Cloud service storage capacity demand prediction method and computer program product

Abstract

The invention discloses a cloud service storage capacity demand prediction method and a computer program product. The method comprises the following steps: acquiring historical data of a to-be-predicted user, wherein the historical data comprises user historical cloud service use behavior data, operator historical service information and user historical operator service use behavior data; and inputting the historical data into a preset cloud service storage capacity demand prediction model for processing, and predicting the cloud service storage capacity demand of the to-be-predicted user, the preset cloud service storage capacity demand prediction model is used for predicting the cloud service storage capacity demand on the basis of feature dependence between user cloud service use behavior data and user service use operator service behavior data; and predicting the cloud service storage capacity demand of the user according to the user cloud service use behavior data and the trend dependency of adjacent time between the user cloud service use behavior data and the user operator service use behavior data.

Description

Technical Field

[0001] The present application relates to the field of 5G message technology, and in particular to a cloud service storage capacity demand prediction method and computer program product. Background Art

[0002] In related technologies, cloud storage capacity prediction and planning are generally based on cloud storage resource pool status, cloud storage node distribution, status and performance and other operation and maintenance data or external emergencies, predicting future capacity consumption based on historical trends, and providing capacity assessment suggestions and alarm functions to help users plan capacity and optimize storage resources in advance. However, most of the above methods implement large-scale resource scheduling on a global scale based on the historical operation data of hardware resources at the infrastructure layer, and cannot be accurately adjusted according to actual user needs.

[0003] Therefore, it is urgent to provide a method that can solve the above problems. Summary of the invention

[0004] The embodiments of the present application provide a cloud service storage capacity demand prediction method and a computer program product to solve the problem of inaccurate user cloud storage capacity prediction in the related art.

[0005] In a first aspect, an embodiment of the present application provides a method for predicting cloud service storage capacity demand, comprising: Obtaining historical data of the user to be predicted, the historical data including historical cloud service usage behavior data of the user, historical service information of the operator, and historical operator service usage behavior data of the user; The historical data is input into a preset cloud service storage capacity demand prediction model for processing to predict the cloud service storage capacity demand of the user to be predicted. The preset cloud service storage capacity demand prediction model is used to predict the user's cloud service storage capacity demand based on the feature dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data, and the trend dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data at adjacent times.

[0006] In a second aspect, an embodiment of the present application provides a cloud service storage capacity demand prediction device, the device comprising: An acquisition module is used to acquire historical data of the user to be predicted, the historical data including the user's historical cloud service usage behavior data, the operator's historical service information and the user's historical operator service usage behavior data; The prediction module is used to input historical data into a preset cloud service storage capacity demand prediction model for processing, and predict the cloud service storage capacity demand of the user to be predicted. The preset cloud service storage capacity demand prediction model is used to predict the user's cloud service storage capacity demand based on the feature dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data, and the trend dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data at adjacent times.

[0007] In a third aspect, an embodiment of the present application provides an electronic device, comprising a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction, when executed by the processor, implements the steps of the method described in the first aspect.

[0008] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described in the first aspect are implemented.

[0009] In a fifth aspect, an embodiment of the present application provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions, which, when executed by a computer, implement the steps of the method described in the first aspect.

[0010] In an embodiment of the present application, historical data of the user to be predicted is first obtained, the historical data includes the user's historical cloud service usage behavior data, the operator's historical service information and the user's historical operator service usage behavior data, and then the historical data is input into a preset cloud service storage capacity demand prediction model for processing to predict the cloud service storage capacity demand of the user to be predicted. The preset cloud service storage capacity demand prediction model is used to predict the user's cloud service storage capacity demand based on the feature dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data, and the trend dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data at adjacent times. By predicting the cloud service storage capacity demand from the perspective of business and user behavior, the embodiment of the present application can more intuitively reflect the user's real needs, thereby making the prediction more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings: Figure 1 is a flow chart of a method for predicting cloud service storage capacity demand provided by an embodiment of the present application; Figure 2 It is a schematic diagram of the structure of the cloud service storage capacity demand prediction model provided in the embodiment of the present application; Figure 3 is a schematic diagram of a cloud service storage capacity demand prediction device provided in an embodiment of the present application; Figure 4 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0012] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0013] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here. In addition, the "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally represents that the objects associated with each other are in an "or" relationship.

[0014] The following is combined with Figures 1 to 4 , a cloud service storage capacity demand prediction method and a computer program product provided in an embodiment of the present application are described in detail through specific embodiments and application scenarios.

[0015] like Figure 1 As shown, it is a flow chart of a cloud service storage capacity demand prediction method provided by an embodiment of the present application. Figure 1 As shown, the cloud service storage capacity demand prediction method may include: the contents shown in S101 to S102.

