An order management method and system based on machine learning model

Abstract

The present invention provides an order management method and system based on a machine learning model, which belongs to the technical field of alarm devices, and specifically includes: determining the distribution data of the extraction time deviation date of a matching e-commerce platform within a preset time period, and based on the distribution data and the historical order data of the matching e-commerce platform, determining a matching e-commerce platform with a variable extraction strategy, and using it as a strategy-variable platform, and based on the commodity listing data in the strategy-variable platform, determining the intersection of commodities in the strategy-variable platform and other matching e-commerce platforms with different extraction processing strategies, and determining the extraction processing strategies of the matching e-commerce platform in different time periods in combination with the inventory data of different commodities, thereby improving the operating stability of the background server.

Description

Technical Field

[0001] The present invention belongs to the technical field of order management, and in particular, relates to an order management method and system based on a machine learning model. Background Art

[0002] In order to realize order management of e-commerce enterprises on different e-commerce platforms, in the invention patent application CN202011629374.1 "Order Interaction Management System Based on Multi-Platform", Java multi-threading technology and database are used to realize concurrent downloading to support data real-time and automation at the current level, which effectively solves the bottleneck brought by the growth of cross-border e-commerce business volume. However, the existing technical solutions have the following technical problems:

[0003] In the process of order extraction and processing, the existing technical solutions only take into account the limitations of hardware resources. However, on different e-commerce platforms, due to differences in data such as the number of products on the shelves, there are differences in the order update requirements of different platforms. Therefore, how to generate differentiated order update processing strategies based on the order update requirements of different platforms cannot guarantee the timeliness and effectiveness of order processing.

[0004] In response to the above technical problems, the present application specifically provides an order management method and system based on a machine learning model. Summary of the invention

[0005] To achieve the purpose of the present invention, the present invention adopts the following technical solutions:

[0006] Specifically, in a first aspect, the present application provides an order management method based on a machine learning model, which specifically includes:

[0007] S1 obtains the number of matching e-commerce platforms of the e-commerce enterprise, and combines the abnormal data of the server load in the process of extracting and processing orders using the asynchronous extraction strategy. When it is determined that the asynchronous extraction strategy cannot be used at the same time to extract and process the order data, proceed to the next step;

[0008] S2 determines the extraction time deviation date of the matching e-commerce platform using a machine learning model based on the extraction processing time of the matching e-commerce platform under the asynchronous extraction strategy and the preset time threshold;

[0009] S3 determines the distribution data of the extraction duration deviation date of the matching e-commerce platform within a preset period, and based on the distribution data and the historical order data of the matching e-commerce platform, determines the matching e-commerce platform with variable extraction strategy, and uses it as the strategy variable platform;

[0010] S4 determines the intersection of the products in the strategy variable platform and other matching e-commerce platforms with different extraction and processing strategies based on the product listing data in the strategy variable platform, and determines the extraction and processing strategies of the matching e-commerce platforms in different time periods in combination with the inventory data of different products.

[0011] The beneficial effects of the present invention are:

[0012] Based on the distribution data of the extraction time deviation date of the matching e-commerce platform within the preset time period and the historical order data, the matching e-commerce platform with a variable extraction strategy is determined, thereby realizing the screening of matching e-commerce platforms with a small number of historical orders and a small proportion of the number of extraction time deviation dates, avoiding the technical problems of excessive server load and overly complex operating logic of the software system caused by the use of asynchronous extraction strategy for all matching e-commerce platforms, thereby ensuring the operating reliability of the software system and the operating reliability of the server.

[0013] The extraction and processing strategies of the e-commerce platform in different periods are determined based on the intersection of the products in the strategy-variable platform with other matching e-commerce platforms with different extraction and processing strategies, and the inventory data of different products. This not only takes into account the differences in the real-time requirements for updating the sales data of the products on the strategy-variable platform due to the differences in the intersection with other matching e-commerce platforms, but also further combines the inventory data to realize the differences in the real-time requirements for updating processing on the matching e-commerce platforms due to the differences in inventory quantities. This realizes the extraction and processing strategies of matching e-commerce platforms from multiple dimensions, which ensures the overall operation security and stability while reducing the operating load of the server.

[0014] A further technical solution is that the matching e-commerce platform is an e-commerce platform where the e-commerce enterprise has products on the shelves.

[0015] A further technical solution is that the abnormal data of the server load includes the historical operating time period of the server load within a preset load rate range and the number of load rate abnormality alarms.

