004] Currently, before pushing advertising content to specific users through information flow advertising platforms, it takes a long time to determine the user group to which the advertisement will be delivered, and the user conversion effect cannot be guaranteed during the advertising delivery process.

[0005] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure, and therefore may include information that does not constitute the prior art known to ordinary technicians in the field. Summary of the invention

[0006] Based on the above problems, the inventors have conducted corresponding thinking and made targeted improvements, providing a method for determining the target population, a device for determining the target population, an electronic device, and a computer-readable storage medium. The method can generate multi-level population packages for multiple industries based on user feature data and advertisement feature data, wherein each level of the multi-level population package contains a preset number of user populations; determine the initial level population package of the target industry corresponding to the advertisement to be delivered; obtain the delivery data of the advertisement to be delivered during the delivery process, and determine the level of the target level population package based on the delivery data; and expand the delivery population of the advertisement to be delivered from the initial level population package to the target level population package based on the level of the target level population package. Advertisements can be delivered based on the industry multi-level population package during the cold start phase of the advertisement, ensuring the user conversion effect while realizing the intelligent promotion of the advertisement.

[0007] According to a first aspect of an embodiment of the present disclosure, a method for determining a target population is disclosed, comprising:

[0008] Generate multi-level population packages for multiple industries based on user feature data and advertisement feature data, wherein each level of the multi-level population package contains a preset number of user populations;

[0009] Determine the initial group of people in the target industry corresponding to the advertisement to be placed;

[0010] Acquire the delivery data of the advertisement to be delivered during the delivery process, and determine the order of the target group package according to the delivery data;

[0011] The population to be delivered the advertisement is expanded from the initial population package to the target population package according to the order of the target population package.

[0012] According to a second aspect of an embodiment of the present disclosure, a device for determining a target population is disclosed, comprising: a population package generation module, a population package matching module, and a population package upgrade module, wherein:

[0013] A crowd package generation module, used to generate multi-level crowd packages for multiple industries based on user feature data and advertisement feature data, wherein each level of the multi-level crowd package contains a preset number of user groups;

[0014] A crowd package matching module is used to determine the initial crowd package of the target industry corresponding to the advertisement to be placed;

[0015] A target order determination module is used to obtain the delivery data of the advertisement to be delivered during the delivery process, and determine the order of the target order population package according to the delivery data;

[0016] The crowd package upgrade module is used to expand the delivery crowd of the to-be-delivered advertisement from the initial-order crowd package to the target-order crowd package according to the order of the target-order crowd package.

[0017] According to a third aspect of an embodiment of the present disclosure, an electronic device is disclosed, including: a processor; and a memory, on which computer-readable instructions are stored, and when the computer-readable instructions are executed by the processor, the method for determining a target population as disclosed in the first aspect is implemented.

[0018] According to a fourth aspect of an embodiment of the present disclosure, a computer program medium is disclosed, on which computer-readable instructions are stored. When the computer-readable instructions are executed by a processor of a computer, the computer executes the method for determining a target population disclosed in the first aspect of the present disclosure.

[0019] The disclosed embodiment generates multi-level population packages for multiple industries according to user characteristic data and advertisement characteristic data, wherein each level population package in the multi-level population package contains a preset number of user populations; determines an initial level population package for the target industry corresponding to the advertisement to be delivered; obtains delivery data of the advertisement to be delivered during delivery, and determines the level of the target level population package according to the delivery data; and expands the delivery population of the advertisement to be delivered from the initial level population package to the target level population package according to the level of the target level population package. On the one hand, multiple industry population packages are generated according to user characteristic data and advertisement characteristic data, and by matching with the industry in which each advertisement to be delivered is located, an industry population package for each advertisement to be delivered can be quickly generated, thereby enhancing the timeliness of the population package; on the other hand, the target level of the industry population package corresponding to the advertisement to be delivered can be determined according to the delivery data of the advertisement to be delivered, and the population package is expanded according to the target level. The expansion of the population package indicates that the user conversion effect of the current level population package is good. Therefore, while ensuring the user conversion effect, the intelligent promotion of the advertisement is realized.

[0020] Other features and advantages of the present disclosure will become apparent from the following detailed description, or may be learned in part by the practice of the present disclosure.

[0021] It is to be understood that the foregoing general description and the following detailed description are exemplary only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The above and other objects, features and advantages of the exemplary embodiments of the present disclosure will become readily understood by reading the detailed description below with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present disclosure are shown in an exemplary and non-limiting manner, in which:

[0023] Figure 1 A schematic diagram showing a system architecture of a method and device for determining a target population to which the present disclosure embodiments can be applied;

[0024] Figure 2 Shown is a flow chart of a method for determining a target population according to an exemplary embodiment of the present disclosure;

[0025] Figure 3 Shown is a schematic diagram of a process for generating an industry population package according to an example implementation of the present disclosure;

[0026] Figure 4 Shown is a schematic diagram of a process for generating an industry population package according to another example implementation of the present disclosure;

[0027] Figure 5 FIG. 1 is a flow chart of determining a target group package according to an exemplary embodiment of the present disclosure;

[0028] Figure 6 FIG. 1 is a schematic diagram of a process for determining a target group package according to another exemplary embodiment of the present disclosure;

[0029] Figure 7 A schematic diagram showing the upgraded delivery of industry group packages according to an exemplary embodiment of the present disclosure is shown;

[0030] Figure 8 A schematic diagram of a process of upgrading the delivery of industry group packages according to another exemplary embodiment of the present disclosure is shown;

[0031] Fig. 9 A structural block diagram of a device for determining a target population according to an exemplary embodiment of the present disclosure is shown;

[0032] Fig.10 FIG. 1 is a schematic diagram of a computer system of an electronic device according to an exemplary embodiment of the present disclosure.

[0033] In the drawings, the same or corresponding reference numerals represent the same or corresponding parts. DETAILED DESCRIPTION

[0034] The principles and spirit of the present disclosure will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are provided only to enable those skilled in the art to better understand and implement the present disclosure, and are not intended to limit the scope of the present disclosure in any way. On the contrary, these embodiments are provided to make the present disclosure more thorough and complete, and to fully convey the scope of the present disclosure to those skilled in the art.

[0035] Those skilled in the art will appreciate that the embodiments of the present disclosure may be implemented as a device, apparatus, method or computer program product. Therefore, the present disclosure may be specifically implemented in the following forms, namely: complete hardware, complete software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0036] According to an embodiment of the present disclosure, a method for determining a target population, a device for determining a target population, an electronic device, and a computer-readable storage medium are proposed.

[0037] In this document, any number of elements in the drawings is used for illustration rather than limitation, and any naming is used only for distinction and does not have any limiting meaning.

