Method and device for detecting cost on game cloud, storage medium and electronic equipment

Abstract

This application discloses a method, apparatus, storage medium, and electronic device for detecting cloud costs in games. The method includes: responding to a cloud cost detection request triggered by a virtual game, acquiring at least one type of game data for the virtual game, wherein the cloud cost detection request is used to request detection of cloud costs incurred by the virtual game using cloud resources or cloud services; determining key game data categories from the at least one type of game data based on the correlation between various types of game data and cloud costs; and using the key game data categories to detect cloud costs, obtaining a detection result, wherein the detection result indicates whether the cloud costs of the virtual game are normal or abnormal. This application solves the technical problem of low accuracy in detecting game cloud costs.

Description

Technical Field

[0001] This application relates to the field of computers, and more specifically, to a method, apparatus, storage medium, and electronic device for detecting cloud gaming costs. Background Technology

[0002] In scenarios involving cloud cost assessment for games, cloud costs are often estimated by acquiring relevant game data. However, due to the high complexity of game scenarios and the diverse and varied types of related data, significant errors can easily occur in the estimated cloud costs, affecting the accuracy of game cloud cost assessment and resulting in low accuracy. Therefore, the accuracy of game cloud cost assessment remains low.

[0003] There is currently no effective solution to the above problems. Summary of the Invention

[0004] This application provides a method, apparatus, storage medium, and electronic device for detecting game cloud costs, in order to at least solve the technical problem of low accuracy in detecting game cloud costs.

[0005] According to one aspect of the embodiments of this application, a method for detecting cloud costs of a game is provided, comprising: responding to a cloud cost detection request triggered by a virtual game, acquiring at least one type of game data of the virtual game, wherein the cloud cost detection request is used to request detection of cloud costs incurred by the virtual game using cloud resources or cloud services; determining key game data from the at least one type of game data based on the degree of correlation between various types of game data in the at least one type of game data and the cloud costs; and detecting the cloud costs using the key game data to obtain a detection result, wherein the detection result is used to indicate whether the cloud costs of the virtual game are normal or abnormal.

[0006] According to another aspect of the embodiments of this application, a device for detecting cloud costs of a game is also provided, comprising: an acquisition unit, configured to acquire at least one type of game data of the virtual game in response to a cloud cost detection request triggered by a virtual game, wherein the cloud cost detection request is used to request detection of cloud costs incurred by the virtual game using cloud resources or cloud services; a determination unit, configured to determine key game data from the at least one type of game data based on the degree of correlation between various types of game data in the at least one type of game data and the cloud costs; and a detection unit, configured to detect the cloud costs using the key game data and obtain a detection result, wherein the detection result is used to indicate whether the cloud costs of the virtual game are normal or abnormal.

[0007] As an optional solution, the aforementioned determining unit includes: a first acquisition module, used to acquire a first degree of correlation between a first type of game data in the at least one type of game data and the cloud cost before determining key game data from the at least one type of game data based on the correlation between various types of game data in the at least one type of game data and the cloud cost, wherein the first type of game data is static game data; and a second acquisition module, used to acquire a second degree of correlation between a second type of game data in the at least one type of game data and the cloud cost, wherein the second type of game data is dynamic game data.

[0008] As an optional solution, the second acquisition module includes: a first acquisition submodule for acquiring first trend information of the second type of game data changing over time; a second acquisition submodule for acquiring second trend information of the cloud cost changing over time; and a third acquisition submodule for acquiring the trend similarity between the first trend information and the second trend information, wherein the trend similarity is positively correlated with the second correlation degree.

[0009] As an optional solution, the third acquisition submodule includes: an acquisition subunit, used to acquire the curve fitting degree between the first trend curve and the second trend curve, wherein the curve fitting degree is positively correlated with the trend similarity.

[0010] As an optional solution, the detection unit includes: an input module for inputting the time-series game data into a pre-trained time-series large-scale model, wherein the pre-trained time-series large-scale model is trained using multiple samples and is used to identify the time dependencies in the input data to output corresponding prediction results; a third acquisition module for acquiring the predicted cloud cost output by the pre-trained time-series large-scale model, wherein the predicted cloud cost is the predicted cloud cost incurred by the virtual game using cloud resources or cloud services in the future time period; and a fourth acquisition module for acquiring the detection results based on the predicted cloud cost.

[0011] As an optional solution, the fourth acquisition module includes: a fourth acquisition submodule, used to acquire the cost difference between the predicted cloud cost and the actual cloud cost; a fifth acquisition submodule, used to acquire a first detection result when the cost difference is less than a preset threshold, wherein the first detection result indicates that the cloud cost of the virtual game is normal; and a sixth acquisition submodule, used to acquire a second detection result when the cost difference is greater than or equal to the preset threshold, wherein the second detection result indicates that the cloud cost of the virtual game is abnormal.

[0012] As an optional solution, the above input module includes: a first scaling and quantization submodule, used to scale and quantize the time-series game data after inputting it into the trained time-series large model, and convert it into multiple discrete first marker words; a sampling submodule, used to sample the multiple discrete first marker words using an autoregressive prediction distribution through the trained time-series large model, and obtain a first probability prediction; and a second scaling and quantization submodule, used to map the first probability prediction back to the actual value, and scale and quantize it to obtain the predicted cloud cost.

[0013] As an optional solution, the input module includes: a seventh acquisition submodule, used to acquire first time-series data as the current sample before inputting the time-series game data into the trained time-series large model, wherein the multiple samples include the first time-series data; a third scaling and quantization submodule, used to scale and quantize the first time-series data to obtain multiple discrete second tokens; a prediction submodule, used to predict the next token for the tokens among the multiple discrete second tokens using the current time-series large model, to obtain the loss function corresponding to the current time-series large model, wherein the loss function is used to measure the difference between the result predicted by the current time-series large model and the actual sequence corresponding to the tokens among the multiple discrete second tokens; a selection submodule, used to select the second time-series data among the multiple samples as the current sample and the updated time-series large model as the current time-series large model if the loss function does not meet the convergence condition; and a determination submodule, used to determine the current time-series large model as the trained time-series large model if the loss function meets the convergence condition.

[0014] As an optional solution, the input module includes: a pre-training submodule, used to pre-train the initial time-series large model using multiple first sample data before inputting the time-series game data into the trained time-series large model, to obtain a pre-trained time-series large model, wherein the multiple samples include the multiple first sample data, and the first sample data are sample data from a general domain; and a deep training submodule, used to perform deep training on the pre-trained time-series large model using multiple second sample data, to obtain the trained time-series large model, wherein the multiple samples include the multiple second sample data, and the second sample data are sample data from the game domain, and the game domain is the domain in which the virtual game is located.

[0015] As an optional solution, the detection unit includes: a display module, used to display an abnormality prompt message when the detection result indicates that the cloud cost of the virtual game is abnormal, after the detection result is obtained by using the key game data to detect the cloud cost, wherein the abnormality prompt message is used to indicate that the cloud cost of the virtual game is abnormal, and the abnormality prompt message is set to be prohibited from being displayed repeatedly within a preset time.

[0016] As an optional solution, the aforementioned determining unit includes: a fifth acquisition module, used to, when the aforementioned key game data includes third and fourth game data, after determining the key game data from the at least one type of game data based on the correlation degree between various types of game data in the at least one type of game data and the aforementioned cloud costs, acquire the third correlation degree between the third type of game data and the aforementioned cloud costs; a first allocation module, used to allocate a first weight to the third type of game data according to the aforementioned third correlation degree; a sixth acquisition module, used to acquire the fourth correlation degree between the fourth type of game data and the aforementioned cloud costs; a second allocation module, used to allocate a second weight to the fourth type of game data according to the aforementioned fourth correlation degree; an integration module, used to integrate the third type of game data and the fourth type of game data using the aforementioned first weight and the aforementioned second weight to obtain key game data; and a detection module, used to detect the aforementioned cloud costs using the aforementioned key game data.

[0017] According to another aspect of the embodiments of this application, a computer program product or computer program is provided, which includes computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform the above-described method for detecting game cloud costs.

[0018] According to another aspect of the embodiments of this application, an electronic device is also provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the above-described method for detecting cloud-based game costs through the computer program.

[0019] In this embodiment of the application, in response to a cloud cost detection request triggered by a virtual game, at least one type of game data of the virtual game is obtained, wherein the cloud cost detection request is used to request the detection of cloud costs incurred by the virtual game using cloud resources or cloud services; based on the degree of correlation between various types of game data in the at least one type of game data and the cloud costs, key game data is determined from the at least one type of game data; using the key game data, the cloud costs are detected to obtain a detection result, wherein the detection result is used to indicate whether the cloud costs of the virtual game are normal or abnormal.

