A transaction guarantee management system and method

Abstract

The application provides a transaction guarantee management system and method, belonging to the technical field of guarantee management, specifically comprising: identification processing of guarantee reference jewelry based on identification strategy for guarantee analysis jewelry, and determination of guarantee amount of the guarantee analysis jewelry using the guarantee reference jewelry; determination of a transaction guarantee management method of a merchant based on guarantee amount of the guarantee analysis jewelry and on-shelf data of transaction jewelry of the merchant; when the transaction guarantee management method is determined to have risks based on guarantee analysis jewelry data in different merchants, the merchant on which the guarantee analysis jewelry is on-shelf is determined as a related on-shelf merchant; determination of an image recognition analysis strategy of the guarantee analysis jewelry based on identification deviation in the transaction jewelry on-shelf in the related on-shelf merchant of the guarantee analysis jewelry and on-shelf composition data of the guarantee analysis jewelry, thereby reducing the risk of insufficient guarantee amount.

Description

Technical Field

[0001] This invention belongs to the field of guarantee management technology, and in particular relates to a transaction guarantee management system and method. Background Technology

[0002] The trading of in-game items has become a massive digital asset market. Taking CS:GO / CS2 as an example, some rare skins are worth thousands or even tens of thousands of dollars. However, the trading of in-game items is fraught with serious security risks and trust issues. When buyers and sellers use third-party platforms, common fraudulent methods include the seller failing to ship after the buyer pays, the buyer failing to pay after the seller ships, sending fake transfer screenshots, and exploiting time differences in payment platforms. Therefore, existing trading platforms often have transaction guarantee systems in place to reduce the occurrence of transaction disputes.

[0003] However, existing transaction guarantee systems often suffer from the following technical shortcomings: Existing transaction guarantee systems on trading platforms often use fixed guarantee amounts or determine the guarantee amount based on the transaction value of the traded items. However, in reality, to reduce the guarantee amount, merchants may lower the listed price of traded items and process transactions through external channels. In the event of a transaction dispute, there is a high risk of insufficient guarantee amount. Therefore, determining different guarantee amount management strategies for traded items based on the recognition deviation risk of image recognition models, and optimizing the guarantee amount for traded items with correlations in recognition deviation, thereby reducing the risk of insufficient guarantee amount, has become an urgent technical problem to be solved.

[0004] In summary, to solve the above-mentioned technical problems, this application provides a transaction guarantee management system and method. Summary of the Invention

[0005] To achieve the objectives of this invention, the following technical solution is adopted: Specifically, this application provides a transaction guarantee management method, which includes: S1 uses the recognition results of the image recognition model to determine the recognition deviation of the image recognition model in different trading jewelry, and determines the collateral analysis jewelry in the trading jewelry based on the recognition deviation in different trading jewelry; S2 determines the identification strategy of the collateral reference jewelry based on the degree of correlation between the collateral analysis jewelry and the image recognition model in different transaction jewelry. S3 identifies the reference guarantee item for the guaranteed analysis item based on the identification strategy, and uses the reference guarantee item to determine the guarantee amount of the guaranteed analysis item. Based on the guarantee amount of the guaranteed analysis item and the merchant's transaction item listing data, the merchant's transaction guarantee management method is determined. If the transaction guarantee management method is found to have risks based on the guaranteed analysis item data of different merchants, proceed to the next step. S4 identifies merchants whose guaranteed analysis jewelry is listed as associated listing merchants. Based on the identification deviation of the transaction jewelry listed in the associated listing merchants of the guaranteed analysis jewelry and the upper structure composition data of the guaranteed analysis jewelry, the image recognition analysis strategy of the guaranteed analysis jewelry is determined.

[0006] The beneficial effects of this invention are as follows: Based on the recognition deviation of the image recognition model in different traded items, this study determines the risk of insufficient collateral amount due to the transaction amount of the traded items when subsequent conflicts occur due to recognition deviation. Based on the risk of insufficient collateral amount due to the transaction amount of the traded items, the study identifies the collateral analysis items in the traded items, thus laying the foundation for optimizing the collateral amount analysis of the collateral analysis items and reducing the risk of insufficient collateral amount during after-sales conflicts.

[0007] Based on the identification deviations of the traded jewelry listed by the associated merchants of the guaranteed analysis product, and the upper structure data of the guaranteed analysis jewelry in the associated merchants, the merchants whose listed products include guaranteed analysis products are identified. That is, the risk of insufficient guarantee amount caused by the use of the guaranteed reference jewelry in the associated merchants of guaranteed analysis products is identified. Based on the risk of insufficient guarantee amount caused by the use of the guaranteed reference jewelry in the associated merchants of guaranteed analysis products, the image recognition analysis strategy for guaranteed analysis jewelry is determined, laying the foundation for further improving the update reliability of the guaranteed reference jewelry for guaranteed analysis products.

[0008] Furthermore, the recognition results of the image recognition model are based on the type recognition results included in different traded jewelry.

[0009] Furthermore, the method for determining the collateral analysis jewelry in the transaction jewelry is as follows: S11 determines the number of deviations in the type identification result of the trading jewelry based on the identification deviations in different trading jewelry; S12 determines the identification deviation risk type of the trading jewelry based on the number of deviations in the type identification results; S13 determines the collateral analysis jewelry among the trading jewelry based on the identification bias risk type of different trading jewelry.

[0010] Specifically, the number of deviations in the type identification result of the traded item is determined based on the number of times other traded items are incorrectly identified as the traded item.

[0011] Furthermore, the identification deviation risk type of the trading jewelry is determined based on the number of deviations in the type identification results of the trading jewelry. Specifically, the identification deviation risk type includes a first-class deviation type, a second-class deviation type, and a third-class deviation type, which are determined according to the preset identification deviation risk type corresponding to the number of deviations.

[0012] Furthermore, the method for determining the image recognition analysis strategy for the guaranteed analysis jewelry is as follows: Based on the identification deviation of the transaction accessories listed in the associated merchants of the guaranteed product, the number of identification risk accessories in the associated merchants is determined. Using the upper structure of guaranteed analysis jewelry in different associated merchants as data, determine the proportion of guaranteed analysis jewelry in the identified analysis jewelry in the associated merchants, and use it as the identification analysis ratio of the associated merchants. Based on the number and proportion of risky jewelry items identified in different associated merchants, an image recognition and analysis strategy for guaranteed analysis jewelry items is determined.

[0013] Secondly, this application provides a transaction guarantee management system, which employs the aforementioned transaction guarantee management method, specifically including: Jewelry screening module, guarantee management module, image recognition management module; The jewelry screening module is responsible for identifying the collateralized jewelry among the traded jewelry items. The guarantee management module is responsible for determining the transaction guarantee management method for merchants; The image recognition management module is responsible for determining the image recognition analysis strategy for the guaranteed analysis jewelry.

