Multi-task model modeling method, promotion content processing method and related device

Abstract

The present disclosure provides a modeling method of a multi-task model, a promotion content processing method and related devices, and relates to the field of artificial intelligence technology. The modeling method of the multi-task model includes: obtaining features for constructing tasks, constructing an initial task set according to the features for constructing tasks; determining the mutual information between different tasks in the initial task set; obtaining a related task set according to the mutual information between the different tasks, and the mutual information of the tasks included in the related task set meets a first preset condition; generating samples of the multi-task model according to the features corresponding to each task in the related task set, and performing model training using the samples of the multi-task model to obtain the multi-task model. In this way, the method can make the multiple tasks in the multi-task model have a strong correlation, and improve the learning efficiency of the multi-task model.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence (AI) technology, and in particular to a modeling method of a multi-task model, a promotion content processing method, an apparatus, a device, a computer-readable storage medium, and a computer program product. Background Art

[0002] With the development of computer technology, especially artificial intelligence technology, the application scenarios of artificial intelligence technology are becoming more and more extensive. For example, in the scenario of pushing promotional content (such as advertisements), the conversion rate of the promotional content can be predicted based on artificial intelligence technology, and then the promotional content can be pushed to users based on the conversion rate.

[0003] In order to improve the generalization ability of the model and the accuracy of the predicted conversion rate, multi-task learning (MTL) is usually used to build a multi-task model. At present, multiple tasks are selected based on artificial subjective experience and multi-task learning is performed so that the learned features can be shared between different tasks, thereby improving the learning efficiency and generalization ability of the multi-task model.

[0004] However, multiple tasks selected based on subjective experience may have poor correlation, which may lead to negative impacts between multiple tasks and reduce the learning efficiency of the multi-task model. Summary of the invention

[0005] The purpose of the present disclosure is to provide a modeling method of a multi-task model, a promotion content processing method, an apparatus, a device, a computer-readable storage medium, and a computer program product, which can improve the learning efficiency of the multi-task model.

[0006] In a first aspect, the present disclosure provides a modeling method for a multi-task model, comprising:

[0007] Acquire features for constructing tasks, and construct an initial task set according to the features for constructing tasks; the initial task set includes at least two of: a conversion rate of promotional content, a play time of the promotional content, a presentation type of the promotional content, and information of promotion objects in the promotional content;

[0008] Determining mutual information between different tasks in the initial task set;

[0009] Obtaining a related task set according to the mutual information between the different tasks, wherein the mutual information of the tasks included in the related task set satisfies a first preset condition;

[0010] According to the features corresponding to each task in the related task set, samples of the multi-task model are generated, and the samples of the multi-task model are used to perform model training to obtain the multi-task model.

[0011] In a second aspect, the present disclosure provides a promotion content processing method, comprising:

[0012] Obtaining attributes of user behavior towards promoted content;

[0013] According to the attributes of the user's behavior on the promotional content and the multi-task model, an inference result of the multi-task model is obtained; the multi-task model is obtained by generating a sample of the multi-task model based on the features corresponding to the tasks in the related task set, the related task set is obtained based on the mutual information between different tasks, and the different tasks are tasks in the initial task set constructed by the features used to construct the tasks; the inference result includes a conversion rate of the promotional content, a playback time of the promotional content, a presentation type of the promotional content, or information of the promotion object in the promotional content;

[0014] According to the inference result, the promotion strategy for the promotion content is adjusted.

[0015] In a third aspect, the present disclosure provides a modeling device for a multi-task model, comprising:

[0016] an acquisition module, configured to acquire features for constructing tasks, and construct an initial task set according to the features for constructing tasks; the initial task set includes at least two of: a conversion rate of promotional content, a play duration of the promotional content, a presentation type of the promotional content, and information on promotion objects in the promotional content;

[0017] A mutual information determination module, used to determine the mutual information between different tasks in the initial task set;

[0018] A related task determination module, configured to obtain a related task set according to the mutual information between the different tasks, wherein the mutual information of the tasks included in the related task set satisfies a first preset condition;

[0019] The training module is used to generate samples of the multi-task model according to the features corresponding to each task in the related task set, and perform model training using the samples of the multi-task model to obtain the multi-task model.

[0020] In a fourth aspect, the present disclosure provides a promotion content processing device, characterized by comprising:

[0021] An acquisition module, used to obtain the attributes of the user's behavior on the promotion content;

[0022] An inference module, configured to obtain an inference result of the multi-task model according to the attributes of the user's behavior on the promotional content and the multi-task model; the multi-task model is obtained by generating a sample of the multi-task model based on the features corresponding to the tasks in the related task set, the related task set is obtained based on the mutual information between different tasks, and the different tasks are tasks in the initial task set constructed by the features used to construct the tasks; the inference result includes a conversion rate of the promotional content, a playback time of the promotional content, a presentation type of the promotional content, or information of the promotion object in the promotional content;

[0023] A processing module is used to adjust the promotion strategy for the promotion content according to the inference result.

[0024] In a fifth aspect, the present disclosure provides a computer-readable medium having a computer program stored thereon, which, when executed by a processing device, implements the steps of the method described in any one of the first aspect or the second aspect of the present disclosure.

[0025] In a sixth aspect, the present disclosure provides an electronic device, including:

[0026] a storage device having a computer program stored thereon;

[0027] A processing device is used to execute the computer program in the storage device to implement the steps of the method described in any one of the first aspect or the second aspect of the present disclosure.

