Information pushing method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention provides an information push method and device, electronic equipment and a storage medium, and the method comprises the steps: predicting whether a user responds to push information or not through an incentive intention prediction model based on target node information sent by a client and first historical behavior data corresponding to the client, obtaining a prediction result, and sending the prediction result to the client; the target node information represents node information of content display; when the prediction result indicates that the user responds to the push information, a target incentive strategy is obtained, and the target incentive strategy comprises target push information and a target right type; and pushing the target incentive strategy to the client, the target incentive strategy being used for being displayed at a target node represented by the target node information to obtain target rights and interests under the target rights and interests type. According to the embodiment of the invention, the accuracy of the pushing opportunity can be improved.

Description

Technical Field

[0001] This invention relates to the field of information processing technology, and in particular to an information push method, apparatus, electronic device, and storage medium. Background Technology

[0002] As the mobile internet enters an era of fierce competition for existing users, various content platforms, video platforms, and social platforms are generally facing a situation of "peaked user growth and slowing daily active user growth." Against this backdrop, user activity has become a core operational metric, especially for infrequent users. These users may still retain their accounts and possibly the app technically, but they haven't generated any effective activity for a long time, or only access it very infrequently. Therefore, activating these users can significantly improve user activity. Traditional activation methods, such as SMS wake-up calls, coupon pushes, and membership discounts, are becoming increasingly ineffective for converting these users. However, research and behavioral analysis show that a significant proportion of users are "non-paying, infrequent, but highly tolerant of push notifications." If these users are properly guided through push notification incentives, their activity can be effectively stimulated.

[0003] Current technologies mostly focus on users periodically tagging and configuring delivery rules, which cannot adapt to different users. This results in the information pushed and the timing of the push not matching users' interests, leading to low accuracy in push timing. Summary of the Invention

[0004] The purpose of this invention is to provide an information push method, device, electronic device, and storage medium to improve the accuracy of push timing. The specific technical solution is as follows:

[0005] In a first aspect of this invention, an information push method is provided, comprising:

[0006] Based on the target node information sent by the client and the first historical behavior data corresponding to the client, the incentive intent prediction model is used to predict whether the user will respond to the push information, and the prediction result is obtained. The target node information represents the node information of the content display.

[0007] When the prediction result indicates that the user should respond to the push notification, a target incentive strategy is obtained, the target incentive strategy including the target push notification and the target benefit type;

[0008] The target incentive strategy is pushed to the client. The target incentive strategy is used to display the target node represented by the target node information in order to obtain the target rights under the target rights category.

[0009] In a second aspect of this invention, an information push method is also provided, comprising:

[0010] Get the target node information for the current content display;

[0011] Send the target node information to the server;

[0012] The system receives a target incentive strategy sent by the server. The target incentive strategy includes target push information and target benefit types. The target incentive strategy is obtained by predicting the user's response to the push information through an incentive intent prediction model.

[0013] The target incentive strategy is displayed at the target node represented by the target node information, and the target incentive strategy is used to obtain the target rights under the target rights category.

[0014] In a third aspect of the present invention, an information push device is also provided, comprising:

[0015] The response prediction module is used to predict whether a user will respond to push information based on the target node information sent by the client and the first historical behavior data corresponding to the client, and obtain the prediction result by using an incentive intent prediction model. The target node information represents the node information of the content display.

[0016] The strategy acquisition module is used to acquire a target incentive strategy when the prediction result indicates that the user should respond to the push information. The target incentive strategy includes the target push information and the target benefit type.

[0017] The strategy push module is used to push the target incentive strategy to the client. The target incentive strategy is used to display the target node represented by the target node information in order to obtain the target rights under the target rights category.

[0018] In a fourth aspect of the present invention, an information push device is also provided, comprising:

[0019] The node information acquisition module is used to acquire the target node information for the currently displayed content;

[0020] A node information sending module is used to send the target node information to the server;

[0021] The strategy receiving module is used to receive the target incentive strategy sent by the server. The target incentive strategy includes target push information and target benefit types. The target incentive strategy is obtained by predicting the user's response to the push information through an incentive intent prediction model.

[0022] The strategy display module is used to display the target incentive strategy at the target node represented by the target node information, and the target incentive strategy is used to obtain the target rights under the target rights type.

[0023] In another aspect of the present invention, an electronic device is also provided, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus.

[0024] Memory, used to store computer programs;

[0025] When a processor executes a program stored in memory, it implements any of the steps described above.

[0026] In another aspect of the present invention, a computer-readable storage medium is also provided, wherein instructions are stored therein, which, when executed on a computer, cause the computer to perform any of the information push methods described above.

[0027] In another aspect of the present invention, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to execute any of the information push methods described above.

[0028] The information push method, apparatus, electronic device, and storage medium provided in this invention, based on target node information sent by the client and first historical behavior data, predict whether the user will respond to the push information through an incentive intent prediction model, obtain a prediction result, and when the prediction result indicates that the user should respond to the push information, obtain a target incentive strategy and push the target incentive strategy to the client. The target incentive strategy can be displayed on the target node of the client. By combining the target node information when predicting whether the user will respond to the push information, the degree of acceptance of the user at the target node can be accurately predicted. The target node information represents the push timing, thereby improving the accuracy of the push timing. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0030] Figure 1 This is a flowchart of an information push method provided in an embodiment of the present invention;

[0031] Figure 2 This is a flowchart of an information push method provided in an embodiment of the present invention;

[0032] Figure 3 This is a system implementation block diagram of the information push method provided in the embodiments of the present invention;

[0033] Figure 4 This is a schematic diagram of the structure of an information push device provided in an embodiment of the present invention;

[0034] Figure 5 This is a schematic diagram of the structure of an information push device provided in an embodiment of the present invention;

[0035] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0036] The technical solutions of the present invention will now be described with reference to the accompanying drawings in the embodiments of the present invention.

