Resource niche traffic distribution method, device and equipment, readable storage medium and computer program product

Abstract

The invention relates to the technical field of media traffic distribution, and provides a resource niche traffic distribution method, device and equipment, a readable storage medium and a computer program product. The method comprises the steps of obtaining a traffic estimation value of a target time period according to historical traffic decomposition data and a traffic estimation model; according to a target user preference factor corresponding to the target time period and a target traffic distribution proportion between the operation plans, under the condition that a user click rate threshold and a traffic contract display amount are satisfied, distributing the traffic estimated value of the target time period, and obtaining a first traffic distribution target value of each type of content in the target time period; wherein different operation plans correspond to different types of contents; and for any type of content, determining a plurality of resource nichs of any type of content, and obtaining a second traffic distribution target value of the plurality of resource nichs in the target time period according to the first traffic distribution value of any type of content in the target time period and the traffic matching model. By adopting the method, the flow utilization rate and the user click rate can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of media traffic allocation, and in particular to a method, apparatus, computer equipment, computer-readable storage medium and computer program product for allocating traffic of resource positions. Background Art

[0002] Traffic is a valuable resource on applications. Traffic allocation plays a vital role in operations management and directly affects the growth and success of the business. Operations need to retain and convert traffic to promote good traffic allocation of applications while ensuring revenue.

[0003] At present, traffic is mainly allocated to each resource position based on empirical rules. However, the existing resource position traffic allocation method lacks dynamic allocation, the traffic utilization rate is not high, and the user click rate is low. Summary of the invention

[0004] Based on this, it is necessary to provide a resource bit traffic allocation method, device, computer equipment, computer readable storage medium and computer program product to address the above technical problems.

[0005] In a first aspect, the present application provides a method for allocating resource traffic, including:

[0006] Based on the historical traffic decomposition data and traffic estimation model, the traffic estimation value for the target period is obtained;

[0007] According to the target user preference factor corresponding to the target time period and the target traffic allocation ratio between the operation plans, the traffic estimate value of the target time period is allocated under the condition that the user click rate threshold and the traffic contract display volume are met, so as to obtain the first traffic allocation target value of each type of content in the target time period; wherein different operation plans correspond to different types of content;

[0008] For any type of content, a number of resource locations of the content are determined, and a second flow allocation target value of the resource locations in the target time period is obtained according to a first flow allocation value of the content in the target time period and a flow matching model.

[0009] In one embodiment, before obtaining the traffic estimation value of the target period according to the historical traffic decomposition data and the traffic estimation model, the method further includes:

[0010] Perform feature decomposition on historical traffic data to obtain seasonal component data, trend component data and random component data;

[0011] The historical traffic decomposition data is obtained according to the seasonal component data, trend component data and random component data.

[0012] In one embodiment, the step of obtaining the target user preference factor corresponding to the target time period and the target traffic allocation ratio between the operation plans includes:

[0013] Obtaining a flow difference value of the previous period according to the flow estimate value of the previous period of the target period and the actual flow value of the previous period;

[0014] When the traffic difference value of the previous time period is less than the difference threshold and the operation plans have not changed, the initial user preference factor and the initial traffic allocation ratio between the operation plans are used as the target user preference factor corresponding to the target time period and the target traffic allocation ratio between the operation plans;

[0015] When the traffic difference value of the previous time period is greater than the difference threshold, or when any operation plan changes, the initial user preference factor and the initial traffic allocation ratio between the operation plans are updated to obtain the target user preference factor corresponding to the target time period and the target traffic allocation ratio between the operation plans.

[0016] In one embodiment, the updating of the initial user preference factor and the initial traffic allocation ratio between the operation plans includes:

[0017] For any operation plan, obtain the traffic contract display volume of any operation plan and the traffic display volume up to the target period;

[0018] According to the traffic contract display volume and the traffic displayed volume, the initial traffic allocation ratio of any operation plan is updated;

[0019] The initial user preference factor is updated according to the updated initial traffic allocation ratio.

[0020] In one embodiment, according to the target user preference factor corresponding to the target time period and the target traffic allocation ratio between the operation plans, the traffic estimate value of the target time period is allocated while satisfying the user click rate threshold and the traffic contract display volume to obtain the first traffic allocation target value of each type of content in the target time period, including:

[0021] According to the target user preference factor corresponding to the target time period and the target traffic allocation ratio between the operation plans, a first traffic allocation initial value of each type of content in the target time period is obtained;

[0022] If the initial value of the first traffic allocation of each type of content in the target time period does not meet the traffic contract display volume of each type of content, then optimize the initial value of the first traffic allocation of each type of content in the target time period until the traffic contract display volume of each type of content is met, and obtain the intermediate value of the first traffic allocation of each type of content in the target time period;

[0023] According to the first traffic distribution median value of each type of content in the target time period and the user click rate prediction model, obtaining a user click rate prediction value corresponding to the first traffic distribution median value; the user click rate prediction model is obtained by combining a deep factor decomposition machine model and a weighted loss function;

[0024] If the user click rate estimate is not greater than the user click rate threshold, the first traffic allocation intermediate value of each type of content in the target time period is optimized until the user click rate estimate is greater than the user click rate threshold, thereby obtaining the first traffic allocation target value of each type of content in the target time period.

