Hardware server-based dynamic discount rate calculation and feedback adjustment method and system for auctioned items

Abstract

This invention discloses a method and system for dynamic discount rate calculation and feedback adjustment of auction unsold items based on a hardware server. By constructing a hardware architecture consisting of a storage server, an algorithm server, and a feedback control module, it achieves real-time collection, calculation, and feedback of multi-dimensional data without manual intervention. It defines a quantitative standard for the ideality of auction results and establishes a closed-loop feedback mechanism from discount rate calculation to auction implementation to result evaluation to discount rate adjustment.

Description

Technical Field

[0001] This invention relates to the field of auction service technology, specifically to a method and system for calculating and adjusting the dynamic discount rate of unsold auction items based on a hardware server. Background Technology

[0002] like Figure 1 As shown, the current auction platform primarily uses a "fixed 0.8 discount + manual assistance" scheme to handle unlisted items. The hardware relies solely on basic storage servers to store auction records, without dedicated data processing and control hardware. The specific process is as follows:

[0003] Hardware support: Only ordinary storage servers (such as Dell PowerEdge R750) are used to store basic data such as the initial auction reserve price and unsold auction records, without dedicated algorithm servers and feedback control modules;

[0004] Discount calculation: After an item fails to sell at auction, the initial auction reserve price is retrieved manually from the storage server, and the reserve price for the second auction is calculated at a fixed ratio of 0.8, or a non-standardized calculation is performed using a simple Excel spreadsheet.

[0005] Subsequent handling: If the second auction also fails to sell, the fixed discount logic is repeated without data feedback or optimization mechanism. For example, if a batch of Pu'er tea that failed to sell is auctioned at 0.8 discount in the second auction and still fails to sell, the third auction will continue to calculate the discount at 0.8 discount, which is 0.64 discount, completely ignoring market feedback.

[0006] For example, a small-to-medium-sized auction platform disposed of unsold pure silver jewelry. The initial reserve price was 100,000 yuan. After the failed auction, the data was manually exported from the storage server and a calculator was used to calculate the second reserve price of 80,000 yuan at a 0.8 discount. After the second failed auction, the reserve price was calculated again at a 0.8 discount to a third reserve price of 64,000 yuan. The entire process was completely dependent on manual operation without any hardware automatic calculation or feedback adjustment.

[0007] The existing solution has obvious flaws in both hardware architecture and software logic, and cannot achieve "automated, precise, and closed-loop" handling of unsold items. Specifically, these flaws are reflected in the following aspects:

[0008] Insufficient hardware support and low degree of automation: It relies solely on storage servers and lacks dedicated algorithm servers and feedback control hardware. Price calculation and data updates require manual intervention, which is inefficient and prone to errors. For example, when manually entering the base price, 100,000 yuan is mistakenly entered as 10,000 yuan, resulting in a completely incorrect price calculation result.

[0009] Fixed discounts with no feedback fail to adapt to market changes: There is no feedback mechanism of "discount rate - auction result". Even if the auction fails after being auctioned at 0.8 discount, the price will still be reduced by a fixed percentage in the future, resulting in the reserve price being too low or too high. For example, if a popular watch fails to sell at auction, it will be quickly sold in a second auction at 0.8 discount. However, because there is no feedback, similar watches will still be sold at 0.8 discount in the next auction, resulting in lost revenue.

[0010] Weak data processing capabilities and lack of real-time computing functions: Ordinary storage servers cannot handle real-time calculations of multi-dimensional data (historical, same batch, third-party), and cannot achieve rapid output of dynamic discount rates. The calculation of the reserve price for the second auction needs to wait 1-2 working days, delaying the processing opportunity.