[0028] In a seventh aspect, the present disclosure provides a computer program product comprising instructions, which, when executed on a device, enables the device to execute the method described in any one of the implementations of the first or second aspect.

[0029] It can be seen from the above technical solutions that the present disclosure has the following advantages:

[0030] The present disclosure provides a modeling method for a multi-task model, in which the features used to construct the tasks are first obtained, an initial task set is constructed based on the features used to construct the tasks, and then the mutual information between different tasks in the initial task set is determined, and the initial task set is screened based on the mutual information to obtain a set of related tasks with strong correlation; then, based on the features corresponding to the tasks in the related task set, a sample of the multi-task model is generated to train the multi-task model. Compared with multiple tasks selected solely based on subjective experience, the correlation between the tasks in the related task set obtained after the present disclosure screens the initial task set is strong. In this way, the learning efficiency of the multi-task model can be improved during multi-task learning.

[0031] Other features and advantages of the present disclosure will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical method of the embodiments of the present disclosure, the drawings required for use in the embodiments are briefly introduced below.

[0033] Figure 1 A flowchart of a modeling method of a multi-task model provided in an embodiment of the present disclosure;

[0034] Figure 2 A schematic diagram of obtaining a set of related tasks provided in an embodiment of the present disclosure;

[0035] Figure 3 A schematic diagram of a multi-task model provided in an embodiment of the present disclosure;

[0036] Figure 4 A flowchart of a modeling method of another multi-task model provided in an embodiment of the present disclosure;

[0037] Figure 5 A flowchart of a promotion content processing method provided by an embodiment of the present disclosure;

[0038] Figure 6 A schematic diagram of a modeling device for a multi-task model provided in an embodiment of the present disclosure;

[0039] Figure 7 A schematic diagram of a promotion content processing device provided by an embodiment of the present disclosure;

[0040] Figure 8 A schematic diagram of the structure of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0041] The terms "first" and "second" in the embodiments of the present disclosure are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features.

[0042] First, some technical terms involved in the embodiments of the present disclosure are introduced.

[0043] A multi-task model refers to a model built based on multi-task learning. The core of multi-task learning is that multiple tasks are trained in parallel and the learned features are shared with each other. In the promotional content push scenario, it is usually necessary to predict multiple contents. Each content can be abstracted as a task, such as predicting conversion rate, predicting potential users, predicting whether users click on promotional content, etc. Based on this, a multi-task model can be built based on multi-task learning to predict multiple contents.

[0044] In the multi-task learning process, when the correlation between multiple tasks is strong, multiple tasks share the learned features, which can improve the learning efficiency of the multi-task model; when the correlation between multiple tasks is weak, multiple tasks share the learned features, and negative transfer will occur between multiple tasks, that is, multiple tasks have a negative impact on each other, which reduces the prediction accuracy of each task and reduces the learning efficiency of the multi-task model.

[0045] At present, we mainly rely on subjective experience to select multiple tasks and build multi-task models. Multiple tasks selected based on subjective experience will have poor correlation, which will reduce the learning efficiency of the multi-task model.

[0046] In view of this, an embodiment of the present disclosure provides a modeling method of a multi-task model, which can be performed by an electronic device. The electronic device can be a server. The server can be a cloud server, such as a central server in a central cloud computing cluster, or an edge server in an edge cloud computing cluster. Of course, the server can also be a server in a local data center. A local data center refers to a data center directly controlled by a user.

[0047] The modeling method of the multi-task model includes: the electronic device obtains features for constructing tasks, constructs an initial task set according to the features for constructing tasks, then determines the mutual information between different tasks in the initial task set, obtains a related task set based on the mutual information between different tasks, the mutual information in the related task set satisfies a first preset condition, generates samples of the multi-task model according to the features corresponding to each task in the related task set, and uses the samples of the multi-task model to perform model training to obtain the multi-task model.

[0048] It can be seen that in the modeling method of the multi-task model provided by the embodiment of the present disclosure, the electronic device first screens the initial task set, and the correlation between the tasks in the obtained related task set is strong, which improves the correlation between the multiple tasks corresponding to the multi-task model compared with the multiple tasks selected simply by subjective experience. In this way, the learning efficiency of the multi-task model can be improved during the multi-task learning process.

[0049] The multi-task model obtained by the modeling method of the multi-task model provided by the embodiment of the present disclosure can be applied to a variety of scenarios. For example, in the promotion content push scenario, the multi-task model can be used to predict the conversion rate of the promotion content and the playback time of the promotion content (the time from the start of the promotion content until the user closes the promotion content). The electronic device can input the attributes of the user's behavior on the promotion content into the multi-task model, and then obtain the above-mentioned reasoning result. Then the electronic device can adjust the promotion strategy of the promotion content based on the above-mentioned reasoning result. For example, when the conversion rate of the promotion content is greater than or equal to the conversion rate threshold, and the playback time of the promotion content is greater than or equal to the duration threshold, the number of promotions of the promotion content is increased; when the conversion rate of the promotion content is less than the conversion rate threshold, and the playback time of the promotion content is less than the duration threshold, the number of promotions of the promotion content is reduced.