[0038] The principle and spirit of the present disclosure are explained in detail below with reference to several representative embodiments of the present disclosure. SUMMARY OF THE INVENTION

[0040] In the process of delivering information flow advertisements, in order to ensure the delivery effect of advertisements, it is usually necessary to optimize the advertising delivery strategy. In the related technology, one method is to associate the delivery population with labels. For example, the DMP (Data Management Platform) platform labels each user by analyzing the user's historical behavior data and establishes a user label system. Advertisers can select label keywords at the delivery end to circle the user population through label keywords for advertising delivery. Another method is to use look-alike (ad targeting technology) to expand the delivery population. For example, advertisers upload high-quality seed population packages, and the DMP platform searches for the historical behavior data of seed users in the historical delivery data to construct the user features of seed users, and uses the user features of seed users as positive samples. Then, a part of non-seed users is negatively sampled from all users, and the user features of non-seed users are constructed based on the historical behavior data of non-seed users, and the user features of non-seed users are used as negative samples. After the sample processing is completed, the obtained positive and negative samples are used to train the regression model. After the training is completed, the regression model is used to predict the similarity between the candidate users and the seed users. Finally, the candidate users with high similarity to the seed users are regarded as similar groups, and advertisements are delivered to the group package consisting of seed users and similar groups.

[0041] However, the first method only considers user interests in the advertising process, but does not evaluate the advertising effect, so it cannot guarantee the advertising effect, such as user conversion effect. Moreover, the process of associating users with tags needs to be submitted to the background for execution, and it takes a certain amount of time to circle the crowd package, and the timeliness of the crowd package is poor. For the second method, this method must upload the seed population to obtain the similar population. From submitting the seed population package to determining the similar population, it also takes a certain amount of time to wait, which makes the timeliness of the crowd package poor. Moreover, among the users obtained by negative sampling, there may also be users with high similarity to the seed users, which is not conducive to the training of the regression model, reduces the accuracy of the similar population, and thus cannot guarantee the advertising effect.

[0042] Based on the above problems, the inventors of the present invention have come up with the idea of ​​generating multi-level population packages for multiple industries based on user feature data and advertisement feature data, wherein each level population package in the multi-level population package contains a preset number of user populations; determining the initial level population package of the target industry corresponding to the advertisement to be delivered; obtaining the delivery data of the advertisement to be delivered during the delivery process, and determining the level of the target level population package based on the delivery data; and expanding the delivery population of the advertisement to be delivered from the initial level population package to the target level population package based on the level of the target level population package. On the one hand, multiple industry population packages are generated based on user feature data and advertisement feature data, and by matching with the industry in which each advertisement to be delivered is located, an industry population package for each advertisement to be delivered can be quickly generated, thereby enhancing the timeliness of the population package; on the other hand, the target level of the industry population package corresponding to the advertisement to be delivered can be determined based on the delivery data of the advertisement to be delivered, and the population package can be expanded based on the target level. The expansion of the population package indicates that the user conversion effect of the current level population package is good. Therefore, while ensuring the user conversion effect, the intelligent promotion of the advertisement is realized.

[0043] After introducing the basic principles of the present disclosure, various non-limiting embodiments of the present disclosure are described in detail below.

[0044] Application Scenario Overview

[0045] It should be noted that the following application scenarios are only shown to facilitate understanding of the spirit and principle of the present disclosure, and the embodiments of the present disclosure are not limited in this respect. On the contrary, the embodiments of the present disclosure can be applied to any applicable scenario.

[0046] The embodiments of the present disclosure can automatically expand the target population for advertising during the cold start phase of advertising, thereby realizing intelligent promotion of advertising. For example, when an e-commerce platform delivers product advertisements to a user population, it can generate multi-order population packages for multiple industries based on user feature data and advertising feature data. Among them, each order population package in the multi-order population package contains a preset number of user populations. Determine the initial order population package of the target industry corresponding to the advertisement to be delivered, obtain the delivery data of the advertisement to be delivered during the delivery process, and calculate the order of the target order population package based on the real-time delivery data, so as to expand from the initial order population package to the target order population package according to the order of the target order population package, while ensuring the advertising delivery effect, making the delivery process more intelligent. It can be understood that the method for determining the target population in the present disclosure is applicable to but not limited to the advertising cold start delivery scenario, and can also be scenarios such as product recommendation and navigation route recommendation for multiple users, which are not limited by the present disclosure.

[0047] Exemplary Methods

[0048] The exemplary embodiment of the present disclosure first provides a method for determining a target population. Figure 1 The system architecture diagram of the operating environment of this method is shown. Figure 1 As shown, the system architecture 100 may include one or more of terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc. The terminal devices 101, 102, 103 may be various electronic devices, including but not limited to desktop computers, portable computers, smart phones, and tablet computers, etc. It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. According to the implementation requirements, there may be any number of terminal devices, networks and servers. For example, the server 105 may be a server cluster composed of multiple servers.

[0049] The target population determination method provided in the exemplary embodiment of the present disclosure is generally executed by the server 105, and accordingly, the target population determination device is generally set in the server 105. For example, the terminal device 101 can be a computer used by an operator, and the server 105 can send the current target population to the terminal device 101 after executing the target population determination method to determine the current target population for delivery, so as to display it to the operator. However, it is easy for a person skilled in the art to understand that the target population determination method provided in the exemplary embodiment of the present disclosure can also be executed by one or more of the terminal devices 101, 102, and 103, and accordingly, the target population determination device can also be set in the terminal devices 101, 102, and 103. For example, after the terminal device 101 executes the target population determination method and determines the current target population for delivery, the current target population for delivery can be directly displayed on the display screen of the terminal device 101 to display it to the operator, and this is not particularly limited in this exemplary embodiment.

[0050] The following is an explanation of how to determine the target population.

[0051] Figure 2 FIG. 2 is a flow chart of a method for determining a target population according to an exemplary embodiment of the present disclosure. Figure 2 As shown, the method for determining the target population may include:

[0052] Step S210: generating multi-level population packages of multiple industries according to user characteristic data and advertisement characteristic data, wherein each level population package in the multi-level population package includes a preset number of user populations;

[0053] Step S220: determining the initial group of people in the target industry corresponding to the advertisement to be placed;

[0054] Step S230: obtaining the delivery data of the advertisement to be delivered during the delivery process, and determining the order of the target class group package according to the delivery data;

[0055] Step S240: expanding the population to be delivered the advertisement from the initial population package to the target population package according to the order of the target population package.

[0056] Implementation Figure 2 The method for determining the target population shown, on the one hand, generates multiple industry population packages based on user characteristic data and advertising characteristic data, and by matching them with the industry in which each advertisement to be placed is located, the industry population package for each advertisement to be placed can be quickly generated, thereby enhancing the timeliness of the population package; on the other hand, the target order of the industry population package corresponding to the advertisement to be placed can be determined based on the placement data of the advertisement to be placed, and the population package can be expanded according to the target order. The expansion of the population package indicates that the user conversion effect of the current order population package is better. Therefore, while ensuring the user conversion effect, the intelligent promotion of the advertisement is achieved.

[0057] These steps are described in detail below.

[0058] In step S210, multi-level population packages of multiple industries are generated according to user characteristic data and advertisement characteristic data, and each level population package in the multi-level population package includes a preset number of user populations.