[0025] In one embodiment, for any type of content, determining a number of resource bits of the any type of content, and obtaining a second traffic allocation target value of the number of resource bits in the target time period according to a first traffic allocation value of the any type of content in the target time period and a traffic matching model, includes:

[0026] Inputting the first traffic allocation value of any type of content in the target time period into a traffic matching model to obtain a second traffic allocation initial value of the plurality of resource bits in the target time period; the traffic matching model is a deep semantic matching model including a convolutional neural network layer;

[0027] According to the second traffic allocation initial values ​​of the plurality of resource positions in the target time period and the user conversion rate estimation model, obtaining a user conversion rate estimation value corresponding to the second traffic allocation initial value; the user conversion rate estimation model is obtained by combining a deep factor decomposition machine model and a weighted loss function;

[0028] If the user conversion rate estimate is not greater than the user conversion rate threshold, the second traffic allocation initial values ​​of the plurality of resource locations in the target time period are optimized until the user conversion rate estimate is greater than the user conversion rate threshold, thereby obtaining the second traffic allocation target values ​​of the plurality of resource locations in the target time period.

[0029] In a second aspect, the present application further provides a resource bit flow allocation device, including:

[0030] The traffic estimation value acquisition module is used to obtain the traffic estimation value of the target period based on the historical traffic decomposition data and the traffic estimation model;

[0031] A first target value acquisition module is used to allocate the traffic estimate value of the target period according to the target user preference factor corresponding to the target period and the target traffic allocation ratio between the operation plans, while satisfying the user click rate threshold and the traffic contract display volume, to obtain the first traffic allocation value of each type of content in the target period; wherein different operation plans correspond to different types of content;

[0032] The second target value acquisition module is used to determine a number of resource locations of any type of content, and obtain a second flow allocation target value of the number of resource locations in the target time period according to the first flow allocation value of the any type of content in the target time period and the flow matching model.

[0033] In a third aspect, the present application further provides a computer device, wherein the computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor executes the above method.

[0034] In a fourth aspect, the present application further provides a computer-readable storage medium, wherein a computer program is stored on the computer-readable storage medium, and the computer program is executed by a processor to execute the above method.

[0035] In a fifth aspect, the present application further provides a computer program product, wherein the computer program product comprises a computer program, and the computer program is executed by a processor to execute the above method.

[0036] The traffic allocation method, device, computer equipment, computer-readable storage medium and computer program product of the above-mentioned resource positions obtain the traffic estimate value of the target time period based on the historical traffic decomposition data and the traffic estimation model; allocate the traffic estimate value of the target time period based on the target user preference factor corresponding to the target time period and the target traffic allocation ratio between the operation plans, while satisfying the user click-through rate threshold and the traffic contract display volume, to obtain the first traffic allocation target value of each type of content in the target time period; wherein different operation plans correspond to different categories of content; for any type of content, determine a number of resource positions for any type of content, and obtain the second traffic allocation target value of the number of resource positions in the target time period based on the first traffic allocation value of any type of content in the target time period and the traffic matching model. This application allocates the traffic estimate of the target time period while satisfying the user click rate threshold and the traffic contract display volume, and obtains the first traffic allocation target value for each type of content in the target time period; for any type of content, a number of resource positions for any type of content are determined, and based on the first traffic allocation value and the traffic matching model for any type of content in the target time period, a second traffic allocation target value for a number of resource positions in the target time period is obtained, which can dynamically adjust the distribution of traffic, reasonably allocate traffic, and improve traffic utilization and user click rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0038] Figure 1 It is an application environment diagram of a method for allocating flow of resource positions in one embodiment;

[0039] Figure 2 It is a flow chart of a method for allocating flow of resource positions in one embodiment;

[0040] Figure 3 A schematic diagram of an automatic encoder network structure of a long short-term memory network model in one embodiment;

[0041] Figure 4 Schematic diagram of the architecture of a deep factorization machine model in one embodiment;

[0042] Figure 5 A schematic diagram of a network structure of a conventional deep semantic matching model in an embodiment;

[0043] Figure 6 A schematic diagram of the network structure of a deep semantic matching model with a convolutional neural network layer added in one embodiment;

[0044] Figure 7 It is a flow chart of a method for allocating flow of resource bits in another embodiment;

[0045] Figure 8 A schematic diagram of a business process of a method for allocating traffic to resource locations in an embodiment;

[0046] Fig. 9 A schematic diagram of a model architecture of a method for allocating traffic to resource locations in an embodiment;

[0047] Fig.10 It is a structural block diagram of a flow distribution device for resource positions in one embodiment;

[0048] Fig.11 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0050] The present application embodiment provides a method for allocating resource traffic. The present application embodiment can be executed by a computer device, such as Figure 1 As shown, the computer device can obtain historical traffic decomposition data, obtain the traffic estimate for the target period, obtain the target user adjustment factor, user preference factor and target traffic allocation ratio between the operation plans corresponding to the target period, and then obtain the second traffic allocation target value for a number of resource locations in the target period. It is understandable that the computer device can be implemented through a server, a terminal, or an interactive system between a terminal and a server. In this embodiment, the method includes Figure 2 The steps shown are:

[0051] Step S201, obtaining a traffic estimation value for a target period based on historical traffic decomposition data and a traffic estimation model.

[0052] Based on the historical traffic data, historical traffic decomposition data can be obtained, where traffic can be the number of daily active users (Daily Active User, DAU) of the application.