[0050] In order to make the technical solution of the present disclosure clearer and easier to understand, the modeling method of the multi-task model provided by the embodiment of the present disclosure is introduced from the perspective of electronic equipment. Figure 1 As shown, the figure is a flow chart of a modeling method of a multi-task model provided by an embodiment of the present disclosure, the method comprising:

[0051] S101: The electronic device obtains features for constructing tasks, and constructs an initial task set according to the features for constructing tasks.

[0052] The electronic device may obtain initial data, which may include multiple features. For example, in a promotional content push scenario, the initial data may include features such as the conversion rate of the promotional content, the play time of the promotional content, the presentation type of the promotional content, information about the promotion object in the promotional content, and the generation time of the promotional content.

[0053] The electronic device may obtain features for constructing the task based on the initial data. In some examples, the electronic device may receive multiple features configured by the developer as features for constructing the task. For example, the electronic device may present multiple candidate features to the user through a display device (e.g., a display screen), such as the features included in the initial data, and then, based on the developer's selection operation of the multiple candidate features, use the features selected by the developer as features for constructing the task. Then, the electronic device constructs an initial task set based on the features for constructing the task.

[0054] S102: The electronic device determines mutual information between different tasks in the initial task set.

[0055] Mutual information (MI) refers to the degree of association between two random variables, that is, the degree to which the uncertainty of another random variable is weakened after a given random variable. For example, when the mutual information value is 0 (minimum value), it indicates that a given random variable has no relationship with the determination of another random variable; when the mutual information value is the entropy of the random variable (maximum value), it indicates that a given random variable can completely eliminate the uncertainty of another random variable.

[0056] like Figure 2 As shown, the electronic device can calculate the mutual information between different tasks in the initial task set. For example, the electronic device can calculate the mutual information between different tasks based on the characteristics of the tasks in the initial task set. Specifically, the electronic device can calculate the mutual information between different tasks in the initial task set based on the following formula:

[0057]

[0058] Where I(X; Y) is the mutual information between task X and task Y in the initial task set; x is the feature corresponding to task X, and y is the feature corresponding to task Y; p(x, y) is the probability of task X and task Y occurring at the same time, p(x) is the probability of task X occurring, and p(y) is the probability of task Y occurring.

[0059] It should be noted that the mutual information between different tasks in the initial task set is non-negative, that is, I(X; Y) ≥ 0, and I(Y; X) = I(X; Y), both representing the mutual information between task X and task Y in the initial task set.

[0060] S103: The electronic device obtains a related task set according to mutual information between different tasks, where the mutual information of tasks included in the related task set meets a first preset condition.

[0061] As described above, mutual information can characterize the uncertainty of one random variable in multiple random variables to another random variable. Based on this, the electronic device can screen the initial task set based on the mutual information between different tasks in the initial task set, and then obtain a related task set, and the mutual information of the tasks included in the related task set meets the first preset condition. For example, the first preset condition can be that the mutual information between different tasks is greater than or equal to the mutual information threshold.

[0062] In some examples, the electronic device may receive a mutual information threshold configured by a developer to determine a set of related tasks based on the configured mutual information threshold, thereby allowing the developer to adjust tasks in the set of related tasks within a certain range.

[0063] The embodiments of the present disclosure do not specifically limit the mutual information threshold. As mentioned above, the mutual information threshold may be a value configured by a developer or a default value. For example, the mutual information threshold may be 0.1, 0.2, etc.

[0064] S104: The electronic device generates samples of the multi-task model according to the features corresponding to each task in the related task set, and performs model training using the samples of the multi-task model to obtain the multi-task model.

[0065] After determining the related task set, the electronic device can construct a multi-task model based on the tasks in the related task set. Since the mutual information between the tasks in the related task set meets the first preset condition, it indicates that the correlation between the tasks in the related task set is strong. Therefore, after constructing the multi-task model based on the tasks with strong correlation, the learning efficiency of the multi-task model is high.

[0066] The electronic device can generate samples of the multi-task model based on the features corresponding to the tasks in the related task set. The samples of the multi-task model include the feature vector of the sample and the label of the feature vector of the sample. As mentioned above, the initial data includes multiple features, and the features corresponding to the tasks in the related task set are some of the multiple features of the initial data. Based on this, the electronic device can use the features corresponding to the tasks in the related task set as the labels of the feature vectors of the samples, and after removing the features corresponding to the tasks in the related task set from the initial data, the initial data obtained is used as the feature vector of the sample.

[0067] For example, the initial data includes feature 1, feature 2, feature 3, feature 4, and feature 5. In the related task set, task 1 corresponds to feature 1, and task 2 corresponds to feature 2. The electronic device can use feature 1 as a label for the feature vector of the sample for monitoring task 1, and feature 2 as a label for the feature vector of the sample for monitoring task 2, remove feature 1 and feature 2 from the initial data, and obtain initial data including feature 3, feature 4, and feature 5, and use the initial data including feature 3, feature 4, and feature 5 as the feature vector of the sample of the multi-task model.

[0068] Then, the electronic device can use the samples of the multi-task model to perform model training to obtain the multi-task model. Figure 3As shown, the figure is a schematic diagram of a multi-task model provided by an embodiment of the present disclosure. The multi-task model includes a task-exclusive network 310 and an output network 330, and a shared network 320 corresponding to each of the multiple tasks in the related task set. Among them, the task-exclusive network 310 can be a deep neural network (DNN), a convolutional neural network (CNN) or a self-attention network; similarly, the shared network 320 can also be a deep neural network, a convolutional neural network or a self-attention network.