[0059] In the exemplary implementation of the present disclosure, user feature data may include user portrait data, user behavior data, user context data, etc. Among them, user portrait data is also called user basic attribute data, such as user age, gender, education level and preferences. User behavior data may be user exposure data, user click data and user conversion data generated during the advertising process. It should be noted that user exposure data is data generated when an advertisement is displayed to a user, such as the number of times an advertisement is displayed to a user; user click data is data generated when a user clicks on an advertisement, such as the number of times a user clicks on an advertisement; user conversion data is data generated when a user purchases, downloads, registers or pays through an advertisement. User context data may be data related to user behavior data, such as the time and place when a user clicks on an advertisement. For example, the user feature data of a certain user may be "Zhang San, 28 years old, clicked on an advertisement 3 times, displayed an advertisement 10 times, downloaded an advertisement 1 time". Advertising feature data is also advertising side data, which may include data such as the advertising delivery channel, delivery time, delivery area, industry, and advertising effect of the advertisement. Among them, the delivery channel refers to the specific advertising delivery platform; the delivery time refers to the specific time of the advertisement delivery, whether it is a holiday, etc.; the delivery area refers to the country and city covered by the advertisement; the industry refers to the industry in which the advertisement is located, such as the clothing industry, food industry and software industry, etc.; the advertising effect refers to the advertising exposure, click, conversion and other data on the platform side. It is understandable that user behavior data can also include user portrait data, and basic information of users, such as user age, gender, etc., can also be obtained when obtaining user behavior data. In addition, the user's exposure, click, and conversion data of advertisements can include both user feature data and advertising feature data, and this disclosure does not make specific limitations on this.

[0060] In one embodiment, during the cold start phase of information flow advertising, an oCPX delivery strategy may be used for advertising delivery. The o in the term oCPX stands for optimized, and CPX stands for Cost Per X, indicating a mode of settlement according to different methods. For example, it may be CPC (Cost Per Click), CPM (Cost Per Mille), or CPA (Cost Per Action), where the conversion effect refers to download and installation, new registered users, receiving coupons, adding to shopping carts, placing orders, and other behaviors. Therefore, the oCPX delivery strategy refers to various CPXs that can optimize conversion effects. The oCPX delivery strategy may include two stages. The first stage is the data accumulation period, during which the advertising delivery platform can expose the crowd along the trajectory of user clicks to form conversions; the second stage is the intelligent delivery period, during which the delivery of advertisements can be adjusted according to the user conversion rate.

[0061] In the example implementation of the present disclosure, before advertising is placed, an industry crowd package for each industry can be generated. Among them, the crowd package refers to the classification of user populations, and the industry crowd package refers to the classification of user populations according to industry categories, such as software industry crowd packages, food industry crowd packages, etc. The industry crowd package is a multi-order crowd package, that is, each industry crowd package can be divided into sub-industry crowd packages of multiple stages, and each stage of the sub-industry crowd package (i.e., each order crowd package) includes user populations with similar or the same degree of interest in placing advertisements in the industry. Correspondingly, each order crowd package in the multi-order crowd package can contain a preset number of user populations, and the number of users in each order crowd package increases successively. For example, when the multi-order crowd package is divided into a first-order crowd package, a second-order crowd package, a third-order crowd package, and all crowd packages, the number of user populations in the second-order crowd package is greater than the number of user populations in the first-order crowd package, and it contains the first-order crowd package, the number of user populations in the third-order crowd package is greater than the user population data of the second-order crowd package, and it contains the second-order crowd package, and so on. When advertisements related to a user's industry are served based on industry population packages, there is a high probability that user conversion behavior will occur, thus improving the cold start effect of the advertisement.

[0062] For example, Figure 3 FIG. 1 is a flow chart of generating a multi-stage crowd package of an industry according to an exemplary embodiment of the present disclosure, as shown in FIG. Figure 3 As shown, it may include step S310 and step S320:

[0063] In step S310, a deep learning model is constructed based on the user feature data and the advertisement feature data.

[0064] Exemplarily, user characteristic data and advertisement characteristic data within a preset time period can be obtained, such as user characteristic data and advertisement characteristic data within the past six months. Among them, user characteristic data include basic attribute data such as user age, gender, education level and preference, user exposure, click and conversion and other behavioral data, and context data such as the time and place when the user clicks on the advertisement. Advertisement characteristic data include advertisement delivery channels, delivery time, advertisement corresponding exposure, click and conversion and other advertisement side data. Further, the user characteristic data and advertisement characteristic data obtained within the past six months can be used to form a CTCVR (post-view Click-Through & ConVersion Rate, user click-through rate multiplied by user click conversion rate) sample set, and the sample set can be used to build a deep learning model. Among them, the positive samples in the CTCVR sample set can be the user characteristic data and advertisement characteristic data corresponding to the advertisements that the user clicks and eventually forms a conversion in the exposed advertisements, and the negative samples can be the user characteristic data and advertisement characteristic data corresponding to the advertisements that the user clicks and eventually forms a conversion in the exposed advertisements. The constructed deep learning model can be a DSSM (Deep Structured Semantic Model) model, or a CNN (Convolutional Neural Network)-DSSM, LSTM (Long Short Term Memory)-DSSM and other models, and the present disclosure does not make specific limitations on this.

[0065] The DSSM model can be used as an example for illustration. The DSSM model can include two interconnected neural networks, and the two neural networks can share weights. Correspondingly, the two inputs of the DSSM model are the user feature data and the advertising side feature data in the CTCVR sample set. The two neural networks can map the input to a new space to form a representation of the input in the new space. For example, by inputting the user feature data into the neural network, a user embedding (mapping) vector of a fixed dimension can be obtained. Similarly, by inputting the advertising side feature data into the neural network, an advertising embedding vector of the same dimension can be obtained. By calculating the similarity between the user embedding vector and the advertising embedding vector, the output of the DSSM model, that is, the predicted value of the advertising click-through conversion rate, can be obtained. The predicted value can be used to characterize the user's interest in the advertisement. For example, when the predicted advertising click-through conversion rate is high, it indicates that the user has a high interest in the advertisement, and the similarity between the user embedding vector and the advertising embedding vector is also high. After the model is built, the loss value can be calculated based on the label value and the predicted value of each set of inputs using the loss function, and the model parameters of the DSSM model can be updated to continuously improve the prediction performance of the DSSM model. The training of the DSSM model is completed when the loss function converges or a certain number of iterations are met.

[0066] In the example implementation of the present disclosure, the deep learning model can not only obtain the vector expression of users and the vector expression of advertisements, but also mine the commonalities of users and the association between users and advertisements, so as to facilitate accurate advertising delivery to similar groups of people.

[0067] Step S320: Generate a multi-order population package for each of the multiple industries through the deep learning model.

[0068] After training the deep learning model using the CTCVR sample set, the user embedding vector corresponding to the user feature data in the CTCVR sample set and the advertising embedding vector corresponding to the advertising feature data in the CTCVR sample set can be obtained. It is also possible to obtain user feature data and advertising feature data for any historical time period other than the CTCVR sample set, and obtain the corresponding user embedding vector and advertising embedding vector through the deep learning model, which is not specifically limited in this disclosure. It should be noted that the obtained user embedding vector and advertising embedding vector can be stored in an online redis memory database for subsequent calls.