[0053] The traffic estimation model can be obtained based on the Long Short-Term Memory (LSTM) network model. The autoencoder network structure of the LSTM network model is as follows: Figure 3 shown.

[0054] For example, the historical traffic decomposition data of the previous day can be input into the traffic estimation model to obtain the traffic estimation value of the current day. The current day can be divided into multiple time periods, and the time period for which resource bit traffic allocation is required is called the target time period.

[0055] The traffic estimate value for the current day includes traffic estimate values ​​for multiple time periods, and the traffic estimate value for the target time period is determined from the traffic estimate values ​​for the multiple time periods.

[0056] Step S202, according to the target user preference factor corresponding to the target time period and the target traffic allocation ratio between the operation plans, the traffic estimate value of the target time period is allocated while satisfying the user click rate threshold and the traffic contract display volume, to obtain the first traffic allocation target value of each type of content in the target time period; different operation plans correspond to different categories of content.

[0057] The user preference factor (also called the user adjustment factor) represents the user's preference for each operation plan (also called the operation strategy). Different operation plans correspond to different types of content (also called the content body). For example, operation plan 1 corresponds to advertising content (also called the advertising content body), operation plan 2 corresponds to activity content (also called the activity content body), and operation plan 3 corresponds to room content (also called the room content body).

[0058] The user preference factors corresponding to different time periods and the traffic allocation ratios among various operation plans are different. Based on the target user preference factors corresponding to the target time period and the target traffic allocation ratios among various operation plans, the traffic estimate of the target time period can be allocated while satisfying the user click rate threshold and the traffic contract display volume to obtain the first traffic allocation target value for each type of content in the target time period.

[0059] Step S203, for any type of content, determine a number of resource locations of any type of content, and obtain a second flow allocation target value of the number of resource locations in the target period according to the first flow allocation value of any type of content in the target period and the flow matching model.

[0060] The traffic matching model may be a deep semantic matching model including a convolutional neural network layer.

[0061] For any type of content, determine a number of resource locations for the content, input the first traffic allocation value of the content in the target period into the traffic matching model, and obtain a second traffic allocation target value for the resource locations in the target period.

[0062] In the above-mentioned traffic allocation method for resource positions, the traffic estimate value of the target time period is allocated while satisfying the user click rate threshold and the traffic contract display volume, and the first traffic allocation target value of each type of content in the target time period is obtained; for any type of content, a number of resource positions for any type of content are determined, and based on the first traffic allocation value of any type of content in the target time period and the traffic matching model, the second traffic allocation target values ​​of several resource positions in the target time period are obtained, which can dynamically adjust the distribution of traffic, reasonably allocate traffic, and improve traffic utilization and user click rate.

[0063] In one of the embodiments, before obtaining the traffic estimate value for the target time period based on the historical traffic decomposition data and the traffic estimation model, the method provided by the present application also includes: performing feature decomposition on the historical traffic data to obtain seasonal component data, trend component data and random component data; obtaining the historical traffic decomposition data based on the seasonal component data, trend component data and random component data.

[0064] Historical traffic data can be time series data. The historical traffic data y(t) can be feature decomposed to obtain seasonal component data S(t), trend component data T(t) and random component data R(t); the seasonal component data S(t), trend component data T(t) and random component data R(t) can be used as historical traffic decomposition data.

[0065] The seasonal component data S(t) represents the periodic fluctuations in the historical flow data y(t) due to seasonal changes. Seasonality is usually a fixed period (for example, annual seasonality or weekly seasonality). The seasonal component data S(t) can be obtained by seasonal decomposition or fitting a periodic function.

[0066] The trend component data T(t) represents the long-term trend or change direction of the historical flow data y(t). The trend can be estimated by different methods, such as moving average, exponential smoothing, and polynomial fitting.

[0067] Random component data R(t), also known as residual data, represents the remaining features after removing trend and seasonal features from historical traffic data y(t), that is, fluctuations that cannot be explained by trend and seasonality. Random component data R(t) is usually assumed to be white noise or random error.

[0068] In this embodiment, feature decomposition is performed on historical traffic data to obtain historical traffic decomposition data for traffic estimation, which can improve the accuracy of traffic estimation.

[0069] In one of the embodiments, the steps of obtaining the target user preference factor corresponding to the target time period and the target traffic allocation ratio between each operation plan are as follows: the traffic difference value of the previous time period is obtained according to the traffic estimate value of the previous time period of the target time period and the actual traffic value of the previous time period; when the traffic difference value of the previous time period is less than the difference threshold and the operation plans have not changed, the initial user preference factor and the initial traffic allocation ratio between the operation plans are used as the target user preference factor corresponding to the target time period and the target traffic allocation ratio between the operation plans; when the traffic difference value of the previous time period is greater than the difference threshold, or when any operation plan changes, the initial user preference factor and the initial traffic allocation ratio between the operation plans are updated to obtain the target user preference factor corresponding to the target time period and the target traffic allocation ratio between the operation plans.

[0070] The flow difference value of the previous period can be obtained by subtracting the flow estimate of the previous period from the target period from the actual flow value of the previous period.

[0071] According to the number of operation plans, the initial traffic allocation ratio between the operation plans can be obtained, usually starting from equal shares. For example, the initial traffic allocation ratio of operation plan j is shown in formula (1).