[0037] This application's embodiment constructs a "value substitution chain": replacing "paid subscription" behavior with "ad viewing" behavior, and replacing "membership benefits" with "trial viewing rights." Its core logic is as follows: users can exchange certain platform service rights by completing advertising tasks set by the platform; the system determines whether the incentive cost of a particular campaign is lower than the estimated loss from user churn or the original user's revenue; AI behavioral modeling is used to identify whether the user's current behavior offers a chance for "recovery"; the rights distribution record is linked to the next user behavior, achieving link attribution. The specific solution is as follows:

[0038] Figure 1 This is a flowchart of an information push method provided by an embodiment of the present invention. The information push method can be executed by electronic devices such as servers. The information push method can be applied to any of the following scenarios: a reactivation delivery system for silent or marginal users in a video platform; a user churn warning and incentive module in a game application, combined with a mechanism to continue playing the game by receiving gift packs or watching advertisements; and a retention strategy system in information products that implements "reading for benefits".

[0039] like Figure 1 As shown, the information push method may include:

[0040] Step 110: Based on the target node information sent by the client and the first historical behavior data corresponding to the client, predict whether the user will respond to the push information through the incentive intent prediction model to obtain the prediction result. The target node information represents the node information of the content display.

[0041] The target node information can include at least one of the following: a time point, a content location, etc. The time point can be the moment the video plays. For example, the target node information can include at least one of the following: before content playback, in the middle of content (such as an episode), after content completion, or a dwell time exceeding X seconds (X is a preset value). The target node information can be determined based on the user's historical behavior data. For example, if a user prefers to interact with push notifications or exit the video when it reaches a target time point, that target time point can be determined as the target node information. The first historical behavior data is the client's historical behavior data obtained before receiving the target node information sent by the client, and can include content browsing time, push notification clicks, video completion rate, page dwell time, login frequency, bounce rate, etc. The push notification can be advertisements or other information.

[0042] The client triggers a call to the judgment interface at the target node to determine whether the user should respond to the push notification, i.e., the target node information is sent to the server. After receiving the target node sent by the client, the server obtains the first historical behavior data corresponding to the client, and inputs the first historical behavior data and the target node information into the incentive intent prediction model. The incentive intent prediction model then predicts whether the user will respond to the push notification, and obtains the prediction result.

[0043] The incentive intent prediction model can be a decision tree model, such as LightGBM or XGBoost, or a neural network model, such as an MLP (Multilayer Perceptron). It is a binary classification model targeting whether or not to respond to push notifications, trained using labeled data including user historical behavior data, historical node information, and whether the user responded to the push notifications. Input features can include user historical behavior data, historical node information, and device type (e.g., mobile phone, computer, tablet, etc.). During training, the precision and recall of the incentive intent prediction model can be evaluated. Training is considered complete when precision and recall meet requirements, and the trained model is then used for online prediction.

[0044] Step 120: When the prediction result indicates that the user should respond to the push notification, obtain the target incentive strategy, which includes the target push notification and the target benefit type.

[0045] The types of target benefits may include trial viewing benefits, VIP benefits (membership benefits), points benefits, etc.

[0046] User responses to push notifications can include clicking on or viewing the notification. When the prediction is that the user will respond to the push notification, a targeted incentive strategy that aligns with the user's interests can be derived based on historical user behavior data. For example, if the prediction is that the user will respond to the push notification, a user behavior profile corresponding to the client can be retrieved from a user profile database. Then, push resources corresponding to the user behavior profile can be matched from existing push resources. Furthermore, a target benefit category corresponding to the user behavior profile can be matched from all benefit categories. This generates the targeted push notification, and the target push notification and target benefit category serve as the target incentive strategy.

[0047] Step 130: Push the target incentive strategy to the client. The target incentive strategy is used to display the target node represented by the target node information in order to obtain the target rights under the target rights category.

[0048] After obtaining the target incentive strategy, the target incentive strategy is sent to the client. The client can display the target incentive strategy at the target node, that is, display the target push information and the types of target benefits. This can attract users to view the material resources corresponding to the target push information, and then obtain the target benefits under the target benefit types, thereby incentivizing users to continue using the client and thus increasing user activity.

[0049] The information push method provided in this embodiment of the invention, based on target node information sent by the client and first historical behavior data, predicts whether the user will respond to the push information through an incentive intent prediction model, obtains the prediction result, and when the prediction result indicates that the user should respond to the push information, obtains the target incentive strategy and pushes the target incentive strategy to the client. The target incentive strategy can be displayed on the target node of the client. When predicting whether the user will respond to the push information, the target node information is combined for prediction, which can accurately predict the user's acceptance level of the push information at the target node. The target node information represents the push timing, thereby improving the accuracy of the push timing.

[0050] Based on the above technical solution, after pushing the target incentive strategy to the client, the method may further include: acquiring user behavior data related to the target push information; determining the target benefits under the target benefit category based on the user behavior data; and allocating the target benefits to the client.

[0051] Among them, user behavior data for targeted push information is user behavior data on the targeted push information or the material resources corresponding to the targeted push information. It can include behavioral data such as exposure, clicks, and completion. For material resources of online shopping platforms, user behavior data for targeted push information can also include behavioral data such as purchases.

[0052] The client can record user behavior data such as exposure, clicks, and completion rates corresponding to the target push notification, and send this data to the server. The server can then determine the revenue generated by this user behavior data (such as advertising revenue from user clicks or views, or advertising revenue from user purchases). Furthermore, under the target benefit category, the server can determine the target benefit whose cost matches the push notification's revenue and allocate it to the client, thus incentivizing continued user engagement. Cost refers to the cost of the target benefit. Cost-revenue matching means that the difference between the cost and the push notification's revenue is within a preset range.

[0053] By analyzing user behavior data based on targeted push notifications, the benefits derived from this data can be assessed. This allows for the determination of target benefits within specific benefit categories, thus providing a dynamic benefit / cost balance mechanism that can accurately evaluate ROI (Return on Investment).

[0054] Based on the above technical solution, after allocating the target rights to the client, it may further include: continuously collecting the client's usage behavior data, which is used as training data for incremental training of the incentive intention prediction model.

[0055] Among them, the client usage behavior data is the user behavior data after the target benefits are issued to the client.