[0014] Other features and advantages will be set forth in the following description, and the objects and other advantages of the invention are realized and obtained through the structures particularly pointed out in the description and the drawings.

[0015] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0016] The above and other features and advantages of the present invention will become more apparent from a detailed description of exemplary embodiments thereof with reference to the accompanying drawings.

[0017] Figure 1 This is a flowchart of a transaction guarantee management method; Figure 2 This is a flowchart illustrating the method for determining the collateral in the trading of jewelry. Figure 3 This is a flowchart illustrating the method for determining the identification strategy of the collateral reference jewelry in collateral analysis; Figure 4 This is a framework diagram of a transaction guarantee management system. Detailed Implementation

[0018] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this specification.

[0019] Example 1 like Figure 1 As shown, this application provides a transaction guarantee management method, specifically including: S1 uses the recognition results of the image recognition model to determine the recognition deviation of the image recognition model in different trading jewelry, and determines the collateral analysis jewelry in the trading jewelry based on the recognition deviation in different trading jewelry; S2 determines the identification strategy of the collateral reference jewelry based on the degree of correlation between the collateral analysis jewelry and the image recognition model in different transaction jewelry. S3 identifies the reference guarantee item for the guaranteed analysis item based on the identification strategy, and uses the reference guarantee item to determine the guarantee amount of the guaranteed analysis item. Based on the guarantee amount of the guaranteed analysis item and the merchant's transaction item listing data, the merchant's transaction guarantee management method is determined. If the transaction guarantee management method is found to have risks based on the guaranteed analysis item data of different merchants, proceed to the next step. S4 identifies merchants whose guaranteed analysis jewelry is listed as associated listing merchants. Based on the identification deviation of the transaction jewelry listed in the associated listing merchants of the guaranteed analysis jewelry and the upper structure composition data of the guaranteed analysis jewelry, the image recognition analysis strategy of the guaranteed analysis jewelry is determined.

[0020] Furthermore, the recognition results of the image recognition model are based on the type recognition results included in different traded jewelry.

[0021] The recognition result of the image recognition model refers to the output result after the image recognition model performs type recognition on different trading items, including the type recognition label of the trading items and the record of misclassification that occurred during the recognition process; the recognition deviation refers to the historical record of the image recognition model misidentifying other types of trading items as that type of trading items when recognizing a certain type of trading item.

[0022] Assuming there are various types of items available for trading on the platform, after the image recognition model identifies the type of each item, some items may have a history of being misidentified due to their similar appearance. By summarizing these misidentification records, the identification deviation of each item can be determined.

[0023] This step serves as the data entry point for S1. Its significance lies in concretizing the recognition error of the image recognition model into deviation records for each traded item, providing a data foundation for subsequent deviation count statistics and risk classification, and ensuring that the identification of guaranteed analysis items is based on real recognition error data.

[0024] Specifically, such as Figure 2 As shown, the method for determining the collateral analysis jewelry in the transaction jewelry is as follows: In this embodiment, based on the recognition deviation of the image recognition model in different traded items, the risk of insufficient guarantee amount due to the transaction amount of the traded items is determined when subsequent conflicts occur due to recognition deviation. Based on the risk of insufficient guarantee amount due to the transaction amount of the traded items, the guarantee analysis items in the traded items are determined, thereby laying the foundation for optimizing the guarantee amount analysis of the guarantee analysis items and reducing the risk of insufficient guarantee amount in the event of after-sales conflicts.

[0025] S11 determines the number of deviations in the type identification result of the trading jewelry based on the identification deviations in different trading jewelry; The deviation count of the type identification result refers to the number of times in history that other traded items were incorrectly identified as this traded item by the image recognition model, that is, the cumulative number of incorrect classifications with this traded item as the target of incorrect identification.

[0026] Suppose that a certain piece of jewelry A has been misidentified as jewelry A multiple times in the historical identification records by other pieces of jewelry (such as jewelry B, jewelry C, and other similar-looking categories). The total number of these misidentifications is the number of deviations in the type identification result of jewelry A.

[0027] This step quantifies the number of times each traded item is misidentified. Its significance lies in transforming the abstract image recognition deviation into a sortable and comparable numerical indicator, providing a unified measurement benchmark for subsequent classification of recognition deviation risk types, and ensuring that risk classification has an objective data basis.

[0028] S12 determines the identification deviation risk type of the trading jewelry based on the number of deviations in the type identification results; The identified deviation risk type refers to the type corresponding to the preset risk level based on the number of deviations in the traded jewelry. It includes three types of deviations: Type 1, Type 2, and Type 3. The higher the number of deviations, the higher the risk level. Type 1 deviation corresponds to the highest risk, and Type 3 deviation corresponds to the lowest risk.

[0029] Assume the preset classification rules are as follows: when the number of deviations is greater than a certain high threshold, it is classified as a type 1 deviation; when the number of deviations is in the middle range, it is classified as a type 2 deviation; when the number of deviations is low but deviations still exist, it is classified as a type 3 deviation. If a piece of jewelry has a high number of deviations, then the jewelry is classified as a type 1 deviation.

[0030] This step completes the classification of risk types for identifying deviations in various trading items. Its significance lies in discretizing the number of consecutive deviations into a finite number of risk levels, providing a unified classification basis for subsequent quantitative assessment through risk weight values, and also providing a classification basis for differentiated processing strategies for guaranteed analysis items.

[0031] S13 determines the collateral analysis jewelry among the trading jewelry based on the identification bias risk type of different trading jewelry.

[0032] The term "collateral analysis item" refers to any item among all traded items that, due to significant identification bias or high trading activity, has a substantial risk of insufficient collateral amount determined based on the listed price, requiring a specific collateral amount optimization analysis.

[0033] Assuming that each traded item has been categorized into different identification deviation risk types based on the number of deviations in S12, the system comprehensively determines which items should be included in the guarantee analysis based on the average level of the identification risk weight values ​​of all traded items and the proportion of identification risk items in all items.

[0034] This step is the core output step of S1. Its significance lies in accurately screening out the items that require special collateral amount optimization from all traded items. This avoids the unnecessary management costs of including all items in the collateral analysis and ensures that the items with higher risks can receive sufficient collateral protection.