[0016] A further technical solution is to determine that it is not possible to simultaneously use an asynchronous extraction strategy to extract and process order data, specifically including:

[0017] Based on the abnormal data of the server load, determine the historical operating time ratio of the server load within a preset load rate range, and use it as the high load rate time ratio;

[0018] Obtaining the number of matching e-commerce platforms of the e-commerce enterprise, and determining the extraction management complexity coefficient of the asynchronous extraction strategy according to the number of matching e-commerce platforms;

[0019] The average abnormality coefficient is determined according to the proportion of the high load rate time and the average value of the extraction management complexity coefficient, and the average abnormality coefficient is used to determine whether an asynchronous extraction strategy can be used simultaneously to extract and process the order data.

[0020] A further technical solution is that the extraction management complexity coefficient is determined according to the ratio of the number of the matched e-commerce platforms to a preset proportional factor.

[0021] A further technical solution is that, when the average abnormality coefficient is greater than a preset abnormality coefficient threshold, it is determined that the asynchronous extraction strategy cannot be used simultaneously to extract and process the order data.

[0022] A further technical solution is that the method for determining the extraction and processing strategy of the matching e-commerce platform at different time periods is:

[0023] The matching e-commerce platform adopting the asynchronous extraction strategy is used as the asynchronous extraction platform, and the total number of cross-commodities with the asynchronous extraction platform and the total number of cross-commodities with the strategy variable platform are determined based on the cross-commodities between the products in the strategy variable platform and other matching e-commerce platforms;

[0024] Determine the asynchronous extraction cross coefficient and the strategy variable cross coefficient based on the ratio of the total quantity to the quantity of commodities in the strategy variable platform;

[0025] Determine the preset weight value of the asynchronous extraction platform and the preset weight value of the strategy variable platform, and determine the weight sum of the asynchronous extraction cross coefficient and the strategy variable cross coefficient in combination with the asynchronous extraction cross coefficient and the strategy variable cross coefficient to determine the commodity cross coefficient of the matching e-commerce platform;

[0026] Based on the commodity cross-coefficient and the proportion of commodities whose inventory in the matching e-commerce platform is less than the preset inventory, the extraction and processing strategy of the matching e-commerce platform in different time periods is determined.

[0027] A further technical solution is to determine the extraction and processing strategy of the matching e-commerce platform in different time periods based on the product cross-coefficient and the proportion of the number of products whose inventory in the matching e-commerce platform is less than the preset inventory, which specifically includes:

[0028] When the commodity cross coefficient is greater than a preset cross coefficient threshold, it is determined to use an asynchronous extraction strategy to extract the order data of the matching e-commerce platform in different time periods;

[0029] When the commodity cross-coefficient is not greater than the preset cross-coefficient threshold, it is also necessary to determine whether the proportion of the number of commodities in the matching e-commerce platform in the time period whose inventory is less than the preset inventory is greater than the proportion of the preset number of commodities. If so, determine to use the asynchronous extraction strategy to extract the order data of the matching e-commerce platform in the time period; if not, determine to use the synchronous extraction strategy to extract the order data of the matching e-commerce platform in the time period.

[0030] A further technical solution is that the synchronous extraction strategy is to extract the order data of the matching e-commerce platform after receiving an instruction to obtain the order data of the matching e-commerce platform.

[0031] A further technical solution is that the cross-products are products that are listed on the strategy variable platform and the matching e-commerce platform at the same time.

[0032] In a second aspect, the present invention provides a computer system comprising: a memory and a processor that are communicatively connected, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the above-mentioned order management method based on a machine learning model when running the computer program.

[0033] Other features and advantages will be described in the following description. The objects and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description and drawings.

[0034] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The above and other features and advantages of the present invention will become more apparent by describing in detail exemplary embodiments thereof with reference to the attached drawings.

[0036] Figure 1 is a flow chart of an order management method based on a machine learning model;

[0037] Figure 2 It is a flowchart that determines that the asynchronous extraction strategy cannot be used simultaneously to extract and process order data;

[0038] Figure 3 It is a flow chart of a method for determining a deviation date of extraction duration matching an e-commerce platform;

[0039] Figure 4 It is a flowchart of a method for extracting a determination of a matching e-commerce platform with a variable extraction strategy;

[0040] Figure 5 It is a framework diagram of a computer system. DETAILED DESCRIPTION

[0041] In order 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 in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this specification.

[0042] In the existing technical solutions, an asynchronous extraction strategy is used to extract and process order data for different matching e-commerce platforms. The asynchronous extraction strategy is to extract and process order data for different matching e-commerce platforms in parallel during the software operation, thereby ensuring the real-time extraction and processing of order data. However, as the number of matching e-commerce platforms increases, it will inevitably cause excessive operating pressure on the server.

[0043] In this application, the intersection of the listed products of the matching order platform and other matching order platforms and the inventory of the listed products are utilized to determine the extraction and processing strategy of the order data of different matching order platforms, thereby avoiding the technical problem of excessive operating pressure on the server caused by the use of all asynchronous extraction strategies.