[0011] Without evaluation standards, feedback and adjustments lack a basis: There is no quantitative standard for "ideal / unideal auction". Even if you want to adjust the discount rate, the lack of clear indicators (such as transaction time, premium rate) leads to subjective and arbitrary adjustments. For example, one person may think that "a transaction in 10 days is ideal", while another person may think that "a transaction in 5 days is ideal", resulting in chaotic standards. Summary of the Invention

[0012] The purpose of this invention is to propose a method and system for calculating and adjusting the dynamic discount rate of auctioned items that fail to sell, based on a hardware server, so as to achieve fully automated disposal of auctioned items.

[0013] To achieve the above-mentioned technical objectives, the technical solution of the present invention is implemented as follows:

[0014] A method for calculating and adjusting the dynamic discount rate of unsold auction items based on a hardware server, the method comprising the following steps:

[0015] 1) Read the historical transaction rate R1, historical premium rate P1, transaction rate R2, and premium rate P2 of the same batch of auction items from the storage server, as well as the third-party transaction rate R3 and third-party premium rate P3 obtained from the API gateway of the third-party platform.

[0016] 2) Determine if the data read in step 1) exceeds a reasonable range. If so, replace it with the average value of the same product category over the past 3 months. If not, proceed to the following steps.

[0017] 3) Calculate the comprehensive index value Z according to the following formula:

[0018]

[0019] in, , , , , , ;

[0020] 4) Calculate the initial dynamic discount rate using the following formula. :

[0021] ;

[0022] 5) Output the reserve price for the second auction. The reserve price for the second auction = the reserve price for the first auction × D0;

[0023] 6) Determine whether the result of the second auction is ideal. Specifically, determine whether the auction result satisfies both condition one and condition two. If both conditions are met, the result is considered ideal, and step 7) is executed. Otherwise, the result is considered unsatisfactory, and step 8) is executed. Condition one is that the transaction time is less than or equal to the historical average transaction time of the same category, and condition two is that the actual premium rate is greater than or equal to the historical average premium rate of the same category.

[0024] 7) Calculate the adjusted discount rate D1 for the next unsold item of the same type using the following formula:

[0025] And D1≤1

[0026]

[0027] in, This represents the premium rate of the items actually sold in the secondary auction. This represents the historical average premium rate for the same product category.

[0028] 8) Calculate the adjusted discount rate D1 for the next unsold item of the same type using the following formula:

[0029] ,and ≥0.8

[0030] or

[0031] in, T represents the historical average transaction time for the same product category, and T represents the actual transaction time of the second auction. Take the larger value obtained from the two calculation methods above;

[0032] 9) Store D1 on the server as the initial reference discount rate when similar items fail to sell at auction next time.

[0033] A hardware server-based system for calculating and adjusting the dynamic discount rate of unsold auction items, the system comprising:

[0034] The storage server is used to store the historical transaction rate R1, the historical premium rate P1, the transaction rate R2, the premium rate P2, the third-party transaction rate R3, the third-party premium rate P3, and the auction result data.

[0035] The data acquisition module is used to read the historical transaction rate R1, historical premium rate P1, transaction rate R2, and premium rate P2 of the same batch of auction items from the storage server, as well as the third-party transaction rate R3 and third-party premium rate P3 obtained from the API gateway of the third-party platform.

[0036] The data preprocessing module is used to determine whether the data read from the data acquisition module exceeds a reasonable range. If so, it replaces the data with the average value of the same product category over the past three months and sends the processed data to the calculation module.

[0037] The calculation module is used to calculate the comprehensive index value Z based on the following formula:

[0038]

[0039] in, , , , , , ,

[0040] And calculate the initial dynamic discount rate according to the following formula. :

[0041]

[0042] And output the reserve price for the second auction, where the reserve price for the second auction = the reserve price for the first auction × D0;

[0043] The auction result ideality evaluation module is used to determine whether the secondary auction result is ideal. Specifically, it determines whether the auction result meets both condition one and condition two at the same time. If both conditions are met, the result is considered ideal, and instruction one is sent to the feedback adjustment module. Otherwise, the result is considered unideal, and instruction two is sent to the feedback adjustment module. Condition one is that the transaction time is less than or equal to the historical average transaction time of the same category, and condition two is that the actual premium rate is greater than or equal to the historical average premium rate of the same category.