[0069] It should be noted that the present application does not specifically limit the types of task-exclusive network 310 and shared neural network 320. In the image scenario, in order to build the invariance of content information, a convolutional neural network can be used to build a shared network 320 and a task-exclusive network 310; in the promotion content scenario, in order to meet the demand for feature crossover, a deep neural network can be used to build a shared network 320 and a task-exclusive network 310; in the text scenario, in order to meet the demand for parallel processing of temporal information, a self-attention network can be used to build a shared network 320 and a task-exclusive network 310. The output network 330 can use activation functions such as sigmoid and relu to obtain the output corresponding to each task (for example, it can be a classification value or a regression value).

[0070] In some examples, the electronic device may input the feature vector of the sample of the multi-task model into the shared network 320 to obtain a shared component, and then input the shared component into the task-exclusive network of each task in the related task set to obtain the output of each task-exclusive network. The electronic device may then determine a loss value based on the label value of the feature vector of the sample of the multi-task model and the output of the task-exclusive network, and update the weight of the task-exclusive network based on the loss value to perform model training, thereby obtaining the multi-task model.

[0071] Based on the above description, the embodiment of the present disclosure provides a modeling method of a multi-task model. Compared with multiple tasks selected solely based on subjective experience, the correlation between tasks in the related task set obtained after the electronic device screens the initial task set is stronger. In this way, the learning efficiency of the multi-task model can be improved during the multi-task learning process. Furthermore, it meets the business needs in various scenarios and achieves more accurate prediction of multiple contents.

[0072] The present disclosure also provides a method for modeling a multi-task model. Figure 2 Based on the embodiment shown, the tasks in the related task set are further grouped and aggregated, so as to balance the number of positive samples in multiple task groups in the related task set after aggregation. Figure 4As shown, the modeling method of the multi-task model also includes:

[0073] S401: The electronic device aggregates tasks that meet a second preset condition in a set of related tasks.

[0074] The second preset condition may be a plurality of tasks in the related task set whose task relevance is greater than a relevance threshold. The electronic device may aggregate the tasks in the related task set based on the second preset condition, and the aggregated related task set includes a plurality of task groups.

[0075] Continue to see Figure 3 , the electronic device may obtain the correlation between the outputs of the task exclusive network 310 corresponding to the multiple tasks in the related task set, and then aggregate the multiple tasks in the related task set based on the correlation between the outputs of the task exclusive network 310. Specifically, the electronic device may calculate the correlation between the outputs of the task exclusive network 310 corresponding to the multiple tasks in the related task set based on the following formula:

[0076]

[0077] Among them, r mn is the correlation between task m and task n in the related task set; im is the output of the task-exclusive network 310 corresponding to task m for sample i, is the mean value of the output of all samples of the task exclusive network 310 corresponding to task m; in is the output of the task-exclusive network 310 corresponding to task n for sample i, is the mean of the outputs of all samples of the task exclusive network 310 corresponding to task n, and K is the total number of samples.

[0078] In some examples, the electronic device may calculate the correlation r between task m and task n. mn Compared with the correlation threshold, r mn ≥ the correlation threshold, task m and task n are aggregated into one group, so that the electronic device can obtain the aggregated related task set including multiple task groups.

[0079] S402: The electronic device determines, for each task group, a plurality of target tasks according to the number of positive samples of tasks in the task group.

[0080] Among them, the sum of the number of positive samples of multiple target tasks is greater than the positive sample number threshold, and the positive sample number threshold is the minimum value of the sum of the number of positive samples in each task group. For example, the aggregated related task set includes multiple task groups. For example, multiple task groups include task group G1 and task group G2, where task group G1 includes task T1, task T2, and task T3, the number of positive samples of task T1 is 100, the number of positive samples of task T2 is 200, and the number of positive samples of task T3 is 300; task group G2 includes task T4, task T5, and task T6, the number of positive samples of task T4 is 200, the number of positive samples of task T5 is 400, and the number of positive samples of task T6 is 500. It can be seen that the sum of the number of positive samples of task group G1 is 600, and the sum of the number of positive samples of task group G2 is 1100. The electronic device can use the sum of the number of positive samples of task group G1, which is 600, as the positive sample number threshold.

[0081] In some examples, the electronic device may sort the tasks T4, T5, and T6 in the task group G2, such as adding them one by one in the order of the number of positive samples from small to large, until the sum of the number of positive samples of the tasks in the task group G2 is greater than or equal to the above-mentioned positive sample number threshold 600. The electronic device may determine that when the sum of the number of positive samples of the tasks in the task group is greater than or equal to the positive sample number threshold, the multiple tasks in the task group are multiple target tasks. In this way, after the electronic device adds the number of positive samples of task T4 and the number of positive samples of task T5, the sum is equal to the positive sample number threshold, and the electronic device may regard task T4 and task T5 as multiple target tasks in the task group.

[0082] In other examples, the electronic device may randomly select the number of positive samples of multiple tasks in task group G2 and add them up until the number of positive samples of the multiple tasks is greater than or equal to the threshold number of positive samples, and then determine the randomly selected multiple tasks as multiple target tasks. Continuing with the above example, the electronic device may add the number of positive samples of task T5 and task T6 in task group G2, and the sum is greater than the threshold number of positive samples, so the electronic device may determine task T5 and task T6 as multiple target tasks.