[0072] (1)

[0073] The initial user preference factor can be obtained according to the initial traffic allocation ratio among the operation plans, as shown in formula (2).

[0074] (2)

[0075] in, It reflects the proportion of each operation plan that has the opportunity to be displayed among all possible operation plans displayed to user i. Indicates user Operational plan The preference for the ad, which can be determined based on historical data, such as the user's click-through rate, viewing time, conversion rate, and other indicators for a specific campaign; The higher the value, the more likely the user Operational plan The higher the interaction tendency. represents the traffic allocation ratio of operation plan j, and also represents the operation plan The importance or priority of the impression contract. An operation plan with a higher value means that the contract requirements of the operation plan are more urgent or important and require more impressions.

[0076] When the traffic difference value of the current time period is less than the difference threshold and the operation plans have not changed, it indicates that the accuracy of traffic allocation based on the initial user preference factor and the initial traffic allocation ratio among the operation plans is high. At this time, the initial user preference factor and the initial traffic allocation ratio among the operation plans can be used as the target user preference factor corresponding to the target time period and the target traffic allocation ratio among the operation plans.

[0077] The change in the operation plan may be a change in the resource position corresponding to the operation plan, for example, the resource position corresponding to the operation plan 1 changes from resource position 1 to resource position 2.

[0078] When the traffic difference value of the previous time period is greater than the difference threshold, or when any operation plan changes, it indicates that the accuracy of traffic allocation based on the initial user preference factor and the initial traffic allocation ratio among the operation plans is low. At this time, the initial user preference factor and the initial traffic allocation ratio among the operation plans can be updated according to the traffic contract display volume of the operation plan and the traffic display volume up to the target time period to obtain the target user preference factor corresponding to the target time period and the target traffic allocation ratio among the operation plans.

[0079] In this embodiment, the traffic difference value of the previous time period is obtained based on the traffic estimate value of the previous time period before the target time period and the actual traffic value of the previous time period; based on whether the traffic difference value of the previous time period is less than the difference threshold and whether the operation plans have changed, it is determined whether to update the initial user preference factor and the initial traffic allocation ratio between the operation plans, so as to obtain a more accurate target user preference factor corresponding to the target time period and the target traffic allocation ratio between the operation plans, so as to reasonably allocate traffic and improve traffic utilization and user click-through rate.

[0080] In one of the embodiments, the initial user preference factor and the initial traffic allocation ratio between each operation plan are updated, and the specific steps are as follows: for any operation plan, the traffic contract display volume and the traffic displayed volume up to the target period of time of any operation plan are obtained; according to the traffic contract display volume and the traffic displayed volume, the initial traffic allocation ratio of any operation plan is updated; according to the updated initial traffic allocation ratio, the initial user preference factor is updated.

[0081] For operation plan j, get the traffic contract display volume of operation plan j j and traffic impressions up to the target period j ; Display volume according to traffic contract j and traffic displayed j , the initial traffic allocation ratio of operation plan j is updated as shown in formula (3).

[0082] (3)

[0083] In the formula, represents the initial traffic distribution ratio after the operation plan j is updated, represents the initial traffic allocation ratio of operation plan j.

[0084] The initial user preference factor can be updated according to the updated initial traffic allocation ratio, as shown in formula (2).

[0085] The initial user preference factor and the initial traffic allocation ratio between the operation plans can be updated multiple times until the convergence condition is met, and then the update is stopped. The convergence condition can be that the display contract is met, the traffic allocation ratio is stable, the number of iterations is met, the performance index is close to the optimal, the stability of the real-time data is high, or the stability of the user preference factor is high.

[0086] Specifically, satisfying the display contract means that the gap between the actual display volume of each operation plan and the display volume of the traffic contract is less than the preset tolerance error; stable traffic allocation ratio means that the traffic allocation ratio between operation plans does not change much in several consecutive time periods, indicating that the system has adapted to the current traffic pattern and user behavior; satisfying the number of iterations means that a maximum number of iterations can be set as a convergence condition; performance indicators are close to optimality means that if some performance indicators (such as maximizing revenue, minimizing costs, etc.) are used, when these indicators reach the optimal or close to the optimal level, it can be considered that the convergence conditions are met; high stability of real-time data means that if real-time data (such as traffic, click-through rate, etc.) remains stable for a period of time, it can be considered that the convergence conditions are met; high stability of user preference factors means that if user preference factors remain stable for several consecutive time periods, it can be considered that the convergence conditions are met.

[0087] In this embodiment, the initial traffic allocation ratio of any operation plan can be updated according to the traffic contract display volume and traffic displayed volume of any operation plan; the initial user preference factor is updated according to the updated initial traffic allocation ratio, so as to obtain a more accurate target user preference factor corresponding to the target time period and the target traffic allocation ratio between each operation plan, so as to reasonably allocate traffic and improve traffic utilization and user click-through rate.