[0020] Specifically, by acquiring at least one type of game data from virtual games, and then by analyzing the correlation between this at least one type of game data and cloud costs, key game data is obtained. This key game data is then used to detect cloud costs. In other words, by filtering through diverse and complex game data, key game data is obtained. This ensures that cloud cost estimation is performed using only key data, and the estimated cloud costs are compared with actual cloud costs as the detection result. This avoids the influence of non-key data on cloud cost estimation, thus improving the technical efficiency of game cloud cost detection and solving the technical problem of low accuracy in game cloud cost detection. Attached Figure Description

[0021] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0022] Figure 1 This is a schematic diagram of an application environment for an optional method for detecting game cloud costs according to an embodiment of this application;

[0023] Figure 2 This is a schematic diagram of the flow of an optional method for detecting game cloud costs according to an embodiment of this application;

[0024] Figure 3 This is a schematic diagram of an optional method for detecting game cloud costs according to an embodiment of this application;

[0025] Figure 4 This is a schematic diagram of another optional method for detecting game cloud costs according to an embodiment of this application;

[0026] Figure 5 This is a schematic diagram of another optional method for detecting game cloud costs according to an embodiment of this application;

[0027] Figure 6This is a schematic diagram of another optional method for detecting game cloud costs according to an embodiment of this application;

[0028] Figure 7 This is a schematic diagram of another optional method for detecting game cloud costs according to an embodiment of this application;

[0029] Figure 8 This is a schematic diagram of another optional method for detecting game cloud costs according to an embodiment of this application;

[0030] Figure 9 This is a schematic diagram of another optional method for detecting game cloud costs according to an embodiment of this application;

[0031] Figure 10 This is a schematic diagram of another optional method for detecting game cloud costs according to an embodiment of this application;

[0032] Figure 11 This is a schematic diagram of another optional method for detecting game cloud costs according to an embodiment of this application;

[0033] Figure 12 This is a schematic diagram of an optional game cloud cost detection device according to an embodiment of this application;

[0034] Figure 13 This is a schematic diagram of the structure of an optional electronic device according to an embodiment of this application. Detailed Implementation

[0035] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0036] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0037] According to one aspect of the embodiments of this application, a method for detecting game cloud costs is provided. Optionally, as an optional implementation, the above-described method for detecting game cloud costs can be applied to, but is not limited to, [examples of other methods]. Figure 1 The environment shown may include, but is not limited to, user equipment 102 and server 112. User equipment 102 may include, but is not limited to, a display 104, a processor 106 and a memory 108. Server 112 includes a database 114 and a processing engine 116.

[0038] The specific process can be summarized in the following steps:

[0039] Step S102: User equipment 102 obtains a cloud cost detection request;

[0040] Step S104: Send the cloud cost detection request to server 112 via network 110;

[0041] In steps S106-S110, server 112 obtains at least one type of game data of virtual game through processing engine 116, then determines key type of game data from the at least one type of game data, and then uses the key type of game data to detect cloud cost and obtain detection results.

[0042] In step S112, the detection result is sent to the user equipment 102 via the network 110. The user equipment 102 obtains the detection result through the processor 106, displays the detection result on the display 104, and stores the detection result in the memory 108.

[0043] remove Figure 1 Beyond the examples shown, the terminal devices described above can be terminal devices configured with a target client, including but not limited to at least one of the following: mobile phones (such as Android phones, iOS phones, etc.), laptops, tablets, PDAs, MIDs (Mobile Internet Devices), PADs, desktop computers, smart TVs, etc. The target client can be a video client, instant messaging client, browser client, educational client, etc. The networks described above can include, but are not limited to, wired networks and wireless networks. The wired networks include local area networks (LANs), metropolitan area networks (MANs), and wide area networks (WANs). The wireless networks include Bluetooth, Wi-Fi, and other networks that enable wireless communication. The server described above can be a single server, a server cluster consisting of multiple servers, or a cloud server. The above is merely an example, and no limitations are imposed in this embodiment.

[0044] Alternatively, as an alternative implementation method, such as Figure 2As shown, the method for detecting game cloud costs can be performed by an electronic device, such as... Figure 1 The user equipment or server shown includes the following specific steps:

[0045] S202, in response to a cloud cost detection request triggered by a virtual game, obtain at least one type of game data of the virtual game, wherein the cloud cost detection request is used to request the detection of cloud costs incurred by the virtual game using cloud resources or cloud services;

[0046] S204, Based on the correlation between various types of game data in at least one type of game data and cloud costs, identify key types of game data from at least one type of game data;

[0047] S206. Using key game data, cloud costs are detected and the results are obtained. The results indicate whether the cloud costs of virtual games are normal or abnormal.

[0048] Optionally, virtual games can be, but are not limited to, any form of game software or program that runs on an electronic device.

[0049] Optionally, cloud resources may be, but are not limited to, computing resources provided as a service over a network, and may include, but are not limited to, virtual machine instances, storage space, network bandwidth, etc.

[0050] Optionally, cloud services may include, but are not limited to, various services provided by cloud resource providers over a network.

[0051] Optionally, cloud costs may, but are not limited to, the costs incurred by a cloud user for using a specific amount of cloud resources or cloud services within a specific time period.

[0052] Optionally, a cloud cost detection request can be understood, but is not limited to, as a request to estimate the cloud costs incurred by the currently used cloud services or cloud resources by obtaining historical or operational data.

[0053] Optionally, the detection results may, but are not limited to, be obtained by comparing the estimated cloud costs with the actual cloud costs to determine whether the cloud costs of virtual games are normal or abnormal.

[0054] It's important to note that cloud costs for virtual games differ from those in conventional sectors. Cloud costs can fluctuate significantly due to operational events or game anomalies, leading to substantial changes in game data. Therefore, relying solely on historical cloud costs to estimate virtual game costs may result in significant errors.

[0055] To illustrate further, suppose a virtual game A updates its version every three months. During a particular update, a significant amount of new game content is added, attracting many players to play online simultaneously. As a result, the number of concurrent online players increases significantly, as do other data related to the virtual game. This increase in data will have a substantial impact on actual cloud costs, leading to a significant increase in actual cloud costs.

[0056] Therefore, relying solely on historical cloud costs to estimate cloud costs for version updates may lead to significant errors. For example, if only data from the past month is used for cost estimation, and no version updates have occurred in the month preceding the update, with relatively unchanged game content, user game data will be stable. Consequently, the estimated cloud costs will differ considerably from the actual cloud costs on the day of the update. This demonstrates that relying solely on historical cloud costs to estimate the cloud costs of virtual games can result in substantial errors.

[0057] In this embodiment, when making cloud cost estimates, in addition to using historical cloud costs for preliminary cloud cost estimates, key game data that have a significant impact on cloud costs are identified from various types of game data in the at least one type of game data. Then, based on the preliminary cloud cost estimates and the key game data, a secondary cloud cost estimate is made, which improves the accuracy of cloud cost prediction in game scenarios.

[0058] In an optional embodiment, when the cloud cost management system receives a cloud cost detection request, it first obtains at least one type of game data from the game operation database. The at least one type of game data may include, but is not limited to, at least one of the following: number of online players, number of registered users, user online time, user operations, active user ratio, and game version update information. This data can reflect the game's operating status and user activities, and has a direct or indirect impact on the consumption of cloud resources. It can be used as one of the bases for cloud cost estimation and detection.

[0059] It should be noted that when only one type of data exists in the at least one type of game data obtained, such as type A data, if the correlation between type A data and cloud costs meets the preset conditions, the related data will be identified as type A data; if the correlation between type A data and cloud data does not meet the preset conditions, the historical cloud costs and type A data will be weighted and merged to form key game data; or, at least one type of game data of virtual games will be re-obtained, and the correlation of the newly obtained data will be re-determined to meet the preset threshold.

[0060] In addition, when there are two or more data types among the acquired data, such as data type B and data type C, the correlation between data type B and data type C and cloud costs is determined. If the correlation between only one of the data types B and data type C reaches a preset threshold, then that data type that reaches the preset threshold is designated as key game data. If both data types B and data type C reach the preset threshold, then according to the correlation between data types B and data type C, both data types can be designated as key data, or the data with the highest correlation can be designated as key data, or the two data types can be weighted and merged to form key data. When the correlation between data type B and data type C does not reach the preset threshold, the historical cloud costs and data types B and C are weighted and merged to form key game data, or at least one type of game data from the virtual game is reacquired, and the correlation of the newly acquired data is re-determined to meet the preset threshold.

[0061] Furthermore, based on various game data within at least one category of game data, and by analyzing the correlation between these acquired data and cloud costs, key data categories are determined according to the degree of correlation between these data and cloud costs. Optionally, key data categories can be understood, but are not limited to, game data categories that have a specific correlation with changes in cloud costs during the operation of virtual games.