[0056] After distributing the target benefits to the client, user behavior data is continuously collected. This data can include next-day visits, new visit depth, and bounce rate changes. Continuous collection of this data allows for tracking user behavior changes and generating user behavior change reports. This enables the evaluation of the incentive effect on the user. Specifically, the incentive effect is determined based on the user's behavior data regarding the target benefits and the revenue from push notifications. If the cost of the user's behavior data regarding the target benefits does not exceed the revenue from push notifications, the incentive (sending the target benefits to the user based on their behavior data regarding the target push notification) is considered positive. If the cost of the user's behavior data regarding the target benefits exceeds the revenue from push notifications, the incentive is considered negative. In addition to evaluating the incentive effect on individual users, the effect can also be evaluated at fixed intervals (e.g., daily, weekly) for all users incentivized within a target time period. This evaluation can be assessed using metrics such as increased DAU (Daily Active Users), improved conversion rates, and changes in retention. When the incentive effect on all users meets the incremental training conditions (such as conversion rate lower than target conversion rate, DAU lower than target DAU, etc.), user behavior data can be fed back into the model training process. This allows for incremental training of the incentive intent prediction model and the large model (used to obtain target incentive strategies based on user behavior profiles). Alternatively, user behavior data can be used for strategy A / B testing, comparing the incentive effect of the information push method provided in this embodiment with the incentive effect of other methods. Specifically, during incremental training of the large model, the incentive effect on individual users can be labeled as positive or negative, and the incentive strategy and user behavior profile with this label can be saved to the knowledge base corresponding to the large model. This allows the large model to learn the relationship between user behavior profiles, incentive strategies, and incentive effect labels, improving the accuracy of the incentive strategies determined by the large model. When incrementally training the incentive intent prediction model, the usage behavior data and node information corresponding to the positive incentive effect on a single user can be defined as "user responds to push information", and the usage behavior data of a single user with a negative incentive effect or the user does not respond after being pushed information, along with the node information corresponding to the label, can be defined as "user does not respond to push information". Based on this data, the incentive intent prediction model can be incrementally trained to make the prediction results of the incentive intent prediction model more accurate.

[0057] By continuously collecting user behavior data after the target benefits are distributed to the client, the user status can be dynamically tracked. Based on the user behavior data, the incentive value can be judged in real time. Incremental training of incentive intent prediction models and large models can be performed based on the user behavior data, forming a complete closed-loop system of incentive behavior → user feedback → strategy optimization, so as to further improve the accuracy of incentive strategies and increase user activity.

[0058] Based on the above technical solution, the step of obtaining the target incentive strategy, i.e., step 120 above, may include: obtaining the target incentive strategy through a large model based on the target user behavior profile, wherein the large model corresponds to a knowledge base including the user behavior profile and the incentive strategy.

[0059] The knowledge base stores the correspondence between user behavior profiles and incentive strategies, which facilitates the large model to learn the correspondence and determine the target incentive strategies that match user interests.

[0060] User behavior profiles are determined based on users' historical behavioral data and can include payment willingness scores, push notification acceptance scores, content preference categories, churn risk levels, push notification interaction habits, and behavioral change trends. A time window sliding algorithm can be used to construct multi-dimensional user behavior sequences, such as behavioral trends over the past 7, 30, and 90 days. A target number of days can be used as the time window length, and behavioral data (such as clicks and browsing) within each period (e.g., daily) can be collected and arranged chronologically to obtain the multi-dimensional user behavior sequence. After constructing the user behavior profile, it can be written to a user profile database for real-time querying by the model service and strategy judgment modules. Payment willingness score refers to the user's rating of purchasing rights (such as membership rights, points rights, etc.). Push notification acceptance score refers to the user's acceptance of push notifications, such as whether they open the notification or purchase the item associated with it. Content preference categories refer to the user's preferred content categories, such as daily necessities and clothing under the shopping category. Churn risk level refers to the rating of whether a user will no longer log in to the target platform. Push notification interaction habits refer to users' interactive behaviors with push notifications, such as clicking on push notifications, the length of time they watch the push notification resources after clicking on them, and purchasing items corresponding to the push notifications.

[0061] The target user behavior profile corresponding to the current client can be obtained from the profile database. The target user behavior profile is then input into the large model. Based on the learning of the correspondence between user behavior profiles and incentive strategies in the knowledge base, the large model determines the target incentive strategy corresponding to the target user behavior profile.

[0062] By using a large model to determine target incentive strategies based on target user behavior profiles, we can identify target incentive strategies that align with user interests and improve the accuracy of incentive strategies.

[0063] In some embodiments of the present invention, obtaining the target incentive strategy based on the target user behavior profile through a large model may include: obtaining push information under the content preference category in the target user behavior profile; inputting the target user behavior profile and the push information into a large model, and determining the target incentive strategy through the large model.

[0064] The target user behavior profile includes content preference categories, which represent the categories of push information that users are interested in, such as daily necessities and clothing under the shopping category, or the preferred video types (such as TV series, TV programs, short videos, etc.) under the video viewing category.

[0065] The system retrieves push information under content preference categories from the target user behavior profile in the database. The target user behavior profile and push information are then input into a large model. Based on learning the correspondence between user behavior profiles, push information and incentive strategies stored in the knowledge base, the large model determines the target incentive strategy that matches the user's interests.

[0066] By inputting push notifications under content preference categories and user behavior profiles into a large model, the accuracy of determining target incentive strategies can be further improved.

[0067] Based on the above technical solution, before obtaining the target incentive strategy through a large model according to the user behavior profile, the method may further include: determining the user behavior profile corresponding to the client based on the historical behavior data corresponding to the client.

[0068] Historical behavioral data may include content browsing time, push notification clicks, video completion rate, page dwell time, login frequency, bounce rate, and other data.

[0069] In one optional implementation, historical behavioral data corresponding to the client can be statistically analyzed to determine a user behavior profile. This user behavior profile can include a willingness-to-pay score, push notification acceptance score, content preference categories, churn risk level, and push notification interaction habits; it can also include behavioral trends over target days (e.g., 7 days, 30 days, 90 days). For example, the proportion of time a user spends after purchasing benefits within their total usage time can be used as a willingness-to-pay score based on their historical behavioral data; and the proportion of times a user interacts with push notifications out of all push notifications can be used as a push notification acceptance score.