[0016] In S101, historical data of the user to be predicted is obtained, where the historical data includes the user's historical cloud service usage behavior data, the operator's historical service information, and the user's historical operator service usage behavior data.

[0017] Among them, the user's historical cloud service usage behavior data includes the frequency, frequency, duration, file size, file type, etc. of the user's use of cloud services to upload, download or share files, the type and frequency of the user's active or passive capacity changes, and the storage capacity size involved in each change. Based on the user's historical cloud service usage behavior data, the user's stickiness, activeness and long-term retention probability for cloud service products can be determined to judge the user's long-term value.

[0018] The operator's historical service information includes the type of call package used by the user, the package price, and the basic service content of the package (such as the data capacity, voice duration, number of text messages, etc. within the package).

[0019] The user's historical usage of operator service behavior data includes the user's traffic consumption, the user's call duration, the number of text messages received and sent by the user, etc. Based on this data, the user's communication consumption habits, network stability, consumption potential, etc. can be determined.

[0020] In one example, in order to ensure the accuracy of the cloud service storage capacity demand prediction, after obtaining the historical data of the user to be predicted, the obtained historical data may be preprocessed.

[0021] Among them, preprocessing includes cleaning, adding, deleting, and replacing abnormal data.

[0022] In one example, the user's historical cloud service usage behavior data, the operator's historical service information, and the user's historical operator service usage behavior data may be as shown in Table 1.

[0023] Table 1

[0024] In S102, the historical data is input into a preset cloud service storage capacity demand prediction model for processing to predict the cloud service storage capacity demand of the user to be predicted.

[0025] Among them, the preset cloud service storage capacity demand prediction model is used to predict the user's cloud service storage capacity demand based on the characteristic dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data, and the trend dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data in adjacent times.

[0026] In an embodiment of the present application, historical data of the user to be predicted is first obtained, the historical data includes the user's historical cloud service usage behavior data, the operator's historical service information and the user's historical operator service usage behavior data, and then the historical data is input into a preset cloud service storage capacity demand prediction model for processing to predict the cloud service storage capacity demand of the user to be predicted. The preset cloud service storage capacity demand prediction model is used to predict the user's cloud service storage capacity demand based on the feature dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data, and the trend dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data at adjacent times. By predicting the cloud service storage capacity demand from the perspective of business and user behavior, the embodiment of the present application can more intuitively reflect the user's real needs, thereby making the prediction more accurate.

[0027] like Figure 2 As shown, a cloud service storage capacity demand prediction model provided in an embodiment of the present application is provided.

[0028] like Figure 2 As shown, the cloud service storage capacity demand prediction model includes a feature dependency module and a time dependency fusion pyramid attention module.

[0029] Among them, the feature dependency module includes multiple stacked convolutional neural network layers (i.e., the layer in Figure 2). Specifically, each stacked convolutional neural network layer includes multiple time convolutional neural networks, which can be 3 or other numbers, depending on the actual application. The feature dependency module is used to obtain feature information and feature dependency between user cloud service usage behavior data and user operator service usage behavior data based on user historical cloud service usage behavior data, user historical operator service usage behavior data, user historical cloud service usage behavior decomposition data, and user historical operator service usage behavior decomposition data.

[0030] It should be noted that multiple stacked convolutional neural network layers are used in this scheme in order to extract features of different levels and abstraction levels based on information such as the decomposition data of users' cloud service usage behavior and the decomposition data of operators' service usage behavior data, thereby improving the generalization ability of the cloud service storage capacity prediction model.

[0031] The temporal-dependent fusion pyramid attention module consists of multiple stack blocks (i.e. Figure 2 satck in) and a fully connected layer (i.e. Figure 2liner in the example), which is used to obtain the temporal features of the user's cloud service usage behavior and the operator's service usage behavior over time, where the temporal features can be understood as the trend dependency of adjacent times, and to obtain the prediction results. Specifically, each stack block includes a reversible normalization layer, a multi-scale feature attention pyramid, a denormalization attention layer, and an anti-reversible normalization layer. Among them, the reversible normalization layer is used to reduce the impact of the distribution shift of the user's cloud service usage behavior and the operator's service usage behavior data and make the sequence prediction more accurate. The multi-scale feature attention pyramid and denormalization attention are used to eliminate the non-stationary factors in the user's historical cloud service usage behavior data and the user's historical operator service usage behavior data, and give greater weights to subsequences with important feature information in the user's historical cloud service usage behavior data and the user's historical operator service usage behavior data.