[0069] In one example, reference Figure 4As shown, a schematic diagram of a process of generating a multi-stage crowd package of an industry through a deep learning model is shown, and the process may include steps S410 to S430:

[0070] Step S410: Inputting user feature data of multiple users and advertisement feature data of multiple advertisements in each industry into the deep learning model to obtain multiple user vectors and multiple advertisement vectors in each industry.

[0071] Exemplarily, user feature data of multiple users and advertising feature data of all advertisements in each industry related to each user may be obtained. For example, user feature data of user A and advertising feature data of advertisement B related to user A may be simultaneously input into the DSSM model to obtain a user vector of user A and an advertising vector of advertisement B. Among them, the industry in which advertisement B is located is industry C. Similarly, advertising vectors of all advertisements related to user A in industry C may be obtained in sequence, and further, user vectors of multiple users and advertising vectors of multiple advertisements in each industry may be obtained.

[0072] Step S420: Aggregate multiple advertisement vectors in each industry to obtain an industry vector for each industry.

[0073] For a certain industry, there may be multiple advertisements. For example, after obtaining the advertisement vectors of all advertisements in industry C, the industry vector of industry C may be obtained by aggregating the advertisement vectors of all advertisements. For example, all advertisement vectors in industry C may be weighted summed to generate a vector expression of industry C.

[0074] Specifically, it can be based on:

[0075] I j =∑w1A1+w2A2+w3A3+… (1)

[0076] Among them, I j represents the industry vector of the jth industry. Different values ​​of j correspond to different industries. A1, A2, A3, etc. represent the advertising vectors of each advertisement in the jth industry. a, b, c, etc. represent the weights corresponding to each advertisement in the jth industry. The weights can be set according to factors such as the historical exposure, consumption, and delivery effect of the advertisement. For example, when the historical exposure of advertisement A1 is large, the consumption of advertising delivery is small, and the delivery effect is good, the value of weight a corresponding to advertisement A1 is large.

[0077] Step S430: Generate a multi-order population package for each industry according to the industry vector of each industry and the multiple user vectors.

[0078] It is understandable that each level of population package in each industry can be composed of user groups with similar interests in the industry. Furthermore, the user's interest in the industry can be measured by the industry vector of each industry and the user vector of each user. Exemplarily, the similarity between each industry vector and the multiple user vectors can be calculated; and the multi-level population package of each industry is generated by multiple users whose similarity sizes meet a preset similarity threshold.

[0079] For example, when calculating the similarity between the industry vector of industry C and multiple user vectors, multiple similarity thresholds can be set, such as setting three similarity thresholds, and the threshold values ​​decrease in sequence, that is, the first similarity threshold> the second similarity threshold> the third similarity threshold. The first-order population package of industry C can be composed of users whose similarity is greater than the first similarity threshold, the second-order population package of industry C can be composed of users whose similarity is greater than the second similarity threshold, and the third-order population package of industry C can be composed of users whose similarity is greater than the third similarity threshold. All users are the entire population package. It can be seen that the smaller the similarity threshold, the more similar user populations there are, and the more users make up the population package. Correspondingly, the number of users in the second-order population package of industry C is greater than the number of users in the first-order population package, and the number of users in the third-order population package is greater than the number of users in the second-order population package. Similarly, the generated multi-order population packages of each industry can also be stored in an online redis memory database for subsequent calls and updates. For example, the industry population package of each industry can be stored in the form of a per-order population vocabulary, and the advertisement to be placed and the industry population package corresponding to the advertisement can be matched through the vocabulary inverted index, so that the industry population package of each advertisement to be placed can be quickly generated, thereby enhancing the timeliness of the population package.

[0080] In addition, after the industry population package is generated, the delivery effect of the industry population package can be evaluated offline. For example, a preset number of user populations can be obtained, and the user population belonging to a certain industry population package can be obtained by filtering, and advertisements can be delivered to the user population of the industry population package for experimentation. The same advertisements are delivered to all unfiltered user populations for comparison to evaluate the delivery effect of the industry population package. Among them, the evaluation indicators can be exposure, click volume, consumption, conversion volume, user coverage, click rate, conversion rate and cost, etc., which are not limited in this disclosure.

[0081] In one implementation, when obtaining user behavior data from user feature data, a real-time streaming data channel can be established to collect user exposure, click and conversion data generated during the advertising process. For example, real-time user behavior data can be processed through a kafka (streaming processing platform) channel. The process of calculating the similarity between users and industries and generating industry population packages is run offline in the background. By continuously obtaining the latest user behavior data, the industry population package can be updated, such as replacing the old population package with the generated new population package. The update of the industry population package is completely transparent to advertisers, and the industry population package can be obtained without waiting. At the same time, it can be guaranteed that the generated industry population package contains the latest active users, thereby improving user coverage.

[0082] In step S220, an initial group of people in the target industry corresponding to the advertisement to be placed is determined.

[0083] After generating multi-level crowd packages for multiple industries, for a certain advertisement to be placed, the corresponding industry crowd package can be matched for the advertisement to be placed during the cold start phase of the advertisement. Exemplarily, the target industry corresponding to the advertisement to be placed, that is, the industry in which the advertisement to be placed is located, can be determined first, so as to match the corresponding industry crowd package for the advertisement to be placed in the redis memory database according to the industry in which the advertisement to be placed is located. In order to ensure the effect of advertising placement, the initial-level crowd package of the corresponding industry can be matched for the advertisement to be placed, such as the first-level crowd package with high crowd similarity. Placing advertisements to the first-level crowd package can improve the user conversion effect.

[0084] For example, when the advertisement to be delivered is a promotional advertisement for a multimedia application, the advertisement belongs to the multimedia software industry. At this time, the first-order population package of the multimedia software industry can be used as the initial delivery population of the advertisement in the cold start phase. It can be seen that when the industry population package is matched for delivery at the initial stage of advertisement delivery, the users in the industry population package have a high probability of converting to the delivered advertisement, so the cold start effect of the advertisement can be guaranteed.

[0085] In step S230, the delivery data of the advertisement to be delivered during the delivery process is obtained, and the order of the target class group package is determined according to the delivery data.

[0086] In the early stage of advertising, the advertisement can be delivered to the initial group package (such as the first-level group package) of the industry in which the advertisement is located, so as to accurately deliver the advertisement to the user group that is interested in it, thereby ensuring the user conversion effect. Figure 5 As shown, the order of the target-order population package can be determined according to step S510 and step S520 to expand from the first-order population package to the target-order population package, while ensuring the user conversion effect, and realizing intelligent promotion of advertising.

[0087] Step S510: Counting the real-time delivery data of the advertisement to be delivered during the delivery process, wherein the real-time delivery data at least includes the current average exposure frequency per person, the number of people exposed, the current conversion volume and the current delivery duration.

[0088] During the advertising delivery process, real-time delivery data such as the average exposure frequency per person of the advertisement to be delivered, the number of exposed people, the current conversion volume, and the current delivery duration can be counted to evaluate the user conversion effect of the first-order population package based on the real-time delivery data, and then determine whether it is necessary to upgrade the delivery based on the first-order population package based on the evaluation results.