[0044] The proportion of historical operating time of the server load within the preset load rate range is taken as the proportion of high load rate time, and the product of the number of matching e-commerce platforms of the e-commerce enterprise and the proportion of high load rate time is used to determine the strategy matching anomaly coefficient of the asynchronous extraction strategy. When the strategy matching anomaly coefficient of the asynchronous extraction strategy is greater than 10, it is determined that the asynchronous extraction strategy cannot be used at the same time to extract and process order data.

[0045] The number of extraction processing times when the extraction processing time is greater than the preset time threshold, and the average value of the extraction processing time under different extraction processing times in the date are taken as input, and the machine learning model is used to determine the extraction time deviation factor of the date. When the time deviation factor is greater than 0.6, the date is determined to be the extraction time deviation date.

[0046] The date when the historical extraction times that exceed the preset time threshold account for more than 0.6 of the current date is taken as the extraction time deviation date.

[0047] Based on the distribution data, the proportion of the number of deviation dates of the extraction duration of the matching e-commerce platform within the preset time period is determined, and it is used as the proportion of the number of deviation dates. The unit time period in which the number of historical orders is greater than the preset number of orders is used as the order aggregation period. According to the proportion of the number of order aggregation periods within the preset time period and the average proportion of the number of deviation dates, the order distribution clustering coefficient of the matching e-commerce platform is determined. When the order distribution clustering coefficient is greater than 0.5, the matching e-commerce platform is determined to be a matching e-commerce platform with a variable extraction strategy.

[0048] Determine the total number of cross-products with other matching e-commerce platforms. When the ratio of cross-products to listed products in the strategy-variable platform is greater than 0.6, determine to use an asynchronous extraction strategy to extract order data of the matching e-commerce platform in different time periods;

[0049] When the ratio of cross-products to listed products in the strategy-variable platform is not greater than 0.6, it is also necessary to determine whether the proportion of products in the matching e-commerce platform in the time period whose inventory is less than the preset inventory is greater than 0.7. If so, determine to use the asynchronous extraction strategy to extract the order data of the matching e-commerce platform in the time period; if not, use the synchronous extraction strategy to extract the order data of the matching e-commerce platform in the time period.

[0050] Example 1 Figure 1 As shown, the present application provides an order management method based on a machine learning model, which specifically includes:

[0051] S1 obtains the number of matching e-commerce platforms of the e-commerce enterprise, and combines the abnormal data of the server load in the process of extracting and processing orders using the asynchronous extraction strategy. When it is determined that the asynchronous extraction strategy cannot be used at the same time to extract and process the order data, proceed to the next step;

[0052] S2 determines the extraction time deviation date of the matching e-commerce platform using a machine learning model based on the extraction processing time of the matching e-commerce platform under the asynchronous extraction strategy and the preset time threshold;

[0053] S3 determines the distribution data of the extraction duration deviation date of the matching e-commerce platform within a preset period, and based on the distribution data and the historical order data of the matching e-commerce platform, determines the matching e-commerce platform with variable extraction strategy, and uses it as the strategy variable platform;

[0054] S4 determines the intersection of the products in the strategy variable platform and other matching e-commerce platforms with different extraction and processing strategies based on the product listing data in the strategy variable platform, and determines the extraction and processing strategies of the matching e-commerce platforms in different time periods in combination with the inventory data of different products.

[0055] Furthermore, the matching e-commerce platform is an e-commerce platform where the e-commerce enterprise has products on the shelves.

[0056] Specifically, the abnormal data of the server load includes the historical operation time period of the server load within a preset load rate range and the number of load rate abnormality alarms.

[0057] It should be noted that if Figure 2 As shown, it is determined that the asynchronous extraction strategy cannot be used to extract and process order data at the same time, including:

[0058] Based on the abnormal data of the server load, determine the historical operating time ratio of the server load within a preset load rate range, and use it as the high load rate time ratio;

[0059] Obtaining the number of matching e-commerce platforms of the e-commerce enterprise, and determining the extraction management complexity coefficient of the asynchronous extraction strategy according to the number of matching e-commerce platforms;

[0060] The average abnormality coefficient is determined according to the proportion of the high load rate time and the average value of the extraction management complexity coefficient, and the average abnormality coefficient is used to determine whether an asynchronous extraction strategy can be used simultaneously to extract and process the order data.

[0061] Furthermore, the extraction management complexity coefficient is determined according to the ratio of the number of the matched e-commerce platforms to a preset proportional factor.

[0062] It can be understood that when the average abnormality coefficient is greater than the preset abnormality coefficient threshold, it is determined that the asynchronous extraction strategy cannot be used simultaneously to extract and process the order data.

