A dynamic pricing method and system for auction-based cloud servers based on threshold segmentation

By initializing the segmented threshold matrix, comprehensively evaluating user behavior, and generating equivalent mapping coefficients, the adaptability problem of the bidding cloud server pricing strategy was solved, resource utilization and user retention were improved, and the pricing strategy was matched with the market environment.

CN123118162APending Publication Date: 2026-11-17SHANGHAI XIAOCHI TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611558899.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-09
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In existing technologies, the pricing method for auction cloud servers lacks adaptive adjustment capabilities, cannot flexibly respond to changes in supply and demand, has low resource utilization, makes it difficult to reasonably reflect the contributions of high-quality users, and results in poor user retention and pricing fairness.

Method used

By initializing the segmented threshold matrix, comprehensively evaluating user behavior data, generating equivalent mapping coefficients, adjusting pricing strategies, forming a closed-loop feedback of reputation data, optimizing the boundaries of the segmented threshold matrix, and achieving dynamic pricing.

Benefits of technology

It improved resource utilization and supplier revenue, enhanced the retention rate of high-quality users and pricing fairness, and ensured that the pricing strategy was aligned with the market environment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123118162A_ABST
    Figure CN123118162A_ABST
Patent Text Reader

Abstract

This invention relates to the field of cloud computing technology, specifically disclosing a dynamic pricing method and system for auction cloud servers based on threshold segmentation. The method includes: initializing a segmented threshold matrix based on bidding request data and cloud resource idle status data in the current bidding cycle; comprehensively evaluating the target user's behavior data over historical periods to obtain dynamic reputation data; correcting the threshold matrix boundary based on the dynamic reputation data and generating equivalent mapping coefficients; performing an equivalent transformation on the original bid data using the equivalent mapping coefficients, replacing the original bid data with equivalent bid data to participate in the bidding ranking; and updating the dynamic reputation data based on the target user's resource release time information and correcting the threshold matrix boundary again. This invention can achieve closed-loop adaptive adjustment of the pricing strategy, improving the resource utilization and pricing fairness of auction cloud servers.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of cloud computing technology, and in particular to a dynamic pricing method and system for auction-based cloud servers based on threshold segmentation. Background Technology

[0002] In existing technologies, pricing methods for auction cloud servers typically employ fixed price adjustment steps or single threshold comparison strategies, lacking a dynamic segmentation mechanism based on cloud resource supply and demand and market bidding distribution. Such methods struggle to adaptively adjust segmentation strategies according to resource saturation and price dispersion, resulting in pricing rules that cannot flexibly respond to changes in supply and demand, thus reducing overall resource utilization and the potential for optimizing supplier revenue.

[0003] In existing technologies, resource allocation for auction-based cloud servers is mostly based solely on user bids, lacking a comprehensive evaluation of users' historical behavioral characteristics and a credit quantification mechanism. Even in the few schemes that introduce credit evaluation, credit data is not quantitatively linked to pricing threshold adjustments and equivalent bid transformations, and there is a lack of closed-loop feedback based on user resource release behavior to dynamically update credit data. This makes it difficult to reasonably reflect the long-term contributions of high-quality users in the competition, affecting user retention rates and pricing fairness. Therefore, there is an urgent need to develop a threshold-segmented dynamic pricing method and system for auction-based cloud servers to address the technical problems of the lack of adaptive adjustment between pricing strategies and supply and demand, the lack of quantitative linkage between credit evaluation and pricing parameters, and the lack of closed-loop feedback correction, thereby improving resource utilization, operational revenue, and user retention rates. Summary of the Invention

[0004] This invention provides a dynamic pricing method and system for auction cloud servers based on threshold segmentation, in order to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, this invention provides a dynamic pricing method for auction-based cloud servers based on threshold segmentation, comprising: S1. Initialize the segmented threshold matrix based on the bidding request data and cloud resource idle status data of the current bidding cycle; S2. A comprehensive evaluation of the target user's behavioral data over a historical period is performed to obtain the target user's dynamic reputation data; S3. Based on the dynamic reputation data, correct the boundaries of the segmented threshold matrix and generate equivalent mapping coefficients; S4. Based on the equivalent mapping coefficient, perform an equivalent transformation on the original bid data of the target user to obtain equivalent bid data, and use the equivalent bid data to replace the original bid data to participate in the bidding ranking of the current bidding cycle. S5. Based on the resource release time information of the target user, update the dynamic reputation data, and based on the updated dynamic reputation data, correct the boundary of the segmented threshold matrix again.

[0006] In a preferred embodiment, initializing the segmented threshold matrix based on the bidding request data and cloud resource idle status data of the current bidding cycle includes: Statistical fitting is performed on the bidding request data of the current bidding period to obtain the benchmark price data of the current bidding period; Based on the cloud resource idle status data of the current bidding period, determine the resource saturation level of the current bidding period; Based on the resource saturation level, determine the segmentation strategy; Based on the benchmark price data and the segmentation strategy, a segmentation threshold matrix is ​​generated, which contains multiple level boundary values.

[0007] In a preferred embodiment, determining the segmentation strategy based on the resource saturation level includes: Perform dispersion analysis on the bidding request data to obtain the price dispersion characteristics of the current bidding period; Based on the combination of the resource saturation level and the price dispersion characteristics, one of the preset multiple segmentation strategies is selected as the target segmentation strategy. The multiple segmentation strategies include at least the equilibrium segmentation strategy, the compression segmentation strategy, and the expansion segmentation strategy. The segmentation strategy includes the number of segments, the width ratio of each segment, and the upward and downward floating ratios of each segment boundary relative to the benchmark price data. When the resource saturation level is the first level and the price dispersion feature is the first dispersion feature, the equilibrium segmentation strategy is selected. When the resource saturation level is the second level and the price dispersion feature is the second dispersion feature, the compression segmentation strategy is selected; When the resource saturation level is level three and the price dispersion feature is the third dispersion feature, the expansion segmentation strategy is selected; The balanced segmentation strategy is configured to have a baseline number of segments and a proportionally distributed segment width. The compression segmentation strategy is configured such that the number of segments is less than the baseline number of segments and the width of each segment is in a contraction distribution relative to the balanced segmentation strategy. The expansion segmentation strategy is configured such that the number of segments is greater than the baseline number of segments and the width of each segment is expanded relative to the balanced segmentation strategy.