[0089] In one example, when evaluating the user conversion effect of a first-order population package based on real-time delivery data, the real-time delivery data of the first-order population package can be compared with the target delivery data of the conversion in the industry to determine the user conversion effect of the first-order population package. For example, the historical data of each order population package in the target industry can be obtained first, and the target exposure population number can be obtained by fitting the historical data, and the target delivery data corresponding to the conversion of the target industry can be calculated to determine the order of the target order population package according to the target exposure population number, target delivery data, the current per capita exposure frequency, the number of exposed people, the current conversion amount and the current delivery duration. For example, when the number of exposed people in the first-order population package meets the target exposure population number, the first-order population package can be upgraded to the target order population package, or when the real-time delivery data of the first-order population package other than the number of exposed people meets the target delivery data corresponding to the conversion of the target industry, the first-order population package can also be upgraded to the target order population package. Among them, the target delivery data can include the target delivery duration, the target per capita exposure frequency and the target conversion amount.

[0090] For example, the historical delivery data of each group package in each industry can be obtained. By statistically analyzing the historical delivery data, the following can be obtained:

[0091] 1) The minimum delivery time required for a conversion to occur in each industry, that is, the target delivery time for conversion to occur, denoted as p_time.

[0092] 2) The number of target exposed people for each order of population package: For example, the number of exposed people and the attenuation of each order of population package in a historical period (such as the past 3 days) can be counted, and the population coverage rate p_cover of each order of population package on that day can be obtained by fitting. i , where i is the order of the crowd package. Further, the crowd coverage rate p_cover can be obtained according to the prediction i And the number of users in each group uv i Calculate the target exposure population of each level of population package, which is p_cover i *uv i .

[0093] 3) The target conversion amount for each group of people to be converted once, recorded as p_convert i :

[0094]

[0095] Among them, i is the order of the crowd package, α is the target conversion volume set by the oCPX delivery strategy when entering the second stage, and convert i represents the conversion volume of the i-th order group package, and ∑convert represents the total conversion volume of all order group packages. Formula (2) can be used to characterize the conversion volume of the current order group package during the advertising delivery process to reach the target conversion volume p_convert i You can enter the target level population package, such as the next level population package.

[0096] 4) The target average exposure frequency per converted person in each industry, recorded as avg:

[0097] avg=pv / uv (3)

[0098] Among them, pv is the exposure of the converted population, and uv is the number of users in the converted population. Formula (3) can be used to represent that when the average exposure frequency of the current-order population package reaches the target average exposure frequency avg during the advertising delivery process, it can enter the target-order population package, such as the next-order population package.

[0099] The target delivery duration, target average exposure frequency per person, target conversion volume, and target number of exposed people can be used as hyperparameters to control the automatic upgrade of industry population packages during the actual delivery process. Among them, determining the target delivery duration, target average exposure frequency per person, target conversion volume, and target number of exposed people can be achieved through Spark offline tasks. Spark offline tasks refer to the user having stored the data to be processed on object storage or HDFS (Hadoop Distributed File System). For example, the data to be processed can be a CTCVR sample set, which was generated in the past, not in real time. After the user submits the task, the cluster can automatically complete the calculation and obtain the results. Obtaining the hyperparameters required for advertising promotion through offline tasks can improve the efficiency of intelligent advertising promotion.

[0100] Step S520: Determine the order of the target-order population package according to the current average exposure frequency per person, the number of exposed people, the current conversion volume and the current delivery duration.

[0101] The order of the target group package can be determined based on the current average exposure frequency, the number of exposed people, the current conversion volume, the current delivery time and other real-time delivery data and the number of target exposed people and the target delivery data of conversions in the industry. Figure 6 As shown, the order of the target-order population package can also be determined according to steps S610 to S640 to expand from the first-order population package to the target-order population package, thereby ensuring user conversion effects and realizing intelligent promotion of advertisements.

[0102] In step S610, the first order of the target population package is calculated according to the current average exposure frequency per person and the target average exposure frequency per person, the current delivery duration and the target delivery duration.

[0103] The first order of the population package can be calculated based on the current average exposure frequency and the target average exposure frequency, the current delivery duration and the target delivery duration, such as referring to the following formula:

[0104]

[0105] Among them, i is the order of the industry population package, T is the time period, cur_uv / cur_pv is the current average exposure frequency per capita of the i-th order population package, avg is the target average exposure frequency per capita for conversion; cur_time is the current delivery duration of the i-th order population package, and p_time is the target delivery duration for conversion. It can be seen from formula (4) that when the average exposure frequency per capita of the i-th order population package in the T-1 time period is greater than or equal to the target average exposure frequency per capita for conversion, and the delivery duration of the i-th order population package in the T-1 time period is greater than the target delivery duration for conversion, the advertising delivery population can be upgraded from the i-th order population package to the i+1-th order population package, that is, the advertisement can be delivered to the i+1-th order population package in the T time period. Otherwise, the advertisement will continue to be delivered to the i-th order population package in the T time period to improve the user conversion effect of the i-th order population package.

[0106] In step S620, the second order of the target population package is calculated according to the number of exposed populations and the number of target exposed populations.

[0107] The second order of the population package can be calculated based on the number of exposed populations and the number of target exposed populations, such as referring to the following formula:

[0108]

[0109] Among them, i is the order of the industry population package, T is the time period, cur_uv is the number of exposed people in the i-th order population package, p_cover k *uv k is the number of target exposure groups that have converted in the k-th group package, is the number of target exposed populations that have been converted in all the order population packages that have been delivered so far. From formula (5), it can be seen that when the number of exposed populations of the i-th order population package in the T-1 time period is greater than the number of target exposed populations that have been converted, the advertising population can be upgraded from the i-th order population package to the i+1-th order population package, that is, the advertisement can be delivered to the i+1-th order population package in the T time period. Otherwise, the advertisement will continue to be delivered to the i-th order population package in the T time period to improve the user conversion effect of the i-th order population package.

[0110] In step S630, the third order of the target population package is calculated according to the current conversion amount and the target conversion amount.

[0111] The third order of the population package can be calculated based on the current conversion volume and the target conversion volume, such as referring to the following formula:

[0112]

[0113] Among them, i is the order of the industry population package, T is the time period, cur_convert is the current conversion volume of the i-th order population package, p_convert k is the target conversion volume of the k-th group of people, is the target conversion volume of all the order population packages that have been delivered so far. From formula (6), it can be seen that when the current conversion volume of the i-th order population package in the T-1 time period is greater than the target conversion volume of all the order population packages that have been delivered, the advertising delivery population can be upgraded from the i-th order population package to the i+1-th order population package, that is, the advertisement can be delivered to the i+1-th order population package in the T time period. Otherwise, the advertisement will continue to be delivered to the i-th order population package in the T time period to improve the user conversion effect of the i-th order population package.

[0114] It is understandable that the crowd package order calculated in step S610 according to the current per capita exposure frequency and the target per capita exposure frequency, the current delivery time and the target delivery time can also be used as the second order of the target crowd package. or the third order Similarly, the crowd package order calculated in step S620 based on the number of exposed people and the number of target exposed people can also be used as the first order of the target crowd package. or the third order Also, the group package order calculated according to the current conversion amount and the target conversion amount in step S630 may be used as the first order of the target group package. or second order This disclosure does not make any specific limitation to this.