[0063] In another possible embodiment, determining that it is impossible to simultaneously use an asynchronous extraction strategy to extract order data specifically includes:

[0064] Based on the abnormal data of the server load, determine the historical operating time ratio of the server load within a preset load rate range, and use it as the high load rate time ratio;

[0065] Obtain the number of matching e-commerce platforms of the e-commerce enterprise, and determine the strategy matching anomaly coefficient of the asynchronous extraction strategy by multiplying the number of matching e-commerce platforms by the proportion of the high load rate duration;

[0066] The strategy matching anomaly coefficient is used to determine whether an asynchronous extraction strategy can be used simultaneously to extract and process order data.

[0067] Furthermore, when the strategy matching anomaly coefficient does not meet the requirement, it is determined that the asynchronous extraction strategy cannot be used simultaneously to extract and process the order data.

[0068] In another possible embodiment, determining that it is impossible to simultaneously use an asynchronous extraction strategy to extract order data specifically includes:

[0069] Obtaining the number of matching e-commerce platforms of the e-commerce enterprise, and when the number of matching e-commerce platforms of the e-commerce enterprise is less than the preset number of platforms, determining that an asynchronous extraction strategy can be simultaneously used to extract and process order data;

[0070] When the number of matching e-commerce platforms of the e-commerce enterprise is not less than the preset number of platforms:

[0071] Based on the abnormal data of the server load, determine the historical operating time ratio of the server load within the preset load rate range, and use it as the high load rate time ratio. When the high load rate time ratio is less than the preset time ratio threshold, determine that the asynchronous extraction strategy can be used to extract the order data at the same time;

[0072] When the high load rate duration ratio is not less than the preset duration ratio threshold:

[0073] When the high load rate duration ratio does not meet the requirement, it is determined that the asynchronous extraction strategy cannot be used to extract and process the order data at the same time;

[0074] When the high load rate duration ratio meets the requirement:

[0075] Obtaining the number of abnormal load rate alarms of the server, and when the number of abnormal load rate alarms of the server does not meet the requirement, determining that the asynchronous extraction strategy cannot be used to extract and process the order data at the same time;

[0076] When the number of abnormal load rate alarms of the server meets the requirement:

[0077] Based on the high load rate duration ratio of the server and the number of load rate abnormality alarms, the load rate abnormality coefficient of the server is determined. When the load rate abnormality coefficient of the server does not meet the requirements, it is determined that the asynchronous extraction strategy cannot be used to extract and process the order data at the same time;

[0078] When the server's load rate abnormality coefficient meets the requirement:

[0079] The strategy matching anomaly coefficient of the asynchronous extraction strategy is determined by multiplying the number of the matched e-commerce platforms by the load rate anomaly coefficient, and the strategy matching anomaly coefficient is used to determine whether the asynchronous extraction strategy can be used simultaneously to extract and process order data.

[0080] Specifically, Figure 3 As shown, the method for determining the extraction time deviation date of the matching e-commerce platform is:

[0081] Based on the extraction processing time of the matching e-commerce platform under different extraction processing times under the asynchronous extraction strategy, determine the number of extraction processing times in the date whose extraction processing time is greater than the preset time threshold, and use it as the delayed extraction processing times;

[0082] Based on the extraction processing time of the matching e-commerce platform under different extraction processing times under the asynchronous extraction strategy, determine the average value of the extraction processing time under different extraction processing times on the date, and use it as the average extraction processing time;

[0083] The number of delayed extraction processing times and the average extraction processing time in the date are taken as input quantities, and the machine learning model adopted is used to determine the extraction time deviation factor of the date, and the extraction time deviation factor is used to determine whether the date is an extraction time deviation date.

[0084] Furthermore, the machine learning model is constructed using one or more of a supervised learning model, an unsupervised learning model, a reinforcement learning model, and a deep learning model.

[0085] It can be understood that when the extraction duration deviation factor of the date is greater than the preset deviation factor threshold, the date is determined to be an extraction duration deviation date.

[0086] Specifically, the preset time period is determined according to the number of matching e-commerce platforms of the e-commerce enterprise, wherein the more the number of matching e-commerce platforms of the e-commerce enterprise is, the longer the preset time period is.

[0087] It should be noted that the distribution data of the extraction duration deviation dates includes the number of extraction duration deviation dates within the preset period and the date intervals between different extraction duration deviation dates.

[0088] Furthermore, the historical order data of the matching e-commerce platform includes the historical order quantity of the matching e-commerce platform and the historical order quantity in different unit time periods.

[0089] Specifically, Figure 4 As shown, the method for determining the extraction strategy to match the e-commerce platform is as follows:

[0090] Determine the proportion of the number of deviation dates of the extraction duration of the matching e-commerce platform within a preset period based on the distribution data, and use it as the proportion of the number of deviation dates;

[0091] Determine the number of historical orders of the matching e-commerce platform in different unit time periods within a preset time period based on the historical order data of the matching e-commerce platform, and use the unit time period in which the number of historical orders is greater than the preset number of orders as the order aggregation period;

[0092] According to the proportion of the number of order clustering periods within the preset time period and the average proportion of the number of deviation dates, the order distribution clustering coefficient of the matching e-commerce platform is determined, and the order distribution clustering coefficient is used to determine whether the matching e-commerce platform is a matching e-commerce platform with a variable extraction strategy.