[0032] In a possible implementation of the present application, historical data is input into a preset cloud service storage capacity demand prediction model for processing, and predicting the cloud service storage capacity demand of the user to be predicted may include: determining the decomposition data of the user's historical cloud service usage behavior and the decomposition data of the user's historical operator service usage behavior based on the analysis of the user's historical cloud service usage behavior data and the user's historical operator service usage behavior data; inputting the user's historical cloud service usage behavior data, the user's historical operator service usage behavior data, the decomposition data of the user's historical cloud service usage behavior, and the decomposition data of the user's historical operator service usage behavior into a feature dependency module in the preset cloud service storage capacity demand prediction model to obtain feature information and feature dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data; inputting the feature information and feature dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data into a time dependency fusion pyramid attention module in the preset cloud service storage capacity demand prediction model to obtain the trend dependency of the user's cloud service usage behavior data and the user's operator service usage behavior data at adjacent times; based on the trend dependency of the user's cloud service usage behavior data and the user's operator service usage behavior data at adjacent times, predicting the cloud service storage capacity demand of the user to be predicted.

[0033] That is to say, in this embodiment, the historical data is first decomposed to obtain decomposed data, and then the historical data and the decomposed data are input into the feature dependency module for feature extraction to obtain the feature information and feature dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data, and then the obtained features and feature dependencies are input into the time dependency fusion pyramid attention module for processing to obtain the trend dependency of adjacent time, and finally the cloud service storage capacity demand of the user to be predicted is predicted based on the above feature dependency. The feature dependency module provided by this embodiment can make the cloud service storage capacity demand prediction model have higher generalization ability, and the time dependency fusion pyramid attention module makes the prediction result more accurate.

[0034] In one possible implementation of the present application, based on the analysis of the user's historical cloud service usage behavior data and the user's historical operator service usage behavior data, the decomposition data of the user's historical cloud service usage behavior and the decomposition data of the user's historical operator service usage behavior are determined, which may include the contents shown in steps one and two.

[0035] Step 1: Based on the user's historical cloud service usage behavior data, the operator's historical service information and the user's historical operator service usage behavior data, determine a first trend graph corresponding to the user's historical cloud service usage behavior data and a second trend graph corresponding to the user's historical operator service usage behavior data.

[0036] In one example, taking the frequency of the user's historical use of cloud services for file uploads in the user's historical cloud service use behavior data as an example, a trend graph of the user's historical use of cloud services for file uploads can be determined based on the frequency data of the user's historical file upload behavior at different times. Correspondingly, a corresponding trend graph can also be determined based on other behaviors in the user's historical cloud service use behavior data, which are not listed one by one in this embodiment.

[0037] Step 2: Decompose the first trend graph and the second trend graph respectively to determine the decomposed data of the user's historical cloud service usage behavior and the decomposed data of the user's historical operator service usage behavior, the decomposed data of the user's historical cloud service usage behavior includes a first steady trend sequence and a first non-steady trend sequence, and the decomposed data of the user's historical operator service usage behavior includes a second steady trend sequence and a second non-steady trend sequence, wherein the steady trend sequence refers to the data information in the trend graph whose trend is steady, and the non-steady trend sequence refers to the data information in the trend graph whose trend fluctuation is greater than the fluctuation threshold.

[0038] That is to say, the trend graph obtained above can be decomposed to determine a steady trend sequence corresponding to data information with a steady trend in the trend graph and a non-steady trend sequence corresponding to data information with a trend fluctuation greater than a fluctuation threshold in the trend graph.

[0039] In this embodiment, the trend graphs corresponding to different data changing over time are determined through historical data, and then the trend graphs are decomposed to obtain stationary trend sequences and non-stationary trend sequences, thereby determining the trends of various data changing over time, so that feature information can be subsequently extracted based on the trend sequences, making the final prediction results more accurate.

[0040] In a possible implementation of the present application, the user's historical cloud service usage behavior data, the user's historical operator service usage behavior data, the decomposition data of the user's historical cloud service usage behavior, and the decomposition data of the user's historical operator service usage behavior are input into a feature dependency module in a preset cloud service storage capacity demand prediction model to obtain feature information and feature dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data, which may include: obtaining a first feature vector corresponding to a first stationary trend sequence and a first non-stationary trend sequence, and a second feature vector corresponding to a second stationary trend sequence and a second non-stationary trend sequence; inputting the first feature vector and the second feature vector into the feature dependency module in the preset cloud service storage capacity demand prediction model for analysis to obtain feature information and feature dependency, wherein the feature dependency module includes multiple stacked convolutional neural network layers.