[0062] Finally, after identifying key game data, cloud costs are monitored. The monitoring method involves comparing the estimated cloud costs of virtual games with the actual cloud costs incurred. If the deviation exceeds a preset threshold, the system determines the cost is abnormal, generates an alarm, notifies operations personnel, or automatically triggers cost optimization measures. Conversely, if the estimated and actual values ​​are within a reasonable range, the system considers the cost normal and no special action is required.

[0063] Further examples, such as Figure 3 As shown, after receiving the detection request 304 sent by the user device 302, the game data 308 corresponding to the game 306 is first obtained. The game data 308 contains registered user data 310, online user data 312 and version information data 314. Finally, the online user data 312 is identified as key data 316, and the detection result 318 is obtained based on the key data 316.

[0064] According to the embodiments provided in this application, in response to a cloud cost detection request triggered by a virtual game, at least one type of game data of the virtual game is obtained, wherein the cloud cost detection request is used to request the detection of cloud costs incurred by the virtual game using cloud resources or cloud services; based on the degree of correlation between various types of game data in the at least one type of game data and cloud costs, key types of game data are determined from the at least one type of game data; using the key types of game data, cloud costs are detected to obtain detection results, wherein the detection results are used to indicate whether the cloud costs of the virtual game are normal or abnormal.

[0065] Specifically, by acquiring at least one type of game data from virtual games, and then by analyzing the correlation between this at least one type of game data and cloud costs, key game data is obtained. This key game data is then used to detect cloud costs. In other words, by filtering through diverse and complex game data, key game data is obtained. This ensures that cloud cost estimation is performed using only key data, and the estimated cloud costs are compared with actual cloud costs as the detection result. This avoids the influence of non-key data on cloud cost estimation, thus improving the technical efficiency of game cloud cost detection.

[0066] As an optional approach, before determining the key game data category from at least one type of game data based on the correlation between various game data types within that category and cloud costs, the method further includes:

[0067] S1-1, Obtain the first correlation degree between the first type of game data in at least one type of game data and cloud cost, wherein the first type of game data is static game data;

[0068] S1-2, Obtain the second type of game data from at least one type of game data, and the second degree of correlation between the second type of game data and cloud costs, wherein the second type of game data is dynamic game data.

[0069] In optional embodiments, the first type of game data may be, but is not limited to, static game data. It may refer to data categories in a virtual game that do not dynamically change with the game's running state, and may include, but is not limited to, game type, game version information, etc. Optionally, the first degree of correlation may be, but is not limited to, used to express the strength of the correlation between static game data and cloud costs.

[0070] In optional embodiments, the second type of game data may, but is not limited to, dynamic game data, and may, but is not limited to, data categories showing real-time changes in the game's running status, including, but not limited to, the number of online players, average online time, and behavioral data. This type of data can instantly reflect the game's activity and behavior, and is directly related to the real-time demand for cloud resources and cloud services. Optionally, the second degree of correlation may, but is not limited to, be used to express the strength of the correlation between dynamic game data and cloud costs.

[0071] It's important to note that static game data, such as game version information and game type, while relatively stable over a certain period, can still significantly impact cloud costs. For example, different game versions may use different cloud service configurations, leading to varying cost bases; and game type can also affect player online time and resource consumption.

[0072] Furthermore, dynamic game data, such as real-time online player counts and behavioral data, directly reflects the game's current activity level and user interaction. Due to the instantaneous and volatile nature of this data, it is directly related to fluctuations in cloud costs. For example, a sudden increase in the number of online players may require more computing resources, leading to an increase in cloud costs.

[0073] The embodiments provided in this application obtain a first degree of correlation between a first type of game data (static game data) and cloud costs from at least one type of game data; and obtain a second degree of correlation between a second type of game data (dynamic game data) and cloud costs from at least one type of game data. By obtaining the degree of correlation between static and dynamic game data and cloud costs, the technical objective of providing a basis for filtering key types of data is achieved, thereby improving the accuracy of key type data filtering.

[0074] As an optional approach, obtain a second type of game data from at least one type of game data, and determine the second degree of correlation between this second type of game data and cloud costs, including:

[0075] S2-1, Obtain the first trend information of the second type of game data changing over time;

[0076] S2-2, Obtain the second trend information on cloud costs over time;

[0077] S2-3, obtain the trend similarity between the first trend information and the second trend information, wherein the trend similarity is positively correlated with the second correlation degree.

[0078] In an optional embodiment, the first trend information may, but is not limited to, refer to the trend of the second type of game data changing over time, and may, but is not limited to, identifying the increase or decrease trend, fluctuation cycle, outliers, etc. of the data through time series analysis.

[0079] In optional embodiments, the second trend information may, but is not limited to, the trend of cloud cost data over time, which can reveal the long-term change pattern and short-term fluctuation of cloud costs.

[0080] In optional embodiments, trend similarity may be, but is not limited to, measuring the degree of similarity between first trend information and second trend information, and is positively correlated with the second correlation degree. By calculating the similarity index between trend information, the correlation between the second type of game data and cloud costs in terms of trends can be evaluated.

[0081] It should be noted that, in order to accurately assess the impact of dynamic game data on cloud costs, a time series analysis was first performed on the second type of game data to obtain its primary trend information over time. Then, a similar time series analysis was performed on the cloud cost data to obtain its secondary trend information over time. After obtaining the trend information for both the second type of game data and the cloud cost data, the correlation between the two sets of data in terms of trends was quantified by calculating trend similarity. The higher the trend similarity, the more significant the impact of dynamic game data on cloud costs.

[0082] To illustrate further, suppose an MMORPG (Massively Multiplayer Online Role-Playing Game) needs to optimize resource usage and cost control through cloud cost monitoring during its operation. First, dynamic game data needs to be collected, such as the number of online players and participation in in-game activities. Then, time series analysis is used to obtain the primary trend information of these data changes over time. Simultaneously, a similar analysis is performed on the cloud cost data to extract the secondary trend information of costs changing over time. Then, by calculating the trend similarity between the primary and secondary trend information, it is found that the trend of changes in the number of online players highly matches the fluctuations in cloud costs. This means that the number of online players is a key data type for cloud costs, indicating a high degree of correlation between the number of online players and cloud costs.

[0083] The embodiments provided in this application obtain first trend information of the change of a second type of game data over time; obtain second trend information of the change of cloud costs over time; and obtain the trend similarity between the first trend information and the second trend information, wherein the trend similarity and the second correlation degree are positively correlated. By obtaining the trend information of dynamic game data and cloud costs changing over time, the technical objective of assessing the impact of dynamic game data on cloud costs is achieved, thereby improving the technical effect of improving the accuracy of cloud cost detection.

[0084] As an optional approach, the first trend information includes a first trend curve of the second type of game data changing over time in a preset chart, and the second trend information includes a second trend curve of cloud costs changing over time in a preset chart. The trend similarity between the first and second trend information is obtained, including:

[0085] Obtain the curve fit between the first trend curve and the second trend curve, where the curve fit is positively correlated with the trend similarity.

[0086] In an optional embodiment, the first trend curve may be, but is not limited to, a curve representing the change of dynamic game data over time, and is a visualization of the dynamic characteristics of game data in the time dimension.

[0087] In optional embodiments, the second trend curve may, but is not limited to, be a curve representing the change of cloud costs over time, and may, but is not limited to, display information such as long-term trends, seasonal fluctuations, and short-term anomalies in cloud costs.

[0088] In optional embodiments, the curve fit can be, but is not limited to, used to measure the degree of agreement between the first trend curve and the second trend curve, and can be, but is not limited to, positively correlated with the trend similarity.

[0089] It's important to note that there's a positive correlation between curve fit and trend similarity. A high curve fit indicates that the second type of game data closely matches the changing trends of cloud costs, meaning there's a high degree of trend similarity and that this type of game data has a significant impact on cloud costs. Conversely, a low curve fit indicates a lower degree of agreement between this type of game data and the changing trends of cloud costs, meaning there's a low degree of trend similarity and that this type of game data may have a smaller direct impact on costs.

[0090] The embodiments provided in this application obtain the curve fitting degree between a first trend curve and a second trend curve, wherein the curve fitting degree is positively correlated with the trend similarity. By obtaining the curve fitting degree between the first trend curve and the second trend curve, the technical objective of representing the correlation between the second type of game data and cloud costs through the curve fitting degree is achieved, thereby improving the accuracy of cloud cost detection.