[0093] Further, when the order distribution clustering coefficient is greater than a preset clustering coefficient threshold, it is determined that the matching e-commerce platform does not belong to a matching e-commerce platform with a variable extraction strategy.

[0094] It can be understood that when the matching e-commerce platform does not belong to the matching e-commerce platform with a variable extraction strategy, the order data of the matching e-commerce platform is extracted and processed by adopting an asynchronous extraction strategy.

[0095] In another possible embodiment, the method for determining the extraction strategy to be variable and matching the e-commerce platform is:

[0096] Determine the proportion of the number of deviation dates of the extraction duration of the matching e-commerce platform within a preset period based on the distribution data, and use it as the proportion of the number of deviation dates;

[0097] Determine the number of historical orders of the matching e-commerce platform within a preset time period based on the historical order data of the matching e-commerce platform;

[0098] The corrected order quantity of the matching e-commerce platform is determined based on the product of the number of historical orders within a preset period and the proportion of the number of deviation dates, and the corrected order quantity is used to determine whether the matching e-commerce platform is a matching e-commerce platform with a variable extraction strategy.

[0099] Further, when the corrected order quantity of the matching e-commerce platform does not meet the requirement, it is determined that the matching e-commerce platform does not belong to the matching e-commerce platform with a variable extraction strategy.

[0100] Optionally, the extraction strategy may be variably matched to determine the e-commerce platform by:

[0101] S31: determining the number of historical orders of the matching e-commerce platform within a preset period of time based on the historical order data of the matching e-commerce platform, and determining the order distribution clustering coefficient of the matching e-commerce platform based on the number of historical orders in different unit time periods;

[0102] S32 determines the number of extraction duration deviation dates of the matching e-commerce platform within a preset period based on the distribution data, and uses it as the number of deviation dates, and determines the extraction duration abnormality coefficient of the matching e-commerce platform in combination with the interval dates between different extraction duration deviation dates;

[0103] S33 determines the extraction time deviation factor of the matching e-commerce platform according to the order distribution clustering coefficient within the preset time period and the extraction time abnormality coefficient, and uses the extraction time deviation factor to determine whether the matching e-commerce platform is a matching e-commerce platform with a variable extraction strategy.

[0104] It should be noted that the method for determining the extraction and processing strategy of the matching e-commerce platform at different time periods is:

[0105] The matching e-commerce platform adopting the asynchronous extraction strategy is used as the asynchronous extraction platform, and the total number of cross-commodities with the asynchronous extraction platform and the total number of cross-commodities with the strategy variable platform are determined based on the cross-commodities between the products in the strategy variable platform and other matching e-commerce platforms;

[0106] Determine the asynchronous extraction cross coefficient and the strategy variable cross coefficient based on the ratio of the total quantity to the quantity of commodities in the strategy variable platform;

[0107] Determine the preset weight value of the asynchronous extraction platform and the preset weight value of the strategy variable platform, and determine the weight sum of the asynchronous extraction cross coefficient and the strategy variable cross coefficient in combination with the asynchronous extraction cross coefficient and the strategy variable cross coefficient to determine the commodity cross coefficient of the matching e-commerce platform;

[0108] Based on the commodity cross-coefficient and the proportion of commodities whose inventory in the matching e-commerce platform is less than the preset inventory, the extraction and processing strategy of the matching e-commerce platform in different time periods is determined.

[0109] Furthermore, based on the commodity cross-coefficient and the proportion of commodities whose inventory in the matching e-commerce platform is less than the preset inventory, the extraction processing strategy of the matching e-commerce platform in different time periods is determined, which specifically includes:

[0110] Furthermore, based on the commodity cross-coefficient and the proportion of commodities whose inventory in the matching e-commerce platform is less than the preset inventory, the extraction processing strategy of the matching e-commerce platform in different time periods is determined, which specifically includes:

[0111] When the commodity cross coefficient is greater than a preset cross coefficient threshold, it is determined to use an asynchronous extraction strategy to extract the order data of the matching e-commerce platform in different time periods;

[0112] When the commodity cross-coefficient is not greater than the preset cross-coefficient threshold, it is also necessary to determine whether the proportion of the number of commodities in the matching e-commerce platform whose inventory is less than the preset inventory in the time period is greater than the proportion of the preset commodity quantity. If so, the product of the proportion of the number of commodities in the matching e-commerce platform whose inventory is less than the preset inventory and the commodity cross-coefficient is used to determine the duration weight coefficient, and the product of the duration weight coefficient and the preset duration value is used to determine the scheduled acquisition duration. The extraction duration interval of the asynchronous extraction strategy is determined by the scheduled acquisition duration, and the asynchronous extraction strategy is used to extract the order data of the matching e-commerce platform in the time period. If not, the synchronous extraction strategy is used to extract the order data of the matching e-commerce platform in the time period.