[0088] In one of the embodiments, according to the target user preference factor corresponding to the target time period and the target traffic allocation ratio among the operation plans, the traffic estimate value of the target time period is allocated under the condition that the user click rate threshold and the traffic contract display volume are met, and the first traffic allocation target value of each type of content in the target time period is obtained. The specific steps are as follows: according to the target user preference factor corresponding to the target time period and the target traffic allocation ratio among the operation plans, the first traffic allocation initial value of each type of content in the target time period is obtained; if the first traffic allocation initial value of each type of content in the target time period does not meet the traffic contract display volume of each type of content, the first traffic allocation initial value of each type of content in the target time period is allocated; The initial allocation value is optimized until the traffic contract display volume of each type of content is met, and the first traffic allocation median value of each type of content in the target time period is obtained; according to the first traffic allocation median value of each type of content in the target time period and the user click rate estimation model, the user click rate estimation value corresponding to the first traffic allocation median value is obtained; the user click rate estimation model is obtained by combining the deep factor decomposition machine model and the weighted loss function; if the user click rate estimation value is not greater than the user click rate threshold, the first traffic allocation median value of each type of content in the target time period is optimized until the user click rate estimation value is greater than the user click rate threshold, and the first traffic allocation target value of each type of content in the target time period is obtained.

[0089] According to the target user preference factor corresponding to the target time period and the target traffic allocation ratio between the operation plans, the first traffic allocation initial value of each type of content in the target time period is obtained. The specific steps are as follows:

[0090] According to the target traffic allocation ratio between the operation plans corresponding to the target period and the traffic estimate of the target period, the traffic display volume of each operation plan in the target period can be obtained, as shown in formula (4).

[0091] (4)

[0092] In the formula, represents the traffic display volume of operation plan j in the target period, Indicates the estimated traffic volume during the target period. represents the target flow distribution ratio of operation plan j.

[0093] The target user preference factor represents the user's preference for each operation plan, and the degree of the user's preference for each operation plan can be determined based on the user preference factor.

[0094] Traffic can be the number of daily active users of an application. The operation plan assigned to each user can be determined based on the target user preference factor and the traffic display volume of each operation plan in the target period. The user can be preferentially assigned to the operation plan with the highest preference. When the traffic allocated to the operation plan with the highest preference is greater than the traffic display volume in the target period, the user is assigned to an operation plan with a medium to high preference. Similarly, the estimated daily active users in the target period are allocated to obtain the number of daily active users allocated to each operation plan in the target period. The first traffic allocation initial value of each type of content in the target period can be obtained based on the estimated number of daily active users allocated to each operation plan in the target period and the content category corresponding to each operation plan.

[0095] If the initial value of the first traffic allocation of each type of content in the target time period does not meet the traffic contract display volume of each type of content, the initial value of the first traffic allocation of each type of content in the target time period is optimized until the traffic contract display volume of each type of content is met, and the intermediate value of the first traffic allocation of each type of content in the target time period is obtained.

[0096] The Deep Factorization Machine (DeepFM) model can be combined with the weighted loss function to obtain a user click rate prediction model to improve the accuracy of user click rate prediction. The architecture diagram of the Deep Factorization Machine model is shown in the figure below. Figure 4 The weighted loss function can be a weighted binary cross-entropy loss function (Weighted Binary Cross-Entropy Loss), as shown in formula (5),

[0097] (5)

[0098] In the formula, is the total number of samples, It is The weight of each sample is usually = 1) will be given a higher weight to compensate for the fact that the number of positive samples is less than that of negative samples (e.g. =0). Among them, it can be calculated in The normalized result of the historical user payment amount within a day is taken as Weight; It is The true labels of the samples. It is The predicted probability of a sample is usually obtained from the output of the user click rate prediction model through the sigmoid function.

[0099] By inputting the middle value of the first traffic distribution of each type of content in the target time period into the user click rate prediction model, the user click rate prediction value corresponding to the middle value of the first traffic distribution can be obtained; if the user click rate prediction value is not greater than the user click rate threshold, the middle value of the first traffic distribution of each type of content in the target time period is optimized until the user click rate prediction value is greater than the user click rate threshold, thereby obtaining the first traffic distribution target value of each type of content in the target time period.

[0100] In this embodiment, when the first traffic allocation initial value of each type of content in the target time period meets the traffic contract display volume of each type of content, the first traffic allocation intermediate value of each type of content in the target time period is obtained; according to the user click rate prediction model obtained by combining the deep factor decomposition machine model and the weighted loss function, the user click rate estimation value corresponding to the first traffic allocation intermediate value is obtained, so that when the user click rate estimation value is greater than the user click rate threshold, the first traffic allocation target value of each type of content in the target time period is obtained, so that when the user click rate threshold and the traffic contract display volume are met, the first traffic allocation target value of each type of content in the target time period is obtained, so as to reasonably allocate traffic and improve traffic utilization and user click rate.

[0101] In one of the embodiments, for any type of content, a number of resource locations of any type of content are determined, and based on the first traffic allocation value of any type of content in the target time period and the traffic matching model, the second traffic allocation target values ​​of the number of resource locations in the target time period are obtained. The specific steps are as follows: the first traffic allocation value of any type of content in the target time period is input into the traffic matching model to obtain the second traffic allocation initial values ​​of the number of resource locations in the target time period; the traffic matching model is a deep semantic matching model including a convolutional neural network layer; based on the second traffic allocation initial values ​​of the number of resource locations in the target time period and the user conversion rate estimation model, the user conversion rate estimation value corresponding to the second traffic allocation initial value is obtained; the user conversion rate estimation model is obtained by combining the deep factor decomposition machine model and the weighted loss function; if the user conversion rate estimation value is not greater than the user conversion rate threshold, the second traffic allocation initial values ​​of the number of resource locations in the target time period are optimized until the user conversion rate estimation value is greater than the user conversion rate threshold, and the second traffic allocation target values ​​of the number of resource locations in the target time period are obtained.