[0091] As an optional approach, when key game data includes time-series game data, the key game data can be used to detect cloud costs, and the detection results include:

[0092] S3-1, Input the time series game data into the trained time series large model, where the trained time series large model is obtained by training multiple samples and is used to identify the time dependence in the input data in order to output the corresponding prediction results;

[0093] S3-2, Obtain the predicted cloud cost output by the trained time series large model, where the predicted cloud cost is the cloud cost incurred by the virtual game using cloud resources or cloud services in the future time period.

[0094] S3-3, based on predicted cloud costs, obtains detection results.

[0095] In optional embodiments, time-series game data may refer to game data that changes over time, and may include, but is not limited to, dynamic data such as the number of online players, user in-game spending behavior, and user activity.

[0096] In optional embodiments, the trained time series model may be, but is not limited to, a machine learning model that has been pre-trained with a large amount of sample data and is capable of identifying and predicting complex patterns and trends in time series data, and may be, but is limited to, used to process time series game data and predict the cloud costs of virtual games.

[0097] In optional embodiments, time dependence can refer to, but is not limited to, the inherent regularity and pattern exhibited by data changes over time, such as periodicity, trends, and seasonality. In time-series game data, time dependence is a key factor in cost prediction because it reflects the impact of player behavior and game activities on the demand for cloud resources or cloud services.

[0098] In optional embodiments, predicting cloud costs can be, but is not limited to, processing game data through a time-series large model to predict the costs incurred by virtual games using cloud resources or cloud services.

[0099] It should be noted that by collecting and organizing time-series game data and inputting it into a pre-trained time-series model, this model possesses the ability to identify complex patterns and predict trends in time-series data. Under the model's processing, the time dependencies of the time-series game data are identified, and corresponding prediction results are output. Furthermore, the predicted cloud costs output by the model are compared with the actual cloud costs to determine whether any cost anomalies exist, serving as the detection result.

[0100] The embodiments provided in this application input time-series game data into a pre-trained time-series large-scale model. This model, trained using multiple samples, identifies time dependencies in the input data and outputs corresponding prediction results. The predicted cloud costs output by the pre-trained model are then obtained. These predicted cloud costs are the cloud costs incurred by the virtual game using cloud resources or services in the future. Based on the predicted cloud costs, detection results are obtained. By inputting time-series data into the pre-trained time-series large-scale model, the technical objective of predicting cloud costs based on the time-series data and obtaining detection results based on the predicted cloud costs is achieved, thereby improving the accuracy of cloud cost detection results.

[0101] As an optional approach, given the actual cloud costs of the virtual game over a future time period, the detection results can be obtained based on the predicted cloud costs, including:

[0102] S4-1, Obtain the cost difference between the predicted cloud cost and the actual cloud cost;

[0103] S4-2, if the cost difference is less than a preset threshold, obtain the first detection result, wherein the first detection result is used to indicate that the cloud cost of the virtual game is normal;

[0104] S4-3, if the cost difference is greater than or equal to a preset threshold, obtain a second detection result, wherein the second detection result is used to indicate that the cloud cost of the virtual game is abnormal.

[0105] In an optional embodiment, the actual cloud cost may be, but is not limited to, the actual cost incurred by the virtual game in using cloud resources or cloud services during actual operation.

[0106] In an optional embodiment, the first detection result may be used, but is not limited to, to indicate that the cloud costs of the virtual game are normal. Optionally, the second detection result may be used, but is not limited to, to indicate that the cloud costs of the virtual game are abnormal.

[0107] It should be noted that the process involves obtaining the predicted cloud cost output by the model and comparing this predicted value with the actual cloud cost to determine if any cost anomalies exist. This process calculates the cost difference between the predicted and actual cloud costs, which can be achieved, but is not limited to, using root mean square error or mean square error. If the deviation between the predicted and actual costs exceeds a preset anomaly threshold, it indicates an anomaly in the virtual game's cloud costs, requiring an alert to be issued to management personnel to ensure timely notification of the anomaly.

[0108] The embodiments provided in this application obtain the cost difference between the predicted cloud cost and the actual cloud cost. If the cost difference is less than a preset threshold, a first detection result is obtained, indicating that the cloud cost of the virtual game is normal. If the cost difference is greater than or equal to the preset threshold, a second detection result is obtained, indicating that the cloud cost of the virtual game is abnormal. By determining whether there is an anomaly in the cloud cost through the cost difference between the predicted and actual cloud costs, the technical objective of issuing an alarm to management personnel when an anomaly exists is achieved, thereby ensuring the timeliness of alarms issued to management personnel.

[0109] As an alternative approach, after inputting time-series game data into a pre-trained time-series large-scale model, the method also includes:

[0110] S5-1 uses a trained temporal large model to scale and quantize temporal game data, transforming it into multiple discrete first-label words;

[0111] S5-2: Using a well-trained time series model, autoregression is used to sample multiple discrete first-label words from the prediction distribution to obtain the first probability prediction.

[0112] S5-3 maps the first probability prediction back to the actual value, and then scales and quantizes it to obtain the predicted cloud cost.

[0113] It should be noted that when processing time-series game data, the first step is to convert the trained time-series model into multiple discrete first-order tokens as a preprocessing step. Next, in the model prediction stage, the trained time-series model performs autoregressive sampling on the discrete first-order tokens based on the prediction distribution to obtain the first probability prediction. This first probability prediction reflects the probability of the tokens appearing after the time-series game data is converted. Finally, the first probability prediction generated by the model is mapped back to the actual numerical representation and further scaled and quantized to obtain the final estimate of the predicted cloud cost. This converts the discrete prediction result into a continuous numerical representation, facilitating comparison and analysis with the actual cloud cost to form the detection result.

[0114] The embodiments provided in this application utilize a trained temporal series model to scale and quantize temporal game data, converting it into multiple discrete first markers. The trained temporal series model then uses autoregression to sample these discrete first markers from the prediction distribution, obtaining a first probability prediction. This first probability prediction is mapped back to the actual value, scaled, and quantized to obtain the predicted cloud cost. This achieves the technical objective of converting discrete prediction results into continuous numerical representations, facilitating comparison and analysis with actual cloud costs as detection results, thereby improving the efficiency of obtaining cloud cost prediction results.

[0115] As an alternative approach, before inputting the time-series game data into the pre-trained time-series large-scale model, the method also includes:

[0116] Perform the following steps until a well-trained time-series model is obtained:

[0117] S6-1, Obtain the first time series data as the current sample, where multiple samples include the first time series data;

[0118] S6-2, scales and quantizes the first time series data to obtain multiple discrete second tokens;

[0119] S6-3, using the current time series big model, predict the next token for the token in the discrete multiple second tokens, and obtain the loss function corresponding to the current time series big model. The loss function is used to measure the difference between the result predicted by the current time series big model and the actual sequence corresponding to the token in the discrete multiple second tokens.

[0120] S6-4, if the loss function does not meet the convergence condition, the second time series data among multiple samples is taken as the current sample, and the updated time series model is taken as the current time series model.

[0121] S6-5, if the loss function satisfies the convergence condition, the current time series model is determined as the trained time series model.

[0122] In optional embodiments, the first time-series data may be, but is not limited to, a subset of time-series data, and may be, but is not limited to, the initial input for model training. Optionally, the loss function may be, but is not limited to, a mathematical function used to evaluate the difference between the model's predictions and the actual data. Optionally, the convergence condition may be, but is not limited to, a criterion used during model training to determine whether the model has reached a stable state.

[0123] Optionally, in the early stages of model training, it is necessary to first obtain the initial time-series data as the current sample. This initial sample is selected from multiple samples and used for the first training of the model, from which time-series features and patterns are extracted.

[0124] After acquiring the initial time-series data, preprocessing is required, including scaling and quantization, to convert it into a discrete representation suitable for model processing. Scaling can be mean scaling, maximum scaling, etc., to scale the data to a uniform range; quantization converts the scaled data into discrete second-order tokens, which helps the model recognize and remember patterns in the time series.

[0125] After initial data processing, the current large-scale time-series model is used to predict the next token for the discrete second token. This prediction process generates prediction results through an autoregressive mechanism based on the model's current parameters. Subsequently, a loss function is calculated based on the difference between the predicted token sequence and the actual sequence.

[0126] It should be noted that model training is an iterative optimization process. If the loss function does not meet the convergence condition, i.e., its rate of change is still higher than the preset threshold, it indicates that the model parameters have not yet reached their optimal state. When the loss function meets the convergence condition during model training, i.e., its rate of change is lower than the preset threshold, it indicates that the model parameters have stabilized, the expected training objective has been achieved, and the model can be used for cloud cost detection while ensuring the accuracy of the detection results.