[0115] In step S640, the final order of the target group package is determined according to the first order, the second order and the third order of the target group package.

[0116] Formulas (4), (5) and (6) used to calculate the order of the target population package can be used as the three order-up conditions of the industry population package during the advertising delivery process, and then the final order of the target population package can be determined according to the calculated first order, second order and third order of the target population package. For example, it can be based on:

[0117]

[0118] Among them, i T is the final order of the target order population package, and are the first order, second order and third order of the target group package respectively. For example, it means taking the smallest order between the first order and the second order of the target group package. Formula (7) means that when at least two of the three upgrading conditions are met at the same time, the current group package can be upgraded, otherwise the advertisement is still delivered to the current group package.

[0119] In step S240, the population to be delivered the advertisement is expanded from the initial population package to the target population package according to the order of the target population package.

[0120] After determining the final order of the target group package, the group to be placed on the advertisement can be expanded from the initial group package to the target group package. For example, the initial group package for advertisement placement is the first group package, i.e. T-1 =1. According to formula (4), when the average exposure frequency of the first-order group package in the current time period (i.e., time period T-1) is greater than or equal to the target average exposure frequency that can be converted, and the delivery duration of the first-order group package in the current time period is greater than the target delivery duration that can be converted, the advertisement can be delivered to the second-order group package in the next time period (i.e., time period T), and the corresponding otherwise According to formula (5), when the number of exposed people in the first-order group package in the current time period is greater than the number of target exposed people that can be converted, advertisements can be placed in the second-order group package in the next time period. otherwise According to formula (6), when the current conversion volume of the first-order group package in the current time period is greater than the target conversion volume, advertisements can be delivered to the second-order group package in the next time period. otherwise

[0121] For example, when and When , it means that the current delivery data has not reached the upgrade condition. According to formula (7), we can get i T =1, the next time period still needs to deliver ads to the first-order group. Moreover, if the upgrade conditions are not met, it also means that the current delivery effect is poor, such as low user conversion volume. When the current delivery data meets one of the three upgrade conditions, such as and According to formula (7), we can get i T =1, the next time period still needs to deliver advertisements to the first-order group. When the current delivery data meets at least two of the three upgrade conditions, such as and or and According to formula (7), we can get i T =2. by i T-1 =1,i T =2, we can see that advertisements can be automatically delivered to the second-order group in the next time period.

[0122] The order of the industry population package is calculated based on real-time delivery data such as the current average exposure frequency, the number of exposed people, the current conversion volume, and the current delivery duration. The industry population package can be automatically upgraded without the need for advertisers to operate, ensuring the user conversion effect while also realizing intelligent promotion of advertisements. Among them, the delivery data can be counted in real time through Flink (distributed big data processing engine). Flink data processing has the advantages of low latency, high throughput, and accurate results. In addition, the order of the target population package can be quickly determined based on the real-time delivery data to deliver advertisements to the target population package, further improving the efficiency of intelligent advertising promotion.

[0123] Exemplary, reference Figure 7 As shown, a schematic diagram of the upgraded delivery of industry population packages is given. Figure 7 In the figure, the industry multi-level population packages include the first-level population ①, the second-level population ②, the third-level population ③ and the entire population ④. It can be seen that the number of each-level population packages increases in sequence. When the cost of advertising, conversion rate decay, click-through rate decay, population coverage, average exposure frequency and delivery time meet the conditions for industry conversion, the industry population package can be automatically upgraded to ensure the user conversion effect while realizing intelligent advertising promotion.

[0124] In one example implementation, Figure 8 FIG. 2 is a flow chart showing the process of upgrading the delivery of industry group packages in another exemplary embodiment of the present disclosure. Figure 8As shown, the method may include two parts: model building and data processing, and the data processing part may include an offline process and a real-time process. The offline process is used to generate the parameters required for upgrading the industry population package, and the real-time process is used to determine the target population package after the industry population package is upgraded.

[0125] Step S801: Build a DSSM model. User feature data and advertisement feature data of a historical time period may be obtained, and a CTCVR sample set is formed by the obtained user feature data and advertisement feature data, and the DSSM model is built using the sample set;

[0126] Step S802. Generate user vectors and industry vectors. Use the CTCVR sample set to train the DSSM model. After the training is completed, the user embedding vector corresponding to the user feature data in the CTCVR sample set and the advertisement embedding vector corresponding to the advertisement feature data are obtained. Further, by aggregating the advertisement embedding vectors in each industry, the industry vector of each industry is obtained.

[0127] Step S803. Generate a vocabulary for each industry and each level of population. Generate a vocabulary for each industry and each level of population by calculating the similarity between each industry vector and multiple user vectors, so as to match the advertisement to be delivered with the industry population package corresponding to the advertisement by means of the vocabulary inverted index. After generating the vocabulary for each industry and each level of population, step S809 can be executed;

[0128] Step S804. Offline evaluation of the crowd package effect. The crowd package flow can be filtered as an experiment, and the non-filtering flow can be used as a baseline to evaluate the delivery effect of the industry crowd package;

[0129] Step S805. Count the hyperparameters required for upgrading the crowd package. Use Spark offline tasks to count the hyperparameters required for upgrading the crowd package, including the target duration of advertising, target average exposure frequency per person, target conversion volume, and target number of exposed people;

[0130] Step S806. Offline storage of the hyper parameters required for upgrading the crowd package;

[0131] Step S807. Count the real-time delivery data. Use Flink to count the real-time delivery data in real time, including the current average exposure frequency per person, the number of people exposed, the current conversion volume, and the current delivery duration;

[0132] Step S808: Determine whether the crowd package is upgraded. If the crowd package meets the upgrade conditions, execute step S809;

[0133] Step S809. Store the multi-level crowd packages of industries online. The generated multi-level crowd packages of each industry can be stored in the online redis memory database for subsequent calls and updates. Also, when the target crowd package corresponding to the crowd package after upgrading is determined, the target crowd package can be obtained from the redis memory database to deliver advertisements to the target crowd package;

[0134] Step S810: Place advertisements to the target group package. After the group package is upgraded, the placement of the target group package can be monitored to further determine whether the target group package needs to be upgraded.

[0135] In the implementation of the embodiments of the present disclosure, multi-level population packages of multiple industries are generated according to user characteristic data and advertisement characteristic data, and each level population package in the multi-level population package contains a preset number of user populations; the initial level population package of the target industry corresponding to the advertisement to be delivered is determined; the delivery data of the advertisement to be delivered during the delivery process is obtained, and the level of the target level population package is determined according to the delivery data; and the delivery population of the advertisement to be delivered is expanded from the initial level population package to the target level population package according to the level of the target level population package. On the one hand, multiple industry population packages are generated according to user characteristic data and advertisement characteristic data, and by matching with the industry in which each advertisement to be delivered is located, an industry population package for each advertisement to be delivered can be quickly generated, thereby enhancing the timeliness of the population package; on the other hand, the target level of the industry population package corresponding to the advertisement to be delivered can be determined according to the delivery data of the advertisement to be delivered, and the population package is expanded according to the target level. The expansion of the population package indicates that the user conversion effect of the current level population package is good. Therefore, while ensuring the user conversion effect, the intelligent promotion of the advertisement is realized.