[0035] It is understandable that, based on the identification bias risk type of different trading instruments, the collateral analysis instruments among the trading instruments are determined, specifically including: Based on the identification bias risk type of the traded jewelry, determine the identification risk weight value of the traded jewelry; The identification risk weight value refers to the preset weight mapping value corresponding to the identification deviation risk type of the traded jewelry, which is used to quantify the risk degree of insufficient collateral amount due to identification deviation of the traded jewelry. The weight value corresponding to the first type of deviation is higher than that of the second type, and the weight value corresponding to the second type is higher than that of the third type.

[0036] Assume the preset weight mapping rule is: Type I deviation type corresponds to weight value 3, Type II deviation type corresponds to weight value 2, and Type III deviation type corresponds to weight value 1; if a certain trading item belongs to Type I deviation type, then its risk identification weight value is 3.

[0037] This step transforms the risk level of each trading item into a uniform numerical weight. Its significance lies in providing a calculable data basis for subsequently calculating the average risk weight of all trading items, thus enabling the identification and judgment of guaranteed analysis items to have quantitative support.

[0038] It should be noted that, in case 1: if the average risk weight value of different trading items is greater than the preset weight threshold, then the guaranteed analysis item among the trading items is determined to be a trading item with a deviation number greater than the preset deviation number threshold, and a trading item with an average daily trading number greater than the preset trading number threshold. The mean of the identification risk weight value refers to the arithmetic mean of the identification risk weight values ​​of all trading items with identification bias, reflecting the severity of the overall identification bias risk; the trading items with identification risk whose average daily number of transactions is greater than the preset transaction number threshold refer to trading items with identification risk whose number of deviations may not exceed the deviation number threshold, but whose frequent transactions lead to a higher risk of insufficient collateral amount.

[0039] If the average identification risk weight value of all trading items with identification bias in the platform exceeds the preset weight threshold, it indicates that the overall identification bias risk is high. Therefore, all trading items with a number of biases greater than the preset bias number threshold will be included in the guarantee analysis. At the same time, trading items with high daily trading frequency and identification risk will also be included to ensure the guarantee sufficiency of high trading frequency items.

[0040] This step adopts a lenient inclusion strategy when the overall risk level is high. The significance of this strategy is that when the overall risk of platform identification bias is significant, relying solely on the bias frequency threshold for screening may miss high-frequency items. By simultaneously including items with active daily average transaction frequency, the risk of insufficient collateral can be more comprehensively covered.

[0041] Additionally, it should be noted that in Case 2: if the average risk weight value of different trading items is not greater than the preset weight threshold, trading items with identification deviations will be identified as risky trading items. Based on the proportion of risky trading items in the trading items, it will be determined whether the proportion of risky trading items in the trading items is greater than the preset risky item proportion threshold. If so, the guarantee analysis items in the trading items will be determined as trading items with a deviation number greater than the preset deviation number threshold and risky trading items with an average daily trading number greater than the preset trading number threshold. If not, the guarantee analysis items in the trading items will be determined as trading items with a deviation number greater than the preset deviation number threshold.

[0042] The term "risky trading items" refers to trading items that have been misidentified by other items at least once in the historical identification records; the term "risky trading items as a percentage of trading items" refers to the ratio of the number of items with identification errors to the total number of all trading items on the platform.

[0043] Assuming the overall risk weight average does not exceed the preset weight threshold, the overall risk is acceptable. The system further determines the proportion of risky jewelry items. If the proportion is high (exceeding the preset risky jewelry item proportion threshold), it means that the range of jewelry items affected by the identification deviation is wide, and it is still necessary to include jewelry items with daily average trading activity. If the proportion is low, only jewelry items with deviation times exceeding the threshold need to be included.

[0044] This step introduces the identification of the proportion of risky jewelry for a second judgment when the overall risk level is moderate. Its significance lies in the fact that, under the condition that the overall risk is still acceptable, the breadth of the distribution of risky jewelry can be used to further distinguish whether the scope of guarantee analysis needs to be expanded. This achieves accurate selection of guarantee analysis jewelry under the dual constraints of overall risk level and the scope of risk distribution.

[0045] This embodiment achieves adaptive and accurate identification of guaranteed analysis items through S1. Its core value is reflected in three aspects: First, by quantifying and statistically analyzing the number of deviations and classifying the types of deviation risks, the abstract image recognition error is transformed into an operable classification indicator; second, by making dual judgments based on the average risk weight and the proportion of risky items, the scope of guaranteed analysis items is dynamically adjusted; and third, corresponding inclusion strategies are designed for both high-risk and medium-risk scenarios to ensure a balance between the sufficiency and accuracy of the guaranteed analysis coverage.

[0046] Assuming there are 200 types of traded items on the platform (numbered F001 to F200), the image recognition model has performed historical recognition processing on all items, forming a recognition record.

[0047] Step S11: Count the number of deviations in the type identification results for each traded item (i.e., the historical number of times other items were misidentified as this item). Assume that in the statistical results, 80 items have at least one deviation record, while the remaining 120 items have no deviation records. The deviation counts for some high-deviation items are as follows (only typical examples are listed): F012: 42 deviations; F035: 28 deviations; F067: 15 deviations; F089: 8 deviations; F123: 3 deviations.

[0048] Step S12: Classify and identify the deviation risk types of the 80 types of deviation jewelry according to the preset classification rules. Let the threshold for deviation type 1 be ≥20 deviations, for type 2 be 10 ≤ deviations < 20 deviations, and for type 3 be 1 ≤ deviations < 10 deviations. Then: there are 12 types of deviation type 1 (including F012, F035, etc.), 18 types of deviation type 2 (including F067, etc.), and 50 types of deviation type 3 (including F089, F123, etc.).

[0049] Let the weight mapping be: Class 1 corresponds to a weight value of 3, Class 2 corresponds to a weight value of 2, and Class 3 corresponds to a weight value of 1.

[0050] Step S13: Calculate the mean of the identified risk weights. The average risk weight value is calculated as follows: (12×3 + 18×2 + 50×1)÷80 = (36 + 36 + 50)÷80 = 122÷80 = 1.525.

[0051] If the preset weight threshold is 1.50, then 1.525 > 1.50, which satisfies condition 1.

[0052] Determine the guaranteed analysis items according to situation 1: Include trading items with a deviation number greater than the preset deviation number threshold (set to 10 times) (i.e., 12 types in category 1 + 18 types in category 2 = 30 types), as well as risk-identifying trading items with an average daily trading number greater than the preset trading number threshold (set to 50 times / day) in the scope of guaranteed analysis.

[0053] If we assume that there are 8 additional risky trading items with an average daily trading frequency of >50 times / day (including Category III and some Category II items), then the final number of items for guarantee analysis is 38 (30 items with deviation frequency exceeding the threshold + 8 items with high trading activity, 38 after deduplication).