[0008] In a preferred embodiment, the step of comprehensively evaluating the target user's behavioral data over a historical period to obtain the target user's dynamic reputation data includes: The target user's behavioral data within a preset historical backtracking window is obtained. The multiple behavioral data includes at least task execution continuity feature data, interruption response initiative feature data, and resource consumption stability feature data. The task execution continuity feature data, interruption response initiative feature data, and resource consumption stability feature data are normalized and mapped respectively to obtain standardized feature values ​​for each dimension. Based on the preset weight coefficients of each dimension, the standardized feature values ​​of each dimension are weighted and aggregated to obtain the comprehensive reputation score of the target user. Based on the comprehensive reputation score, the reputation level corresponding to the target user is determined, and the comprehensive reputation score and reputation level are used together as dynamic reputation data.

[0009] In a preferred embodiment, the plurality of behavioral dimension data includes: The task execution continuity characteristic data includes at least the average duration of continuous resource occupation for a single bidding task by the target user within the historical backtracking window; The proactive interruption response data includes at least the frequency with which the target user actively saved checkpoints before the bidding task was forcibly interrupted within the historical backtracking window and the average response delay time when rejoining the bidding after the interruption. The resource occupancy stability characteristic data includes at least the average deviation rate between the actual occupancy time and the promised occupancy time of each bidding task by the target user within the historical backtracking window, as well as the task completion rate.

[0010] In a preferred embodiment, the step of correcting the boundaries of the segmented threshold matrix based on the dynamic reputation data and generating equivalent mapping coefficients includes: When the target user's credit rating is high, the predetermined boundary value corresponding to the high-price region in the segmented threshold matrix is ​​offset to expand the credit privilege data range in the segmented threshold matrix. The offset amount of the offset operation is positively mapped to the comprehensive credit score. Based on the target user's comprehensive reputation score, an equivalent mapping coefficient for the target user is generated.

[0011] In a preferred embodiment, generating the equivalent mapping coefficient of the target user based on the target user's comprehensive reputation score includes: The comprehensive credit score is mapped to a plurality of preset continuous credit score intervals, each credit score interval corresponding to a preset mapping coefficient value, and the mapping coefficient value increases monotonically as the credit score interval increases. The mapping coefficient between adjacent credit score intervals increases in a non-uniform step size, and the mapping coefficient is constrained to be no less than the lower limit of the first preset coefficient and no greater than the upper limit of the second preset coefficient.

[0012] In a preferred embodiment, the step of performing an equivalent transformation on the original bid data of the target user based on the equivalent mapping coefficient to obtain equivalent bid data, and using the equivalent bid data to replace the original bid data in the bidding ranking of the current bidding cycle, includes: When the credit rating of the target user is determined to be high, the original bid data and equivalent mapping coefficient of the target user are obtained. Based on the equivalent mapping coefficient, the original bid data is weighted to generate the equivalent bid data for the target user; The original bid data is replaced with the equivalent bid data and used as the bidding sorting key value for the target user in the current bidding cycle; Based on the bidding sorting key value, the target user is included in a global descending sorting queue along with other bidding users, and the resource allocation object for the current bidding period is determined according to the sorting result of the global descending sorting queue.

[0013] In a preferred embodiment, updating the dynamic reputation data based on the resource release time information of the target user, and then correcting the boundaries of the segmented threshold matrix again based on the updated dynamic reputation data, includes: Obtain the promised resource release time data and the actual resource release time data of the target user, and calculate the time deviation value between the actual resource release time data and the promised resource release time data; When the time deviation value is less than zero, it is determined that the target user has released resources in advance, and a positive reputation correction amount is generated based on the absolute value of the time deviation value. When the time deviation value is greater than zero, it is determined that the target user is delaying the release of resources, and a negative reputation correction amount is generated based on the absolute value of the time deviation value. The dynamic reputation data is incrementally updated based on the positive reputation correction amount or the negative reputation correction amount to obtain the updated dynamic reputation data. Based on the credit rating of the updated dynamic credit data, the predetermined boundary values ​​corresponding to the high-price region in the segmented threshold matrix are offset again to adjust the boundary range of the credit privilege data interval in the next bidding cycle.

[0014] To address the aforementioned problems, this invention also provides a dynamic pricing system for auction-based cloud servers based on threshold segmentation, the system comprising: The segmented threshold matrix initialization module is used to initialize the segmented threshold matrix based on the bidding request data and cloud resource idle status data of the current bidding cycle. The dynamic reputation assessment module is used to comprehensively evaluate the target user's behavioral data over a historical period to obtain the target user's dynamic reputation data. The mapping coefficient generation module is used to correct the boundaries of the segmented threshold matrix based on the dynamic reputation data and generate equivalent mapping coefficients. The bid equivalence transformation module is used to perform an equivalent transformation on the original bid data of the target user based on the equivalent mapping coefficient to obtain equivalent bid data, and use the equivalent bid data to replace the original bid data to participate in the bidding ranking of the current bidding cycle. The threshold iteration correction module is used to update the dynamic reputation data based on the resource release time information of the target user, and to correct the boundary of the segmented threshold matrix again according to the updated dynamic reputation data.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. In this invention, a target segmentation strategy is selected based on a combination of resource saturation levels and price dispersion characteristics, enabling the segmentation threshold matrix to adaptively adjust with supply and demand conditions. Benchmark price data is determined through statistical fitting, and a segmentation threshold matrix containing multiple level boundary values ​​is generated in conjunction with the segmentation strategy, ensuring that the pricing benchmark matches the actual market bidding distribution. By differentiating between compression and expansion segmentation strategies, the price range distribution pattern adapts to the current market environment, improving the resource utilization rate of the bidding cloud server and the supplier's revenue.