[0044] The feedback adjustment module, upon receiving instruction one, calculates the adjusted discount rate D1 for the next unsold item of the same type according to the following formula, and stores D1 in the storage server as the initial reference discount rate for the next unsold item of the same type:

[0045] And D1≤1

[0046]

[0047] in, This represents the premium rate of the items actually sold in the secondary auction. This represents the historical average premium rate for the same product category.

[0048] Upon receiving instruction two, calculate the adjusted discount rate D1 for the next unsold item of the same type according to the following formula, and store D1 in the storage server as the initial reference discount rate for the next unsold item of the same type:

[0049] ,and ≥0.8

[0050] or

[0051] in, T represents the historical average transaction time for the same product category, and T represents the actual transaction time of the second auction. Take the larger value obtained from the two calculation methods above.

[0052] The beneficial effects of this invention are: hardware architecture collaboration to achieve full automation: for the first time, a dedicated hardware combination of "storage server + algorithm server + feedback control module" is constructed to replace traditional manual operation. The algorithm server carries out real-time calculation of multi-dimensional data, and the feedback control module realizes automatic transmission of instructions, ensuring that there is no manual intervention from data collection to discount rate adjustment, and improving efficiency by more than 90%.

[0053] Quantify the ideal standard to solve the problem of feedback without basis: Define a dual indicator evaluation system of "transaction time + actual premium rate" and use specific values ​​(such as T≤3 days, P actual ≥12%) to replace subjective judgment, so that feedback adjustment has a clear basis and avoids confusion of standards among different personnel;

[0054] The adjustment range is dynamically adjusted to balance flexibility and stability: the adjustment range is dynamically calculated based on the "degree of unsatisfaction" of the auction results (if the results are severely unsatisfactory, the adjustment range will be lowered more, and if the results are slightly unsatisfactory, the adjustment range will be lowered less), rather than a fixed range adjustment. This ensures that the discount rate can adapt to the market and avoids large fluctuations that could cause the reserve price to get out of control.

[0055] Deep integration of hardware and software: The feedback control module receives auction results in real time through an embedded chip and sends control commands directly to the algorithm server. The response time is ≤1 second, which is much faster than manual processing (1-2 days). It also avoids manual input errors and improves data accuracy to 99.9%.

[0056] Historical data reuse for continuous model optimization: Adjusted discount rate The data is stored on a storage server as an initial reference for the next similar item, forming a closed loop of "calculation-evaluation-adjustment-reuse". This allows the algorithm model to be continuously optimized as the number of auctions increases, and the calculation accuracy to be gradually improved. Attached Figure Description

[0057] Figure 1 This is a flowchart of the process for handling unsold auction items;

[0058] Figure 2 This is a flowchart illustrating the process of the method described in the embodiments of the present invention;

[0059] Figure 3 This is a hardware architecture diagram of the system described in an embodiment of the present invention. Detailed Implementation

[0060] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0061] like Figure 3 As shown, this invention requires the deployment of three types of hardware devices, which are interconnected via a network to achieve data interaction and automated control:

[0062] Storage server: Huawei FusionServer Pro, configured with 32GB of memory and 2TB SSD, used to store: ① Historical transaction rate R1 and historical premium rate P1 of similar items; ② Transaction rate R2 and premium rate P2 of the same batch; ③ Data from third-party platforms R3 and P3; ④ Auction result data (transaction time, transaction price, and ideality evaluation results).

[0063] Algorithm server: Uses Inspur NF5280M6, configured with 64GB memory and 4TB SSD hard drive, deploys core algorithm program, realizes data preprocessing, dynamic discount rate calculation, ideality evaluation, and feedback adjustment range calculation;

[0064] Feedback control module: Based on the embedded chip STM32F407, it communicates with the algorithm server and storage server through the RS485 interface. After receiving the auction results, it sends the control command of "increasing / decreasing the discount rate" to the algorithm server and updates the historical data in the storage server.