[0102] In the deep semantic matching model, a convolutional neural network layer can be added to extract the context features under the sliding window, thereby obtaining a traffic matching model with better matching performance. Among them, the network structure of the conventional deep semantic matching model is as follows: Figure 5 As shown in Figure 2, the network structure of the deep semantic matching model with the convolutional neural network layer is as follows: Figure 6 shown.

[0103] The Deep Factorization Machine (DeepFM) model and the weighted loss function can be combined to obtain a user conversion rate prediction model to improve the accuracy of user conversion rate prediction.

[0104] The first traffic allocation value of any type of content in the target time period is input into the traffic matching model to obtain the second traffic allocation initial values ​​of several resource positions in the target time period; the second traffic allocation initial values ​​of several resource positions in the target time period are input into the user conversion rate estimation model to obtain the user conversion rate estimation value corresponding to the second traffic allocation initial values.

[0105] If the user conversion rate estimate is not greater than the user conversion rate threshold, the second traffic allocation initial values ​​of the resource locations in the target time period are optimized until the user conversion rate estimate is greater than the user conversion rate threshold, thereby obtaining the second traffic allocation target values ​​of the resource locations in the target time period.

[0106] In this embodiment, based on the traffic matching model and the user conversion rate estimation model, when the user conversion rate estimation value is greater than the user conversion rate threshold, a second traffic allocation target value for a number of resource positions in the target time period is obtained to reasonably allocate traffic and improve traffic utilization and user click rate.

[0107] In order to better understand the above method, an application example of the flow allocation method of resource bits of the present application is described in detail below. Figure 7 The business process of the traffic allocation method of the resource position of this application is as follows Figure 8 As shown, the model architecture is as Fig. 9 shown.

[0108] In the offline calculation stage, the historical traffic data is feature decomposed to obtain seasonal component data, trend component data and random component data; based on the seasonal component data, trend component data and random component data, the historical traffic decomposition data is obtained; based on the historical traffic decomposition data and the traffic estimation model, the traffic estimation value for each period is obtained. Based on the number of operation plans, the initial traffic allocation ratio between the operation plans is obtained, and based on the initial traffic allocation ratio between the operation plans, the initial user preference factor is obtained.

[0109] In the near real-time calculation stage, the traffic estimate for the target period is obtained. The traffic difference value for the previous period is obtained based on the traffic estimate for the previous period of the target period and the actual traffic value for the previous period; when the traffic difference value for the previous period is less than the difference threshold, and when the operation plans have not changed, the initial user preference factor and the initial traffic allocation ratio between the operation plans are used as the target user preference factor corresponding to the target period and the target traffic allocation ratio between the operation plans; when the traffic difference value for the previous period is greater than the difference threshold, or when any operation plan changes, for any operation plan, the traffic contract display volume of any operation plan and the traffic displayed volume as of the target period are obtained; based on the traffic contract display volume and the traffic displayed volume, the initial traffic allocation ratio of any operation plan is updated; based on the updated initial traffic allocation ratio, the initial user preference factor is updated to obtain the target user preference factor corresponding to the target period and the target traffic allocation ratio between the operation plans.

[0110] According to the target user preference factor corresponding to the target time period and the target traffic allocation ratio between each operation plan, the initial value of the first traffic allocation of each type of content in the target time period is obtained; if the initial value of the first traffic allocation of each type of content in the target time period does not meet the traffic contract display volume of each type of content, the initial value of the first traffic allocation of each type of content in the target time period is optimized until the traffic contract display volume of each type of content is met, and the intermediate value of the first traffic allocation of each type of content in the target time period is obtained; according to the intermediate value of the first traffic allocation of each type of content in the target time period and the user click rate prediction model, the user click rate prediction value corresponding to the first traffic allocation intermediate value is obtained; the user click rate prediction model is obtained by combining the deep factor decomposition machine model and the weighted loss function; if the user click rate prediction value is not greater than the user click rate threshold, the intermediate value of the first traffic allocation of each type of content in the target time period is optimized until the user click rate prediction value is greater than the user click rate threshold, and the first traffic allocation target value of each type of content in the target time period is obtained; different operation plans correspond to different categories of content.

[0111] The first traffic allocation value of any type of content in the target time period is input into the traffic matching model to obtain the second traffic allocation initial values ​​of several resource locations in the target time period; the traffic matching model is a deep semantic matching model including a convolutional neural network layer; according to the second traffic allocation initial values ​​of several resource locations in the target time period and the user conversion rate estimation model, the user conversion rate estimation value corresponding to the second traffic allocation initial value is obtained; the user conversion rate estimation model is obtained by combining the deep factor decomposition machine model and the weighted loss function; if the user conversion rate estimation value is not greater than the user conversion rate threshold, the second traffic allocation initial values ​​of several resource locations in the target time period are optimized until the user conversion rate estimation value is greater than the user conversion rate threshold, so as to obtain the second traffic allocation target values ​​of several resource locations in the target time period.

[0112] During the online calculation phase, the second traffic allocation target values ​​of several resource locations in the target time period are used to determine whether the contract requirements are met. When the contract requirements are met, the traffic is allocated to each resource location according to the second traffic allocation target values ​​of several resource locations in the target time period.