[0054] Specifically, such as Figure 3 As shown, the method for determining the identification strategy of the guarantee reference jewelry for the guarantee analysis jewelry is as follows: In this embodiment, based on the recognition deviation of the image recognition model in different traded jewelry and the correlation between the collateral analysis jewelry and the collateral risk traded jewelry, the deviation risk of determining the collateral amount using the image recognition model is determined. Based on the deviation risk of determining the collateral amount using the image recognition model, the identification strategy of the collateral reference jewelry for collateral analysis jewelry is determined. This not only avoids the technical problem of poor transaction flexibility for merchants caused by excessive collateral amount, but also reduces the risk of insufficient collateral amount when there are transaction conflicts in the later stage by optimizing the collateral amount.

[0055] S21 Based on the recognition deviation of the image recognition model in different trading items, the trading items with recognition deviation are identified as trading items with risk. The term "risky traded items" refers to traded items that, in historical identification records, have been mistakenly identified as such by other types of traded items at least once. Because of the risk of misidentification, the guarantee amount determined based on the actual price of these items may be used by merchants to circumvent higher guarantee obligations.

[0056] If, among the 200 types of trading items on the platform, 80 types have a history of being misidentified by other items, then these 80 types of items are identified as risky trading items.

[0057] This step completes the scope definition of risky trading items. Its significance lies in separating items with identification bias risk from all trading items, focusing subsequent analysis resources, and avoiding unnecessary optimization calculations of collateral amounts for items without identification bias.

[0058] S22 determines the proportion of the guarantee analysis jewelry in the risk identification jewelry based on the degree of correlation between the guarantee analysis jewelry and the risk identification transaction jewelry, and uses the proportion of the guarantee analysis jewelry in the risk identification transaction jewelry as the guarantee analysis proportion. The percentage of guaranteed analysis refers to the ratio of the number of guaranteed analysis items to the total number of risky transaction items identified. It reflects the concentration of guaranteed analysis items in the group of risky transaction items identified. The higher the percentage, the wider the coverage of guaranteed analysis items in the identification of risky transaction items, and the more prominent the importance of guarantee management.

[0059] Assuming there are 80 types of risky trading items and 38 types of items undergoing guarantee analysis, then the proportion of guarantee analysis = 38 ÷ 80 = 0.475.

[0060] This step quantifies the distribution density of collateralized analysis items in identifying risky trading items as the proportion of collateralized analysis. Its significance lies in providing a core proportional indicator for subsequent identification strategy allocation and judgment. The level of the collateralized analysis proportion directly determines the strictness of the collateralized reference item identification strategy adopted.

[0061] It is understood that the above steps include the following: Determine whether the proportion of the guarantee analysis is less than the preset guarantee analysis proportion threshold. If yes, determine that the identification strategy of the guarantee reference jewelry of the guarantee analysis jewelry is the first identification strategy. If no, proceed to step S23. The first identification strategy is to select the transaction item with the highest average transaction amount among the other transaction items that are incorrectly identified as the collateral analysis item as the collateral reference item for the collateral analysis item.

[0062] The preset guarantee analysis ratio threshold is the critical value for distinguishing between high and low density of guarantee analysis jewelry; the first identification strategy, by selecting the jewelry with the highest average transaction amount as a reference, can cover the risk of insufficient guarantee amount to the maximum extent, and is the most conservative guarantee reference jewelry selection strategy.

[0063] Assuming the guarantee analysis ratio is 0.30 and the preset guarantee analysis ratio threshold is 0.40, if 0.30 < 0.40, then the first identification strategy is directly adopted—from all other jewelry that were incorrectly identified as the guarantee analysis jewelry, the jewelry with the highest average transaction amount is selected as the guarantee reference jewelry.

[0064] This step employs the most conservative strategy when the proportion of guaranteed analysis items is low. The significance of this is that when the density of guaranteed analysis items in identifying risky items is low, it indicates that the identification bias is relatively dispersed. The scenarios in which each type of guaranteed analysis item is misidentified are different, and using the highest amount as a reference can maximize the guarantee protection effect.

[0065] S23 uses the proportion of the guarantee analysis and the identification deviation data of the risk-identifying transaction jewelry excluding the guarantee analysis jewelry to determine the identification strategy of the guarantee reference jewelry.

[0066] The risk identification trading items other than the guaranteed analysis items (hereinafter referred to as other trading items) refer to the risk identification items remaining after deducting the guaranteed analysis items from the risk identification trading items. The size and risk weight of these items reflect the complexity of the current platform's identification bias problem.

[0067] If the proportion of guaranteed analysis is greater than or equal to the preset guaranteed analysis proportion threshold, it indicates that the proportion of guaranteed analysis items in identifying risky items is relatively high, and it is necessary to further combine the characteristics of other trading items to refine the identification strategy.

[0068] This step involves refined judgment when the proportion of collateral analysis does not meet the conditions for rapid diversion. Its significance lies in further optimizing the selection strategy of collateral reference items by analyzing the scale and risk level of the remaining risk identification items.

[0069] Specifically, the above steps include the following: S231 takes the risk-identifying trading items other than the guarantee analysis items as other trading items, and determines whether the proportion of the other trading items in the trading items is greater than the preset value of the item proportion. If so, the identification strategy of the guarantee reference item of the guarantee analysis item is determined to be the second identification strategy. If not, proceed to step S232. The proportion of other trading items in the total number of trading items refers to the ratio of the number of remaining items after deducting the guarantee analysis item from the identified risk trading items to the total number of all trading items. The specific content of the second identification strategy is as follows: if the guarantee analysis item belongs to a type of deviation, then the trading item with the highest average transaction amount among the other trading items that were incorrectly identified as the guarantee analysis item is used as the guarantee reference item for the guarantee analysis item; if the guarantee analysis item does not belong to a type of deviation, then the trading item with the highest average transaction amount among the other trading items that were incorrectly identified as the guarantee analysis item more than a preset threshold number is used as the guarantee reference item for the guarantee analysis item.

[0070] Assuming that among the 80 risky trading items identified, 38 are for collateral analysis, then the number of other trading items is 80 - 38 = 42. The percentage of other trading items in the total 200 trading items is 42 ÷ 200 = 0.21. Assuming the preset percentage of items is 0.20, and 0.21 > 0.20, the second identification strategy is directly determined.

[0071] This step uses the scale of other traded items as a criterion. Its significance lies in the fact that when there are a large number of non-guaranteed analysis items among the identified risky items, it indicates that the platform's overall identification bias problem is relatively common. A second identification strategy that differentiates based on the risk type of guaranteed analysis items can be adopted to balance the efficiency of selecting guarantee references for high-risk and medium-to-low-risk items.