[0016] 2. In this invention, a comprehensive reputation score and reputation level are obtained by acquiring multiple behavioral dimension data, performing normalization mapping and weighted aggregation, thus enabling multi-dimensional objectivity in reputation assessment. The reputation level triggers a threshold matrix boundary offset to expand the reputation privilege data range, and the comprehensive reputation score generates an equivalent mapping coefficient to perform an equivalent transformation on the original bid data. This equivalent bid replaces the original bid in the bidding ranking, ensuring that the long-term contributions of high-quality users are reasonably reflected in the bidding ranking. Positive or negative reputation correction values ​​are generated by acquiring the time deviation between the resource release time and the promised release time, and the dynamic reputation data is updated. Based on the updated reputation level, the threshold boundary is offset again, forming a closed-loop iteration of reputation data with actual behavior, improving the retention rate of high-quality users and the fairness of pricing. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating a dynamic pricing method for auction cloud servers based on threshold segmentation, provided in an embodiment of the present invention. Figure 2 This is a functional block diagram of a dynamic pricing system for auction cloud servers based on threshold segmentation, provided in an embodiment of the present invention. The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0020] This application provides a threshold-based segmented dynamic pricing method for cloud hosting. The execution entity of this threshold-based segmented dynamic pricing method for cloud hosting includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the threshold-based segmented dynamic pricing method for cloud hosting can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0021] Reference Figure 1 The diagram shown is a flowchart illustrating a dynamic pricing method for auction-based cloud servers based on threshold segmentation, according to an embodiment of the present invention. In this embodiment, the dynamic pricing method for auction-based cloud servers based on threshold segmentation includes: S1. Initialize the segmented threshold matrix based on the bidding request data and cloud resource idle status data of the current bidding cycle; In this embodiment of the invention, the bidding request data of the current bidding period is statistically fitted to obtain the benchmark price data of the current bidding period.

[0022] The statistical fitting method uses the median, which sorts all the original bid data submitted by users in the current bidding period in ascending order of numerical value, and takes the value in the middle position after sorting as the benchmark price data.

[0023] When the total number of original bid data is odd, the middle value is taken as the benchmark price data.

[0024] When the total number of original bid data is even, the arithmetic mean of the two middle values ​​is taken as the benchmark price data.

[0025] The benchmark price data represents the concentrated trend of user bids during the current bidding cycle and serves as the central reference value for subsequently constructing the segmented threshold matrix.

[0026] Based on the cloud resource idle status data of the current bidding period, determine the resource saturation level of the current bidding period.

[0027] The cloud resource idle status data is the percentage of unoccupied resources in the cloud resource pool out of the total number of resources during the current bidding period.

[0028] The percentage value is compared with multiple preset saturation threshold ranges to determine the resource saturation level.

[0029] When the percentage value is greater than the first saturation threshold, the resource saturation level is determined to be the first level, which indicates that the cloud resources are in a state of supply exceeding demand.

[0030] When the percentage value is less than or equal to the first saturation threshold and greater than the second saturation threshold, the resource saturation level is determined to be the second level, which indicates that the cloud resources are in a state of supply and demand equilibrium.

[0031] When the percentage value is less than or equal to the second saturation threshold, the resource saturation level is determined to be the third level, which indicates that cloud resources are in short supply.

[0032] The value of the first saturation threshold is greater than the value of the second saturation threshold.

[0033] By performing a dispersion analysis on the bidding request data for the current bidding period, the price dispersion characteristics of the current bidding period can be obtained.

[0034] The dispersion analysis uses standard deviation statistics to calculate the degree of dispersion of all user-submitted original bid data relative to the benchmark price data within the current bidding period.

[0035] For each original bid data in the bidding request data, calculate the difference between it and the benchmark price data. Square each difference separately, sum them up, and then divide by the total number of bidding request data to obtain the variance. Take the square root of the variance to obtain the standard deviation.

[0036] The standard deviation is compared with multiple preset dispersion threshold intervals to determine the price dispersion characteristics.

[0037] When the standard deviation is less than the first dispersion threshold, the price dispersion feature is determined as the first dispersion feature. The first dispersion feature indicates that the price dispersion of the bidding request data is low.

[0038] When the standard deviation is greater than or equal to the first dispersion threshold and less than the second dispersion threshold, the price dispersion feature is determined to be the second dispersion feature. The second dispersion feature indicates that the price dispersion of the bidding request data is moderate.

[0039] When the standard deviation is greater than or equal to the second dispersion threshold, the price dispersion feature is determined to be the third dispersion feature. The third dispersion feature indicates that the price dispersion of the bidding request data is high.

[0040] The value of the first dispersion threshold is less than the value of the second dispersion threshold.

[0041] Based on the combination of resource saturation level and price dispersion characteristics, one of the preset segmentation strategies is selected as the target segmentation strategy.

[0042] Various segmentation strategies include balanced segmentation, compressed segmentation, and expanded segmentation.

[0043] When the resource saturation level is the first level and the price dispersion characteristic is the first dispersion characteristic, the equilibrium segmentation strategy is selected as the target segmentation strategy.

[0044] When the resource saturation level is the second level and the price dispersion characteristic is the second dispersion characteristic, the compression segmentation strategy is selected as the target segmentation strategy.

[0045] When the resource saturation level is level three and the price dispersion characteristic is the third dispersion characteristic, the expansion segmentation strategy is selected as the target segmentation strategy.

[0046] The equilibrium segmentation strategy includes the number of equilibrium segments, the equilibrium width ratio of each segment, and the equilibrium upward and downward multiples of each segment boundary relative to the benchmark price data.