[0113] In this embodiment, under the condition that the user click rate threshold and the traffic contract display volume are met, the traffic estimate value of the target time period is allocated to obtain the first traffic allocation target value of each type of content in the target time period; for any type of content, a number of resource positions of any type of content are determined, and according to the first traffic allocation value of any type of content in the target time period and the traffic matching model, the second traffic allocation target value of several resource positions in the target time period is obtained, which can dynamically adjust the distribution of traffic, reasonably allocate traffic, and improve traffic utilization and user click rate.

[0114] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0115] Based on the same inventive concept, the embodiment of the present application also provides a resource bit flow distribution device for implementing the resource bit flow distribution method involved in the above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in the embodiment of the flow distribution device for one or more resource bits provided below can refer to the limitations of the resource bit flow distribution method above, and will not be repeated here.

[0116] In an exemplary embodiment, Fig.10 As shown, a resource bit flow distribution device is provided, wherein:

[0117] The traffic estimation value acquisition module 1001 is used to obtain the traffic estimation value of the target period according to the historical traffic decomposition data and the traffic estimation model;

[0118] The first target value acquisition module 1002 is used to allocate the traffic estimate value of the target time period according to the target user preference factor corresponding to the target time period and the target traffic allocation ratio between the operation plans, while satisfying the user click rate threshold and the traffic contract display volume, to obtain the first traffic allocation value of each type of content in the target time period; wherein different operation plans correspond to different types of content;

[0119] The second target value acquisition module 1003 is used to determine a number of resource locations of any type of content, and obtain a second traffic allocation target value of the number of resource locations in the target time period according to the first traffic allocation value of the any type of content in the target time period and the traffic matching model.

[0120] In one of the embodiments, the device also includes a historical traffic decomposition data acquisition module, which is used to: perform feature decomposition on historical traffic data to obtain seasonal component data, trend component data and random component data; and obtain historical traffic decomposition data based on the seasonal component data, trend component data and random component data.

[0121] In one of the embodiments, the device also includes a factor ratio decomposition data acquisition module, which is used to: obtain the traffic difference value of the previous time period based on the traffic estimate value of the previous time period of the target time period and the actual traffic value of the previous time period; when the traffic difference value of the previous time period is less than the difference threshold, and the operation plans have not changed, use the initial user preference factor and the initial traffic allocation ratio between the operation plans as the target user preference factor corresponding to the target time period and the target traffic allocation ratio between the operation plans; when the traffic difference value of the previous time period is greater than the difference threshold, or any operation plan changes, update the initial user preference factor and the initial traffic allocation ratio between the operation plans to obtain the target user preference factor corresponding to the target time period and the target traffic allocation ratio between the operation plans.

[0122] In one of the embodiments, the device also includes a factor ratio decomposition data acquisition module, which is used to: for any operation plan, obtain the traffic contract display volume of any operation plan and the traffic displayed volume up to the target period; update the initial traffic allocation ratio of any operation plan based on the traffic contract display volume and the traffic displayed volume; and update the initial user preference factor based on the updated initial traffic allocation ratio.

[0123] In one embodiment, the first target value acquisition module 1002 is used to: obtain the first traffic allocation initial value of each type of content in the target time period according to the target user preference factor corresponding to the target time period and the target traffic allocation ratio between each operation plan; if the first traffic allocation initial value of each type of content in the target time period does not meet the traffic contract display volume of each type of content, then optimize the first traffic allocation initial value of each type of content in the target time period until the traffic contract display volume of each type of content is met, and obtain the first traffic allocation intermediate value of each type of content in the target time period; obtain the user click rate estimation value corresponding to the first traffic allocation intermediate value according to the first traffic allocation intermediate value of each type of content in the target time period and the user click rate estimation model; the user click rate estimation model is obtained by combining the deep factor decomposition machine model and the weighted loss function; if the user click rate estimation value is not greater than the user click rate threshold, then optimize the first traffic allocation intermediate value of each type of content in the target time period until the user click rate estimation value is greater than the user click rate threshold, and obtain the first traffic allocation target value of each type of content in the target time period.

[0124] In one embodiment, the second target value acquisition module 1003 is used to: input the first traffic allocation value of any type of content in the target time period into the traffic matching model to obtain the second traffic allocation initial values ​​of the several resource locations in the target time period; the traffic matching model is a deep semantic matching model including a convolutional neural network layer; according to the second traffic allocation initial values ​​of the several resource locations in the target time period and the user conversion rate estimation model, obtain the user conversion rate estimation value corresponding to the second traffic allocation initial value; the user conversion rate estimation model is obtained by combining the deep factor decomposition machine model and the weighted loss function; if the user conversion rate estimation value is not greater than the user conversion rate threshold, the second traffic allocation initial values ​​of the several resource locations in the target time period are optimized until the user conversion rate estimation value is greater than the user conversion rate threshold, so as to obtain the second traffic allocation target values ​​of the several resource locations in the target time period.

[0125] Each module in the above-mentioned resource bit flow distribution device can be implemented in whole or in part by software, hardware and their combination. Each module can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to each module above.

[0126] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. Fig.11 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data of an embodiment of the flow allocation method of resource bits. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a flow allocation method of resource bits is implemented.

[0127] Those skilled in the art will understand that Fig.11The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0128] In one embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above method embodiments when executing the computer program.