[0065] Third-party platform API gateway: Connects to the algorithm server via Ethernet to enable real-time crawling and transmission of data from JD.com and Taobao auction platforms, ensuring data timeliness within 30 minutes.

[0066] like Figure 2 As shown, based on the hardware device deployed above, this invention discloses a method for calculating and adjusting the dynamic discount rate of unsold auction items based on a hardware server. The method includes the following steps:

[0067] Phase 1: Calculation of Dynamic Discount Rate

[0068] 1) The auction server retrieves the historical transaction rate (R1) of similar items, the historical premium rate (P1) of similar items, the transaction rate (R2) of the same batch, and the premium rate (P2) of the same batch for the auctioned items, as well as the third-party transaction rate (R3) and the third-party premium rate (P3) obtained from the API gateway of the third-party platform. The reason for selecting the above 6 indicators is as follows: Historical data of similar items (R1, P1) reflects the long-term market acceptance and value stability of the items, which is the core basis for calculating the discount rate; data of the same batch (R2, P2) reflects the market heat in the current auction scenario, avoiding pricing deviations due to batch differences; data from third-party platforms (R3, P3) supplements cross-platform market information, eliminates the limitations of single-platform data, and the 6 indicators work together to cover the full-dimensional market characteristics of "long-term-short-term-cross-platform", ensuring the objectivity and comprehensiveness of the discount rate.

[0069] 2) Determine if the data read in step 1) exceeds a reasonable range (e.g., P1>1 or P1<-1). If so, replace it with the average value of the same category over the past 3 months. If not, proceed to the following steps.

[0070] 3) Based on the analytic hierarchy process (AHP), set the weights and calculate the comprehensive index value Z according to the following formula:

[0071]

[0072] Formula principle and calculation process: The overall goal of "market fit" is decomposed into three criteria layers: "historical performance," "performance in the same batch," and "third-party performance" using the Analytic Hierarchy Process (AHP), and further refined into six indicator layers. Ten auction industry experts were invited to conduct pairwise comparisons and scoring of the weights of each indicator, constructing a judgment matrix and performing a consistency test (CR < 0.1). Ultimately, R2 was determined to have the highest weight (0.3). Since the data in the same batch is most correlated with the current auction scenario, R1 and P1 each account for 0.2 (historical data has the second highest influence), and R3, P3, and P2 each account for 0.1 (supplementary indicators). At the same time, the premium rate indicator was standardized. The value range is mapped to [0,1] to maintain dimensionality consistent with the transaction rate indicator (which naturally falls within the [0,1] range), ensuring the rationality of the weighted calculation. Finally, a comprehensive indicator Z (within the [0,1] range) is obtained through linear weighting. The closer Z is to 1, the higher the market fit.

[0073] , , , , , .

[0074] 4) Calculate the initial dynamic discount rate using the following formula. :

[0075]

[0076] Formula principle and calculation process: Based on the core logic of "the higher the market fit, the higher the discount rate", the benchmark range of the discount rate is set as [0.8, 1] (to avoid the discount being too low and harming the interests of the consignor, or too high and causing the auction to fail again); when Z=1 (optimal market fit), D0=1 (no discount), and when Z=0 (worst market fit), D0=0.8 (minimum discount). The formula is constructed through a linear mapping relationship, and 0.2 is the discount rate adjustment coefficient to ensure that the change of Z can be reflected evenly, so as to achieve the initial pricing that dynamically adapts to the market.

[0077] 5) Output the reserve price for the second auction. The reserve price for the second auction = the reserve price for the first auction × D0. The algorithm server sends D0 to the front-end server of the auction platform.