[0070] In another alternative implementation, the historical behavior data corresponding to the client can be input into a large language model. The large language model can then summarize, identify, and process the historical behavior data to determine the user behavior profile corresponding to the client. The user behavior profile can be stored in a user profile database for use when pushing information to users.

[0071] By pre-determining the user behavior profile corresponding to the client based on the client's historical behavior data, it can be directly used when pushing incentive strategies to users, thereby improving the efficiency of incentive strategy determination.

[0072] Based on the above technical solution, the target incentive strategy also includes a target display method, which is used to instruct the client to display the entry information of the target incentive strategy at the target node in the target display method;

[0073] After pushing the target incentive strategy to the client, the method may further include: in response to the client's request for material acquisition based on the target incentive strategy, sending the material resources corresponding to the target push information to the client.

[0074] After the server sends the target incentive strategy to the client, the client can display the entry information of the target incentive strategy in a target display mode on the target node. The target display mode refers to the way the entry information of the target incentive strategy is displayed. For example, the target display mode can include a floating layer, a pop-up window, or a pre-viewing interception.

[0075] The entry information of the target incentive strategy displayed in the client can be clicked by the user. When the client receives the user's click or other operation, it can send a material retrieval request to the server through the SDK (such as the advertising SDK). Based on the client's material retrieval request, the server retrieves the material resources corresponding to the target push information from the database and sends the material resources corresponding to the target push information to the client. The client can display the material resources. For example, when the material resources are videos, the client can play the videos.

[0076] By displaying the entry information of the target incentive strategy in the client in a target-oriented manner, users can be guided into the incentive process without interrupting their normal usage flow, and user aversion can be reduced through lightweight interaction.

[0077] Based on the above technical solution, the method may further include: when the incremental training conditions are met, using second historical behavior data, historical display nodes and historical user response data to perform incremental training on the incentive intent prediction model.

[0078] The second set of historical behavior data consists of the historical behavior data of each user before this incremental training, which may include data such as content browsing time, push notification clicks, video completion rate, page dwell time, login frequency, and bounce rate. Historical user response data is the user's response information to the historical incentive strategies pushed to them, which may include whether they responded or not.

[0079] The incremental training conditions include periodic or model evaluation metrics failing to meet target conditions. Model evaluation metrics may include convergence, recall coverage, and incentive effect on all users. Convergence may include the proportion of users with a positive incentive effect. Target conditions may include at least one of the following: the proportion of users with a positive incentive effect among all incentivized users is less than the target proportion, recall coverage is less than a threshold, conversion rate is lower than the target conversion rate, and DAU is lower than the target DAU. Recall coverage refers to the ratio of the number of users who frequently use the client after being incentivized to the total number of incentivized users.

[0080] When the incremental training conditions are met, the second historical behavior data and historical display nodes are input into the incentive intention prediction model. Based on the output results of the incentive intention prediction model and the historical user response data, the parameters of the incentive intention prediction model are readjusted to perform incremental training on the incentive intention prediction model, which can ensure the prediction effect of the incentive intention prediction model.

[0081] The strategy engine can also periodically evaluate the effectiveness of different incentive strategies, such as whether the combination of push resource types and benefit types is optimal, providing a basis for determining target incentive strategies.

[0082] Based on the above technical solution, before predicting whether a user will respond to push information using an incentive intent prediction model based on the target node information sent by the client and the first historical behavior data corresponding to the client, the method may further include: processing the original behavior data corresponding to the client through a stream processing system to obtain structured first historical behavior data, wherein the original behavior data is collected by the client based on embedded points.

[0083] Key user behaviors can be tracked in the client, such as content browsing time, push notification clicks, video completion rate, page dwell time, login frequency, and bounce rate. Raw user behavior data is collected and reported to the server's log system in real time via a unified data reporting protocol built using an SDK. The server can then use stream processing systems such as Kafka / Flink to parse, clean, and normalize the raw behavior data to obtain structured first-level historical behavior data.

[0084] By processing the raw behavioral data corresponding to the client through a stream processing system, the raw behavioral data sent by the client can be processed in real time, providing a data foundation for the construction of user behavior profiles and the training of models.

[0085] Before predicting whether a user responds to push notifications based on the target node information sent by the client and the first historical behavior data corresponding to the client using the incentive intent prediction model, the method may further include: performing binary classification training on the initial incentive intent prediction model based on sample data to obtain a trained incentive intent prediction model, wherein the sample data includes sample behavior data, sample node information, and labeled data, and the labeled data includes whether the user responds to push notifications or not.

[0086] The incentive intent prediction model can be a decision tree model, such as LightGBM or XGBoost, or a neural network model, such as MLP. When training the incentive intent prediction model, sample data is used for binary classification training with the objective of "whether to respond to push notifications". The input features of the incentive intent prediction model can include user sample behavior data, sample node information, and device type (such as mobile phone, computer, tablet, etc.). During the training process, the precision and recall of the incentive intent prediction model can be evaluated. Training is considered complete when precision and recall meet the requirements, and the trained incentive intent prediction model is then used for online prediction.

[0087] Taking a neural network-based incentive intent prediction model as an example, during the initial training of the incentive intent prediction model, sample behavior data and sample node information from the sample data can be input into the initial incentive intent prediction model to obtain the prediction result of whether the user will respond to the push notification. Based on this prediction result and labeled data, the parameters of the initial incentive intent prediction model are adjusted, and this training process is iteratively executed until the training termination condition is met, at which point training ends, and the trained incentive intent prediction model is obtained. The training termination condition may include the model's precision being greater than or equal to the target precision, the model's recall being greater than or equal to the target recall, or the model's parameters converging, etc.

[0088] By training the initial incentive intent prediction model using binary classification based on sample data, the incentive intent prediction model can correctly determine whether a user responds to push notifications under the corresponding node information, thereby improving the accuracy of push timing judgment.

[0089] Figure 2This is a flowchart of an information push method provided by an embodiment of the present invention. The information push method can be executed by electronic devices such as mobile phones, computers, and tablets as clients. The information push method can be applied to any of the following scenarios: a reactivation delivery system for silent or marginal users in a video platform; a user churn warning and incentive module in a game application, combined with a mechanism to continue playing the game by receiving gift packs or watching advertisements; and a retention strategy system in information products that implements "reading for benefits".