[0129] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0130] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0131] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0132] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0133] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0134] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be construed as limiting the scope of the present application. It should be noted that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for allocating resource traffic, characterized in that: The method comprises: Based on the historical traffic decomposition data and traffic estimation model, the traffic estimation value for the target period is obtained; According to the target user preference factor corresponding to the target time period and the target traffic allocation ratio between the operation plans, the traffic estimate value of the target time period is allocated under the condition that the user click rate threshold and the traffic contract display volume are met, so as to obtain the first traffic allocation target value of each type of content in the target time period; wherein different operation plans correspond to different types of content; For any type of content, a number of resource locations of the content are determined, and a second flow allocation target value of the resource locations in the target time period is obtained according to a first flow allocation value of the content in the target time period and a flow matching model.

2. The method according to claim 1, characterized in that Before obtaining the traffic estimation value of the target period according to the historical traffic decomposition data and the traffic estimation model, the method further includes: Perform feature decomposition on historical traffic data to obtain seasonal component data, trend component data and random component data; The historical traffic decomposition data is obtained according to the seasonal component data, trend component data and random component data.

3. The method according to claim 1, characterized in that The step of obtaining the target user preference factor corresponding to the target time period and the target traffic distribution ratio between the operation plans includes: Obtaining a flow difference value of the previous period according to the flow estimate value of the previous period of the target period and the actual flow value of the previous period; When the traffic difference value of the previous time period is less than the difference threshold and the operation plans have not changed, the initial user preference factor and the initial traffic allocation ratio between the operation plans are used as the target user preference factor corresponding to the target time period and the target traffic allocation ratio between the operation plans; When the traffic difference value of the previous time period is greater than the difference threshold, or when any operation plan changes, the initial user preference factor and the initial traffic allocation ratio between the operation plans are updated to obtain the target user preference factor corresponding to the target time period and the target traffic allocation ratio between the operation plans.

4. The method according to claim 3, characterized in that The updating of the initial user preference factor and the initial traffic allocation ratio between the operation plans includes: For any operation plan, obtain the traffic contract display volume of any operation plan and the traffic display volume up to the target period; According to the traffic contract display volume and the traffic displayed volume, the initial traffic allocation ratio of any operation plan is updated; The initial user preference factor is updated according to the updated initial traffic allocation ratio.

5. The method according to claim 1, characterized in that The method allocates the traffic estimate of the target time period according to the target user preference factor corresponding to the target time period and the target traffic allocation ratio between the operation plans while satisfying the user click rate threshold and the traffic contract display volume to obtain the first traffic allocation target value of each type of content in the target time period, including: According to the target user preference factor corresponding to the target time period and the target traffic allocation ratio between the operation plans, a first traffic allocation initial value of each type of content in the target time period is obtained; If the initial value of the first traffic allocation of each type of content in the target time period does not meet the traffic contract display volume of each type of content, then optimize the initial value of the first traffic allocation of each type of content in the target time period until the traffic contract display volume of each type of content is met, and obtain the intermediate value of the first traffic allocation of each type of content in the target time period; According to the first traffic distribution median value of each type of content in the target time period and the user click rate prediction model, obtaining a user click rate prediction value corresponding to the first traffic distribution median value; the user click rate prediction model is obtained by combining a deep factor decomposition machine model and a weighted loss function; If the user click rate estimate is not greater than the user click rate threshold, the first traffic allocation intermediate value of each type of content in the target time period is optimized until the user click rate estimate is greater than the user click rate threshold, thereby obtaining the first traffic allocation target value of each type of content in the target time period.

6. The method according to claim 1, characterized in that The step of determining a number of resource locations for any type of content, and obtaining a second traffic allocation target value for the number of resource locations in the target time period according to a first traffic allocation value for the any type of content in the target time period and a traffic matching model, comprises: Inputting the first traffic allocation value of any type of content in the target time period into a traffic matching model to obtain a second traffic allocation initial value of the plurality of resource bits in the target time period; the traffic matching model is a deep semantic matching model including a convolutional neural network layer; According to the second traffic allocation initial values ​​of the plurality of resource positions in the target time period and the user conversion rate estimation model, obtaining a user conversion rate estimation value corresponding to the second traffic allocation initial value; the user conversion rate estimation model is obtained by combining a deep factor decomposition machine model and a weighted loss function; If the user conversion rate estimate is not greater than the user conversion rate threshold, the second traffic allocation initial values ​​of the plurality of resource locations in the target time period are optimized until the user conversion rate estimate is greater than the user conversion rate threshold, thereby obtaining the second traffic allocation target values ​​of the plurality of resource locations in the target time period.

7. A resource bit flow distribution device, characterized in that: The device comprises: The traffic estimation value acquisition module is used to obtain the traffic estimation value of the target period based on the historical traffic decomposition data and the traffic estimation model; A first target value acquisition module is used to allocate the traffic estimate value of the target period according to the target user preference factor corresponding to the target period and the target traffic allocation ratio between the operation plans, while satisfying the user click rate threshold and the traffic contract display volume, to obtain the first traffic allocation value of each type of content in the target period; wherein different operation plans correspond to different types of content; The second target value acquisition module is used to determine a number of resource locations of any type of content, and obtain a second flow allocation target value of the number of resource locations in the target time period according to the first flow allocation value of the any type of content in the target time period and the flow matching model.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images