[0041] That is to say, according to the trend sequence obtained in the above embodiment, feature extraction is performed to obtain the corresponding feature vector, and the feature vector is input into the feature dependency module for analysis. The stationary trend sequence and non-stationary trend sequence in the trend graph can be analyzed through multiple stacked convolutional neural network layers to obtain the final feature information and final feature dependency of the trend graph.

[0042] In a possible implementation of the present application, feature information and feature dependency between user cloud service usage behavior data and user operator service usage behavior data are input into a time-dependent fusion pyramid attention module in a preset cloud service storage capacity demand prediction model to obtain the trend dependency of adjacent times between the user cloud service usage behavior data and the user operator service usage behavior data. This may include: inputting feature information and feature dependency into a time-dependent fusion pyramid attention module in a preset cloud service storage capacity demand prediction model to obtain a first adjacent time trend dependency of a first stationary trend sequence and a first non-stationary trend sequence of a first trend graph corresponding to the user cloud service usage behavior data, and a second adjacent time trend dependency of a second stationary trend sequence and a second non-stationary trend sequence of a second trend graph corresponding to the user operator service usage behavior data, wherein the time-dependent fusion pyramid attention module includes multiple stack blocks.

[0043] That is to say, the final feature information and final feature dependency of the trend graph are input into the time-dependency fusion pyramid attention module, so that the adjacent time trend dependencies of the stationary trend sequence and the non-stationary trend sequence of the first trend graph corresponding to the historical cloud service usage behavior, as well as the adjacent time trend dependencies of the stationary trend sequence and the non-stationary trend sequence of the second trend graph corresponding to the historical usage behavior data can be dynamically obtained through multiple stack blocks.

[0044] Specifically, after the stack block receives the final feature information and final feature dependency of the trend graph, it can first use reversible normalization for processing to reduce the impact of the distribution offset of each trend graph and make the prediction result more accurate. After that, the non-stationary factors of each sequence are eliminated by denormalizing attention, and the subsequences with important feature information in the user cloud service usage behavior and operator service usage behavior data are given a greater weight. Finally, the inverse reversible normalization processing is performed to obtain the adjacent time trend dependency of the stationary trend sequence and non-stationary trend sequence of the first trend graph corresponding to the historical cloud service usage behavior, and the adjacent time trend dependency of the stationary trend sequence and non-stationary trend sequence of the second trend graph corresponding to the historical usage behavior data.

[0045] In order to implement more refined and sensitive resource scheduling for the storage capacity occupied by users, thereby avoiding resource idleness caused by large-scale resource scheduling and reducing related costs. After predicting the user's cloud storage needs, the cloud service storage capacity demand prediction method provided in this embodiment may also include: based on the user's historical cloud service usage behavior data, obtaining the social behavior usage data within the cloud service of the user to be predicted; based on the social behavior usage data, determining the social communication value of the user to be predicted; using a clustering algorithm, dividing all users into multiple sub-user groups with three dimensional characteristics of user value, capacity consumption, and activity level, each sub-user group has a multi-dimensional label coefficient, and users in the same sub-user group correspond to the same label coefficient; based on the multi-dimensional label coefficient of each sub-user group, determining the resource scheduling priority of each user in each sub-user group; based on the resource scheduling priority and the cloud service storage capacity demand of the user to be predicted, allocating cloud service storage capacity to the user to be predicted.

[0046] In this embodiment, the user's real historical behavior data is combined with the user's consumption portrait to explore the user's actual cloud storage capacity needs, reduce unnecessary resource distribution, and reduce resource redundancy and idleness; and through the backward process, ensure that resource scheduling is effective and optimize the resource scheduling strategy in a timely and dynamic manner.

[0047] In a possible implementation of the present application, determining the social communication value of the user to be predicted based on social behavior usage data may include: determining the social activity of the user to be predicted and the resource communication influence of the user to be predicted based on the social behavior usage data; determining the social communication value of the user to be predicted based on the social activity of the user to be predicted and the resource communication influence of the user to be predicted.

[0048] Considering that cloud service resources are lightweight and convenient to use, they have become the main form of people transmitting important files, content, and information in the Internet era. Hot spots and cloud service resources can quickly spread virally among users. The owners of such resources can be considered to have extremely high social and communication value, and are a relatively active and high-value user group among cloud service users. Based on this, after predicting the cloud service storage capacity demand, this solution can further use the traditional customer social value model based on the social behavior usage data in the cloud service domain to evaluate the user's social activity and the user's resource communication influence, and determine the user's social communication value by combining the two indicators.

[0049] In one example, the resource propagation influence can be as follows:

[0050] in, Spread influence for resources; The number of users who browsed the resource after sharing; The number of users whose resources were transferred after sharing; The number of users who shared the resource after it was transferred.