[0047] The compression segmentation strategy includes the number of compression segments, the compression width ratio of each segment, and the compression upward and downward multiples of each segment boundary relative to the benchmark price data.

[0048] The expansion segmentation strategy includes the number of expansion segments, the expansion width ratio of each segment, and the expansion upward and downward multiples of each segment boundary relative to the benchmark price data.

[0049] The balanced segmentation strategy is configured with a baseline number of segments and proportionally distributed segment widths.

[0050] The number of balanced segments is the preset standard number of segments.

[0051] The equilibrium width ratio of each segment is distributed proportionally, that is, each segment occupies an equal width on the price axis.

[0052] The equilibrium upward floating rate is a preset standard upward floating rate, used to determine the maximum limit of each segment boundary above the benchmark price data.

[0053] The equilibrium downward floating rate is a preset standard downward floating rate, used to determine the minimum limit value of each segment boundary below the benchmark price data.

[0054] The compression segmentation strategy is configured such that the number of segments is less than the baseline number of segments and the width of each segment is distributed in a contraction manner relative to the balanced segmentation strategy.

[0055] The number of compressed segments is less than the number of balanced segments.

[0056] The compression width ratio of each segment is in a shrinking distribution relative to the equal distribution, that is, the width of each segment is reduced proportionally according to the preset shrinkage coefficient.

[0057] The compression upward floating ratio is less than the equilibrium upward floating ratio, which causes the price range above the benchmark price data to shrink under the compression segmentation strategy.

[0058] The absolute value of the compression downward floating ratio is less than the absolute value of the equilibrium downward floating ratio, which causes the price range below the benchmark price data to shrink under the compression segmentation strategy.

[0059] The expansion segmentation strategy is configured to have more segments than the baseline number of segments and the width of each segment is expanded relative to the balanced segmentation strategy.

[0060] The number of expanding segments is greater than the number of balanced segments.

[0061] The expansion width ratio of each segment is an expanded distribution relative to the equal distribution, that is, the width of each segment is expanded proportionally according to the preset expansion coefficient.

[0062] The expansion upward floating ratio is greater than the equilibrium upward floating ratio, which expands the width of the price range above the benchmark price data under the expansion segmentation strategy.

[0063] The absolute value of the expansion discount ratio is greater than the absolute value of the equilibrium discount ratio, which expands the width of the price range below the benchmark price data under the expansion segmentation strategy.

[0064] Based on benchmark price data and segmentation strategy, a segmentation threshold matrix is ​​generated.

[0065] The segmented threshold matrix contains multiple level boundary values.

[0066] The total number of level boundary values ​​is determined based on the number of segments in the segmentation strategy.

[0067] Based on the width ratio of each segment in the segmentation strategy and the upward and downward floating ratios of each segment boundary relative to the benchmark price data, the value of each level boundary is calculated.

[0068] For each level boundary value below the benchmark price data, starting from the benchmark price data, the values ​​are accumulated downwards according to the width ratio of each segment, and the product of the benchmark price data and the discount rate is used as the lower limit constraint of the lowest level boundary value to generate the boundary values ​​of each low price level.

[0069] For each level boundary value above the benchmark price data, starting from the benchmark price data, the values ​​are accumulated upwards according to the width ratio of each segment, and the product of the benchmark price data and the upward floating rate is used as the upper limit constraint of the highest level boundary value to generate the boundary values ​​of each high price level.

[0070] Using benchmark price data as the central level boundary value, the boundary values ​​of each low-price level and each high-price level are arranged in ascending order to form a segmented threshold matrix.

[0071] S2. A comprehensive evaluation of the target user's behavioral data over a historical period is performed to obtain the target user's dynamic reputation data; In this embodiment of the invention, multiple behavioral dimensions of the target user within a preset historical backtracking window are obtained.

[0072] The preset history backtracking window is a fixed time interval that is cut back from the current time. This fixed time interval is preset by the system.

[0073] The data from multiple behavioral dimensions includes at least the characteristics of task execution continuity, the characteristics of proactive response to interruptions, and the characteristics of resource usage stability.

[0074] The task execution continuity characteristic data should include at least the average duration of continuous resource occupation for a single bidding task by the target user within the historical backtracking window.

[0075] The method for obtaining the average duration of continuous resource occupation for a single bidding task is as follows: extract the resource release time and resource allocation time corresponding to each bidding task of the target user in the historical backtracking window from the historical behavior log, subtract the resource allocation time from the resource release time of each bidding task to obtain the duration of continuous resource occupation for that bidding task, add up the duration of continuous resource occupation for all bidding tasks in the historical backtracking window and divide by the total number of bidding tasks to obtain the average duration of continuous resource occupation for a single bidding task.

[0076] The proactive characteristics data for interruption response include at least the frequency with which the target user actively saves checkpoints before the bidding task is forcibly interrupted within the historical backtracking window, and the average response delay time when rejoining the bidding after the interruption.

[0077] The frequency of actively saving checkpoints is obtained by extracting whether the target user triggered an actively saving checkpoint operation before each bidding task was forcibly interrupted within the historical backtracking window from the historical behavior log, counting the number of times the actively saving checkpoint operation was triggered, and dividing the number of times the actively saving checkpoint operation was triggered by the total number of times the bidding task was forcibly interrupted within the historical backtracking window to obtain the frequency of actively saving checkpoints.

[0078] The average response latency after rejoining the auction after an interruption is obtained as follows: Extract the time when the target user resubmits the auction request after each auction task is forcibly interrupted within the historical backtracking window from the historical behavior log, and the time when the interruption occurs. Subtract the time when the interruption occurs from the time when the auction request is resubmitted to obtain the response latency after the interruption. Add up the response latency after all interruptions within the historical backtracking window and divide by the total number of interruptions to obtain the average response latency after rejoining the auction after an interruption.