[0078] Phase 2: Evaluation of the Desire for Auction Results

[0079] 6) Determine whether the result of the second auction is ideal. Specifically, determine whether the auction result satisfies both condition one and condition two. If both conditions are met, the result is considered ideal, and step 7) is executed. Otherwise, the result is considered unsatisfactory, and step 8) is executed. Condition one is that the transaction time is less than or equal to the historical average transaction time of the same category, and condition two is that the actual premium rate is greater than or equal to the historical average premium rate of the same category.

[0080] 7) Calculate the adjusted discount rate D1 for the next unsold item of the same type using the following formula:

[0081] And D1≤1

[0082]

[0083] Formula principle and calculation process: When the auction result is ideal, it indicates that the current discount rate is too low (there is still room for price increase), and the initial discount rate for the next auction needs to be appropriately increased; 0.8×P 历史平均 P is the minimum threshold for the ideal premium rate. 实 The portion exceeding this threshold reflects the size of the premium space; multiplying it by 0.1 (adjustment sensitivity coefficient) yields the adjustment range D. 调 To ensure a moderate adjustment and avoid significant price increases that could lead to failed auctions in the next round, a maximum limit of D1 (≤1) is set to prevent unlimited price increases that deviate from market realities.

[0084] in, This represents the premium rate of the items actually sold in the secondary auction. This represents the historical average premium rate for the same product category.

[0085] 8) Calculate the adjusted discount rate D1 for the next unsold item of the same type using the following formula:

[0086] ,and ≥0.8

[0087] or

[0088] Formula principle and calculation process: When the auction result is unsatisfactory, the adjustment range needs to be determined based on two dimensions: "transaction efficiency" and "profit level"; the first calculation method focuses on the transaction time dimension. The first method reflects differences in transaction speed, multiplied by 0.005 (time sensitivity coefficient), reflecting the logic that "the faster the transaction, the smaller the adjustment range"; the second calculation method targets the premium rate dimension, where 0.5 × P1 historical average is the minimum acceptable premium rate, and 0.5 × - The magnitude of the return gap is reflected by multiplying it by 0.05 (return sensitivity coefficient), reflecting the logic that "the larger the return gap, the greater the adjustment." The larger of the two values ​​is taken as the return threshold. This ensures that more serious shortcomings are addressed first, while setting a lower limit of D1≥0.8 to protect the basic interests of the client.

[0089] in, T represents the historical average transaction time for the same product category, and T represents the actual transaction time of the second auction. Take the larger value obtained from the two calculation methods above;

[0090] 9) Store D1 on the server as the initial reference discount rate when similar items fail to sell at auction next time.

[0091] This invention also discloses a dynamic discount rate calculation and feedback adjustment system for auction unsold items based on a hardware server. The above-mentioned functional modules of the dynamic discount rate calculation and feedback adjustment system for auction unsold items based on a hardware server of this invention correspond to the various operation steps of the dynamic discount rate calculation and feedback adjustment method for auction unsold items based on a hardware server of this invention, which will not be repeated here.

[0092] To facilitate understanding of the above technical solutions of the present invention, the following detailed description of the above technical solutions of the present invention will be provided through specific usage methods.

[0093] Taking a gold necklace from a certain brand that failed to sell at auction as an example

[0094] 1. Hardware preparation

[0095] The storage server has stored the following: historical transaction rate for gold (R1=0.9), historical premium rate (P1=15%), and historical average transaction time (T=3 days).

[0096] Auction data for the same batch: Transaction rate R² = 0.85, premium rate P² = 12%;

[0097] According to API gateway data from third-party platforms: JD.com's gold category transaction rate R3=0.88, premium rate P3=14%;

[0098] Initial auction starting price: 200,000 yuan.

[0099] 2. Stage 1: Calculate the initial discount rate D0

[0100] Standardization process:

[0101]

[0102]

[0103]

[0104] Calculate Z:

[0105]

[0106] Calculate D0:

[0107]

[0108] The starting price for the second auction is 200,000 × 0.9528 = 190,560 yuan, which is sent to the front end for display by the algorithm server.