[0127] The embodiments provided in this application obtain first time-series data as the current sample, wherein multiple samples include the first time-series data; the first time-series data is scaled and quantized to obtain multiple discrete second tokens; using the current time-series large model, the next token is predicted for the tokens among the discrete second tokens, resulting in a loss function corresponding to the current time-series large model, wherein the loss function measures the difference between the prediction result obtained by the current time-series large model and the actual sequence corresponding to the tokens among the discrete second tokens; if the loss function does not meet the convergence condition, the second time-series data among the multiple samples is used as the current sample, and the updated time-series large model is used as the current time-series large model; if the loss function meets the convergence condition, the current time-series large model is determined as the trained time-series large model. By iteratively training the time-series large model using the first and second time-series data, the technical objective of obtaining a trained time-series large model when the loss function of the time-series large model reaches the convergence condition is achieved, thereby ensuring the accuracy of the time-series large model in cloud cost detection.

[0128] As an alternative approach, before inputting the time-series game data into the pre-trained time-series large-scale model, the method also includes:

[0129] S7-1 uses multiple first sample data to pre-train the initial time series large model to obtain the pre-trained time series large model. The multiple samples include multiple first sample data, which are sample data from the general domain.

[0130] S7-2 uses multiple second sample data to perform deep training on the pre-trained temporal large model to obtain a trained temporal large model. The multiple samples include multiple second sample data, which are sample data from the game domain, and the game domain is the domain in which virtual games are located.

[0131] In an optional embodiment, the first sample data is sample data from a general domain. Optionally, the second sample data is sample data from the gaming domain, which is the domain in which virtual games exist.

[0132] Optionally, in the initial stage of model training, the initial large-scale time series model can be pre-trained using multiple first-sample data from a general domain. The purpose of pre-training is to improve the model's generalization ability, enabling it to handle time series data from different sources.

[0133] Following the pre-training phase, multiple second-sample datasets from the gaming domain are used to perform deep training on the pre-trained time-series model. The purpose of deep training is to further improve the model's prediction accuracy for game time-series data, enabling it to capture specific patterns and regularities within the gaming domain. After deep training, the model parameters are finely tuned, allowing for more accurate predictions of changes in various data points during virtual game operations.

[0134] The embodiments provided in this application utilize multiple first sample data sets to pre-train an initial time-series large-scale model, resulting in a pre-trained time-series large-scale model. The multiple samples include multiple first sample data sets, which are sample data from a general domain. Then, multiple second sample data sets are used to perform deep training on the pre-trained time-series large-scale model, resulting in a trained time-series large-scale model. These second sample data sets include multiple second sample data sets, which are sample data from the game domain, specifically the domain of virtual games. By performing pre-training and deep training on the time-series large-scale model, the technical objective of enabling the model to more accurately predict changes in various data during virtual game operations is achieved, thereby improving the technical effectiveness of the time-series large-scale model in cloud cost detection.

[0135] As an optional approach, when the detection results indicate abnormal cloud costs for virtual games, after utilizing key game data to detect cloud costs and obtaining the detection results, the method further includes:

[0136] Display error messages, which are used to indicate abnormal cloud costs for virtual games. These error messages are set to not be displayed repeatedly within a preset time period.

[0137] In an optional embodiment, the anomaly alert message may be, but is not limited to, an alert message generated by the system during the cloud cost detection process, and may be used to inform that there may be anomalies in the cloud cost.

[0138] It should be noted that once an anomaly is detected in the cloud costs of the virtual game, the system will display an anomaly message to remind administrators to pay attention to and address the anomaly. However, to avoid frequent anomaly messages interfering with administrators and affecting work efficiency, the system manages the display of anomaly messages by setting a preset time. During this time, the same anomaly will not be displayed repeatedly. This reduces the interference of frequent anomaly messages with administrators, thereby improving the efficiency of administrators in handling cost anomalies.

[0139] The embodiments provided in this application display abnormality alerts, which are used to indicate abnormal cloud costs in virtual games. These alerts are set to be disabled from repeated display within a preset time period. By disabling repeated display of the abnormality alerts within a preset time period, the technical objective of avoiding frequent abnormality alerts from interfering with administrators is achieved, thereby improving the efficiency of administrators in handling cost anomalies.

[0140] As an optional approach, when the key game data includes both third and fourth types of game data, after determining the key game data from at least one type of game data based on the correlation between various game data within that at least one type of game data and cloud costs, the method further includes:

[0141] S8-1, Obtain the third type of game data and the third degree of correlation between it and cloud costs;

[0142] S8-2, assign the first weight to the third type of game data according to the third degree of correlation;

[0143] S8-3, acquiring the fourth type of game data and the fourth degree of correlation between it and cloud costs;

[0144] S8-4, according to the fourth degree of correlation, assign the second weight to the fourth type of game data;

[0145] S8-5 uses the first and second weights to integrate and process the third and fourth types of game data to obtain key game data;

[0146] Using key game data, cloud costs were assessed, and the results included:

[0147] S8-6 uses key game data to detect cloud costs.

[0148] It should be noted that the cloud-based cost of virtual games is often influenced by more than one type of game data. There may be multiple or different categories of game data from various dimensions that can have a significant impact on the cloud-based cost of virtual games. When different dimensions of data exist, it is necessary to integrate the different categories of game data effectively to enhance the intuitive representation of the third and fourth categories of game data and improve the efficiency of cloud-based cost detection.

[0149] The embodiments provided in this application obtain a third type of game data and a third degree of correlation between it and cloud costs; assign a first weight to the third type of game data according to the third degree of correlation; obtain a fourth type of game data and a fourth degree of correlation between it and cloud costs; assign a second weight to the fourth type of game data according to the fourth degree of correlation; integrate the third type of game data and the fourth type of game data using the first weight and the second weight to obtain key game data; and use the key game data to detect cloud costs. By assigning weights to the third and fourth types of data, the technical objective of integrating different categories of game data to obtain key game data is achieved, thereby improving the technical effect of improving the efficiency of cloud cost detection.

[0150] As an alternative, the aforementioned methods for detecting game cloud costs can be applied to scenarios involving anomaly detection in game business costs.

[0151] In optional embodiments, for example Figure 4 The diagram shows the overall process of the cloud cost detection algorithm in the gaming field. First, gaming field indicators and cloud cost indicators are input into the time series large model framework and feature fine-tuning is performed. Then, the predicted cost value is obtained and compared with the obtained actual cost value. If there is an anomaly, a cost threshold alarm will be issued.

[0152] In optional embodiments, for example Figure 5 The diagram illustrates the implementation process of another cloud cost detection algorithm in the computing game field in this embodiment. In the anomaly detection of cloud costs, the cost data for the day and historical data are first obtained. Then, the historical data is input into the machine learning algorithm for training to obtain the cost prediction value. The cost prediction value is then compared with the cost data for the day. If there is an anomaly, a cost threshold alarm is issued.

[0153] In an optional embodiment, this embodiment performs visual analysis on a continuous monitoring system. The analysis involves analyzing third-party cloud time-series data for the game business. First, game-specific, cost-related attributes are added as features to the third-party cloud time-series data, such as Peak Concurrent Users (PCU) and version. Then, the time-series data with added features is fine-tuned using a large time-series model framework. The fine-tuned model is then used to make predictions for the following day. Finally, alarms are configured in the continuous monitoring system to complete anomaly detection.

[0154] In an optional embodiment, this embodiment consists of the following three processes: 1. Adding game-specific attribute features to cloud cost time series data; 2. Fine-tuning and prediction of the time series large model; 3. Sending cost anomaly alarms.

[0155] Optionally, all data in this system originates from game business time-series data. The overall system flow is as follows: Figure 6 As shown, in the feature selection stage, cost time series data is first obtained, then game features related to cost are selected, and then the time series data with added features is input into the time series large model in the cost prediction stage for fine-tuning. Then, the cost estimate is obtained by predicting through the fine-tuned time series data model. In the anomaly alarm stage, the actual cost value is obtained and compared with the estimated value, and then it is determined whether an alarm needs to be issued. If so, an alarm is issued; otherwise, the process will converge.

[0156] To further illustrate, in the gaming industry, which encompasses a wide variety of businesses, the cost anomaly detection algorithm for this embodiment needs to be applicable to as many businesses as possible. Therefore, this embodiment first needs to list common features in the gaming industry. Based on this, this embodiment also requires time-series data as much as possible. This allows this embodiment to observe the trends of various features and see if these trends affect costs. Thus, the entire feature selection process for the gaming industry is as follows: Figure 7 As shown: First, obtain game-related features, then list domain-specific features based on game-related features, then determine whether the domain-specific features are time-series data. If so, determine whether they are related to trends and costs. If so, use the above-mentioned domain-specific features as game domain features.