[0051] The social activity of a user can be shown as follows:

[0052] in, is the user’s social activity, and n is the number of sharing links created by the user.

[0053] In a possible implementation of the present application, a clustering algorithm is used to divide all users into multiple sub-user groups with three dimensional characteristics: user value, capacity consumption, and activity level. The method may include: determining user value based on the operator's historical service information and static data in the cloud service, where the static data in the cloud service includes subscription data of products in the cloud service; determining the capacity consumption corresponding to each user based on the operator's historical service information and the user's historical operator service behavior data; determining the user's activity level based on the user's historical cloud service usage behavior data and the user's historical operator service behavior data; and using a clustering algorithm to divide all users into multiple sub-user groups with three dimensional characteristics: user value, capacity consumption, and activity level based on user value, capacity consumption, and user activity level.

[0054] In this embodiment, a clustering algorithm is used to divide all users into several sub-user groups with three dimensional features: user value, capacity consumption, and activity level. The user value dimension is used as a static feature class, and a historical time range is selected to construct dynamic features around the two dimensions of capacity consumption and activity level. Parameters are adjusted based on feature engineering to determine the multi-dimensional user labels of the sub-user groups and the label coefficients of the corresponding dimensions; all users in the same sub-user group correspond to the same label coefficient.

[0055] Among them, the user value dimension is composed of the user's operator service static data and cloud service static data, including package fixed amount, traffic consumption / call duration / number of text messages / group consumption data, home broadband, international roaming service data, and cloud service value-added service subscription data such as cloud service membership products and rights products; the capacity consumption dimension is composed of operator service over-package behavior data and cloud service space and rights consumption behavior data; the activity level dimension is composed of operator service usage behavior data and cloud service usage behavior data, including operator service consumption data such as traffic consumption / call duration / number of text messages / group consumption and out-of-package consumption data within a certain period, and core behavior activity data such as cloud service login / upload / download / browse.

[0056] Furthermore, for the multi-dimensional label coefficients of each sub-user group, the multi-dimensional label coefficient sorting rules between sub-user groups and the corresponding resource scheduling frequency and strategy are formulated according to the current resource status of the cloud service system; within each sub-user group, the two indicators of retention probability and social communication value are combined to determine the resource scheduling priority of users within the user group.

[0057] In a possible implementation of the present application, the cloud service storage capacity demand prediction method may also include: monitoring the user's usage of the allocated space capacity within a scheduling cycle; if the user does not exhibit preset usage behavior within the scheduling cycle, lowering the user's resource scheduling priority, and recording the number of times the user's priority is lowered, and using the number of times as a new feature for determining the weight coefficient of the user's resource scheduling priority.

[0058] That is to say, within a scheduling cycle, the usage of the space capacity allocated to users can be continuously monitored. For users who do not have specific usage behaviors within the cycle, their allocated space capacity will be postponed to the next ranked user in the scheduling order, and the number of times their allocated space capacity has been postponed due to unused space capacity will be recorded. In the next scheduling cycle, the number of times the allocated space capacity has not been used will be added as a new feature to the scheduling priority weight coefficient, and the scheduling weight and priority of all users will be recalculated.

[0059] In other words, the scheduling weight and priority of the user in the next cycle can be determined based on the user's behavior in the previous scheduling cycle to ensure that users in need can be allocated space capacity first, reducing idle cloud storage resources.

[0060] The embodiments of the present application use the user's real historical behavior data in combination with the user's consumption portrait to explore the user's actual cloud storage capacity needs, reduce unnecessary resource distribution, and reduce resource redundancy and idleness; and ensure effective resource scheduling through a backward process, and optimize the resource scheduling strategy in a timely and dynamic manner.

[0061] like Figure 3 As shown, it is a schematic diagram of a cloud service storage capacity demand prediction device provided by an embodiment of the present application. Figure 3 As shown, the cloud service storage capacity demand prediction device may include: an acquisition module 301 and a prediction module 302.

[0062] Among them, the acquisition module 301 is used to obtain the historical data of the user to be predicted, and the historical data includes the user's historical cloud service usage behavior data, the operator's historical service information and the user's historical operator service usage behavior data; the prediction module 302 is used to input the historical data into the preset cloud service storage capacity demand prediction model for processing, and predict the cloud service storage capacity demand of the user to be predicted. The preset cloud service storage capacity demand prediction model is used to predict the user's cloud service storage capacity demand based on the feature dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data, and the trend dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data in adjacent times.