[0109] Phase 2: Evaluation of the Desire for Auction Results

[0110] Auction Result: The item was sold within two days of being listed online, with an actual transaction price of 195,000 yuan. The actual price (Pactual) is calculated as (195,000 - 190,056) / 190,056 ≈ 2.33%.

[0111] Ideal standard: T≤3 days (compliant), P actual ≥0.8×15%=12% (non-compliant, 2.33%<12%);

[0112] Judgment: The auction was unsatisfactory.

[0113] Phase 3: Feedback Adjustment (Calculation of D1)

[0114] Calculate the adjustment range:

[0115] (TT historical average) × 0.005 = (2-3) × 0.005 = -0.5% (absolute value 0.5%);

[0116] (0.5 × 15% - 2.33%) × 0.05 = (7.5% - 2.33%) × 0.05 ≈ 0.2585%;

[0117] Take the larger value, 0.5%, and adjust by 0.5%.

[0118] Calculate D1:

[0119] Data storage: D1=0.9478 is stored on the storage server and will be used as the initial reference discount rate when gold items fail to sell at auction next time.

[0120] Furthermore, a comparative experiment was conducted using three categories of unsold items (gold jewelry, 53-degree Moutai, and Pu'er tea) from a certain auction platform from January to June 2024. The experimental group used the present invention (hardware + feedback), while the control group used the traditional method (fixed 0.8 discount + manual intervention). The results are shown in Table 1 below:

[0121] Category Experimental group (this invention) Control group (traditional regimen) Improvement in effectiveness Gold jewelry The second auction achieved a 95% success rate, an average premium of 13%, and an average completion time of 2.5 days. Following feedback and adjustments, the average discount rate for similar items in the next auction will be increased by 0.02%. The second auction had a 75% success rate, an average premium of 5%, and an average transaction time of 7 days. The transaction rate increased by 26.7%, the premium rate increased by 160%, and the transaction time was shortened by 64.3%. 53-degree Moutai The second auction achieved a 100% sell-through rate, with an average premium of 45% and an average transaction time of 1 day. Following feedback and adjustments, the average discount rate for similar items in the next auction will be increased by 0.03%. The second auction had a 90% success rate, an average premium of 25%, and an average transaction time of 3 days. The transaction rate increased by 11.1%, the premium rate increased by 80%, and the transaction time was shortened by 66.7%. Pu-erh tea The second auction had an 88% success rate, an average premium of 10%, and an average transaction time of 4 days. Following feedback and adjustments, the average discount rate for similar items in the next auction will be reduced by 0.015%. The second auction had a 60% success rate, an average premium of -2%, and an average completion time of 12 days. The transaction rate increased by 46.7%, the premium rate turned from negative to positive, and the transaction time was shortened by 66.7%.

[0122] Experimental conclusions: The experimental group achieved an average increase of 28.2% in the success rate of secondary auctions, an average increase of 106.7% in the revenue of consignors, and an average reduction of 65.2% in the transaction time. Furthermore, after three rounds of feedback and adjustments, the failure rate of similar items dropped from the initial 30% to below 8%, demonstrating significant continuous optimization effects.

[0123] Long-term application effects:

[0124] Model optimization: As the number of auctions increases (more than 1,000 in total), the algorithm model adjusts through feedback, and the matching degree between the dynamic discount rate and market demand increases from the initial 75% to 92%.

[0125] Platform competitiveness: Due to "automated pricing + guaranteed returns", the number of clients cooperating with the auction platform has increased by 60% compared to traditional solutions, and the platform's activity level has increased by 45% due to "reasonable reserve prices".

[0126] Cost savings: Although a dedicated hardware architecture requires initial investment, in the long run, it can save 200,000 to 300,000 yuan in labor costs per year and avoid revenue loss due to human error (approximately 150,000 yuan per year).