[0072] S232 Based on the identification risk weight values ​​of the other trading jewelry, determine whether the sum of the identification risk weight values ​​of the other trading jewelry is greater than the preset risk weight value. If so, determine the identification strategy of the collateral reference jewelry of the collateral analysis jewelry as the second identification strategy. If not, proceed to step S233. The sum of the identification risk weight values ​​of the other trading items refers to the cumulative risk weight value obtained by adding up the identification risk weight values ​​of all other trading items (Category 1 3, Category 2 2, Category 3 1), which reflects the overall risk intensity of the non-guaranteed analysis items in the identified risk trading items.

[0073] Assuming that among the 42 other trading items, there are 3 in category 1 (weight 3), 12 in category 2 (weight 2), and 27 in category 3 (weight 1), then the sum of the risk weight values ​​is 3×3 + 12×2 + 27×1 = 9 + 24 + 27 = 60; if the preset risk weight is 55, and 60 > 55, then the second identification strategy is adopted.

[0074] This step uses the overall risk weight strength of other traded items as a judgment condition. Its significance is that when the cumulative identified risk weight of other traded items is high, it indicates that the overall risk of the remaining identified risk item group cannot be ignored, and the second identification strategy still needs to be adopted; when the cumulative risk weight is low, it enters S233 to calculate a more refined collateral matching value.

[0075] S233 determines the collateral matching value of the collateral analysis item based on the collateral analysis ratio and the proportion of other trading items in the trading items, and determines whether the collateral matching value is greater than a preset matching threshold. If so, the identification strategy of the collateral reference item of the collateral analysis item is determined to be the third identification strategy; otherwise, the identification strategy of the collateral reference item of the collateral analysis item is determined to be the second identification strategy.

[0076] The third identification strategy is to select the trading item with the highest average transaction amount among other trading items that have been incorrectly identified as the guaranteed analysis item more than a preset threshold number of times as the guaranteed reference item for the guaranteed analysis item. The guaranteed matching value reflects the priority of the guaranteed management of the guaranteed analysis item relative to other trading items. The higher the proportion of guaranteed analysis and the lower the proportion of other trading items, the higher the guaranteed matching value, indicating that the guaranteed management needs of the guaranteed analysis item are relatively more concentrated.

[0077] It should be noted that the higher the proportion of the collateral analysis and the lower the proportion of other traded items in the total traded items, the higher the collateral matching value of the collateral analysis item.

[0078] Assuming the collateral analysis accounts for R1, and other traded items account for R2 of the total traded items, let the formula for calculating the collateral matching value be: Guarantee Matching Value = Guarantee Analysis Ratio ÷ (Guarantee Analysis Ratio + Ratio of Other Traded Trinkets in Total Traded Trinkets) Assuming the collateral analysis ratio R1 = 0.475 and the ratio of other traded items in the total traded items R2 = 0.21, substituting these values ​​into the calculation: Collateral Matching Value = 0.475 ÷ (0.475 + 0.21) = 0.475 ÷ 0.685 ≈ 0.694.

[0079] This step achieves precise switching between the third and second identification strategies through comprehensive calculation of the guarantee matching value. Its significance lies in the fact that the third identification strategy adds the condition constraint of "the number of times reaches more than the preset number threshold" on the basis of the first identification strategy. It is more accurate than the first identification strategy (eliminating the influence of occasional misidentification) and more consistent than the second identification strategy (not relying on the difference in the deviation type of the jewelry for guarantee analysis). The use of the guarantee matching value determines whether to adopt the intermediate strategy, realizing the fine-grained layering of identification strategies.

[0080] Furthermore, the determination of the collateral amount for the collateral analysis jewelry using the aforementioned collateral reference jewelry specifically includes: The maximum value between the average transaction amount of the reference jewelry and the listed price of the jewelry being analyzed is used as the collateral amount for the jewelry being analyzed.

[0081] It is understood that the method for determining the transaction guarantee management method for the aforementioned merchants is as follows: Based on the merchant's listing data for traded jewelry, determine the price of the merchant's traded jewelry; The total guarantee amount for the merchant is determined by summing the guarantee amount of the guaranteed analysis jewelry with the listed price of the traded jewelry excluding the guaranteed analysis jewelry.

[0082] The determination of the transaction guarantee management method refers to replacing the original price with the guarantee amount of the guaranteed analysis of the jewelry, and combining it with the total price of all the jewelry listed by the merchant to determine the total guarantee amount payable by each merchant, thus forming a merchant-level transaction guarantee management method.

[0083] Assuming that S2 has completed the determination of the guarantee amount for each item in the guarantee analysis, and combined with the listing data of each merchant, the total guarantee amount for each merchant can be calculated, thus forming a transaction guarantee management method for merchants.

[0084] This step is the main entry point for S3. Its significance lies in elevating the jewelry-level guarantee amount results of S2 to the overall guarantee management method at the merchant level. At the same time, it introduces a risk verification mechanism at the merchant level to provide a basis for judgment for identifying merchants with insufficient guarantee amount management risks.

[0085] Specifically, the risk identified in the transaction guarantee management method includes: In this embodiment, based on the data of guaranteed jewelry from different merchants, the risk of insufficient guarantee amount in different merchants is determined. Based on the risk of insufficient guarantee amount in different merchants, it is determined whether the transaction guarantee management method has any risk. This lays the foundation for optimizing the image recognition analysis strategy for different guaranteed jewelry when the transaction guarantee management method has any risk, and further improving the efficiency and reliability of updating the guarantee reference jewelry for guaranteed jewelry.

[0086] S31 determines the proportion of guaranteed jewelry in the identification of risky transaction jewelry based on the data of guaranteed jewelry in different merchants, and uses it as the identification and analysis ratio of the merchant. The merchant identification and analysis ratio refers to the ratio of the number of guaranteed analysis items listed by the merchant to the total number of risk-identifying transaction items listed by the merchant. It reflects the proportion of risk-identifying items included in the guarantee analysis management among the merchant's listed items. The higher the ratio, the higher the proportion of risk-identifying items listed by the merchant that are properly included in the guarantee amount optimization management, and the lower the risk of insufficient guarantee amount.

[0087] Suppose merchant Y lists 12 types of risk-identifying transaction items, of which 10 are guaranteed analysis items. Then merchant Y's risk identification and analysis ratio = 10 ÷ 12 ≈ 0.833. Another merchant Z lists 8 types of risk-identifying transaction items, of which only 2 are guaranteed analysis items. Then merchant Z's risk identification and analysis ratio = 2 ÷ 8 = 0.250.