[0113] Furthermore, the asynchronous extraction strategy utilizes corresponding time intervals and adopts a multi-threaded approach to extract and process the order data of the matching e-commerce platform.

[0114] It should also be noted that the synchronous extraction strategy is to extract the order data of the matching e-commerce platform after receiving an instruction to obtain the order data of the matching e-commerce platform.

[0115] Furthermore, the cross-products are products that are listed on the strategy variable platform and the matching e-commerce platform at the same time.

[0116] In another possible embodiment, the method for determining the extraction and processing strategy of the matching e-commerce platform in different time periods is:

[0117] Obtaining the proportion of the number of commodities whose inventory quantity of the matching e-commerce platform in the time period is less than the preset inventory quantity, and when the proportion of the number of commodities whose inventory quantity of the matching e-commerce platform in the time period is less than the preset inventory quantity does not meet the requirement, determining to use an asynchronous extraction strategy to extract the order data of the matching e-commerce platform in the time period;

[0118] When the inventory of the matching e-commerce platform during the period is less than the preset inventory and the proportion of the number of goods meets the requirement:

[0119] Based on the inventory of different commodities, an average value of the inventory of different commodities is determined. When the average value of the inventory of different commodities is less than a preset value of the inventory, it is determined to use an asynchronous extraction strategy to extract the order data of the matching e-commerce platform in the period;

[0120] When the average inventory of different products is not less than the preset inventory value:

[0121] The matching e-commerce platform adopting the asynchronous extraction strategy is used as the asynchronous extraction platform, and the total number of cross-commodities with the asynchronous extraction platform and the total number of cross-commodities with the strategy variable platform are determined based on the cross-commodities between the commodities in the strategy variable platform and other matching e-commerce platforms. When the sum of the total number of cross-commodities with the asynchronous extraction platform and the total number of cross-commodities with the strategy variable platform accounts for a greater proportion of the number of cross-commodities in the matching e-commerce platform than the preset cross-commodity number proportion, it is determined to use the asynchronous extraction strategy to extract the order data of the matching e-commerce platform in the period;

[0122] When the sum of the total number of cross-products on the asynchronous extraction platform and the total number of cross-products on the strategy variable platform accounts for no more than the preset cross-product number ratio in the matching e-commerce platform:

[0123] Determine the asynchronous extraction cross coefficient and the strategy variable cross coefficient by the ratio of the total quantity to the quantity of commodities in the strategy variable platform, determine the preset weight value of the asynchronous extraction platform and the preset weight value of the strategy variable platform, and determine the weight sum of the asynchronous extraction cross coefficient and the strategy variable cross coefficient in combination with the asynchronous extraction cross coefficient and the strategy variable cross coefficient, determine the commodity cross coefficient of the matching e-commerce platform, and when the commodity cross coefficient of the matching e-commerce platform does not meet the requirements, determine to use the asynchronous extraction strategy to extract the order data of the matching e-commerce platform in the said time period;

[0124] When the cross-coefficient of the products on the matching e-commerce platform meets the requirements:

[0125] Determine the inventory of cross-products with other matching e-commerce platforms, and determine the product update demand coefficient of the matching e-commerce platform based on the average inventory of different products and the product cross-coefficient, and use the product update demand coefficient to determine the extraction and processing strategy of the matching e-commerce platform in different time periods.

[0126] Furthermore, the product update demand coefficient is used to determine the extraction and processing strategy of the matching e-commerce platform in different time periods, which specifically includes:

[0127] When the commodity update demand coefficient is greater than a preset update coefficient threshold, an asynchronous extraction processing strategy is used to extract the order data of the matching e-commerce platform;

[0128] When the commodity update demand coefficient is not greater than the preset update coefficient threshold, it is also necessary to determine whether the proportion of the number of commodities in the matching e-commerce platform whose inventory is less than the preset inventory in the time period is greater than the proportion of the preset commodity quantity. If so, the product update demand coefficient and the product of the proportion of the number of commodities in the matching e-commerce platform whose inventory is less than the preset inventory are used to determine the duration weight coefficient, and the product of the duration weight coefficient and the preset duration value is used to determine the scheduled acquisition duration. The scheduled acquisition duration is used to determine the extraction duration interval of the asynchronous extraction strategy, and the asynchronous extraction strategy is used to extract the order data of the matching e-commerce platform in the time period. If not, the synchronous extraction strategy is used to extract the order data of the matching e-commerce platform in the time period.