[0063] In the embodiment of the present application, first, the acquisition module 301 acquires the historical data of the user to be predicted, the historical data includes the user's historical cloud service usage behavior data, the operator's historical service information and the user's historical operator service usage behavior data, and then the prediction module 302 inputs the historical data into the preset cloud service storage capacity demand prediction model for processing, and predicts the cloud service storage capacity demand of the user to be predicted. The preset cloud service storage capacity demand prediction model is used to predict the user's cloud service storage capacity demand based on the feature dependency between the user's cloud service usage behavior data and the user's operator service behavior data, and the trend dependency between the user's cloud service usage behavior data and the user's operator service behavior data at adjacent times. The embodiment of the present application can more intuitively reflect the user's real needs by predicting the cloud service storage capacity demand from the perspective of business and user behavior, thereby making the prediction more accurate.

[0064] In a possible implementation of the present application, the prediction module 302 is used to: determine the decomposition data of the user's historical cloud service usage behavior and the decomposition data of the user's historical operator service usage behavior based on the analysis of the user's historical cloud service usage behavior data and the user's historical operator service usage behavior data; input the user's historical cloud service usage behavior data, the user's historical operator service usage behavior data, the decomposition data of the user's historical cloud service usage behavior and the decomposition data of the user's historical operator service usage behavior into a feature dependency module in a preset cloud service storage capacity demand prediction model to obtain feature information and feature dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data; input the feature information and feature dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data into a time dependency fusion pyramid attention module in a preset cloud service storage capacity demand prediction model to obtain the trend dependency of the user's cloud service usage behavior data and the user's operator service usage behavior data at adjacent times; based on the trend dependency of the user's cloud service usage behavior data and the user's operator service usage behavior data at adjacent times, predict the cloud service storage capacity demand of the user to be predicted.

[0065] In a possible implementation of the present application, the prediction module 302 is used to: determine a first trend graph corresponding to the user's historical cloud service usage behavior data and a second trend graph corresponding to the user's historical operator service usage behavior data based on the user's historical cloud service usage behavior data, the operator's historical service information, and the user's historical operator service usage behavior data; decompose the first trend graph and the second trend graph respectively to determine the decomposed data of the user's historical cloud service usage behavior and the decomposed data of the user's historical operator service usage behavior, the decomposed data of the user's historical cloud service usage behavior including a first stable trend sequence and a first non-stationary trend sequence, and the decomposed data of the user's historical operator service usage behavior including a second stable trend sequence and a second non-stationary trend sequence, wherein a stable trend sequence refers to data information in the trend graph whose trend is stable, and a non-stationary trend sequence refers to data information in the trend graph whose trend fluctuation is greater than a fluctuation threshold.

[0066] In a possible implementation of the present application, the prediction module 302 is used to: obtain a first eigenvector corresponding to the first stationary trend sequence and the first non-stationary trend sequence, and a second eigenvector corresponding to the second stationary trend sequence and the second non-stationary trend sequence; input the first eigenvector and the second eigenvector into a feature dependency module in a preset cloud service storage capacity demand prediction model for analysis to obtain feature information and feature dependency, wherein the feature dependency module includes multiple stacked convolutional neural network layers.

[0067] In a possible implementation of the present application, the prediction module 302 is used to: input feature information and feature dependencies into a time-dependent fusion pyramid attention module in a preset cloud service storage capacity demand prediction model, obtain a first stationary trend sequence and a first adjacent time trend dependency of a first non-stationary trend sequence of a first trend graph corresponding to the user's cloud service usage behavior data, and a second adjacent time trend dependency of a second stationary trend sequence and a second non-stationary trend sequence of a second trend graph corresponding to the user's operator service behavior data, the time-dependent fusion pyramid attention module includes multiple stack blocks.

[0068] In a possible implementation of the present application, the cloud service storage capacity demand prediction device may further include: a second acquisition module, a first determination module, a division module, a second determination module and an allocation module.

[0069] Among them, the second acquisition module is used to obtain the social behavior usage data of the user to be predicted in the cloud service based on the user's historical cloud service usage behavior data; the first determination module is used to determine the social communication value of the user to be predicted based on the social behavior usage data; the division module is used to use a clustering algorithm to divide all users into multiple sub-user groups with three dimensional characteristics of user value, capacity consumption, and activity level, each sub-user group has a multi-dimensional label coefficient, and users in the same sub-user group correspond to the same label coefficient; the second determination module is used to determine the resource scheduling priority of each user in each sub-user group based on the multi-dimensional label coefficient of each sub-user group; the allocation module is used to allocate cloud service storage capacity to the user to be predicted based on the resource scheduling priority and the cloud service storage capacity requirements of the user to be predicted.