[0079] Resource usage stability characteristic data should include at least the average deviation rate between the actual usage time and the promised usage time of each bidding task for the target user within the historical backtracking window, as well as the task completion rate.

[0080] The average deviation rate between the actual and promised duration of each bidding task is obtained as follows: Extract the actual and promised resource release times for each bidding task of the target user within the historical backtracking window from the historical behavior logs. Subtract the promised resource release time from the actual resource release time to obtain the time deviation value of the bidding task. Divide the time deviation value by the promised duration to obtain the deviation rate of the bidding task. Take the absolute value of the deviation rates of all bidding tasks within the historical backtracking window, add them together, and divide by the total number of bidding tasks to obtain the average deviation rate between the actual and promised duration.

[0081] The task completion rate is obtained by extracting the task status identifier corresponding to each bidding task of the target user in the historical backtracking window from the historical behavior log. The task status identifier includes completed and incomplete status. The number of bidding tasks in the completed status is counted. The number of bidding tasks in the completed status is divided by the total number of bidding tasks in the historical backtracking window to obtain the task completion rate.

[0082] Normalization mapping is performed on the task execution continuity feature data, interruption response initiative feature data, and resource consumption stability feature data to obtain standardized feature values ​​for each dimension.

[0083] The normalization mapping uses a minimum-maximum normalization method to linearly map the original feature values ​​of each dimension to a unified numerical range.

[0084] For any dimension, extract the maximum and minimum values ​​of that dimension within a preset historical backtracking window from the historical behavior log. Subtract the minimum value from the original feature value and divide by the difference between the maximum and minimum values ​​to obtain the standardized feature value of that dimension.

[0085] The standardized feature values ​​of each dimension are all within a uniform numerical range.

[0086] Based on the preset weight coefficients of each dimension, the standardized feature values ​​of each dimension are weighted and aggregated to obtain the comprehensive reputation score of the target user.

[0087] The preset weight coefficients are the contribution weight values ​​of each dimension in the credit assessment that are pre-set by the system. The sum of the preset weight coefficients of each dimension is equal to the total preset weight value.

[0088] The weighted aggregation method is as follows: multiply the standardized feature value of the task execution continuity feature data by the preset weight coefficient corresponding to the dimension to obtain the weighted value of the dimension; multiply the standardized feature value of the interruption response initiative feature data by the preset weight coefficient corresponding to the dimension to obtain the weighted value of the dimension; multiply the standardized feature value of the resource consumption stability feature data by the preset weight coefficient corresponding to the dimension to obtain the weighted value of the dimension; and sum the weighted values ​​of each dimension to obtain the comprehensive reputation score.

[0089] The credit rating of the target user is determined based on the comprehensive credit score.

[0090] The system pre-sets multiple consecutive rating intervals, each corresponding to a different credit level.

[0091] The overall credit score is compared with the boundary values ​​of each score interval. When the overall credit score falls within a certain score interval, the credit level corresponding to that score interval is determined as the credit level of the target user.

[0092] The overall credit score and credit rating are used together as dynamic credit data.

[0093] Dynamic credit data is a data object containing two fields: a comprehensive credit score and a credit level. The comprehensive credit score is used to quantitatively reflect the credit level of the target user, while the credit level is used to qualitatively classify the credit level to which the target user belongs.

[0094] S3. Based on the dynamic reputation data, correct the boundaries of the segmented threshold matrix and generate equivalent mapping coefficients; In this embodiment of the invention, the comprehensive credit score and credit level of the target user are extracted from dynamic credit data.

[0095] The credit rating includes a high credit rating and a low credit rating. When the overall credit score is greater than or equal to the preset high credit threshold, the credit rating is a high credit rating. When the overall credit score is less than the preset high credit threshold, the credit rating is a low credit rating.

[0096] The high reputation threshold is preset by the system and is used to distinguish whether a target user is qualified to trigger threshold boundary correction.

[0097] Determine whether the credit rating is high. If the credit rating is low, do not perform offset operation on the boundary of the segmented threshold matrix, and set the equivalent mapping coefficient to the preset baseline coefficient value.

[0098] When the credit rating is high, an offset operation is performed on the predetermined boundary values ​​in the segmented threshold matrix that correspond to the high-price region.

[0099] The segmented threshold matrix contains multiple level boundary values ​​arranged in ascending order, where the level boundary values ​​above the benchmark price data are predetermined boundary values ​​for the high-price region.

[0100] The method for performing offset operations on predetermined boundary values ​​is as follows: each predetermined boundary value is added with an offset amount, so that each predetermined boundary value moves towards a higher price level.

[0101] The offset is determined based on the comprehensive credit score, and there is a positive mapping relationship between the offset and the comprehensive credit score.

[0102] The positive mapping relationship is: the higher the overall reputation score, the larger the offset value.

[0103] The offset value is obtained by multiplying the comprehensive credit score by a preset offset coefficient, which is a fixed value preset by the system.

[0104] After the offset operation is performed on each predetermined boundary value, the boundary of the high-price region in the segmented threshold matrix is ​​expanded in the direction of high price. The interval between the boundary of the expanded high-price region and the boundary of the original high-price region constitutes the credit privilege data interval.

[0105] After the boundary correction of the segmented threshold matrix is ​​completed, the equivalent mapping coefficient of the target user is generated based on the target user's comprehensive reputation score.

[0106] The equivalent mapping coefficient has a positive mapping relationship with the comprehensive credit score, that is, the higher the comprehensive credit score, the larger the value of the equivalent mapping coefficient.

[0107] The minimum value of the numerical range of the equivalent mapping coefficients is greater than or equal to the lower limit of the first preset coefficient.

[0108] The first preset coefficient lower limit is the minimum allowable value of the equivalent mapping coefficient preset by the system.

[0109] The maximum value of the equivalent mapping coefficient range is less than the upper limit of the second preset coefficient.