[0088] This step quantifies the coverage of guaranteed jewelry for each merchant into an identification and analysis ratio. Its significance lies in providing a unified ratio indicator for subsequent risk management of merchant identification. Merchants with a low identification and analysis ratio indicate that most of the risk-identifying jewelry they have listed has not been included in the optimized management of the guarantee amount, and there is a high risk of insufficient guarantee amount.

[0089] S32 uses different merchant identification and analysis ratios to determine whether the transaction guarantee management method poses any risks.

[0090] The risk of the transaction guarantee management method lies in the fact that there are too many merchants with insufficient identification and analysis ratios in the overall merchant group, indicating that the current method of setting the scope of guaranteed analysis items or determining the guarantee amount cannot effectively cover the actual guarantee needs of the platform.

[0091] It is understandable that the risk of the transaction guarantee management method is determined by analyzing different merchant identification ratios, specifically including: Merchants whose identification and analysis ratio is less than a preset analysis ratio threshold are identified as merchants subject to risk control. When the number of such merchants exceeds a preset risk merchant number threshold, the transaction guarantee management method is deemed to have a risk.

[0092] The risk-controlled merchants refer to those whose identification and analysis ratio is lower than the preset analysis ratio threshold. Among the risk-identified jewelry items listed by these merchants, the proportion covered by guarantee analysis and management is insufficient, posing a potential risk of insufficient guarantee amount. The preset risk merchant number threshold is a critical value for judging whether there is systemic risk in the overall guarantee management system.

[0093] Assuming the preset analysis ratio threshold is 0.50, merchants with an analysis ratio of less than 50% are marked as risk-controlled merchants. The number of risk-controlled merchants among all merchants is counted. If the number exceeds the preset risk merchant number threshold (set to 10), it is determined that the transaction guarantee management method has risks, and the process proceeds to S4 to optimize the image recognition analysis strategy for guarantee analysis jewelry.

[0094] This step uses the number of risky merchants as the basis for judging overall risk. Its significance lies in transforming the issue of guarantee analysis coverage at the jewelry level into a quantifiable system-level risk indicator through risk aggregation at the merchant level. This provides a clear basis for judging whether to trigger the S4 optimization process and ensures that the guarantee analysis system can respond promptly to the rise in overall risk.

[0095] This embodiment completes the risk status assessment of the transaction guarantee management method through S3. Its core value is reflected in two aspects: First, by quantifying the merchant identification and analysis ratio, the coverage of guaranteed analysis items is transformed from the item dimension to the merchant dimension, which more intuitively reflects the actual effect of guarantee management; Second, by comparing the number of risky merchants with the preset threshold, clear conditions for triggering further optimization are established, forming a dynamic self-checking mechanism for the guarantee management system.

[0096] Specifically, the method for determining the image recognition analysis strategy for the guaranteed analysis jewelry is as follows: In this embodiment, based on the identification deviations of the traded items listed in the associated merchants of the guaranteed analysis product and the upper structure data of the guaranteed analysis items in the associated merchants, the merchants whose listed items include guaranteed analysis products are identified. That is, the risk of insufficient guarantee amount caused by the guarantee reference items of guaranteed analysis items in the associated merchants of guaranteed analysis products is identified. Based on the risk of insufficient guarantee amount caused by the guarantee reference items of guaranteed analysis items in the associated merchants of guaranteed analysis products, the image recognition analysis strategy of guaranteed analysis items is determined, laying the foundation for further improving the update reliability of the guarantee reference items of guaranteed analysis items.

[0097] S41 determines the number of risky jewelry items in the associated merchants based on the identification deviation of the traded jewelry items listed in the associated merchants of the guaranteed analysis product. The risky jewelry items listed in the associated merchants refer to the jewelry items with identification bias among all the jewelry items listed in the associated merchants. The number of such items reflects the scale of jewelry items affected by identification bias in the associated merchants.

[0098] Assuming that the guaranteed analysis item F012 has 15 associated listed merchants, and among all the traded items listed by these 15 merchants, a certain number of items have identification bias, count the number of these risky items.

[0099] This step completes the basic statistics on the number of risky accessories identified among the associated merchants. Its significance lies in providing the basic data for the subsequent rapid judgment in S41 and the calculation of the basic risk weight value in S412.

[0100] The above steps include the following: S411 Obtain the associated merchants of the identified risky jewelry, and determine whether the number of associated merchants of the identified risky jewelry is greater than the preset threshold for the number of associated merchants. If not, determine that the image recognition analysis strategy of the guaranteed analysis jewelry does not require additional image analysis processing, and only needs to perform image analysis processing on the platform settings including listing and sales. If yes, proceed to step S412. The associated merchants for identifying risky jewelry refer to the set of merchants that have listed the jewelry currently under the guarantee analysis; the strategy of not requiring additional image analysis processing means that image recognition is performed only at standard times specified by the platform (such as when jewelry is listed or when a sales activity occurs), without increasing the recognition frequency.

[0101] Suppose that there are 3 associated merchants for a certain guaranteed analysis item F123. Assume that the preset threshold for the number of associated merchants is 5. If 3 < 5 (not greater than the threshold), the judgment is "if not". It is determined that the guaranteed analysis item does not need additional image analysis processing and can only perform recognition according to the platform's standard time. If the number of associated merchants is 8 (greater than 5), the judgment is "if". Proceed to S412.

[0102] This step involves quickly determining the number of associated merchants to distribute traffic. Its significance lies in the fact that when the number of associated merchants is small, the market circulation coverage of the guaranteed jewelry is limited, the risk of insufficient guarantee amount is low, and there is no need to increase the identification frequency, thus avoiding unnecessary consumption of computing resources.

[0103] S412 determines the basic risk weight value of the associated merchants based on the number of risk-identifying jewelry items in different associated merchants, and judges whether the average value of the basic risk weight value in different associated merchants is above the risk weight threshold. If so, the image recognition analysis strategy for the guaranteed analysis jewelry is determined to be that if the number of image recognitions in the most recent preset time period is less than the preset recognition number threshold, then image recognition analysis processing will be performed according to the preset cycle in the future preset time period, thereby realizing the reliable update of the guaranteed reference jewelry for the identified risk jewelry and further reducing the guarantee risk. If not, proceed to step S42. The basic risk weight value of the associated listed merchants refers to a risk quantification indicator determined based on the number of risky accessories identified in the merchant. The more risky accessories identified, the higher the basic risk weight value. The "triggering periodic identification within a future preset time period when the number of identifications in the most recent preset time period is less than a preset identification threshold" in the image recognition analysis strategy refers to a strategy that is triggered to supplement the identification when the recent identification frequency is insufficient, so as to ensure that the guaranteed reference accessories are updated with sufficient frequency.