[0070] In a possible implementation of the present application, the first determination module is used to determine the social activity of the user to be predicted and the resource dissemination influence of the user to be predicted based on social behavior usage data; and determine the social dissemination value of the user to be predicted based on the social activity of the user to be predicted and the resource dissemination influence of the user to be predicted.

[0071] In a possible implementation of the present application, a partitioning module is used to determine user value based on the operator's historical service information and static data in the cloud service, where the static data in the cloud service includes subscription data of products in the cloud service; determine the capacity consumption corresponding to each user based on the operator's historical service information and the user's historical operator service usage behavior data; determine the user's activity level based on the user's historical cloud service usage behavior data and the user's historical operator service usage behavior data; and use a clustering algorithm to divide all users into multiple sub-user groups with three dimensional characteristics: user value, capacity consumption, and activity level based on user value, capacity consumption, and user activity level.

[0072] In a possible implementation of the present application, the cloud service storage capacity demand prediction device may further include: a monitoring module and a processing module.

[0073] Among them, the monitoring module is used to monitor the user's use of the allocated space capacity within a scheduling cycle; the processing module is used to reduce the user's resource scheduling priority if the user does not perform the preset usage behavior within the scheduling cycle, and record the number of times the user's priority is reduced, and use the number of times as a new feature to determine the weight coefficient of the user's resource scheduling priority.

[0074] In a possible implementation of the present application, the cloud service storage capacity demand prediction device may further include: a preprocessing module.

[0075] Wherein, the preprocessing module is used to preprocess the historical data of the user to be predicted.

[0076] The function of the cloud service storage capacity demand prediction device of this application has been Figure 1 to Figure 2 The method embodiment shown is described in detail, so for any details not provided in the description of this embodiment, please refer to the relevant descriptions in the aforementioned embodiments, which will not be repeated here.

[0077] like Figure 4 As shown, an embodiment of the present application also provides an electronic device 400, including a processor 401, a memory 402, and a program or instruction stored in the memory 402 and executable on the processor 401. When the program or instruction is executed by the processor 401, each process of the above-mentioned cloud service storage capacity demand prediction processing method embodiment is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0078] Optionally, the embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, each process of the above-mentioned cloud service storage capacity demand prediction method embodiment is implemented, and the same technical effect can be achieved. To avoid repetition, it is not repeated here. The computer-readable storage medium is, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0079] Optionally, an embodiment of the present application also provides a computer program product, the computer program product includes a computer program stored on a non-transitory computer-readable storage medium, the computer program includes program instructions, and when the program instructions are executed by a computer, the various processes of the above-mentioned cloud service storage capacity demand prediction method embodiment are implemented, and the same technical effect can be achieved. To avoid repetition, they will not be repeated here.

[0080] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0081] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0082] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.

Claims

1. A method for predicting cloud service storage capacity demand, characterized in that: include: Obtaining historical data of the user to be predicted, the historical data including historical cloud service usage behavior data of the user, historical service information of the operator, and historical operator service usage behavior data of the user; The historical data is input into a preset cloud service storage capacity demand prediction model for processing to predict the cloud service storage capacity demand of the user to be predicted. The preset cloud service storage capacity demand prediction model is used to predict the user's cloud service storage capacity demand based on the feature dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data, and the trend dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data at adjacent times.

2. The method according to claim 1, characterized in that The inputting the historical data into a preset cloud service storage capacity demand prediction model for processing to predict the cloud service storage capacity demand of the to-be-predicted user includes: Determine the decomposition data of the user's historical cloud service usage behavior and the decomposition data of the user's historical operator service usage behavior based on the analysis of the user's historical cloud service usage behavior data and the user's historical operator service usage behavior data; Inputting the user's historical cloud service usage behavior data, the user's historical operator service usage behavior data, the decomposed data of the user's historical cloud service usage behavior, and the decomposed data of the user's historical operator service usage behavior into a feature dependency module in a preset cloud service storage capacity demand prediction model to obtain feature information and feature dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data; Inputting the feature information and feature dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data into the time-dependent fusion pyramid attention module in the preset cloud service storage capacity demand prediction model to obtain the trend dependency of the adjacent time between the user's cloud service usage behavior data and the user's operator service usage behavior data; Based on the trend dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data in adjacent time periods, the cloud service storage capacity demand of the to-be-predicted user is predicted.