[0083] It should be noted that the sum of the number of positive samples of task T4 and task T5 in task group G2 is greater than or equal to the positive sample number threshold, which means that when the number of positive samples of either task T4 or task T5 in task group G2 is subtracted, the sum of the number of positive samples of the remaining tasks will be less than the positive sample number threshold.

[0084] After the electronic device determines the multiple target tasks of the task group, it can remove the tasks other than the target task from the task group. For example, if the multiple target tasks in the task group G2 are tasks T4 and T5, the electronic device can remove task T6 from the task group G2.

[0085] In some embodiments, the electronic device can generate samples of the multi-task model based on the features corresponding to the multiple target tasks in the task grouping. Taking the target tasks included in the task grouping G2 as tasks T4 and T5 as an example, for the task grouping G2, the label of the feature vector of the positive sample corresponding to task T4 is used as the label of the task grouping G2, and the label of the feature vector of the positive sample corresponding to task T5 is also used as the label of the task grouping G2. In this way, in this embodiment, the electronic device can evenly configure the number of positive samples included in the multiple task groups, reducing the situation where the learning efficiency of the multi-task model is reduced due to the large difference in the number of positive samples.

[0086] Based on the above description, the modeling method of the multi-task model analyzes the statistical information of large-scale data containing multiple features, determines the set of related tasks based on correlation, and realizes the aggregation of highly related tasks. Using highly related tasks to train the multi-task model can reduce negative transfer and improve the learning efficiency of the multi-task model.

[0087] Furthermore, the structure of the multi-task model is constructed based on the determined set of related tasks, and the task aggregation with model generalization information is completed in combination with the Pearson coefficient; the number of positive samples corresponding to the task group with the smallest number of positive samples is used as the positive sample number threshold to screen and combine the tasks of each group, thereby achieving a balance in the number of positive samples of different tasks. After grouping and aggregation, the number of positive samples between different tasks can be completely proportional, so that the multi-task model can complete the multi-task learning process without bias.

[0088] The disclosed embodiment also provides a promotion content processing method, such as Figure 5 As shown, the promotion content processing method includes:

[0089] S501: The electronic device obtains attributes of the user's behavior on the promotion content.

[0090] The attributes of the user's behavior towards the promoted content may include the length of time the user watches the promoted content, whether the user clicks on the promoted content, the presentation type of the promoted content clicked by the user (e.g., video type, image type, etc.), whether the user is converted by the promoted content, etc.

[0091] It should be noted that the electronic device needs to obtain the user's authorization in advance. Only after obtaining the user's authorization to use the corresponding data (such as the attributes of the user's behavior towards the promotional content mentioned above) can the electronic device obtain data such as the attributes of the user's behavior towards the promotional content.

[0092] S502: The electronic device obtains an inference result of the multi-task model according to the attributes of the user's behavior on the promotion content and the multi-task model.

[0093] Among them, the multi-task model is obtained based on the sample of the multi-task model generated by the features corresponding to each task in the related task set, the related task set is obtained based on the mutual information between different tasks, and the different tasks are tasks in the initial task set constructed by the features used to construct the tasks; the inference results include the conversion rate of the promotion content, the playback time of the promotion content, the presentation type of the promotion content, or the information of the promotion object in the promotion content. The process of training the multi-task model can be referred to the introduction in the above embodiment, and will not be repeated here.

[0094] S503: The electronic device adjusts the promotion strategy for the promotion content according to the inference result.

[0095] In some examples, when the inference result indicates that the conversion rate of the promotional content is greater than the conversion rate threshold, and the playback time of the promotional content is greater than the time threshold, the electronic device increases the number of promotions of the promotional content; when the conversion rate of the promotional content is less than the conversion rate threshold, and the playback time of the promotional content is less than the time threshold, the electronic device reduces the number of promotions of the promotional content. In other examples, when the presentation type of the promotional content is a preset type (such as a video type), and the information of the promotional object in the promotional content is preset information (such as indicating that the promotional object is a game, a virtual item, or a physical object), the electronic device increases the number of promotions of the promotional content. In this way, the ineffective delivery of promotional content is reduced and the waste of resources is reduced.

[0096] Figure 6 is a schematic diagram of a modeling device for a multi-task model according to an exemplary disclosed embodiment. Figure 6 As shown, the modeling device 600 of the multi-task model includes:

[0097] The acquisition module 601 is used to acquire features for constructing tasks, and construct an initial task set according to the features for constructing tasks; the initial task set includes at least two of: a conversion rate of promotional content, a play time of the promotional content, a presentation type of the promotional content, and information of promotion objects in the promotional content;

[0098] A mutual information determination module 602, used to determine the mutual information between different tasks in the initial task set;

[0099] A related task determination module 603 is used to obtain a related task set according to the mutual information between the different tasks, wherein the mutual information of the tasks included in the related task set meets a first preset condition;

[0100] The training module 604 is used to generate samples of the multi-task model according to the features corresponding to each task in the related task set, and perform model training using the samples of the multi-task model to obtain the multi-task model.

[0101] Optionally, the related task determination module 603 is further configured to aggregate tasks in the related task set that meet a second preset condition.

[0102] Optionally, the related task determination module 603 is specifically used to determine the correlation between the outputs of the task exclusive networks corresponding to multiple tasks in the related task set; and aggregate the multiple tasks in the related task set according to the correlation.