[0110] The second preset coefficient upper limit is the maximum allowable value of the equivalent mapping coefficient preset by the system.

[0111] The equivalent mapping coefficient is generated by mapping the comprehensive credit score to multiple preset continuous credit score intervals.

[0112] Multiple consecutive credit score intervals are preset by the system. Each credit score interval is continuous and does not overlap with the others. All credit score intervals together cover the entire range of values ​​for the comprehensive credit score.

[0113] Each credit score range corresponds to a preset mapping coefficient value.

[0114] The mapping coefficient increases monotonically as the credit score range increases; that is, the higher the credit score range, the larger the corresponding mapping coefficient value.

[0115] The mapping coefficient between adjacent credit score intervals increases in a non-uniform step size, meaning that the difference between the mapping coefficients of two adjacent credit score intervals is not exactly equal.

[0116] Within the low credit score range, the difference in mapping coefficients between adjacent credit score ranges is greater than the difference in mapping coefficients between adjacent credit score ranges within the high credit score range.

[0117] The value of the equivalent mapping coefficient is always between the lower limit of the first preset coefficient and the upper limit of the second preset coefficient.

[0118] When the overall credit score falls into a certain credit score range, the value of the mapping coefficient corresponding to that credit score range is determined as the value of the equivalent mapping coefficient.

[0119] S4. Based on the equivalent mapping coefficient, perform an equivalent transformation on the original bid data of the target user to obtain equivalent bid data, and use the equivalent bid data to replace the original bid data to participate in the bidding ranking of the current bidding cycle. In this embodiment of the invention, the credit rating of the target user is extracted from dynamic credit data.

[0120] The system determines whether the credit rating is high. If the credit rating is low, the system does not perform an equivalent transformation on the target user's original bid data. Instead, it directly uses the original bid data as the equivalent bid data and uses the original bid data to participate in the bidding ranking for the current bidding cycle.

[0121] When the reputation level is high, obtain the target user's original bid data and equivalent mapping coefficient.

[0122] The original bid data is the original bid amount submitted by the target user during the current bidding period.

[0123] The equivalent mapping coefficient is a value generated based on the comprehensive reputation score and is used to perform weighted calculations on the original bid data.

[0124] The original bid data is weighted based on the equivalent mapping coefficient to generate the equivalent bid data for the target user.

[0125] The weighted calculation method is to multiply the original bid data by the equivalent mapping coefficient to obtain the equivalent bid data.

[0126] The equivalent bid data is greater than the original bid data.

[0127] Equivalent bid data is positively correlated with reputation level. When the target user has a high reputation level, the difference between the equivalent bid data and the original bid data is positive.

[0128] The specific value of the difference between the equivalent bid data and the original bid data is determined by the value of the equivalent mapping coefficient. The larger the value of the equivalent mapping coefficient, the larger the difference.

[0129] Replace the original bid data with equivalent bid data and use it as the bidding ranking key for the target user in the current bidding cycle.

[0130] Each user participating in the bidding during the current bidding period has a corresponding bidding sorting key value, which is used to determine the sorting position of each user during the bidding sorting process.

[0131] The target user is included in the global descending sort queue along with other bidding users.

[0132] The global descending sort queue is constructed as follows: obtain the bidding sort key values ​​of all users participating in the bidding during the current bidding period, and arrange all users into a queue in descending order of bidding sort key values.

[0133] In a globally descending sorted queue, users with larger sort key values ​​are placed at the front of the queue.

[0134] The resource allocation target for the current bidding period is determined based on the sorting result of the global descending sort queue.

[0135] The method for determining the resource allocation targets is as follows: cloud resources are allocated to each user in descending order of the global sorted queue, until all available cloud resources in the current bidding period are allocated.

[0136] Users ranked higher in the list will have priority in receiving resource allocation.

[0137] S5. Based on the resource release time information of the target user, update the dynamic reputation data, and based on the updated dynamic reputation data, correct the boundary of the segmented threshold matrix again.

[0138] In this embodiment of the invention, the comprehensive credit score and credit level of the target user are extracted from dynamic credit data.

[0139] Obtain the promised resource release time data and the actual resource release time data of the target user.

[0140] The promised resource release time data refers to the expected release time of the resources that the target user submits to the system when acquiring the auctioned resources.

[0141] The actual resource release time data refers to the time when the target user actually releases the resources after the resource occupation ends.

[0142] Calculate the time deviation between the actual resource release time data and the promised resource release time data.

[0143] The time deviation value is calculated by subtracting the promised resource release time data from the actual resource release time data, and the difference is the time deviation value.

[0144] When the time deviation value is less than zero, it is determined that the target user has released resources prematurely.

[0145] Early resource release refers to the behavior of a target user releasing resources earlier than the promised time in actual time.

[0146] A positive reputation correction is generated based on the absolute value of the time deviation.

[0147] The positive reputation correction is a positive value, and this positive value has a positive mapping relationship with the absolute value of the time deviation value. That is, the larger the absolute value of the time deviation value, the larger the value of the positive reputation correction.

[0148] When the time deviation value is greater than zero, it is determined that the target user is delaying the release of resources.

[0149] Delayed resource release refers to the behavior of the target user releasing resources later than the promised time in reality.

[0150] The negative reputation correction is generated based on the absolute value of the time deviation.

[0151] The negative reputation correction is a negative value. The absolute value of this negative value is positively correlated with the absolute value of the time deviation value. That is, the larger the absolute value of the time deviation value, the larger the absolute value of the negative reputation correction.

[0152] The dynamic credit data is incrementally updated based on the positive or negative credit correction amount to obtain the updated dynamic credit data.

[0153] The incremental update method is to add the positive or negative credit correction amount to the overall credit score in the dynamic credit data to obtain the updated overall credit score.

[0154] When the time deviation value is zero, it is determined that the target user releases resources on time, and no incremental update is made to the dynamic reputation data, and the overall reputation score remains unchanged.