3. The method according to claim 2, characterized in that The determining of the decomposition data of the user's historical cloud service usage behavior and the decomposition data of the user's historical operator service usage behavior based on the analysis of the user's historical cloud service usage behavior data and the user's historical operator service usage behavior data includes: Based on the user's historical cloud service usage behavior data, the operator's historical service information, and the user's historical operator service usage behavior data, determine a first trend graph corresponding to the user's historical cloud service usage behavior data and a second trend graph corresponding to the user's historical operator service usage behavior data; The first trend graph and the second trend graph are respectively decomposed to determine the decomposed data of the user's historical cloud service usage behavior and the decomposed data of the user's historical operator service usage behavior, wherein the decomposed data of the user's historical cloud service usage behavior includes a first steady trend sequence and a first non-steady trend sequence, and the decomposed data of the user's historical operator service usage behavior includes a second steady trend sequence and a second non-steady trend sequence, wherein the steady trend sequence refers to data information with a steady trend in the trend graph, and the non-steady trend sequence refers to data information with a trend fluctuation greater than a fluctuation threshold in the trend graph.

4. The method according to claim 3, characterized in that: The step of inputting the user's historical cloud service usage behavior data, the user's historical operator service usage behavior data, the decomposed data of the user's historical cloud service usage behavior, and the decomposed data of the user's historical operator service usage behavior into a feature dependency module in a preset cloud service storage capacity demand prediction model to obtain feature information and feature dependency between the user's cloud service usage behavior data and the user's operator service usage behavior data includes: Obtaining first eigenvectors corresponding to the first stationary trend sequence and the first non-stationary trend sequence, and second eigenvectors corresponding to the second stationary trend sequence and the second non-stationary trend sequence; The first feature vector and the second feature vector are input into a feature dependency module in a preset cloud service storage capacity demand prediction model for analysis to obtain feature information and feature dependency, wherein the feature dependency module includes multiple stacked convolutional neural network layers.

5. The method according to claim 4, characterized in that The step of inputting the feature information and feature dependency between the user cloud service usage behavior data and the user operator service usage behavior data into the time dependency fusion pyramid attention module in the preset cloud service storage capacity demand prediction model to obtain the trend dependency of the adjacent time between the user cloud service usage behavior data and the user operator service usage behavior data includes: The feature information and feature dependency are input into the time-dependent fusion pyramid attention module in the preset cloud service storage capacity demand prediction model to obtain the first stationary trend sequence and the first adjacent time trend dependency of the first non-stationary trend sequence of the first trend graph corresponding to the user cloud service usage behavior data, and the second adjacent time trend dependency of the second stationary trend sequence and the second non-stationary trend sequence of the second trend graph corresponding to the user's operator service usage behavior data. The time-dependent fusion pyramid attention module includes multiple stack blocks.

6. The method according to claim 1, characterized in that The method further comprises: Based on the user's historical cloud service usage behavior data, obtaining the social behavior usage data of the user to be predicted in the cloud service; Based on the social behavior usage data, determining the social communication value of the user to be predicted; Using a clustering algorithm, all users are divided into multiple sub-user groups with three dimensional characteristics: user value, capacity consumption, and activity level. Each sub-user group has a multi-dimensional label coefficient, and users in the same sub-user group correspond to the same label coefficient. Determining the resource scheduling priority of each user in each of the sub-user groups based on the multi-dimensional label coefficient of each of the sub-user groups; Based on the resource scheduling priority and the cloud service storage capacity demand of the user to be predicted, cloud service storage capacity is allocated to the user to be predicted.

7. The method according to claim 6, characterized in that The determining the social communication value of the user to be predicted based on the social behavior usage data includes: Based on the social behavior usage data, determining the social activity of the user to be predicted and the resource dissemination influence of the user to be predicted; The social communication value of the user to be predicted is determined based on the social activity of the user to be predicted and the resource communication influence of the user to be predicted.

8. The method according to claim 6, characterized in that Using a clustering algorithm, all users are divided into multiple sub-user groups with three dimensional characteristics: user value, capacity consumption, and activity level, including: Determine user value based on the operator's historical service information and static data in the cloud service, wherein the static data in the cloud service includes subscription data of products in the cloud service; Determine the capacity consumption corresponding to each user based on the operator's historical service information and the user's historical operator service behavior data; Determining the user's activity level based on the user's historical cloud service usage behavior data and the user's historical operator service usage behavior data; A clustering algorithm is used to divide all users into multiple sub-user groups with characteristics in three dimensions: user value, capacity consumption, and activity level based on the user value, capacity consumption, and activity level.

9. The method according to claim 6, characterized in that The method further comprises: Monitor the user's usage of the allocated space capacity within a scheduling cycle; During the scheduling period, if the user does not perform preset usage behavior, the resource scheduling priority of the user is lowered, and the number of times the user's priority is lowered is recorded, and the number is used as a new feature for determining the weight coefficient of the user's resource scheduling priority.

10. The method according to claim 1, characterized in that After acquiring the historical data of the user to be predicted, the method further includes: Preprocess the historical data of the user to be predicted.

Images