[0090] like Figure 2 As shown, the information push method may include:

[0091] Step 210: Obtain the target node information for the current content display.

[0092] The target node information can include at least one of the following: a time point, a content location, etc. The time point can be the moment the video plays. For example, target node information can include at least one of the following: before content playback, in the middle of content (such as an episode), after content completion, or a dwell time exceeding X seconds (X is a preset value). Target node information can be determined based on the user's historical behavior data. For example, if a user prefers to interact with push notifications or exit the video when it reaches a target time point, that target time point can be determined as the target node information. The first historical behavior data is the client's historical behavior data obtained by the server before receiving the target node sent by the client, and can include content browsing duration, push notification clicks, video completion rate, page dwell time, login frequency, bounce rate, etc. Push notifications can be advertisements or other information.

[0093] When the current content displayed on the client reaches or is about to reach the target node, the information of that target node can be obtained.

[0094] Step 220: Send the target node information to the server.

[0095] When the current content displayed on the client reaches the target node, a judgment interface is triggered to determine whether the user should respond to the push notification, i.e., the target node information is sent to the server. After receiving the target node information sent by the client, the server obtains the client's corresponding first historical behavior data, and inputs the first historical behavior data and the target node information into the incentive intent prediction model. The incentive intent prediction model then predicts whether the user will respond to the push notification, obtaining the prediction result. The first historical behavior data refers to the client's historical behavior data obtained before receiving the target node information sent by the client, and may include content browsing duration, push notification clicks, video completion rate, page dwell time, login frequency, bounce rate, etc. Push notifications may be advertisements or other similar information.

[0096] The incentive intent prediction model can be a decision tree model, such as LightGBM or XGBoost, or a neural network model, such as an MLP (Multilayer Perceptron). It is a binary classification model targeting whether or not to respond to push notifications, trained using labeled data including user historical behavior data, historical node information, and whether the user responded to the push notifications. Input features can include user historical behavior data, historical node information, and device type (e.g., mobile phone, computer, tablet, etc.). During training, the precision and recall of the incentive intent prediction model can be evaluated. Training is considered complete when precision and recall meet requirements, and the trained model is then used for online prediction.

[0097] Step 230: Receive the target incentive strategy sent by the server. The target incentive strategy includes target push information and target benefit types. The target incentive strategy is obtained by predicting the user's response to the push information through an incentive intent prediction model.

[0098] The types of target benefits may include trial viewing benefits, VIP benefits (membership benefits), points benefits, etc.

[0099] User responses to push notifications can include clicking or viewing them. When the server predicts a user response based on an incentive prediction model, it can obtain a target incentive strategy that aligns with the user's interests based on historical behavioral data. For example, if the prediction is a user response, the server can retrieve the user's behavioral profile from a user profile database, match push resources corresponding to the user's behavioral profile from existing resources, match target benefit types from all benefit types, generate target push notifications corresponding to the user's behavioral profile, and use the target push notifications and target benefit types as the target incentive strategy. The user behavioral profile is determined based on the user's historical behavioral data and can include payment willingness ratings, push notification acceptance ratings, content preference categories, churn risk levels, push notification interaction habits, and behavioral change trends. A time window sliding algorithm can be used to construct a multi-dimensional user behavior sequence, such as behavioral trends over the last 7, 30, and 90 days. The target number of days can be used as the time window length, and behavioral data (such as clicks and browsing) can be collected for each period (e.g., daily) within the time window and arranged chronologically to obtain the multi-dimensional user behavior sequence. After constructing user behavior profiles, these profiles can be written to a user profile database for real-time querying by the model service and strategy judgment modules. The willingness to pay score refers to the user's rating of purchased benefits (such as membership benefits, points benefits, etc.). The push notification acceptance score refers to the user's acceptance of push notifications, such as whether they open the notification or purchase the item associated with it. Content preference categories refer to the user's preferred content categories, such as daily necessities and clothing under the shopping category. Streaming risk level refers to the rating of whether a user will no longer log in to the target platform. Push notification interaction habits refer to the user's interaction habits with push notifications, such as clicking on push notifications, the length of time they watch the push notification content after clicking, and purchasing the item associated with the push notification.

[0100] After obtaining the target incentive strategy, the server sends the target incentive strategy to the client, and the client receives the target incentive strategy sent by the server.

[0101] Step 240: Display the target incentive strategy on the target node represented by the target node information. The target incentive strategy is used to obtain the target rights under the target rights category.

[0102] After receiving the target incentive strategy sent by the server, the client can display the target incentive strategy at the target node, that is, display the target push information and the types of target benefits. This can attract users to view the material resources corresponding to the target push information, and then obtain the target benefits under the target benefit types, thereby incentivizing users to continue using the client and increasing user activity.

[0103] The information push method provided in this invention obtains the target node information of the current content display, sends the target node information to the server, receives the target incentive strategy sent by the server, and displays the target incentive strategy at the target node. The target incentive strategy is obtained based on the user behavior profile when the incentive intent prediction model predicts whether the user will respond to the push information. Since the prediction of whether the user will respond to the push information is combined with the target node information, the user's acceptance level of the push information at the target node can be accurately predicted. The target node information represents the push timing, thereby improving the accuracy of the push timing.

[0104] Based on the above technical solution, after the target incentive strategy is displayed at the target node, the method further includes: sending user behavior data for the target push information to the server; and receiving target benefits under the target benefit category sent by the server.

[0105] Among them, user behavior data for targeted push information is user behavior data on the targeted push information or the material resources corresponding to the targeted push information. It can include behavioral data such as exposure, clicks, and completion. For material resources of online shopping platforms, user behavior data for targeted push information can also include behavioral data such as purchases.

[0106] The client can record user behavior data such as exposure, clicks, and completion rates corresponding to the target push notification, and send this data to the server. The server can then determine the revenue generated by this user behavior data (such as advertising revenue from user clicks or views, or advertising revenue from user purchases). Furthermore, under the target benefit category, the server can determine the target benefit whose cost matches the push notification's revenue and allocate it to the client, thus incentivizing continued user engagement. Cost refers to the cost of the target benefit. Cost-revenue matching means that the difference between the cost and the push notification's revenue is within a preset range.