[0157] Following this logic, for example, we can first list the characteristics of the gaming industry: such as online activity, version, category, and other industry-specific features. Here, online activity is time-series data. Next, this embodiment will observe the impact of online data trends on cost trends. This embodiment selects a real game business PCU and cloud resources. By plotting, it can be seen that the upward and downward trends of PCU are correlated with costs. Therefore, this embodiment adds PCU as a feature of the gaming industry to the cost time-series data. Other features can also be added using the same logic; this time, we only calculate the impact after adding the PCU feature.

[0158] Optional, such as Figure 8 The following figures illustrate the trends in cloud cost (Actual Cost) and gaming sector characteristics (Actual PCU), with specific examples as follows:

[0159] In the following data, the first column is Date, the second column is Cost, and the third column is PCU (Game Domain Features).

[0160] 2024 / 7 / 1, 34051, 166140

[0161] 2024 / 7 / 2, 33212, 171367

[0162] 2024 / 7 / 3, 32716, 170576

[0163] 2024 / 7 / 4, 30954, 170252

[0164] 2024 / 7 / 5, 32635, 198316

[0165] 2024 / 7 / 6, 33146, 179693

[0166] In an optional embodiment, the logic for the entire cost prediction is to fine-tune the time series large model and then select the fine-tuned time series large model for prediction. The time series large model is a language model framework that minimizes adjustments for time series prediction. It segments the time series into discrete intervals by simply scaling and quantizing real values, thus training the language model on the "language of time series" without changing the model architecture.

[0167] Further examples, such as Figure 9 The following is an introduction to the operating principle of a large-scale time series model:

[0168] 1. Time series tokenization: The time series language model scales and quantizes historical time series data, converting it into discrete tokens. This allows the time series data to be treated as text in the language model.

[0169] 2. Training Process: By inputting discrete tokens into the temporal large language model for training, the model gains cost prediction capabilities. Then, the cross-entropy loss function is used to train the language model on these tokenized sequences. The model learns to predict the next token in the sequence using the classification distribution on the token vocabulary.

[0170] 3. Prediction process: During inference, discrete tokens are input into the temporal big language model. The prediction distribution is sampled through autoregression to obtain token samples to generate probability predictions. These predictions are then mapped back to actual values ​​and scaled and quantized to obtain the final prediction.

[0171] It should be noted that in actual data usage, data from 0701 to 1101 was used, with the period before 10.08 selected as training data to fine-tune the time series model, and the subsequent data used as the logic for prediction by the large model.

[0172] For example, the following is a sample of the filtered game business:

[0173] In the following data, the first column is Date, the second column is Cost, and the third column is PCU (Game Domain Features).

[0174] 2024 / 7 / 1, 34051, 166140

[0175] 2024 / 7 / 2, 33212, 171367

[0176] 2024 / 7 / 3, 32716, 170576

[0177] 2024 / 7 / 4, 30954, 170252

[0178] 2024 / 7 / 5, 32635, 198316

[0179] 2024 / 7 / 6, 33146, 179693

[0180] This embodiment primarily involves converting the training data into a time-series data format and reading the configuration parameters. Then, it runs the fine-tuning steps provided by the large-scale time-series model. Running the fine-tuning configuration generates a fine-tuned model based on the large-scale time-series model, which is then loaded for prediction.

[0181] To further illustrate, suppose the cloud cost value (Cost) on October 7th, 2024 (October 7th) is 36760, and the online PCU is 184841. After loading a finely tuned time-series large-scale model, using the cost value and the online PCU as tensor inputs, the final value for October 8th is 37065. However, the estimated value based on historical data is 39743. This demonstrates that the large-scale model prediction incorporating PCU achieves better results than anomaly detection based on historical data. Optional features include... Figure 10 As shown, the final October real value comparison curve is as follows: Figure 10 As shown, A is the future cloud cost prediction based on historical data for anomaly detection of cloud costs, and B is the future cloud cost prediction based on anomaly detection of cloud costs after incorporating PCU.

[0182] It can be observed that, in comparison, the cost estimate based on key game data in this embodiment is more accurate in both trend and numerical value, and the calculation of root mean square error (RMSE or RMSD) indeed proves this point.

[0183]

[0184] Where n represents the number of data samples, y i Represents the true value. This represents the predicted value. The smaller the RMSE value, the better the model's predictive performance, because it means that the deviation between the model's predicted value and the actual value is smaller.

[0185] To address the discrepancy between predicted and actual values ​​calculated across multiple services in game time-series datasets:

[0186] Methods used Root mean square error AWS PRMSE 2844.1131892618714 Chronos RMSE 2195.7840285419693

[0187] Therefore, the cost estimate using this embodiment is more accurate. Once the cost estimate is obtained, the final step is alarm detection.

[0188] Optionally, once a cost estimate is available, the actual cost can be compared with the estimate each day on the visual continuous monitoring system, and an alarm can be triggered based on the user-defined threshold.

[0189] To further illustrate, suppose in actual calculations a certain business incurs an actual cost of 50,000 on November 20th. Compared to the predicted cost of 40,000, a threshold of 8,000 is set as the alarm threshold. In this case, the business will issue an alarm. Simultaneously, the system itself incorporates some time-based detection mechanisms; for example, a new alarm can only be issued five minutes after the current alarm is triggered.

[0190] The above embodiments can better align with the cost anomalies desired by the game business side. The cost anomaly detection proposed in this solution can predict the cost value for the following day, compare it with the actual cost value after it appears, and finally issue an alarm.

[0191] To further illustrate, using visual detection to detect cost-effective alerts, such as... Figure 11 As shown in the figure, this alarm will display the alarm level. This alarm is a critical alarm. The information also includes the time of the first anomaly and the time of the most recent anomaly. The content will specify the total duration (SUM), current value, space, dimension, and related information of this anomaly. Among them, the dimension includes the game parameter, which is used to indicate which game the anomaly occurred in. The related information includes functions such as viewing details, alarm confirmation, alarm blocking, and data retrieval.

[0192] Through the embodiments of this application, by training a large temporal language model and then fine-tuning it, PCU data from the gaming domain can be incorporated into the prediction of cloud costs. The prediction is then compared with actual cloud costs. If anomalies are found, an alert is issued to notify administrators of the abnormal cloud costs. This achieves the technical objective of utilizing gaming domain data in cloud cost prediction, thereby enhancing the technical effect of cloud cost prediction in the gaming domain.

[0193] It is understood that in the specific embodiments of this application, data such as user information are involved. When the above embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0194] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0195] According to another aspect of the embodiments of this application, a device for detecting game cloud costs for implementing the above-described method for detecting game cloud costs is also provided. Figure 12 As shown, the device includes:

[0196] The acquisition unit 1202 is used to acquire at least one type of game data of the virtual game in response to a cloud cost detection request triggered by the virtual game, wherein the cloud cost detection request is used to request the detection of cloud costs generated by the virtual game using cloud resources or cloud services.

[0197] The determining unit 1204 is used to determine key game data from at least one type of game data based on the degree of correlation between various types of game data in at least one type of game data and cloud costs;

[0198] The detection unit 1206 is used to detect cloud costs using key game data and obtain detection results, which indicate whether the cloud costs of virtual games are normal or abnormal.

[0199] For specific implementation examples, please refer to the example shown in the above-described method for detecting game cloud costs; these examples will not be repeated here.

[0200] As an optional solution, the determining unit 1204 includes: a first acquisition module, used to acquire a first correlation degree between a first type of game data in at least one type of game data and cloud costs before determining key game data from at least one type of game data based on the correlation degree between various types of game data in at least one type of game data and cloud costs, wherein the first type of game data is static game data; and a second acquisition module, used to acquire a second correlation degree between a second type of game data in at least one type of game data and cloud costs, wherein the second type of game data is dynamic game data.

[0201] For specific implementation examples, please refer to the example shown in the above-described method for detecting game cloud costs; these examples will not be repeated here.

[0202] As an optional solution, the second acquisition module includes: a first acquisition submodule for acquiring first trend information of the second type of game data changing over time; a second acquisition submodule for acquiring second trend information of cloud costs changing over time; and a third acquisition submodule for acquiring the trend similarity between the first trend information and the second trend information, wherein the trend similarity is positively correlated with the second correlation degree.

[0203] For specific implementation examples, please refer to the example shown in the above-described method for detecting game cloud costs; these examples will not be repeated here.

[0204] As an optional solution, the third acquisition submodule includes: an acquisition subunit, used to acquire the curve fitting degree between the first trend curve and the second trend curve, wherein the curve fitting degree is positively correlated with the trend similarity.

[0205] For specific implementation examples, please refer to the example shown in the above-described method for detecting game cloud costs; these examples will not be repeated here.