[0104] Assuming there are 15 associated merchants F012, count the number of risky accessories identified in each merchant, calculate the basic risk weight value for each merchant (basic risk weight value = number of risky accessories identified by the merchant ÷ preset threshold for the number of risky accessories identified by a single merchant), calculate the average basic risk weight value of the 15 merchants, if the average value is above the risk weight threshold (set to 0.60), then determine the supplementary periodic identification strategy; if the average value is below 0.60, then proceed to S42.

[0105] This step uses the average basic risk weight as the intermediate judgment condition. Its significance is that when the number of risky jewelry items identified by the associated merchants as a whole is relatively large, it indicates that the identification deviation of the guaranteed reference jewelry items is more complicated. It is necessary to increase the identification frequency to ensure the timeliness of the guaranteed reference jewelry item data, thereby reducing the risk caused by the lag in updating the guarantee amount.

[0106] S42 uses the upper structure of the guaranteed analysis jewelry in different associated merchants to form data, determines the proportion of the guaranteed analysis jewelry in the associated merchants in the identification analysis jewelry, and uses it as the identification analysis ratio of the associated merchants. The identification and analysis ratio of associated listed merchants refers to the ratio of the number of guaranteed analysis items to the total number of identified analysis items (i.e., risky transaction items) listed by each associated listed merchant. It is used to assess the coverage of the guaranteed analysis management of associated listed merchants.

[0107] Suppose a certain associated merchant has listed 10 types of risk-identifying transaction items, of which 6 are guaranteed analysis items. Then the merchant's risk identification analysis ratio = 6 ÷ 10 = 0.60.

[0108] This step calculates the identification and analysis ratio of each associated listed merchant. Its significance lies in assessing the coverage balance of guarantee analysis and management from the perspective of associated listed merchants, and providing a data foundation for the subsequent calculation of the comprehensive guarantee risk weight value.

[0109] The above steps include the following: Based on the identification and analysis ratio of different associated merchants, determine whether there are associated merchants whose identification and analysis ratio is greater than the preset distribution ratio threshold. If yes, proceed to step S43. If no, determine that the image identification and analysis strategy of the guaranteed analysis jewelry is that when the number of image identifications in the most recent preset time period is less than the preset number of identifications threshold, then image identification and analysis processing will be performed according to the preset cycle in the future preset time period, thereby realizing the reliable update of the guaranteed reference jewelry for the identified risk jewelry and further reducing the guarantee risk. Assuming the preset distribution ratio threshold is 0.50, the identification and analysis ratio of all associated listed merchants is statistically analyzed. If there are associated listed merchants with an identification and analysis ratio greater than 0.50, then proceed to S43 for refined comprehensive calculation; if the identification and analysis ratio of all merchants is ≤0.50, then a supplementary periodic identification strategy is determined.

[0110] This step uses the distribution of the identification and analysis ratio as a judgment condition. Its significance is that when some merchants among the associated listed merchants have a high degree of coverage of guaranteed analysis jewelry, it indicates that these merchants have a sufficient basis for guaranteed analysis and need to be further evaluated in conjunction with the basic risk weight value. When the coverage of all merchants is insufficient, a supplementary periodic identification strategy is directly adopted.

[0111] S43 determines the image recognition and analysis strategy for guaranteed analysis items based on the number and proportion of risky items identified in different associated merchants.

[0112] The core of the image recognition analysis strategy determined by S43 is to accurately select between two strategies: "triggering periodic recognition for a future preset duration when the number of recognitions is less than a threshold in the most recent preset duration" and "triggering periodic recognition for a second future preset duration when the number of recognitions is less than a threshold in the most recent preset duration" by calculating and judging the comprehensive guarantee risk weight value (a comprehensive risk index combining the recognition analysis ratio and the basic risk weight value).

[0113] If S42 determines that there are associated merchants whose identification and analysis ratio is greater than the preset distribution ratio threshold, then in S43, the system calculates the comprehensive guarantee risk weight value for all associated merchants and determines the final identification and analysis strategy accordingly.

[0114] This step is the final refined decision-making step of S4. Its significance lies in achieving precise switching between different identification frequency strategies through quantitative evaluation of comprehensive guarantee risk weight values. This ensures that the frequency of image recognition analysis is highly matched with the level of guarantee risk, avoiding both the lag in updating guarantee reference items due to too low an identification frequency and the unnecessary consumption of computing resources due to too high an identification frequency.

[0115] Specifically, based on the identification and analysis ratio of risky jewelry in different associated merchants of the guaranteed analysis product and the basic risk weight value, the comprehensive guarantee risk weight value of the associated merchants of the guaranteed analysis product is determined. It is then determined whether the comprehensive guarantee risk value of the associated merchants of the guaranteed analysis product is greater than a preset guarantee risk coefficient threshold. If so, the image recognition analysis strategy for the guaranteed analysis jewelry is determined to be that if the number of image recognitions within the most recent preset time period is less than a preset recognition number threshold, image recognition analysis processing will be performed according to a preset cycle within a future preset time period, thereby achieving reliable updates to the guarantee reference jewelry for the identified risky jewelry and further reducing guarantee risk. If not, the image recognition analysis strategy for the guaranteed analysis jewelry is determined to be that if the number of image recognitions within the most recent preset time period is less than a preset recognition number threshold, image recognition analysis processing will be performed according to a preset cycle within a future second preset time period, thereby achieving reliable updates to the guarantee reference jewelry for the identified risky jewelry and further reducing guarantee risk.

[0116] The comprehensive guarantee risk weight value refers to the overall guarantee risk quantification index calculated by combining the identification and analysis ratio of each associated listed merchant with the basic risk weight value. It is determined based on the sum of the products of (1-identification and analysis ratio) and the basic risk weight value of the identified risk accessories in different associated listed merchants. The difference between the preset duration and the second preset duration is that the second preset duration is longer than the preset duration, corresponding to a lower frequency of identification update strategy.

[0117] Specifically, the comprehensive guarantee risk weight value of the associated merchants of the guaranteed analysis product is determined by the sum of the products of (1 - identification analysis ratio) of the identified risk jewelry in different associated merchants and the basic risk weight value.

[0118] Suppose that among the merchants associated with a certain guaranteed jewelry item, there are merchants whose identification and analysis ratio exceeds a preset distribution ratio threshold (enter S43), calculate the comprehensive guarantee risk weight value for all associated merchants: Comprehensive guarantee risk weight value = Σ[(1 - proportion of each merchant identification and analysis) × basic risk weight value of each merchant] Substituting specific values: Suppose there are n associated merchants, each with an identification and analysis ratio of Ri, and a basic risk weight value of Wi. Then the comprehensive guarantee risk weight value = Σ[(1-Ri)×Wi]. If this comprehensive value is greater than the preset guarantee risk coefficient threshold, then a preset periodic identification strategy with a higher frequency is adopted; otherwise, a second preset periodic identification strategy with a lower frequency is adopted.