[0107] By sending user behavior data for targeted push notifications to the server, the server can assess the benefits brought by the user behavior data based on the user behavior data, and then determine the target benefits under the target benefit category based on the benefits, thus providing a dynamic benefit / cost balance mechanism that can accurately evaluate ROI.

[0108] Based on the above technical solution, after receiving the target benefits under the target benefit category sent by the server, the method further includes: continuously collecting user behavior data and sending the usage behavior data to the server, wherein the usage behavior data is used as training data for incremental training of the incentive intention prediction model.

[0109] Among them, user behavior data refers to the user's behavior data when using the client after the server issues the target rights to the client.

[0110] After the server distributes the target benefits to the client, the client continuously collects user behavior data, including next-day visits, new visit depth, and bounce rate changes. This continuous collection and transmission of user behavior data to the server allows for tracking user behavior changes and generating user behavior change reports. This enables the evaluation of the incentive effect on the user. Specifically, the incentive effect is determined based on the user's behavior data regarding the target benefits and the revenue from push notifications. If the cost of the user's behavior data regarding the target benefits does not exceed the revenue from push notifications, the incentive (sending the target benefits to the user based on the user's behavior data regarding the target push notification) is considered positive. If the cost of the user's behavior data regarding the target benefits exceeds the revenue from push notifications, the incentive is considered negative. In addition to evaluating the incentive effect on individual users, the incentive effect can also be evaluated at fixed intervals (e.g., daily, weekly) for all users incentivized within a target time period. This evaluation can be assessed using metrics such as DAU (Daily Active Users), conversion rate improvement, and retention changes. By feeding behavioral data into the model training process, incremental training can be performed on the incentive intent prediction model and the large model (used to obtain target incentive strategies based on user behavior profiles). Alternatively, behavioral data can also be used for strategy A / B testing evaluation, that is, to compare and evaluate the incentive effect of the information push method provided in this embodiment of the invention with the incentive effect of other methods.

[0111] By continuously collecting user behavior data after receiving the target benefits sent by the server and sending this data back to the server, the server can dynamically track user status. Based on the behavior data, the server can determine the incentive value in real time. This allows for incremental training of the incentive intent prediction model, forming a complete closed-loop system of incentive behavior → user feedback → strategy optimization. This further improves the accuracy of incentive strategies and enhances user activity.

[0112] Based on the above technical solution, the target incentive strategy also includes a target display method;

[0113] The step of displaying the target incentive strategy on the target node includes: displaying the entry information of the target incentive strategy on the target node in the target display mode; in response to the triggering operation of the entry information, sending a material acquisition request for the target incentive strategy to the server; receiving the material resources corresponding to the target push information, and displaying the material resources.

[0114] After receiving the target incentive strategy from the server, the client can display the entry information of the target incentive strategy on the target node in a target display manner. The target display method refers to the way the entry information of the target incentive strategy is displayed. For example, the target display method can include a floating layer, a pop-up window, or a pre-viewing interception.

[0115] The entry information of the target incentive strategy displayed in the client can be clicked by the user. When the client receives the user's click or other operation, it can send a material retrieval request to the server through the SDK (such as the advertising SDK). Based on the client's material retrieval request, the server retrieves the material resources corresponding to the target push information from the database and sends the material resources corresponding to the target push information to the client. The client can display the material resources. For example, when the material resources are videos, the client can play the videos.

[0116] By displaying the entry information of the target incentive strategy in the client in a target-oriented manner, users can be guided into the incentive process without interrupting their normal usage flow, and user aversion can be reduced through lightweight interaction.

[0117] Figure 3 This is a system implementation block diagram of the information push method provided in the embodiments of the present invention, such as... Figure 3As shown, the system implementing this information push method can include: a data acquisition layer, a decision-making layer, a rights management module, and a strategy evaluation module. The data acquisition layer can include a client SDK, a log system, and a data cleaning pipeline. The decision-making layer includes a behavior analysis engine, a trigger judgment engine, and an advertising resource management module. Key user behaviors are acquired by embedding tracking points in the client and uploaded to the backend log system in real time via the SDK. The data cleaning pipeline parses, cleans, and normalizes the user behavior data in the log system to form structured behavior data. The behavior analysis engine analyzes user behavior data, generates user behavior profiles, and uses a tag engine to calculate tags based on the user behavior profiles built using the profile service, determining the corresponding profile tags. The trigger judgment engine uses an online model (incentive intent prediction model) to determine whether a user responds to the push notification based on the target node sent by the client and the user's historical behavior data. The advertising resource management module controls the incentive strategy based on the delivery strategy and allocates advertising resources based on the scheduler. The benefits management module, based on the incentive strategies pushed to users and users' benefit participation data (i.e., user behavior data related to the targeted push information), calculates the target benefits through distribution logic, distributes the target benefits to users, tracks the status of the distributed target benefits, and confirms the receipt of benefits through payment logic. The strategy evaluation module, based on user behavior data after benefit distribution, analyzes the incentive effect using an attribution model, and optimizes the feedback system. The evaluation results are fed back to the user behavior analysis engine, the trigger judgment engine for strategy adjustment, and the advertising resource management module for delivery optimization.

[0118] It should be noted that the user-related data used in the embodiments of the present invention, such as historical behavior data and usage behavior data, are all obtained and processed on the basis of user authorization, and the user is provided with a way to revoke authorization.

[0119] Figure 4 This is a schematic diagram of an information push device provided in an embodiment of the present invention. This device is applied to a server, such as... Figure 4 As shown, the device includes:

[0120] The response prediction module 410 is used to predict whether a user will respond to push information based on the target node information sent by the client and the first historical behavior data corresponding to the client, and obtain the prediction result by using an incentive intent prediction model. The target node information represents the node information of the content display.

[0121] The strategy acquisition module 420 is used to acquire a target incentive strategy when the prediction result indicates that the user should respond to the push information. The target incentive strategy includes the target push information and the target benefit type.