[0155] The updated dynamic reputation data includes the updated overall reputation score and the updated reputation level.

[0156] The updated credit rating is determined as follows: the credit rating is re-determined based on the updated comprehensive credit score. The updated comprehensive credit score is compared with the preset high credit threshold. When the updated comprehensive credit score is greater than or equal to the preset high credit threshold, the updated credit rating is a high credit rating. When the updated comprehensive credit score is less than the preset high credit threshold, the updated credit rating is a low credit rating.

[0157] Based on the credit rating in the updated dynamic credit data, the predetermined boundary values ​​corresponding to the high-price region in the segmented threshold matrix are offset again.

[0158] Performing the offset operation again means using the predetermined boundary value of the high-price region in the current segmented threshold matrix as a reference, and performing the same operation as the aforementioned offset operation, that is, adding an offset to each predetermined boundary value, so that each predetermined boundary value moves towards the high-price direction.

[0159] The offset used for the next offset operation is determined based on the updated comprehensive reputation score, and the offset is positively correlated with the updated comprehensive reputation score.

[0160] The offset value is obtained by multiplying the updated overall credit score by a preset offset coefficient.

[0161] After each predetermined boundary value is offset again, the boundary range of the reputation privilege data interval in the segmented threshold matrix is ​​further expanded towards the higher price level.

[0162] The boundary range of the reputation privilege data interval in the next bidding cycle is determined by the boundary value of the high price area after the offset operation, which serves as the pricing benchmark basis for providing equivalent bid transformation for high reputation level users in the next bidding cycle.

[0163] like Figure 2 The diagram shown is a functional block diagram of a dynamic pricing system for auction cloud servers based on threshold segmentation, provided by an embodiment of the present invention.

[0164] The threshold-segmented dynamic pricing system for cloud servers described in this invention can be installed in electronic devices. Depending on the functions implemented, the threshold-segmented dynamic pricing system for cloud servers may include a segmented threshold matrix initialization module, a dynamic reputation assessment module, a mapping coefficient generation module, a bid equivalence transformation module, and a threshold iteration correction module. The modules described in this invention can also be referred to as units, which are a series of computer program segments that can be executed by the processor of an electronic device and perform a fixed function, stored in the memory of the electronic device.

[0165] In this embodiment, the functions of each module / unit are as follows: The segmented threshold matrix initialization module is used to initialize the segmented threshold matrix based on the bidding request data and cloud resource idle status data of the current bidding cycle. The dynamic credit assessment module is used to comprehensively assess the target user's behavioral data over a historical period to obtain the target user's dynamic credit data. The mapping coefficient generation module is used to correct the boundaries of the segmented threshold matrix based on the dynamic reputation data and generate equivalent mapping coefficients. The bid equivalence transformation module is used to perform an equivalent transformation on the original bid data of the target user based on the equivalent mapping coefficient to obtain equivalent bid data, and use the equivalent bid data to replace the original bid data to participate in the bidding ranking of the current bidding cycle. The threshold iteration correction module is used to update the dynamic reputation data based on the resource release time information of the target user, and to correct the boundary of the segmented threshold matrix again according to the updated dynamic reputation data.

[0166] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0167] Furthermore, the functional modules in the various embodiments of the present invention 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 in the form of hardware plus software functional modules.

[0168] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0169] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A dynamic pricing method for auction-based cloud servers based on threshold segmentation, characterized in that, The method includes: S1. Initialize the segmented threshold matrix based on the bidding request data and cloud resource idle status data of the current bidding cycle; S2. A comprehensive evaluation of the target user's behavioral data over a historical period is performed to obtain the target user's dynamic reputation data; S3. Based on the dynamic reputation data, correct the boundaries of the segmented threshold matrix and generate equivalent mapping coefficients; S4. Based on the equivalent mapping coefficient, perform an equivalent transformation on the original bid data of the target user to obtain equivalent bid data, and use the equivalent bid data to replace the original bid data to participate in the bidding ranking of the current bidding cycle. S5. Based on the resource release time information of the target user, update the dynamic reputation data, and based on the updated dynamic reputation data, correct the boundary of the segmented threshold matrix again.

2. The dynamic pricing method for auction cloud servers based on threshold segmentation as described in claim 1, characterized in that, The initialization of the segmented threshold matrix based on the bidding request data and cloud resource idle status data of the current bidding cycle includes: Statistical fitting is performed on the bidding request data of the current bidding period to obtain the benchmark price data of the current bidding period; Based on the cloud resource idle status data of the current bidding period, determine the resource saturation level of the current bidding period; Based on the resource saturation level, determine the segmentation strategy; Based on the benchmark price data and the segmentation strategy, a segmentation threshold matrix is ​​generated, which contains multiple level boundary values.

3. The dynamic pricing method for auction cloud servers based on threshold segmentation as described in claim 2, characterized in that, The step of determining the segmentation strategy based on the resource saturation level includes: Perform dispersion analysis on the bidding request data to obtain the price dispersion characteristics of the current bidding period; Based on the combination of the resource saturation level and the price dispersion characteristics, one of the preset multiple segmentation strategies is selected as the target segmentation strategy. The multiple segmentation strategies include at least the equilibrium segmentation strategy, the compression segmentation strategy, and the expansion segmentation strategy. The segmentation strategy includes the number of segments, the width ratio of each segment, and the upward and downward floating ratios of each segment boundary relative to the benchmark price data. When the resource saturation level is the first level and the price dispersion feature is the first dispersion feature, the equilibrium segmentation strategy is selected. When the resource saturation level is the second level and the price dispersion feature is the second dispersion feature, the compression segmentation strategy is selected; When the resource saturation level is level three and the price dispersion feature is the third dispersion feature, the expansion segmentation strategy is selected; The balanced segmentation strategy is configured to have a baseline number of segments and a proportionally distributed segment width. The compression segmentation strategy is configured such that the number of segments is less than the baseline number of segments and the width of each segment is in a contraction distribution relative to the balanced segmentation strategy. The expansion segmentation strategy is configured such that the number of segments is greater than the baseline number of segments and the width of each segment is expanded relative to the balanced segmentation strategy.

