A multi-element data-based order information dynamic analysis management system and method

Abstract

The application relates to the technical field of order information management, in particular to an order information dynamic analysis management system and method based on multi-element data. In the system, an order feature analysis module obtains cross-domain feature evaluation coefficients of commodities on sale based on historical order after-sales records of the commodities on sale based on logistics information and social communication information; and obtains order after-sales pressure prediction values of the commodities on sale according to extraction results of the logistics information and the social communication information corresponding to each order of the commodities on sale which have been sold and have not completed transactions. The application not only considers the influence relationship between after-sales order information and to-be-executed order information and the user portrait situation of the to-be-executed order, but also effectively combines the inventory situation of the commodities on sale to realize order management and control, complete dynamic adjustment of an execution risk perception threshold of the to-be-executed order, assist a commodity supplier in completing execution priority decision of the to-be-executed order, and effectively reduce the influence of the inventory commodities on the after-sales return state.

Description

Technical Field

[0001] This invention relates to the field of order information management technology, specifically to a dynamic analysis and management system and method for order information based on multi-source data. Background Technology

[0002] With the explosive growth of e-commerce, supply chain digitalization, and IoT technologies, modern order management systems face the challenge of an exponential expansion of data dimensions. Simultaneously, the volatility of orders over time and the impact of order cancellations significantly affect inventory replenishment cycles and delivery times. Current dynamic order information analysis and management systems based on multi-source data only predict the supply status of inventory goods by integrating data from different order information sources; however, they fail to consider the relationship between after-sales order information and pending order information, as well as the user profiles of pending orders, thus failing to effectively manage orders by combining the inventory status of goods on sale. Summary of the Invention

[0003] The purpose of this invention is to provide a dynamic analysis and management system and method for order information based on multi-source data, so as to solve the problems mentioned in the background art.

[0004] To address the aforementioned technical problems, this invention provides the following technical solution: a method for dynamic analysis and management of order information based on multi-source data, the method comprising the following steps:

[0005] S1. Obtain historical order data for products currently on sale and analyze the order sales trend curve of these products over time;

[0006] S2. Based on the historical order after-sales records of the products on sale, obtain the cross-domain feature evaluation coefficients of the products on sale based on logistics information and social communication information respectively; and based on the extraction results of logistics information and social communication information corresponding to each order of the products on sale that has been sold but not completed, obtain the order after-sales pressure prediction value of the products on sale.

[0007] S3. Obtain the current inventory of goods on sale, the predicted order after-sales pressure of goods on sale, and the predicted order load analysis value of goods on sale.

[0008] S4. Obtain the pending orders for the products on sale and the user information for each pending order, construct the user profile for each pending order, and generate the user anomaly risk assessment coefficient for each pending order; obtain the order transfer fit between the information of each sold but not completed order for the products on sale and each pending order, and obtain the execution risk perception coefficient for each pending order for the products on sale.

[0009] S5. Based on the execution risk perception coefficient of each pending order of the goods on sale, generate the order pending evaluation sequence of the goods on sale; combine the order sales trend curve of the goods on sale over time and the prediction results of the order load analysis value of the current goods on sale, dynamically adjust the execution risk perception threshold of the pending orders, and construct order risk early warning information by combining the order pending evaluation sequence of the goods on sale and the execution risk perception threshold of the pending orders.

[0010] Furthermore, S1 includes:

[0011] S11. Obtain historical order data of products for sale. The historical order data includes each order data statistical time interval and the sum of the number of products for sale corresponding to each order within each order data statistical time interval. The interval lengths corresponding to each order data statistical time interval are the same and are preset values.

[0012] S12. Obtain the data pairs corresponding to the statistical time intervals of each order data in the historical data. The first value in the data pair represents the time point corresponding to the center point within the statistical time interval of the corresponding order data. The second value in the data pair represents the quotient of the sum of the quantities of goods for sale corresponding to each order within the statistical time interval of the corresponding order data divided by the duration of the corresponding interval. Obtain the mapping coordinates of each data pair corresponding to the historical data in the coordinate system of order quantity and time relationship, and connect each adjacent mapping coordinate point in the coordinate system of order quantity and time relationship in chronological order to construct the sales trend curve of sold orders changing over time.