[0206] As an optional solution, the detection unit 1206 includes: an input module for inputting time-series game data into a pre-trained time-series large-scale model, wherein the pre-trained time-series large-scale model is trained using multiple samples and is used to identify the time dependencies in the input data to output corresponding prediction results; a third acquisition module for acquiring the predicted cloud cost output by the pre-trained time-series large-scale model, wherein the predicted cloud cost is the predicted cloud cost incurred by the virtual game using cloud resources or cloud services in the future time period; and a fourth acquisition module for acquiring the detection result based on the predicted cloud cost.

[0207] For specific implementation examples, please refer to the example shown in the above-described method for detecting game cloud costs; these examples will not be repeated here.

[0208] As an optional solution, the fourth acquisition module includes: a fourth acquisition submodule, used to acquire the cost difference between the predicted cloud cost and the actual cloud cost; a fifth acquisition submodule, used to acquire a first detection result when the cost difference is less than a preset threshold, wherein the first detection result is used to indicate that the cloud cost of the virtual game is normal; and a sixth acquisition submodule, used to acquire a second detection result when the cost difference is greater than or equal to the preset threshold, wherein the second detection result is used to indicate that the cloud cost of the virtual game is abnormal.

[0209] For specific implementation examples, please refer to the example shown in the above-described method for detecting game cloud costs; these examples will not be repeated here.

[0210] As an optional solution, the input module includes: a first scaling and quantization submodule, used to scale and quantize the time-series game data after inputting it into a trained time-series large model, and convert it into multiple discrete first tokens; a sampling submodule, used to sample the multiple discrete first tokens using an autoregressive prediction distribution through the trained time-series large model to obtain a first probability prediction; and a second scaling and quantization submodule, used to map the first probability prediction back to the actual value and perform scaling and quantization to obtain the predicted cloud cost.

[0211] For specific implementation examples, please refer to the example shown in the above-described method for detecting game cloud costs; these examples will not be repeated here.

[0212] As an optional approach, the input module includes: a seventh acquisition submodule, used to acquire first time-series data as the current sample before inputting time-series game data into the trained time-series large model, wherein multiple samples include the first time-series data; a third scaling and quantization submodule, used to scale and quantize the first time-series data to obtain multiple discrete second-order tags; a prediction submodule, used to predict the next tag for the tags among the multiple discrete second-order tags using the current time-series large model, obtaining the loss function corresponding to the current time-series large model, wherein the loss function is used to measure the difference between the prediction result obtained by the current time-series large model and the actual sequence corresponding to the tags among the multiple discrete second-order tags; a selection submodule, used to take the second time-series data among the multiple samples as the current sample and the updated time-series large model as the current time-series large model when the loss function does not meet the convergence condition; and a determination submodule, used to determine the current time-series large model as the trained time-series large model when the loss function meets the convergence condition.

[0213] For specific implementation examples, please refer to the example shown in the above-described method for detecting game cloud costs; these examples will not be repeated here.

[0214] As an optional approach, the input module includes: a pre-training submodule, used to pre-train the initial temporal model using multiple first sample data before inputting temporal game data into the trained temporal model, to obtain a pre-trained temporal model, wherein the multiple samples include multiple first sample data, which are sample data from a general domain; and a deep training submodule, used to perform deep training on the pre-trained temporal model using multiple second sample data, to obtain a trained temporal model, wherein the multiple samples include multiple second sample data, which are sample data from the game domain, and the game domain refers to the domain in which the virtual game is located.

[0215] For specific implementation examples, please refer to the example shown in the above-described method for detecting game cloud costs; these examples will not be repeated here.

[0216] As an optional solution, the detection unit 1206 includes: a display module, used to detect cloud costs using key game data when the detection result indicates that the cloud costs of the virtual game are abnormal, and then display an abnormality prompt message after obtaining the detection result. The abnormality prompt message is used to indicate that the cloud costs of the virtual game are abnormal, and the abnormality prompt message is set to be prohibited from being displayed repeatedly within a preset time.

[0217] For specific implementation examples, please refer to the example shown in the above-described method for detecting game cloud costs; these examples will not be repeated here.

[0218] As an optional solution, the determining unit 1204 includes: a fifth acquisition module, used to, when the key game data includes third and fourth game data, after determining the key game data from at least one type of game data based on the correlation degree between various types of game data in at least one type of game data and cloud costs, acquire the third correlation degree between the third type of game data and cloud costs; a first allocation module, used to allocate a first weight to the third type of game data according to the third correlation degree; a sixth acquisition module, used to acquire the fourth correlation degree between the fourth type of game data and cloud costs; a second allocation module, used to allocate a second weight to the fourth type of game data according to the fourth correlation degree; an integration module, used to integrate the third and fourth type of game data using the first and second weights to obtain key game data; and a detection module, used to detect cloud costs using the key game data.

[0219] For specific implementation examples, please refer to the example shown in the above-described method for detecting game cloud costs; these examples will not be repeated here.

[0220] According to another aspect of the embodiments of this application, an electronic device for implementing the above-described method for detecting game cloud costs is also provided. This electronic device may, but is not limited to, […]. Figure 1 The user equipment 102 or server 112 shown in the figure, in this embodiment, is taken as an example of an electronic device, namely user equipment 102. Further, as shown in the figure... Figure 13 As shown, the electronic device includes a memory 1302 and a processor 1304. The memory 1302 stores a computer program, and the processor 1304 is configured to execute the steps of any of the above method embodiments through the computer program.

[0221] In an optional embodiment, the aforementioned electronic device may be located in at least one of a plurality of network devices in a computer network.

[0222] In an optional embodiment, the processor described above may be configured to perform the following steps via a computer program:

[0223] S1, in response to a cloud cost detection request triggered by a virtual game, obtain at least one type of game data of the virtual game, wherein the cloud cost detection request is used to request the detection of cloud costs generated by the virtual game using cloud resources or cloud services;

[0224] S2, based on the correlation between various types of game data in at least one type of game data and cloud costs, identify key types of game data from at least one type of game data;

[0225] S3 uses key game data to detect cloud costs and obtain detection results, which indicate whether the cloud costs of virtual games are normal or abnormal.

[0226] Alternatively, as those skilled in the art will understand, Figure 13 The structure shown is for illustrative purposes only. Figure 13 This does not limit the structure of the aforementioned electronic devices. For example, the electronic device may also include components that are more... Figure 13 The more or fewer components shown (such as network interfaces, etc.), or having the same Figure 13 The different configurations shown.

[0227] The memory 1302 can be used to store software programs and modules, such as the program instructions / modules corresponding to the game cloud cost detection method and apparatus in this embodiment. The processor 1304 executes various functional applications and data processing by running the software programs and modules stored in the memory 1302, thereby realizing the aforementioned game cloud cost detection method. The memory 1302 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 1302 may further include memory remotely located relative to the processor 1304, and these remote memories can be connected to electronic devices via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof. Specifically, the memory 1302 may be used, but is not limited to, to store at least one type of game data, key game data, and detection results, etc. As an example, such as... Figure 13 As shown, the memory 1302 may include, but is not limited to, the acquisition unit 1202, the determination unit 1204, and the detection unit 1206 in the aforementioned game cloud cost detection device. Furthermore, it may include, but is not limited to, other module units in the aforementioned game cloud cost detection device, which will not be elaborated upon in this example.

[0228] Optionally, the transmission device 1306 described above is used to receive or send data via a network. Specific examples of the network described above may include wired networks and wireless networks. In one example, the transmission device 1306 includes a Network Interface Controller (NIC), which can be connected to other network devices and a router via a network cable to communicate with the Internet or a local area network. In another example, the transmission device 1306 is a Radio Frequency (RF) module, used for wireless communication with the Internet.

[0229] In addition, the aforementioned electronic device also includes: a display 1308 for displaying information such as at least one type of game data, key type game data, and detection results; and a connection bus 1310 for connecting various module components in the aforementioned electronic device.

[0230] In other embodiments, the aforementioned user equipment or server can be a node in a distributed system, wherein the distributed system can be a blockchain system, which is a distributed system formed by connecting multiple nodes through network communication. The nodes can form a peer-to-peer network, and any form of computing device, such as a server, user equipment, or other electronic device, can become a node in the blockchain system by joining this peer-to-peer network.

[0231] According to one aspect of this application, a computer program product is provided, comprising a computer program / instructions containing program code for performing the methods shown in the flowchart. In such embodiments, the computer program can be downloaded and installed from a network via a communication component, and / or installed from a removable medium. When the computer program is executed by a central processing unit, it performs various functions provided in embodiments of this application.

[0232] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0233] It should be noted that the computer system of the electronic device is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0234] A computer system includes a Central Processing Unit (CPU), which performs various appropriate actions and processes based on programs stored in Read-Only Memory (ROM) or loaded from RAM. ROM also stores various programs and data required for system operation. The CPU, ROM, and RAM are interconnected via a bus. Input / output interfaces (I / O interfaces) are also connected to the bus.