[0136] Exemplary Devices

[0137] refer to Fig. 9 A device for determining a target population according to an exemplary embodiment of the present disclosure will be described.

[0138] Fig. 9 FIG. 2 is a block diagram of a target population determination device according to an exemplary embodiment of the present disclosure. Fig. 9 As shown, a target population determination device 900 according to an exemplary embodiment of the present disclosure includes: a population package generation module 910, a population package matching module 920, a target level determination module 930, and a population package level enhancement module 940, wherein:

[0139] A crowd package generation module 910 is used to generate multi-level crowd packages for multiple industries based on user feature data and advertisement feature data, wherein each level of the multi-level crowd package contains a preset number of user groups;

[0140] The crowd package matching module 920 is used to determine the initial crowd package of the target industry corresponding to the advertisement to be placed;

[0141] The target order determination module 930 is used to obtain the delivery data of the advertisement to be delivered during the delivery process, and determine the order of the target order group package according to the delivery data;

[0142] The crowd package upgrade module 940 is used to expand the population to be delivered to the advertisement from the initial-stage crowd package to the target-stage crowd package according to the order of the target-stage crowd package.

[0143] In an optional implementation, the crowd package generation module 910 includes:

[0144] A model building submodule, used to build a deep learning model based on the user feature data and the advertisement feature data;

[0145] A crowd package generation submodule is used to generate a multi-order crowd package for each of the multiple industries through the deep learning model.

[0146] In an optional implementation, the crowd package generation submodule includes:

[0147] A first vector generating unit, configured to input user feature data of a plurality of users and advertisement feature data of a plurality of advertisements in each industry into the deep learning model to obtain a plurality of user vectors and a plurality of advertisement vectors in each industry;

[0148] A second vector generating unit, configured to aggregate the multiple advertisement vectors in each industry to obtain an industry vector for each industry;

[0149] A crowd package generating unit is used to generate a multi-order crowd package for each industry according to the industry vector of each industry and the multiple user vectors.

[0150] In an optional implementation, the crowd package generating unit includes:

[0151] A vector operation subunit, used for calculating the similarity between each industry vector and the plurality of user vectors;

[0152] The crowd package generating subunit is used to generate a multi-order crowd package for each industry from a plurality of users whose similarity levels meet a preset similarity threshold.

[0153] In an optional implementation, the crowd package upgrading module 940 includes:

[0154] A real-time delivery data acquisition submodule is used to collect statistics on the real-time delivery data of the advertisement to be delivered during the delivery process, wherein the real-time delivery data at least includes the current average exposure frequency per person, the number of people exposed, the current conversion volume and the current delivery duration;

[0155] The target-order population package determination submodule is used to determine the order of the target-order population package according to the current average exposure frequency per person, the number of exposed people, the current conversion volume and the current delivery duration.

[0156] In an optional implementation, the target population determination device 900 further includes:

[0157] The offline data statistics module is used to obtain the historical data of each order of population packages in the target industry, and to obtain the target number of exposed populations based on the historical data; and to calculate the target delivery data corresponding to the conversion of the target industry, so as to determine the order of the target order population package based on the target number of exposed populations, the target delivery data, the current average exposure frequency per capita, the number of exposed populations, the current conversion volume and the current delivery duration; wherein the target delivery data includes the target delivery duration, the target average exposure frequency per capita and the target conversion volume.

[0158] In an optional implementation, the target group group determination submodule includes:

[0159] A first order determination unit, configured to calculate a first order of a target population package according to the current average exposure frequency per capita and the target average exposure frequency per capita, the current delivery duration and the target delivery duration;

[0160] A second order determination unit, configured to calculate a second order of a target population package according to the number of exposed populations and the number of target exposed populations;

[0161] A third order determination unit, used for calculating the third order of the target population package according to the current conversion amount and the target conversion amount;

[0162] The final order determining unit is used to determine the final order of the target group package according to the first order, the second order and the third order of the target group package.

[0163] Exemplary Media

[0164] After introducing the method according to the exemplary embodiment of the present disclosure, next, a medium according to the exemplary embodiment of the present disclosure will be described.

[0165] In this exemplary embodiment, the above method can be implemented by a program product, such as a portable compact disk read-only memory (CD-ROM) and including program code, and can be run on a device, such as a personal computer. However, the program product of the present disclosure is not limited thereto, and in this document, a readable storage medium can be any tangible medium containing or storing a program, which can be used by or in combination with an instruction execution system, an apparatus, or a device.

[0166] The program product may use any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0167] Computer readable signal media may include data signals propagated in baseband or as part of a carrier wave, in which readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Readable signal media may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0168] The program code contained on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RE, etc., or any suitable combination of the foregoing.

[0169] Program code for performing the operations of the present disclosure may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user computing device, partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (FAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., using an Internet service provider to connect through the Internet).

[0170] Exemplary Electronic Devices

[0171] After introducing the method, apparatus, and medium according to the exemplary embodiment of the present disclosure, next, an electronic device according to an exemplary embodiment of the present disclosure is introduced.

[0172] Those skilled in the art will appreciate that various aspects of the present disclosure may be implemented as systems, methods or program products. Therefore, various aspects of the present disclosure may be specifically implemented in the following forms, namely: complete hardware implementation, complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software, which may be collectively referred to herein as "circuits", "modules" or "systems".

[0173] Fig.10 A schematic diagram of the structure of a computer system suitable for implementing an electronic device of an embodiment of the present disclosure is shown.

[0174] It should be noted that Fig.10 The computer system 1000 of the electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0175] like Fig.10 As shown, the electronic device 1000 is in the form of a general computing device. The components of the electronic device 1000 may include, but are not limited to: at least one processing unit 1010, at least one storage unit 1020, a bus 1030 connecting different system components (including the storage unit 1020 and the processing unit 1010), and a display unit 1040.

[0176] The storage unit stores program codes, which can be executed by the processing unit 1010, so that the processing unit 1010 performs the steps according to various exemplary embodiments of the present disclosure described in the above “Exemplary Method” section of this specification. For example, the processing unit 1010 can perform the following steps: Figure 2 The method steps shown, etc.

[0177] The storage unit 1020 may include a volatile storage unit, such as a random access storage unit (RAM) 1021 and / or a cache storage unit 1022 , and may further include a read-only storage unit (ROM) 1023 .

[0178] The storage unit 1020 may also include a program / utility 1024 having a set (at least one) of program modules 1025, such program modules 1025 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0179] The bus 1030 may include a data bus, an address bus, and a control bus.

[0180] The electronic device 1000 may also communicate with one or more external devices 1100 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), and such communication may be performed via an input / output (I / O) interface 1050. The electronic device 1000 also includes a display unit 1040, which is connected to the input / output (I / O) interface 1050 for display. Furthermore, the electronic device 1000 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter 1060. As shown, the network adapter 1060 communicates with other modules of the electronic device 1000 via a bus 1030. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 1000, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.