[0103] Optionally, the related task determination module 603 is further used to determine, for each task group, a plurality of target tasks according to the number of positive samples of the tasks in the task group, the sum of the number of positive samples of the plurality of target tasks being greater than a positive sample number threshold, and the positive sample number threshold being the minimum value of the sum of the number of positive samples in each task group;

[0104] The training module 604 is specifically configured to generate samples of the multi-task model according to the features corresponding to the target tasks in the task grouping.

[0105] Optionally, the related task determination module 603 is specifically used to add the positive sample numbers of the tasks in the task group one by one in the order from small to large until the sum of the positive sample numbers of the tasks in the task group is greater than or equal to the positive sample number threshold; when it is determined that the sum of the positive sample numbers of the tasks in the task group is greater than or equal to the positive sample number threshold, the multiple tasks in the task group are multiple target tasks

[0106] Optionally, the training module 604 is specifically used to input the feature vectors of the samples of the multi-task model into a shared network to obtain a shared component; input the shared component to the task-exclusive network of each task in the related task set to obtain the output corresponding to the task-exclusive network of each task; and train the multi-task model according to the label values ​​of the feature vectors of the samples of the multi-task model and the output of the task-exclusive network.

[0107] Optionally, the task-exclusive network includes a deep neural network, a convolutional neural network or a self-attention network; the shared network includes a deep neural network, a convolutional neural network or a self-attention network.

[0108] Figure 7 is a schematic diagram of a promotion content processing device according to an exemplary disclosed embodiment. Figure 7 As shown, the promotion content processing device 700 includes:

[0109] The acquisition module 701 is used to acquire the attributes of the user's behavior on the promotion content;

[0110] The inference module 702 is used to obtain the inference result of the multi-task model according to the attributes of the user's behavior on the promotion content and the multi-task model; the multi-task model is obtained by generating a sample of the multi-task model based on the features corresponding to the tasks in the related task set, the related task set is obtained based on the mutual information between different tasks, and the different tasks are tasks in the initial task set constructed by the features used to construct the tasks; the inference result includes a conversion rate of the promotion content, a playback time of the promotion content, a presentation type of the promotion content, or information of the promotion object in the promotion content;

[0111] The processing module 703 is used to adjust the promotion strategy for the promotion content according to the inference result.

[0112] The functions of the above modules have been described in detail in the method steps in the previous embodiment and will not be repeated here.

[0113] Reference below Figure 8 , which shows a schematic diagram of the structure of an electronic device 800 suitable for implementing the embodiment of the present disclosure, the electronic device is used to implement the following Figure 6 The modeling device 600 of the multi-task model shown in FIG. 6 corresponds to the function of the multi-task model, or is used to implement the following Figure 7 The functions corresponding to the promotion content processing device 700 shown. Figure 8 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.

[0114] like Figure 8 As shown, the electronic device 800 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 801, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 802 or a program loaded from a storage device 808 into a random access memory (RAM) 803. In the RAM 803, various programs and data required for the operation of the electronic device 800 are also stored. The processing device 801, the ROM 802, and the RAM 803 are connected to each other via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0115] Typically, the following devices may be connected to the I / O interface 805: input devices 806 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; output devices 807 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 808 including, for example, a magnetic tape, a hard disk, etc.; and communication devices 809. The communication device 809 may allow the electronic device 800 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 8 The electronic device 800 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0116] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device 809, or installed from a storage device 808, or installed from a ROM 802. When the computer program is executed by the processing device 801, the above-mentioned functions defined in the method of the embodiment of the present disclosure are executed.

[0117] It should be noted that the computer-readable medium disclosed above may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-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 computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM 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. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer 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. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0118] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0119] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0120] The computer-readable medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device:

[0121] Acquire features for constructing tasks, and construct an initial task set according to the features for constructing tasks; the initial task set includes at least two of: a conversion rate of promotional content, a play time of the promotional content, a presentation type of the promotional content, and information of promotion objects in the promotional content;

[0122] Determining mutual information between different tasks in the initial task set;

[0123] Obtaining a related task set according to the mutual information between the different tasks, wherein the mutual information of the tasks included in the related task set satisfies a first preset condition;

[0124] Generate samples of a multi-task model according to the features corresponding to each task in the related task set, and perform model training using the samples of the multi-task model to obtain the multi-task model; or

[0125] Obtaining attributes of user behavior towards promoted content;

[0126] According to the attributes of the user's behavior on the promotional content and the multi-task model, an inference result of the multi-task model is obtained; the multi-task model is obtained by generating a sample of the multi-task model based on the features corresponding to the tasks in the related task set, the related task set is obtained based on the mutual information between different tasks, and the different tasks are tasks in the initial task set constructed by the features used to construct the tasks; the inference result includes a conversion rate of the promotional content, a playback time of the promotional content, a presentation type of the promotional content, or information of the promotion object in the promotional content;

[0127] According to the inference result, the promotion strategy for the promotion content is adjusted.

[0128] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages ​​or a combination thereof, including, but not limited to, object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0129] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0130] The modules involved in the embodiments described in the present disclosure may be implemented by software or hardware. The name of a module does not limit the module itself in some cases. For example, the first acquisition module may also be described as a "module for acquiring at least two Internet Protocol addresses".

[0131] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0132] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer 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 foregoing.