[0129] Embodiment 2 In the second aspect, as Figure 5 As shown, the present invention provides a computer system, comprising: a memory and a processor that are communicatively connected, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the above-mentioned order management method based on a machine learning model when running the computer program.

[0130] Optionally, the above step S31 includes the following contents:

[0131] S311 determines the number of historical orders of the matching e-commerce platform within a preset period of time based on the historical order data of the matching e-commerce platform. When the number of historical orders of the matching e-commerce platform within the preset period of time is less than the preset order number, the matching e-commerce platform is determined to be a matching e-commerce platform with a variable extraction strategy. When the number of historical orders of the matching e-commerce platform within the preset period of time is not less than the preset order number, the process proceeds to step S312.

[0132] S312: when the number of historical orders of the matching e-commerce platform within the preset time period is within the preset order quantity range, it is determined that the matching e-commerce platform does not belong to the matching e-commerce platform with a variable extraction strategy; when the number of historical orders of the matching e-commerce platform within the preset time period is not within the preset order quantity range, the process proceeds to step S313;

[0133] S313 determines the number of historical orders of the matching e-commerce platform in different unit time periods within the preset time period, and takes the unit time period in which the number of historical orders is greater than the preset number of orders as the order aggregation time period. When the number of order aggregation time periods of the matching e-commerce platform within the preset time period accounts for a greater proportion than the number of preset aggregation time periods, it is determined that the matching e-commerce platform does not belong to the matching e-commerce platform with a variable extraction strategy. When the number of order aggregation time periods of the matching e-commerce platform within the preset time period accounts for a less than the number of preset aggregation time periods, the process proceeds to step S314.

[0134] S314 determines the order distribution clustering coefficient of the matching e-commerce platform based on the number of historical orders of the matching e-commerce platform within a preset time period and the number of historical orders in different unit time periods. When the order distribution clustering coefficient of the matching e-commerce platform is greater than the preset clustering coefficient threshold, it is determined that the matching e-commerce platform does not belong to the matching e-commerce platform with a variable extraction strategy. When the order distribution clustering coefficient of the matching e-commerce platform is not greater than the preset clustering coefficient threshold, proceed to step S32.

[0135] Optionally, the above step S32 includes the following contents:

[0136] S321 determines the proportion of the number of extraction duration deviation dates of the matching e-commerce platform within the preset period based on the distribution data, and uses it as the proportion of the number of deviation dates. When the proportion of the number of deviation dates is greater than the preset proportion of the number of deviation dates, it is determined that the matching e-commerce platform does not belong to the matching e-commerce platform with a variable extraction strategy. When the proportion of the number of deviation dates is not greater than the preset proportion of the number of deviation dates, it proceeds to step S322;

[0137] S322: When the deviation date quantity ratio is within the preset deviation date ratio interval, proceed to step S323; when the deviation date quantity ratio is not within the preset deviation date ratio interval, proceed to step S324;

[0138] S323: When the order distribution clustering coefficient of the matching e-commerce platform is within the preset clustering coefficient interval, it is determined that the matching e-commerce platform does not belong to the matching e-commerce platform with a variable extraction strategy. When the order distribution clustering coefficient of the matching e-commerce platform is not within the preset clustering coefficient interval, the process proceeds to step S324.

[0139] S324 determines the extraction time anomaly coefficient of the matching e-commerce platform based on the proportion of the deviation date quantity and the interval date between different extraction time deviation dates. When the extraction time anomaly coefficient of the matching e-commerce platform does not meet the requirements, it is determined that the matching e-commerce platform does not belong to the matching e-commerce platform with a variable extraction strategy. When the extraction time anomaly coefficient of the matching e-commerce platform meets the requirements, proceed to step S33.

[0140] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0141] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0142] The above description is only one or more embodiments of this specification and is not intended to limit this specification. For those skilled in the art, one or more embodiments of this specification may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of this specification shall be included in the scope of the claims of this specification.

Claims

1. An order management method based on a machine learning model, characterized in that: Specifically include: Obtain the number of matching e-commerce platforms of the e-commerce enterprise, and combine the abnormal data of the server load in the process of extracting and processing orders using the asynchronous extraction strategy. When it is determined that the asynchronous extraction strategy cannot be used at the same time to extract and process the order data, proceed to the next step; Based on the extraction processing time of the matching e-commerce platform under the asynchronous extraction strategy and the preset time threshold, a machine learning model is used to determine the extraction time deviation date of the matching e-commerce platform; Determine the distribution data of the extraction duration deviation date of the matching e-commerce platform within a preset period, and determine the matching e-commerce platform with a variable extraction strategy based on the distribution data and the historical order data of the matching e-commerce platform, and use it as the strategy variable platform; Based on the product listing data in the strategy-variable platform, the intersection of the products in the strategy-variable platform and other matching e-commerce platforms with different extraction and processing strategies is determined, and the extraction and processing strategies of the matching e-commerce platforms in different time periods are determined in combination with the inventory data of different products.