[0122] The strategy push module 430 is used to push the target incentive strategy to the client. The target incentive strategy is used to display the target node represented by the target node information in order to obtain the target rights under the target rights type.

[0123] Optionally, the device further includes:

[0124] The data acquisition module is used to acquire user behavior data related to the target push information;

[0125] The rights distribution module is used to determine the target rights under the target rights category based on the user behavior data, and to allocate the target rights to the client.

[0126] Optionally, the device further includes:

[0127] The behavior data collection module is used to continuously collect the user behavior data of the client, which is used as training data for incremental training of the incentive intention prediction model.

[0128] Optionally, the strategy acquisition module is specifically used for:

[0129] Obtain push notifications based on content preference categories from the target user's behavior profile;

[0130] The target user behavior profile and the push information are input into a large model, and the target incentive strategy is determined through the large model. The large model has a knowledge base that includes user behavior profiles and incentive strategies.

[0131] Optionally, the target incentive strategy further includes a target display method, which is used to instruct the client to display the entry information of the target incentive strategy on the target node in the target display method;

[0132] The device further includes:

[0133] The material sending module is used to send the material resources corresponding to the target push information to the client in response to the client's material acquisition request for the target incentive strategy.

[0134] Optionally, the device further includes:

[0135] The incremental training module is used to incrementally train the incentive intent prediction model using second historical behavior data, historical display nodes, and historical user response data when the incremental training conditions are met.

[0136] Optionally, the incremental training conditions include periodic conditions or conditions where the model evaluation metrics do not meet the target criteria.

[0137] Optionally, the device further includes:

[0138] The data processing module is used to process the original behavioral data corresponding to the client through the stream processing system to obtain structured first historical behavioral data, wherein the original behavioral data is collected by the client based on embedded points.

[0139] Optionally, the device further includes:

[0140] The model training module is used to perform binary classification training on the initial incentive intention prediction model based on sample data to obtain the trained incentive intention prediction model. The sample data includes sample behavior data, sample node information, and labeled data. The labeled data includes user response push information or user non-response push information.

[0141] The information push device provided in this embodiment of the invention predicts whether a user will respond to the push information based on target node information sent by the client and first historical behavior data, and obtains a prediction result through an incentive intent prediction model. When the prediction result indicates that the user should respond to the push information, a target incentive strategy is obtained and pushed to the client. The target incentive strategy can be displayed on the target node of the client. When predicting whether a user will respond to the push information, the target node information is combined for prediction, which can accurately predict the user's acceptance level of the push information at the target node represented by the target node information. The target node information represents the push timing, thereby improving the accuracy of the push timing.

[0142] Figure 5 This is a schematic diagram of the structure of an information push device provided in an embodiment of the present invention. This device is applied to a client, such as... Figure 5 As shown, the device includes:

[0143] The node information acquisition module 510 is used to acquire the target node information for the current content display.

[0144] The node information sending module 520 is used to send the target node information to the server;

[0145] The strategy receiving module 530 is used to receive the target incentive strategy sent by the server. The target incentive strategy includes target push information and target benefit types. The target incentive strategy is obtained by predicting the user's response to the push information through an incentive intent prediction model.

[0146] The strategy display module 540 is used to display the target incentive strategy at the target node represented by the target node information, and the target incentive strategy is used to obtain the target rights under the target rights type.

[0147] The optional device also includes:

[0148] The first data sending module is used to send user behavior data for the target push information to the server;

[0149] The rights receiving module is used to receive the target rights under the target rights category sent by the server.

[0150] Optionally, the device further includes:

[0151] The second data sending module is used to continuously collect user behavior data and send the user behavior data to the server. The user behavior data is used as training data for incremental training of the incentive intention prediction model.

[0152] Optionally, the target incentive strategy may also include a target display method;

[0153] The strategy display module includes:

[0154] An entry display unit is used to display the entry information of the target incentive strategy at the target node in the target display manner.

[0155] The request sending unit is configured to send a material acquisition request for the target incentive strategy to the server in response to a triggering operation on the entry information.

[0156] The material display unit is used to receive material resources corresponding to the target push information and display the material resources.

[0157] The information push method provided in this invention obtains the target node information of the current content display, sends the target node information to the server, receives the target incentive strategy sent by the server, and displays the target incentive strategy at the target node. The target incentive strategy is obtained based on the user behavior profile when the incentive intent prediction model predicts whether the user will respond to the push information. Since the prediction of whether the user will respond to the push information is combined with the target node information, the user's acceptance level of the push information at the target node can be accurately predicted. The target node information represents the push timing, thereby improving the accuracy of the push timing.

[0158] This invention also provides an electronic device, such as... Figure 6 As shown, it includes a processor 601, a communication interface 602, a memory 603, and a communication bus 604, wherein the processor 601, the communication interface 602, and the memory 603 communicate with each other through the communication bus 604.

[0159] Memory 603 is used to store computer programs;

[0160] When processor 601 executes a program stored in memory 603, it performs the following steps:

[0161] Based on the target node information sent by the client and the first historical behavior data corresponding to the client, the incentive intent prediction model is used to predict whether the user will respond to the push information, and the prediction result is obtained. The target node information represents the node information of the content display.

[0162] When the prediction result indicates that the user should respond to the push notification, a target incentive strategy is obtained, the target incentive strategy including the target push notification and the target benefit type;

[0163] The target incentive strategy is pushed to the client. The target incentive strategy is used to display the target node represented by the target node information in order to obtain the target rights under the target rights category.

[0164] Alternatively, perform the following steps:

[0165] Get the target node information for the current content display;

[0166] Send the target node information to the server;

[0167] The system receives a target incentive strategy sent by the server. The target incentive strategy includes target push information and target benefit types. The target incentive strategy is obtained by predicting the user's response to the push information through an incentive intent prediction model.

[0168] The target incentive strategy is displayed at the target node represented by the target node information, and the target incentive strategy is used to obtain the target rights under the target rights category.

[0169] The communication bus mentioned in the above electronic devices can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not indicate that there is only one bus or one type of bus.

[0170] The communication interface is used for communication between the aforementioned electronic devices and other devices.