4. The dynamic pricing method for auction cloud servers based on threshold segmentation as described in claim 1, characterized in that, The process of comprehensively evaluating the target user's behavioral data over a historical period to obtain the target user's dynamic reputation data includes: The target user's behavioral data within a preset historical backtracking window is obtained. The multiple behavioral data includes at least task execution continuity feature data, interruption response initiative feature data, and resource consumption stability feature data. The task execution continuity feature data, interruption response initiative feature data, and resource consumption stability feature data are normalized and mapped respectively to obtain standardized feature values ​​for each dimension. Based on the preset weight coefficients of each dimension, the standardized feature values ​​of each dimension are weighted and aggregated to obtain the comprehensive reputation score of the target user. Based on the comprehensive reputation score, the reputation level corresponding to the target user is determined, and the comprehensive reputation score and reputation level are used together as dynamic reputation data.

5. The dynamic pricing method for auction cloud servers based on threshold segmentation as described in claim 4, characterized in that, The multiple behavioral dimension data include: The task execution continuity characteristic data includes at least the average duration of continuous resource occupation for a single bidding task by the target user within the historical backtracking window; The proactive interruption response data includes at least the frequency with which the target user actively saved checkpoints before the bidding task was forcibly interrupted within the historical backtracking window and the average response delay time when rejoining the bidding after the interruption. The resource occupancy stability characteristic data includes at least the average deviation rate between the actual occupancy time and the promised occupancy time of each bidding task by the target user within the historical backtracking window, as well as the task completion rate.

6. The dynamic pricing method for auction cloud servers based on threshold segmentation as described in claim 4, characterized in that, The step of correcting the boundaries of the segmented threshold matrix based on the dynamic reputation data and generating equivalent mapping coefficients includes: When the target user's credit rating is high, the predetermined boundary value corresponding to the high-price region in the segmented threshold matrix is ​​offset to expand the credit privilege data range in the segmented threshold matrix. The offset amount of the offset operation is positively mapped to the comprehensive credit score. Based on the target user's comprehensive reputation score, an equivalent mapping coefficient for the target user is generated.

7. The dynamic pricing method for auction cloud servers based on threshold segmentation as described in claim 6, characterized in that, The step of generating the equivalent mapping coefficient for the target user based on the target user's comprehensive reputation score includes: The comprehensive credit score is mapped to a plurality of preset continuous credit score intervals, each credit score interval corresponding to a preset mapping coefficient value, and the mapping coefficient value increases monotonically as the credit score interval increases. The mapping coefficient between adjacent credit score intervals increases in a non-uniform step size, and the mapping coefficient is constrained to be no less than the lower limit of the first preset coefficient and no greater than the upper limit of the second preset coefficient.

8. The dynamic pricing method for auction cloud servers based on threshold segmentation as described in claim 4, characterized in that, The step of performing an equivalent transformation on the original bid data of the target user based on the equivalent mapping coefficient to obtain equivalent bid data, and using the equivalent bid data to replace the original bid data in the bidding ranking of the current bidding cycle, includes: When the credit rating of the target user is determined to be high, the original bid data and equivalent mapping coefficient of the target user are obtained. Based on the equivalent mapping coefficient, the original bid data is weighted to generate the equivalent bid data for the target user; The original bid data is replaced with the equivalent bid data and used as the bidding sorting key value for the target user in the current bidding cycle; Based on the bidding sorting key value, the target user is included in a global descending sorting queue along with other bidding users, and the resource allocation object for the current bidding period is determined according to the sorting result of the global descending sorting queue.

9. The dynamic pricing method for auction cloud servers based on threshold segmentation as described in claim 4, characterized in that, The step of updating the dynamic reputation data based on the resource release time information of the target user, and then correcting the boundaries of the segmented threshold matrix again based on the updated dynamic reputation data, includes: Obtain the promised resource release time data and the actual resource release time data of the target user, and calculate the time deviation value between the actual resource release time data and the promised resource release time data; When the time deviation value is less than zero, it is determined that the target user has released resources in advance, and a positive reputation correction amount is generated based on the absolute value of the time deviation value. When the time deviation value is greater than zero, it is determined that the target user is delaying the release of resources, and a negative reputation correction amount is generated based on the absolute value of the time deviation value. The dynamic reputation data is incrementally updated based on the positive reputation correction amount or the negative reputation correction amount to obtain the updated dynamic reputation data. Based on the credit rating of the updated dynamic credit data, the predetermined boundary values ​​corresponding to the high-price region in the segmented threshold matrix are offset again to adjust the boundary range of the credit privilege data interval in the next bidding cycle.

10. A dynamic pricing system for auction-based cloud servers based on threshold segmentation, characterized in that, The system is used to implement the threshold-based segmented dynamic pricing method for cloud servers as described in claim 1, the system comprising: The segmented threshold matrix initialization module is used to initialize the segmented threshold matrix based on the bidding request data and cloud resource idle status data of the current bidding cycle. The dynamic reputation assessment module is used to comprehensively evaluate the target user's behavioral data over a historical period to obtain the target user's dynamic reputation data. The mapping coefficient generation module is used to correct the boundaries of the segmented threshold matrix based on the dynamic reputation data and generate equivalent mapping coefficients. The bid equivalence transformation module is used to perform an equivalent transformation on the original bid data of the target user based on the equivalent mapping coefficient to obtain equivalent bid data, and use the equivalent bid data to replace the original bid data to participate in the bidding ranking of the current bidding cycle. The threshold iteration correction module is used to update the dynamic reputation data based on the resource release time information of the target user, and to correct the boundary of the segmented threshold matrix again according to the updated dynamic reputation data.