[0133] According to one or more embodiments of the present disclosure, Example 1 provides a modeling method for a multi-task model, which obtains features for constructing tasks, and constructs an initial task set based on the features for constructing tasks; the initial task set includes: at least two of: the conversion rate of the promotional content, the playback time of the promotional content, the presentation type of the promotional content, and the information of the promotional object in the promotional content; determines the mutual information between different tasks in the initial task set; obtains a related task set based on the mutual information between the different tasks, and the mutual information of the tasks included in the related task set meets a first preset condition; generates samples of the multi-task model based on the features corresponding to each task in the related task set, and uses the samples of the multi-task model to perform model training to obtain the multi-task model.

[0134] According to one or more embodiments of the present disclosure, Example 2 provides the method of Example 1, wherein the method further includes:

[0135] Aggregate the tasks that meet the second preset condition in the related task set.

[0136] According to one or more embodiments of the present disclosure, Example 3 provides the method of Example 2, wherein the aggregating the tasks in the related task set that meet the second preset condition includes:

[0137] Determine the correlation between outputs of task-exclusive networks corresponding to multiple tasks in the related task set;

[0138] A plurality of tasks in the related task set are aggregated according to the correlation.

[0139] According to one or more embodiments of the present disclosure, Example 4 provides the method of Example 3, wherein the aggregated set of related tasks includes multiple task groups, and the method further includes:

[0140] For each task group, determine multiple target tasks according to the number of positive samples of the tasks in the task group, the sum of the number of positive samples of the multiple target tasks is greater than a positive sample number threshold, and the positive sample number threshold is the minimum value of the sum of the number of positive samples in each task group;

[0141] The step of generating a sample of the multi-task model according to the features corresponding to each task in the set of related tasks includes:

[0142] Generate a sample of a multi-task model according to the features corresponding to the target tasks in the task grouping.

[0143] According to one or more embodiments of the present disclosure, Example 5 provides the method of Example 4, wherein determining multiple target tasks according to the number of positive samples of tasks in the task group includes:

[0144] The positive sample quantities of the tasks in the task group are added one by one in order from small to large, until the sum of the positive sample quantities of the tasks in the task group is greater than or equal to the positive sample quantity threshold;

[0145] When it is determined that the sum of the number of positive samples of tasks in the task group is greater than or equal to the positive sample number threshold, the multiple tasks in the task group are multiple target tasks.

[0146] According to one or more embodiments of the present disclosure, Example 6 provides the method of Example 1, wherein the model training is performed using samples of the multi-task model, including:

[0147] Inputting the feature vector of the sample of the multi-task model into a shared network to obtain a shared component;

[0148] Inputting the shared component to the task exclusive network of each task in the related task set to obtain the output corresponding to the task exclusive network of each task;

[0149] The multi-task model is trained according to the label values ​​of the feature vectors of the samples of the multi-task model and the outputs of the task-exclusive network.

[0150] According to one or more embodiments of the present disclosure, Example 7 provides the method of Example 6, wherein the task-exclusive network includes a deep neural network, a convolutional neural network, or a self-attention network; and the shared network includes a deep neural network, a convolutional neural network, or a self-attention network.

[0151] According to one or more embodiments of the present disclosure, Example 8 provides a promotional content processing method, including: obtaining attributes of a user's behavior towards promotional content; obtaining an inference result of the multi-task model based on the attributes of the user's behavior towards the promotional content and a multi-task model; the multi-task model is obtained by generating a sample of the multi-task model based on features corresponding to tasks in a related task set, the related task set is obtained based on mutual information between different tasks, and the different tasks are tasks in an initial task set constructed by features used to construct tasks; the inference result includes a conversion rate of the promotional content, a playback duration of the promotional content, a presentation type of the promotional content, or information on a promotion object in the promotional content; and adjusting the promotion strategy for the promotional content based on the inference result.

[0152] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present disclosure is not limited to the technical solutions formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in the present disclosure (but not limited to) by each other to form a technical solution.

[0153] In addition, although each operation is described in a specific order, this should not be understood as requiring these operations to be performed in the specific order shown or in a sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details are included in the above discussion, these should not be interpreted as limiting the scope of the present disclosure. Some features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination mode.

[0154] Although the subject matter has been described in language specific to structural features and / or method logic actions, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. On the contrary, the specific features and actions described above are merely example forms of implementing the claims. Regarding the device in the above embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment related to the method, and will not be elaborated here.

Claims

1. A modeling method for a multi-task model, characterized in that: include: Acquire features for constructing tasks, and construct an initial task set according to the features for constructing tasks; The initial task set includes at least two of: a conversion rate of the promotion content, a play time of the promotion content, a presentation type of the promotion content, and information of a promotion object in the promotion content; Determining mutual information between different tasks in the initial task set; Obtaining a related task set according to the mutual information between the different tasks, wherein the mutual information of the tasks included in the related task set satisfies a first preset condition; Generate samples of a multi-task model according to the features corresponding to each task in the related task set, and perform model training using the samples of the multi-task model to obtain the multi-task model; Determine the correlation between outputs of task-exclusive networks corresponding to multiple tasks in the related task set; aggregate the multiple tasks in the related task set according to the correlation; The aggregated set of related tasks includes a plurality of task groups, and the method further includes: For each task group, determine multiple target tasks according to the number of positive samples of the tasks in the task group, the sum of the number of positive samples of the multiple target tasks is greater than a positive sample number threshold, and the positive sample number threshold is the minimum value of the sum of the number of positive samples in each task group; The step of generating a sample of the multi-task model according to the features corresponding to each task in the set of related tasks includes: Generate a sample of a multi-task model according to the features corresponding to the target tasks in the task grouping.