2. The order management method based on a machine learning model according to claim 1, characterized in that: The matching e-commerce platform is an e-commerce platform where the e-commerce enterprise has products on the shelves.

3. The order management method based on a machine learning model according to claim 1, characterized in that: The abnormal data of the server load includes the historical operation time period of the server load within a preset load rate interval and the number of load rate abnormality alarms.

4. The order management method based on a machine learning model according to claim 1, characterized in that: It is determined that the asynchronous extraction strategy cannot be used to extract and process order data at the same time, including: Based on the abnormal data of the server load, determine the historical operating time ratio of the server load within a preset load rate range, and use it as the high load rate time ratio; Obtaining the number of matching e-commerce platforms of the e-commerce enterprise, and determining the extraction management complexity coefficient of the asynchronous extraction strategy according to the number of matching e-commerce platforms; The average abnormality coefficient is determined according to the proportion of the high load rate time and the average value of the extraction management complexity coefficient, and the average abnormality coefficient is used to determine whether an asynchronous extraction strategy can be used simultaneously to extract and process the order data.

5. The order management method based on a machine learning model according to claim 4, characterized in that: The extraction management complexity coefficient is determined according to the ratio of the number of the matching e-commerce platforms to a preset proportional factor.

6. The order management method based on a machine learning model according to claim 1, characterized in that: The method for determining the extraction time deviation date of the matching e-commerce platform is: Based on the extraction processing time of the matching e-commerce platform under different extraction processing times under the asynchronous extraction strategy, determine the number of extraction processing times in the date whose extraction processing time is greater than the preset time threshold, and use it as the delayed extraction processing times; Based on the extraction processing time of the matching e-commerce platform under different extraction processing times under the asynchronous extraction strategy, determine the average value of the extraction processing time under different extraction processing times on the date, and use it as the average extraction processing time; The number of delayed extraction processing times and the average extraction processing time in the date are taken as input quantities, and the machine learning model adopted is used to determine the extraction time deviation factor of the date, and the extraction time deviation factor is used to determine whether the date is an extraction time deviation date.

7. The order management method based on a machine learning model according to claim 6, characterized in that: The machine learning model is constructed using one or more of a supervised learning model, an unsupervised learning model, a reinforcement learning model, and a deep learning model.

8. The order management method based on a machine learning model according to claim 1, characterized in that: The method for determining the extraction and processing strategy of the matching e-commerce platform in different time periods is: The matching e-commerce platform adopting the asynchronous extraction strategy is used as the asynchronous extraction platform, and the total number of cross-commodities with the asynchronous extraction platform and the total number of cross-commodities with the strategy variable platform are determined based on the cross-commodities between the products in the strategy variable platform and other matching e-commerce platforms; Determine the asynchronous extraction cross coefficient and the strategy variable cross coefficient based on the ratio of the total quantity to the quantity of commodities in the strategy variable platform; Determine the preset weight value of the asynchronous extraction platform and the preset weight value of the strategy variable platform, and determine the weight sum of the asynchronous extraction cross coefficient and the strategy variable cross coefficient in combination with the asynchronous extraction cross coefficient and the strategy variable cross coefficient to determine the commodity cross coefficient of the matching e-commerce platform; Based on the commodity cross-coefficient and the proportion of commodities whose inventory in the matching e-commerce platform is less than the preset inventory, the extraction and processing strategy of the matching e-commerce platform in different time periods is determined.

9. The order management method based on a machine learning model according to claim 8, characterized in that: Based on the commodity cross coefficient and the proportion of commodities whose inventory in the matching e-commerce platform is less than the preset inventory, the extraction and processing strategy of the matching e-commerce platform in different time periods is determined, which specifically includes: When the commodity cross coefficient is greater than a preset cross coefficient threshold, it is determined to use an asynchronous extraction strategy to extract the order data of the matching e-commerce platform in different time periods; When the commodity cross-coefficient is not greater than the preset cross-coefficient threshold, it is also necessary to determine whether the proportion of the number of commodities in the matching e-commerce platform in the time period whose inventory is less than the preset inventory is greater than the proportion of the preset number of commodities. If so, determine to use the asynchronous extraction strategy to extract the order data of the matching e-commerce platform in the time period; if not, determine to use the synchronous extraction strategy to extract the order data of the matching e-commerce platform in the time period.

10. A computer system comprising: A memory and a processor connected in communication, and a computer program stored in the memory and capable of running on the processor, characterized in that the processor executes an order management method based on a machine learning model as described in any one of claims 1-9 when running the computer program.

Images