[0119] This step combines the sum of (1 - identification and analysis ratio) × basic risk weight value to comprehensively quantify the degree of guarantee analysis coverage gap for each merchant (the larger the 1 - identification and analysis ratio, the more insufficient the coverage) and the intensity of identification risk (the higher the basic risk weight value, the greater the identification risk). The higher the comprehensive value, the greater the overall guarantee risk of the associated listed merchants, and the more frequent the image recognition updates are needed to reduce the risk.

[0120] This embodiment achieves multi-level adaptive determination of the guarantee analysis jewelry image recognition strategy through S4. Its core value is reflected in three aspects: First, by quickly judging the number of merchants associated with the listing, unnecessary additional recognition resource investment is avoided when market coverage is limited; second, by making dual judgments based on the average basic risk weight and the distribution of recognition analysis ratios, supplementary periodic recognition strategies are accurately triggered in most medium-risk scenarios; third, by finely calculating the comprehensive guarantee risk weight value, the preset duration strategy and the second preset duration strategy are accurately switched in complex scenarios with a wide distribution of recognition analysis ratios, ensuring a high degree of matching between recognition frequency and guarantee risk level.

[0121] Example 2 Secondly, such as Figure 4 As shown, this application provides a transaction guarantee management system, which employs the above-mentioned transaction guarantee management method, specifically including: Jewelry screening module, guarantee management module, image recognition management module; The jewelry screening module is responsible for identifying the collateralized jewelry among the traded jewelry items. The guarantee management module is responsible for determining the transaction guarantee management method for merchants; The image recognition management module is responsible for determining the image recognition analysis strategy for the guaranteed analysis jewelry.

[0122] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and non-volatile computer storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0123] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0124] The above description is merely one or more embodiments of this specification and is not intended to limit this specification. Various modifications and variations can be made to the one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of this specification.

Claims

1. A method for managing transaction guarantees, characterized in that, Specifically, it includes: Based on the recognition results of the image recognition model, determine the recognition deviation of the image recognition model in different trading jewelry, and determine the collateral analysis jewelry in the trading jewelry based on the recognition deviation in different trading jewelry; Based on the degree of correlation between the collateral analysis item and the image recognition model in the recognition deviation of different traded items, the recognition strategy of the collateral analysis item as a collateral reference item is determined. Based on the identification strategy, the process of identifying the reference jewelry for the guarantee analysis is performed, and the guarantee amount of the guarantee analysis is determined using the reference jewelry. Based on the guarantee amount of the guarantee analysis and the merchant's transaction jewelry listing data, the merchant's transaction guarantee management method is determined. If the transaction guarantee management method is found to have risks based on the guarantee analysis jewelry data of different merchants, the process proceeds to the next step. Merchants whose guaranteed analysis jewelry is listed are designated as associated listing merchants. Based on the identification deviation of the transaction jewelry listed in the associated listing merchants of the guaranteed analysis jewelry and the upper structure composition data of the guaranteed analysis jewelry, the image recognition analysis strategy of the guaranteed analysis jewelry is determined.

2. The transaction guarantee management method as described in claim 1, characterized in that, The recognition result of the image recognition model is based on the type recognition result included in different traded jewelry.

3. The transaction guarantee management method as described in claim 1, characterized in that, The method for determining the collateral analysis items in the traded jewelry is as follows: Based on the identification deviation in different traded jewelry, determine the number of deviations in the type identification results of the traded jewelry; The identification deviation risk type of the traded jewelry is determined by the number of deviations in the type identification results. Based on the identification bias risk type of different trading jewelry, the collateral analysis jewelry among the trading jewelry is determined.

4. The transaction guarantee management method as described in claim 3, characterized in that, The number of deviations in the type identification result of the traded jewelry is determined based on the number of times other traded jewelry is incorrectly identified as the traded jewelry.

5. The transaction guarantee management method as described in claim 3, characterized in that, The identification bias risk type of the traded jewelry is determined based on the number of deviations in the type identification results of the traded jewelry.

6. The transaction guarantee management method as described in claim 3, characterized in that, Based on the different identification bias risk types of the traded items, the collateral analysis items among the traded items are identified, specifically including: Based on the identification deviation risk type of the trading items, the identification risk weight value of the trading items is determined. If the average identification risk weight value of different trading items is greater than the preset weight threshold, then the guaranteed analysis items among the trading items are identified as trading items with a deviation number greater than the preset deviation number threshold, and trading items with identification risk whose average daily trading number is greater than the preset trading number threshold.

7. The transaction guarantee management method as described in claim 1, characterized in that, The determination of the collateral amount for the collateral analysis jewelry using the aforementioned collateral reference jewelry specifically includes: The maximum value between the average transaction amount of the reference jewelry and the listed price of the jewelry being analyzed is used as the collateral amount for the jewelry being analyzed.

8. The transaction guarantee management method as described in claim 1, characterized in that, The method for determining the transaction guarantee management method for the aforementioned merchants is as follows: Based on the merchant's listing data for traded jewelry, determine the price of the merchant's traded jewelry; The total guarantee amount for the merchant is determined by summing the guarantee amount of the guaranteed analysis jewelry with the listed price of the traded jewelry excluding the guaranteed analysis jewelry.

9. The transaction guarantee management method as described in claim 1, characterized in that, The method for determining the image recognition analysis strategy for the guaranteed analysis jewelry is as follows: Based on the identification deviation of the transaction accessories listed in the associated merchants of the guaranteed product, the number of identification risk accessories in the associated merchants is determined. Using the upper structure of guaranteed analysis jewelry in different associated merchants as data, determine the proportion of guaranteed analysis jewelry in the identified analysis jewelry in the associated merchants, and use it as the identification analysis ratio of the associated merchants. Based on the number and proportion of risky jewelry items identified in different associated merchants, an image recognition and analysis strategy for guaranteed analysis jewelry items is determined.

10. A transaction guarantee management system, employing the transaction guarantee management method according to any one of claims 1-9, characterized in that, Specifically, it includes: Jewelry screening module, guarantee management module, image recognition management module; The jewelry screening module is responsible for identifying the collateralized jewelry among the traded jewelry items. The guarantee management module is responsible for determining the transaction guarantee management method for merchants; The image recognition management module is responsible for determining the image recognition analysis strategy for the guaranteed analysis jewelry.

Images