2. The method according to claim 1, characterized in that Determining a plurality of target tasks according to the number of positive samples of tasks in the task grouping includes: The positive sample quantities of the tasks in the task group are added one by one in order from small to large, until the sum of the positive sample quantities of the tasks in the task group is greater than or equal to the positive sample quantity threshold; When it is determined that the sum of the number of positive samples of tasks in the task group is greater than or equal to the positive sample number threshold, the multiple tasks in the task group are multiple target tasks.

3. The method according to claim 1, characterized in that The using the samples of the multi-task model to perform model training includes: Inputting the feature vector of the sample of the multi-task model into a shared network to obtain a shared component; Inputting the shared component to the task exclusive network of each task in the related task set to obtain the output corresponding to the task exclusive network of each task; The multi-task model is trained according to the label values ​​of the feature vectors of the samples of the multi-task model and the outputs of the task-exclusive network.

4. The method according to claim 3, characterized in that The task-exclusive network includes a deep neural network, a convolutional neural network or a self-attention network; the shared network includes a deep neural network, a convolutional neural network or a self-attention network.

5. A promotion content processing method, characterized in that: include: Obtaining attributes of user behavior towards promoted content; According to the attributes of the user's behavior on the promotional content and the multi-task model, an inference result of the multi-task model is obtained; the multi-task model is obtained by generating samples of the multi-task model based on the features corresponding to the target task in the task grouping, the related task set is obtained based on the mutual information between different tasks, the task grouping is obtained by aggregating the related task set, the related task set is obtained by aggregating multiple tasks in the related task set according to the correlation, the correlation is the correlation between the outputs of the task exclusive networks corresponding to the multiple tasks in the related task set, the target task is determined for each of the task groups according to the number of positive samples of the tasks in the task grouping, the sum of the number of positive samples of the multiple target tasks is greater than a threshold value of the number of positive samples, the threshold value of the number of positive samples is the minimum value of the sum of the number of positive samples in each task grouping, and the different tasks are tasks in the initial task set constructed by the features used to construct the tasks; the inference result includes multiple of the conversion rate of the promotional content, the playing time of the promotional content, the presentation type of the promotional content, or the information of the promotion object in the promotional content; According to the inference result, the promotion strategy for the promotion content is adjusted.

6. A modeling device for a multi-task model, characterized in that: include: An acquisition module, used to acquire features for constructing tasks, and construct an initial task set according to the features for constructing tasks; The initial task set includes at least two of: a conversion rate of the promotion content, a play time of the promotion content, a presentation type of the promotion content, and information of a promotion object in the promotion content; A mutual information determination module, used to determine the mutual information between different tasks in the initial task set; A related task determination module, configured to obtain a related task set according to the mutual information between the different tasks, wherein the mutual information of the tasks included in the related task set satisfies a first preset condition; A training module, used to generate samples of the multi-task model according to the features corresponding to each task in the related task set, and perform model training using the samples of the multi-task model to obtain the multi-task model; The related task determination module is further used to determine the correlation between the outputs of the task exclusive networks corresponding to the multiple tasks in the related task set; and aggregate the multiple tasks in the related task set according to the correlation; The related task determination module is further used to determine, for each task group, multiple target tasks according to the number of positive samples of the tasks in the task group, the sum of the number of positive samples of the multiple target tasks is greater than a positive sample number threshold, and the positive sample number threshold is the minimum value of the sum of the number of positive samples in each task group; The training module is specifically used to generate samples of the multi-task model according to the features corresponding to the target tasks in the task grouping.

7. A promotion content processing device, characterized in that: include: An acquisition module, used to obtain attributes of user behavior on promotional content; An inference module, configured to obtain an inference result of the multi-task model according to the attributes of the user's behavior on the promotional content and the multi-task model; the multi-task model is obtained by generating a sample of the multi-task model based on the features corresponding to the target task in the task grouping, the task grouping is obtained by aggregating a set of related tasks, the related task set is obtained by aggregating multiple tasks in the related task set according to the correlation, the correlation is the correlation between the outputs of the task exclusive networks corresponding to the multiple tasks in the related task set, the target task is determined for each of the task groups according to the number of positive samples of the tasks in the task grouping, the sum of the number of positive samples of the multiple target tasks is greater than a threshold of the number of positive samples, the threshold of the number of positive samples is the minimum value of the sum of the number of positive samples in each task group, the related task set is obtained based on the mutual information between different tasks, the different tasks are tasks in the initial task set constructed by the features used to construct the tasks; the inference result includes a plurality of the conversion rate of the promotional content, the playing time of the promotional content, the presentation type of the promotional content, or the information of the promotion object in the promotional content; A processing module is used to adjust the promotion strategy for the promotion content according to the inference result.

8. An electronic device, characterized in that: include: a storage device having a computer program stored thereon; A processing device, used to execute the computer program in the storage device to implement the steps of the method in any one of claims 1 to 4; or, to implement the steps of the method in claim 5.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processing device, the steps of the method described in any one of claims 1 to 4 are implemented; or, when the program is executed by a processing device, the steps of the method described in claim 5 are implemented.

10. A computer program product, characterized in that When the computer program product runs on a computer, the computer executes the steps of the method according to any one of claims 1 to 4; or, when the computer program product runs on a computer, the computer executes the steps of the method according to claim 5.

Images