[0171] The memory may include random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0172] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0173] In another embodiment of the present invention, a computer-readable storage medium is also provided, which stores instructions that, when executed on a computer, cause the computer to perform any of the information push methods described in the above embodiments.

[0174] In another embodiment of the present invention, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to execute any of the information push methods described in the above embodiments.

[0175] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid state disk (SSD)).

[0176] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0177] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0178] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of protection of the present invention.

Claims

1. An information push method characterized by comprising: include: Based on the target node information sent by the client and the first historical behavior data corresponding to the client, the incentive intent prediction model is used to predict whether the user will respond to the push information, and the prediction result is obtained. The target node information represents the node information of the content display. When the prediction result indicates that the user should respond to the push notification, a target incentive strategy is obtained, the target incentive strategy including the target push notification and the target benefit type; The target incentive strategy is pushed to the client. The target incentive strategy is used to display the target node represented by the target node information in order to obtain the target rights under the target rights category.

2. The method of claim 1, wherein, After pushing the target incentive strategy to the client, the method further includes: Obtain user behavior data related to the target push information; Based on the user behavior data, target benefits under the target benefit category are determined, and the target benefits are allocated to the client.

3. The method of claim 2, wherein, After allocating the target benefits to the client, the process further includes: The user behavior data of the client is continuously collected and used as training data for incremental training of the incentive intention prediction model.

4. The method of claim 1, wherein, The strategy for obtaining the target incentive includes: Obtain push notifications based on content preference categories from the target user's behavior profile; The target user behavior profile and the push information are input into a large model, and the target incentive strategy is determined through the large model. The large model has a knowledge base that includes user behavior profiles and incentive strategies.

5. The method according to any one of claims 1 to 4, characterized in that, The target incentive strategy also includes a target display method, which is used to instruct the client to display the entry information of the target incentive strategy at the target node in the target display method; After pushing the target incentive strategy to the client, the method further includes: In response to the client's request to acquire materials for the target incentive strategy, the material resources corresponding to the target push information are sent to the client.

6. The method according to any one of claims 1 to 4, characterized in that, Also includes: When the incremental training conditions are met, the incentive intent prediction model is incrementally trained using the second historical behavior data, historical display nodes, and historical user response data.

7. The method of claim 6, wherein, The incremental training conditions include periodic or model evaluation metrics not meeting the target conditions.

8. The method according to any one of claims 1 to 4, characterized in that, Before predicting whether a user will respond to a push notification using an incentive intent prediction model based on the target node information sent by the client and the first historical behavior data corresponding to the client, the method further includes: The original behavioral data corresponding to the client is processed by the stream processing system to obtain structured first historical behavioral data, which is collected by the client based on embedded points.

9. The method according to any one of claims 1 to 4, characterized in that, Before predicting whether a user will respond to a push notification using an incentive intent prediction model based on the target node information sent by the client and the first historical behavior data corresponding to the client, the method further includes: Based on sample data, the initial incentive intention prediction model is trained for binary classification, and a trained incentive intention prediction model is obtained. The sample data includes sample behavior data, sample node information, and labeled data. The labeled data includes user response push information or user non-response push information.

10. An information push method characterized by comprising: The method comprises: obtaining target node information of current content display; sending the target node information to a server; receiving a target incentive strategy sent by the server, the target incentive strategy including target push information and a target benefit category, the target incentive strategy being obtained when the incentive intention prediction model predicts that the user will respond to the push information; displaying the target incentive strategy at a target node represented by the target node information, the target incentive strategy being used to obtain a target benefit of the target benefit category.

11. The method of claim 10, wherein, After displaying the target incentive strategy at the target node, the method further comprises: sending user behavior data for the target push information to the server; receiving a target benefit of the target benefit category sent by the server.

12. The method of claim 11, wherein, After receiving the target benefit of the target benefit category sent by the server, the method further comprises: continuously collecting user usage behavior data and sending the usage behavior data to the server, the usage behavior data being used as training data for incremental training of the incentive intention prediction model.

13. The method according to any one of claims 10-12, characterized in that, The target incentive strategy further includes a target display mode. Displaying the target incentive strategy at the target node represented by the target node information comprises: displaying entry information of the target incentive strategy at the target node in the target display mode; in response to a triggering operation on the entry information, sending a material acquisition request for the target incentive strategy to the server; receiving a material resource corresponding to the target push information and displaying the material resource.

14. An information push apparatus characterized by comprising: The method comprises: a response prediction module configured to predict, based on target node information sent by a client and first historical behavior data corresponding to the client, whether a user will respond to push information by using an incentive intention prediction model, to obtain a prediction result, the target node information representing node information of content display; a strategy acquisition module configured to, when the prediction result indicates that the user will respond to push information, acquire a target incentive strategy, the target incentive strategy including target push information and a target benefit category; a strategy push module configured to push the target incentive strategy to the client, the target incentive strategy being used to display at a target node represented by the target node information to obtain a target benefit of the target benefit category.

15. An information push apparatus, characterized by comprising: The method comprises: a node information acquisition module configured to acquire target node information of current content display; a node information sending module configured to send the target node information to a server; a strategy receiving module configured to receive a target incentive strategy sent by the server, the target incentive strategy including target push information and a target benefit category, the target incentive strategy being obtained when an incentive intention prediction model predicts that a user will respond to push information; a strategy receiving module configured to receive a target incentive strategy sent by the server, the target incentive strategy including target push information and a target benefit category, the target incentive strategy being obtained when an incentive intention prediction model predicts that a user will respond to push information; A strategy display module is configured to display the target incentive strategy at a target node represented by the target node information, the target incentive strategy being used to obtain a target right of a target right type.

16. An electronic device, comprising: The computer device comprises a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; The memory is configured to store a computer program; The processor is configured to execute the program stored in the memory, and implement the method steps in any one of claims 1-8 or 9-12.

17. A computer readable storage medium having stored thereon a computer program, characterized in that The program is executed by the processor to implement the method in any one of claims 1-9 or 10-13.

18. A computer program product comprising computer program instructions, characterised in that, The computer program instructions, when running on the computer, cause the computer to perform the method in any one of claims 1-9 or 10-13.

Images