[0181] It should be noted that, although several modules or submodules of the device are mentioned in the above detailed description, such division is merely exemplary and not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more units / modules described above can be embodied in one unit / module. Conversely, the features and functions of one unit / module described above can be further divided into multiple units / modules to be embodied.

[0182] In addition, although the operations of the disclosed method are described in a specific order in the drawings, this does not require or imply that the operations must be performed in this specific order, or that all the operations shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps.

[0183] Although the spirit and principle of the present disclosure have been described with reference to several specific embodiments, it should be understood that the present disclosure is not limited to the disclosed specific embodiments, and the division of various aspects does not mean that the features in these aspects cannot be combined to benefit, and such division is only for the convenience of expression. The present disclosure is intended to cover various modifications and equivalent arrangements included in the spirit and scope of the attached claims.

Claims

1. A method for determining a target population, characterized in that: include: Generate multi-level population packages for multiple industries based on user feature data and advertisement feature data, wherein the multi-level population packages are divided into sub-industry population packages at multiple stages, and each sub-industry population package at each stage is a per-level population package, and each level population package in the multi-level population package contains a preset number of user populations; Determine the initial group of people in the target industry corresponding to the advertisement to be placed; Acquire the delivery data of the advertisement to be delivered during the delivery process, and determine the order of the target group package according to the delivery data; Expanding the population to be delivered the advertisement from the initial population package to the target population package according to the order of the target population package; The generating of multi-order crowd packages for multiple industries according to user characteristic data and advertisement characteristic data comprises: determining an industry vector and multiple user vectors for each industry according to the user characteristic data and advertisement characteristic data; calculating the similarity between the industry vector for each industry and the multiple user vectors; generating the multi-order crowd package for each industry from multiple users whose similarity satisfies a preset similarity threshold; Among them, the obtaining of the delivery data of the advertisement to be delivered during the delivery process, and determining the order of the target-order population package according to the delivery data includes: counting the current per capita exposure frequency, the number of exposed people, the current conversion volume and the current delivery duration of the advertisement to be delivered during the delivery process; obtaining the historical data of each order of population packages in the target industry, and fitting the target number of exposed people according to the historical data; calculating the target delivery data corresponding to the conversion of the target industry, so as to determine the order of the target-order population package according to the target number of exposed people, the target delivery data, the current per capita exposure frequency, the number of exposed people, the current conversion volume and the current delivery duration; wherein the target delivery data includes the target delivery duration, the target per capita exposure frequency and the target conversion volume.

2. The method for determining a target population according to claim 1, characterized in that: The method of generating multi-level crowd packages for multiple industries based on user feature data and advertisement feature data includes: Building a deep learning model based on the user feature data and the advertisement feature data; A multi-stage population package for each of the multiple industries is generated through the deep learning model.

3. The method for determining a target population according to claim 2, characterized in that: The generating a multi-stage crowd package for each of the multiple industries by using the deep learning model includes: Inputting user feature data of multiple users and advertisement feature data of multiple advertisements in each industry into the deep learning model to obtain multiple user vectors and multiple advertisement vectors in each industry; Aggregating multiple advertisement vectors in each industry to obtain an industry vector for each industry; A multi-order population package for each industry is generated according to the industry vector of each industry and the multiple user vectors.

4. The method for determining a target population according to claim 1, characterized in that: The step of determining the order of the target-order population package according to the current average exposure frequency per person, the number of exposed people, the current conversion volume, and the current delivery duration includes: The first order of the target population package is calculated based on the current average exposure frequency per person and the target average exposure frequency per person, the current delivery duration and the target delivery duration; The second order of the target population package is calculated based on the number of exposed populations and the number of target exposed populations; Calculate the third order of the target population package according to the current conversion amount and the target conversion amount; The final order of the target group package is determined according to the first order, the second order and the third order of the target group package.

5. A device for determining a target population, characterized in that: include: A crowd package generation module is used to generate multi-level crowd packages for multiple industries based on user feature data and advertisement feature data. The multi-level crowd packages are divided into sub-industry crowd packages of multiple stages. The sub-industry crowd package of each stage is a per-level crowd package. Each level of the multi-level crowd package contains a preset number of user populations. A crowd package matching module is used to determine the initial crowd package of the target industry corresponding to the advertisement to be placed; A target order determination module is used to obtain the delivery data of the advertisement to be delivered during the delivery process, and determine the order of the target order population package according to the delivery data; A crowd package upgrade module, used for expanding the delivery crowd of the to-be-delivered advertisement from the initial-order crowd package to the target-order crowd package according to the order of the target-order crowd package; The generating of multi-order crowd packages for multiple industries according to user characteristic data and advertisement characteristic data comprises: determining an industry vector and multiple user vectors for each industry according to the user characteristic data and advertisement characteristic data; calculating the similarity between the industry vector for each industry and the multiple user vectors; generating the multi-order crowd package for each industry from multiple users whose similarity satisfies a preset similarity threshold; Among them, the obtaining of the delivery data of the advertisement to be delivered during the delivery process, and determining the order of the target-order population package according to the delivery data includes: counting the current per capita exposure frequency, the number of exposed people, the current conversion volume and the current delivery duration of the advertisement to be delivered during the delivery process; obtaining the historical data of each order of population packages in the target industry, and fitting the target number of exposed people according to the historical data; calculating the target delivery data corresponding to the conversion of the target industry, so as to determine the order of the target-order population package according to the target number of exposed people, the target delivery data, the current per capita exposure frequency, the number of exposed people, the current conversion volume and the current delivery duration; wherein the target delivery data includes the target delivery duration, the target per capita exposure frequency and the target conversion volume.

6. The device for determining a target population according to claim 5, characterized in that: The crowd package generation module includes: A model building submodule, used to build a deep learning model based on the user feature data and the advertisement feature data; A crowd package generation submodule is used to generate a multi-order crowd package for each of the multiple industries through the deep learning model.

7. The device for determining a target population according to claim 6, characterized in that: The crowd package generation submodule includes: A first vector generating unit, configured to input user feature data of a plurality of users and advertisement feature data of a plurality of advertisements in each industry into the deep learning model to obtain a plurality of user vectors and a plurality of advertisement vectors in each industry; A second vector generating unit, configured to aggregate the multiple advertisement vectors in each industry to obtain an industry vector for each industry; A crowd package generating unit is used to generate a multi-order crowd package for each industry according to the industry vector of each industry and the multiple user vectors.

8. The device for determining a target population according to claim 5, characterized in that: The target order determination module comprises: A first order determination unit, configured to calculate a first order of a target population package according to the current average exposure frequency per capita and the target average exposure frequency per capita, the current delivery duration and the target delivery duration; A second order determination unit, configured to calculate a second order of a target population package according to the number of exposed populations and the number of target exposed populations; A third order determination unit, used for calculating the third order of the target population package according to the current conversion amount and the target conversion amount; The final order determining unit is used to determine the final order of the target group package according to the first order, the second order and the third order of the target group package.

9. An electronic device, characterized in that: include: processor; A memory, configured to store executable instructions of the processor; The processor is configured to execute the target population determination method according to any one of claims 1 to 4 by executing the executable instructions.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for determining a target population as described in any one of claims 1 to 4 is implemented.

Images