[0127] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for calculating and adjusting the dynamic discount rate of unsold auction items based on a hardware server, characterized in that, The method includes the following steps: 1) Read the historical transaction rate R1, historical premium rate P1, transaction rate R2, and premium rate P2 of the same batch of auction items from the storage server, as well as the third-party transaction rate R3 and third-party premium rate P3 obtained from the API gateway of the third-party platform. 2) Determine if the data read in step 1) exceeds a reasonable range. If so, replace it with the average value of the same product category over the past 3 months. If not, proceed to the following steps. 3) Calculate the comprehensive index value Z according to the following formula: in, , , , , , ; 4) Calculate the initial dynamic discount rate using the following formula. : ； 5) Output the reserve price for the second auction. The reserve price for the second auction = the reserve price for the first auction × D0; 6) Determine whether the result of the second auction is ideal. Specifically, determine whether the auction result satisfies both condition one and condition two. If both conditions are met, the result is considered ideal, and step 7) is executed. Otherwise, the result is considered unsatisfactory, and step 8) is executed. Condition one is that the transaction time is less than or equal to the historical average transaction time of the same category, and condition two is that the actual premium rate is greater than or equal to the historical average premium rate of the same category. 7) Calculate the adjusted discount rate D1 for the next unsold item of the same type using the following formula: And D1≤1 in, This represents the premium rate of the items actually sold in the second auction. This represents the historical average premium rate for the same product category. 8) Calculate the adjusted discount rate D1 for the next unsold item of the same type according to the following formula: ,and ≥0.8 or in, T represents the historical average transaction time for the same product category, and T represents the actual transaction time of the second auction. Take the larger value obtained from the two calculation methods above; 9) Store D1 on the server as the initial reference discount rate when similar items fail to sell at auction next time.

2. A dynamic discount rate calculation and feedback adjustment system for auction unsold items based on a hardware server, characterized in that, The system includes: The storage server is used to store the historical transaction rate R1, the historical premium rate P1, the transaction rate R2, the premium rate P2, the third-party transaction rate R3, the third-party premium rate P3, and the auction result data. The data acquisition module is used to read the historical transaction rate R1, historical premium rate P1, transaction rate R2, and premium rate P2 of the same batch of auction items from the storage server, as well as the third-party transaction rate R3 and third-party premium rate P3 obtained from the API gateway of the third-party platform. The data preprocessing module is used to determine whether the data read from the data acquisition module exceeds a reasonable range. If so, it replaces the data with the average value of the same product category over the past three months and sends the processed data to the calculation module. The calculation module is used to calculate the comprehensive index value Z based on the following formula: in, , , , , , , And calculate the initial dynamic discount rate according to the following formula. : And output the reserve price for the second auction, where the reserve price for the second auction = the reserve price for the first auction × D0; The auction result ideality evaluation module is used to determine whether the secondary auction result is ideal. Specifically, it determines whether the auction result meets both condition one and condition two at the same time. If both conditions are met, the result is considered ideal, and instruction one is sent to the feedback adjustment module. Otherwise, the result is considered unsatisfactory, and instruction two is sent to the feedback adjustment module. Condition one is that the transaction time is less than or equal to the historical average transaction time of the same category, and condition two is that the actual premium rate is greater than or equal to the historical average premium rate of the same category. The feedback adjustment module, upon receiving instruction one, calculates the adjusted discount rate D1 for the next unsold item of the same type according to the following formula, and stores D1 in the storage server as the initial reference discount rate for the next unsold item of the same type: And D1≤1 in, This represents the premium rate of the items actually sold in the second auction. This represents the historical average premium rate for the same product category. Upon receiving instruction two, calculate the adjusted discount rate D1 for the next unsold item of the same type according to the following formula, and store D1 in the storage server as the initial reference discount rate for the next unsold item of the same type: ,and ≥0.8 or in, T represents the historical average transaction time for the same product category, and T represents the actual transaction time of the second auction. Take the larger value obtained from the two calculation methods above.

Images