[0235] The following components are connected to the input / output interface: input sections including keyboards, mice, etc.; output sections including cathode ray tubes (CRTs), liquid crystal displays (LCDs), and speakers; storage sections including hard drives; and communication sections including network interface cards such as LAN cards and modems. The communication section performs communication processing via a network such as the Internet. Drives are also connected to the input / output interface as needed. Removable media, such as disks, optical discs, magneto-optical discs, semiconductor memories, etc., are installed on the drive as needed so that computer programs read from them can be installed into the storage section as required.

[0236] Specifically, according to embodiments of this application, the processes described in the various method flowcharts can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication component, and / or installed from a removable medium. When the computer program is executed by a central processing unit, it performs various functions defined in the system of this application.

[0237] According to one aspect of this application, a computer-readable storage medium is provided, wherein a processor of a computer device reads computer instructions from the computer-readable storage medium, and executes the computer instructions, causing the computer device to perform the methods provided in the various alternative implementations described above.

[0238] In an optional embodiment, the computer-readable storage medium described above may be configured to store a computer program for performing the following steps:

[0239] S1, in response to a cloud cost detection request triggered by a virtual game, obtain at least one type of game data of the virtual game, wherein the cloud cost detection request is used to request the detection of cloud costs generated by the virtual game using cloud resources or cloud services;

[0240] S2, based on the correlation between various types of game data in at least one type of game data and cloud costs, identify key types of game data from at least one type of game data;

[0241] S3 uses key game data to detect cloud costs and obtain detection results, which indicate whether the cloud costs of virtual games are normal or abnormal.

[0242] Optionally, in embodiments of this application, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0243] In optional embodiments, those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware of an electronic device. The program can be stored in a computer-readable storage medium, which may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc.

[0244] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0245] If the integrated units in the above embodiments are implemented as software functional units and sold or used as independent products, they can be stored in the aforementioned computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause one or more computer devices (which may be personal computers, servers, or network devices, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application.

[0246] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0247] In the several embodiments provided in this application, it should be understood that the disclosed user equipment can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between units or modules may be electrical or other forms.

[0248] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0249] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0250] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for detecting cost on a game cloud, characterized in that, include: In response to a cloud cost detection request triggered by a virtual game, at least one type of game data of the virtual game is obtained, wherein the cloud cost detection request is used to request the detection of cloud costs incurred by the virtual game in using cloud resources or cloud services; Based on the correlation between various types of game data in the at least one type of game data and the cloud cost, key game data are determined from the at least one type of game data; Using the key game data, the cloud costs are detected to obtain detection results, wherein the detection results are used to indicate whether the cloud costs of the virtual game are normal or abnormal.

2. The method of claim 1, wherein, Before determining key game data from the at least one type of game data based on the correlation between various game data types within the at least one type of game data and the cloud cost, the method further includes: Obtain the first type of game data from the at least one type of game data and the first degree of correlation between it and the cloud cost, wherein the first type of game data is static game data; Obtain a second type of game data from the at least one type of game data, and a second degree of correlation between it and the cloud cost, wherein the second type of game data is dynamic game data.

3. The method of claim 2, wherein, The second correlation between obtaining the second type of game data from the at least one type of game data and the cloud cost includes: Obtain the first trend information of the second type of game data changing over time; Obtain the second trend information of the cloud cost change over time; Obtain the trend similarity between the first trend information and the second trend information, wherein the trend similarity is positively correlated with the second correlation degree.

4. The method of claim 3, wherein, The first trend information includes a first trend curve of the second type of game data changing over time in a preset chart, and the second trend information includes a second trend curve of the cloud cost changing over time in the preset chart. The step of obtaining the trend similarity between the first trend information and the second trend information includes: Obtain the curve fitting degree between the first trend curve and the second trend curve, wherein the curve fitting degree is positively correlated with the trend similarity.

5. The method of claim 1, wherein, When the key game data includes time-series game data, the step of using the key game data to detect the cloud cost and obtain the detection result includes: The time-series game data is input into a pre-trained time-series large model, wherein the pre-trained time-series large model is obtained by training multiple samples and is used to identify the time dependencies in the input data in order to output the corresponding prediction results. Obtain the predicted cloud cost output by the trained time series large model, wherein the predicted cloud cost is the cloud cost incurred by the virtual game in using cloud resources or cloud services in the future time period. The detection results are obtained based on the predicted cloud costs.

6. The method of claim 5, wherein, Given the actual cloud cost of the virtual game in the future time period, obtaining the detection result based on the predicted cloud cost includes: Obtain the cost difference between the predicted cloud cost and the actual cloud cost; If the cost difference is less than a preset threshold, a first detection result is obtained, wherein the first detection result is used to indicate that the cloud cost of the virtual game is normal; If the cost difference is greater than or equal to the preset threshold, a second detection result is obtained, wherein the second detection result is used to indicate that the cloud cost of the virtual game is abnormal.

7. The method of claim 5, wherein, After inputting the time-series game data into the trained time-series large model, the method further includes: The trained temporal model is used to scale and quantize the temporal game data, and convert it into multiple discrete first marker words. Using the trained time-series large model, an autoregressive method is used to sample the discrete multiple first tokens from the prediction distribution to obtain a first probability prediction; The first probability prediction is mapped back to the actual value, and then scaled and quantized to obtain the predicted cloud cost.

8. The method of claim 5, wherein, Before inputting the time-series game data into the trained time-series large model, the method further includes: Perform the following steps until the trained time series model is obtained: Obtain first time-series data as the current sample, wherein the plurality of samples include the first time-series data; The first time-series data is scaled and quantized to obtain multiple discrete second tokens; Using the current time series model, the next token is predicted for the tokens in the discrete multiple second tokens, and the loss function corresponding to the current time series model is obtained. The loss function is used to measure the difference between the result predicted by the current time series model and the actual sequence corresponding to the tokens in the discrete multiple second tokens. If the loss function does not meet the convergence condition, the second time series data among multiple samples is taken as the current sample, and the updated time series large model is taken as the current time series large model. If the loss function satisfies the convergence condition, the current temporal large model is determined as the trained temporal large model.

9. The method of claim 5, wherein, Before inputting the time-series game data into the trained time-series large model, the method further includes: The initial time series model is pre-trained using multiple first sample data to obtain the pre-trained time series model. The multiple samples include the multiple first sample data, which are sample data from a general domain. The pre-trained temporal large model is deep-trained using multiple second sample data to obtain the trained temporal large model. The multiple samples include the multiple second sample data, which are sample data from the game domain, and the game domain is the domain in which the virtual game is located.

10. The method according to any one of claims 1 to 9, characterized in that, When the detection result indicates that the cloud costs of the virtual game are abnormal, after using the key game data to detect the cloud costs and obtaining the detection result, the method further includes: The system displays an error message, which indicates an anomaly in the cloud costs of the virtual game. The error message is set to be disabled from being displayed repeatedly within a preset time period.

11. The method according to any one of claims 1 to 9, characterized in that, When the key game data includes a third type of game data and a fourth type of game data, after determining the key game data from the at least one type of game data based on the correlation between various types of game data in the at least one type of game data and the cloud cost, the method further includes: Obtain the third type of game data and the third degree of correlation between it and the cloud cost; Based on the third degree of correlation, a first weight is assigned to the third type of game data; Obtain the fourth type of game data and the fourth degree of correlation between it and the cloud cost; According to the fourth degree of correlation, a second weight is assigned to the fourth type of game data; Using the first weight and the second weight, the third type of game data and the fourth type of game data are integrated and processed to obtain key game data; The step of using the key game data to detect the cloud costs and obtain detection results includes: using the key game data to detect the cloud costs.

12. A device for detecting cost on a game cloud, characterized by, include: The acquisition unit is configured to acquire at least one type of game data of the virtual game in response to a cloud cost detection request triggered by the virtual game, wherein the cloud cost detection request is used to request the detection of cloud costs generated by the virtual game using cloud resources or cloud services; The determining unit is used to determine key game data from the at least one type of game data based on the degree of correlation between various types of game data in the at least one type of game data and the cloud cost; The detection unit is used to detect the cloud costs using the key game data and obtain detection results, wherein the detection results are used to indicate whether the cloud costs of the virtual game are normal or abnormal.

13. A computer readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein the program is executed by an electronic device to perform the method according to any one of claims 1 to 11.

14. A computer program product comprising computer programs / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method described in any one of claims 1 to 11.

15. An electronic device comprising a memory and a processor, characterized in that The memory stores a computer program, and the processor is configured to execute the method described in any one of claims 1 to 11 through the computer program.

Images