[0013] S13. Based on the preset period duration in the database, segment the sales trend curve of sold orders for the products on sale over time in ascending order of the time interval from the current time, generating a set of sales trend segmentation fragments of sold orders for the products on sale, denoted as {A1, A2, A3, ..., A...}. n A n+1}, where A n This represents the nth segment of the sales trend curve of sold orders for products on sale over time, based on a preset period in the database; the number of segments is denoted as n+1; the result is the set of sales trend curve segments of sold orders for products on sale excluding A1 and A... n+1 The trend fit deviation between each element other than A1 is calculated using the following formula:

[0014] Q n =|K (T,1) -K (T,n) |+μ·P n

[0015] Q n A represents n Trend adaptation deviation between A1 and A1; T represents the time interval between the current time and the nearest extreme point in A1 (the extreme points include the maximum and minimum values ​​in the segments of the corresponding sold order sales trend change curves, and the maximum and minimum values ​​are obtained based on the change trends in the corresponding segments); K (T，1) K represents the slope between the coordinate point in A1 with an interval T from the current time and the starting coordinate point in A1; (T，n) A represents n The coordinates of the starting coordinates at a time interval of T and A n The slope between the initial coordinate points; μ is the preset weighting coefficient; P n Indicates A n The average deviation between the functions corresponding to A1 when the horizontal translation is performed so that the corresponding time interval is the same as the time interval of A1.

[0016]

[0017] F1(t) represents the function of A1 within the corresponding time interval; F n (t) represents A n The function within the corresponding time interval when the horizontal shift is the same as the time interval of A1; TY represents the preset period duration in the database;

[0018] S14. Extract the element preceding the element with the smallest trend fit deviation between the sold order sales trend change segment set and A1, denoted as AX; obtain the symmetrical result of AX about the vertical axis, denoted as AXD; translate AXD so that the minimum value coordinate in the time interval coincides with the coordinate point of the current time in the sold order sales trend change curve of the sold goods over time, and obtain the order sales trend curve of the sold goods over time.

[0019] Furthermore, in the process of obtaining the cross-domain feature evaluation coefficients of the products for sale based on logistics information and social communication information in S2, the number of returned orders in the historical order after-sales records of the products for sale due to logistics reasons and social communication reasons is obtained. The cross-domain feature evaluation coefficient of the products for sale based on logistics information represents the ratio of the number of returned orders due to logistics reasons to the total number of returned orders in the historical order after-sales records of the products for sale; the cross-domain feature evaluation coefficient of the products for sale based on social communication information represents the ratio of the number of returned orders due to social communication reasons to the total number of returned orders in the historical order after-sales records of the products for sale.

[0020] Furthermore, feature extraction is performed on the logistics information and social communication information corresponding to each sold but not yet completed order of goods for sale. The feature extraction result of the logistics information corresponding to the i-th sold but not yet completed order of goods for sale is denoted as Bsi, and the feature extraction result of the social communication information corresponding to the i-th sold but not yet completed order of goods for sale is denoted as Bci. The feature extraction result of the logistics information includes the difference between the actual execution time of each logistics link and the preset execution time reference value of the corresponding logistics link. The feature extraction result of the social communication information includes each preset keyword in the social communication information that belongs to the preset form of the database and the frequency of each preset keyword.

[0021] For returned orders due to logistics reasons in the historical order after-sales records of products on sale, obtain the summary set of differences between the actual execution time of each logistics link and the preset execution time reference value of the corresponding logistics link in the logistics information of each returned order, denoted as BLs; obtain the summary set of each preset keyword in the database preset form and the frequency of occurrence of each preset keyword in the social communication information of each returned order in the historical order after-sales records of products on sale, denoted as Blc;

[0022] The formula for calculating the predicted post-sales pressure of orders for goods on sale in S2 is as follows:

[0023]

[0024] OFP represents the projected post-order pressure for products currently on sale; COU i Rs represents the number of goods available for sale in the i-th order that has been sold but not yet completed; Rc represents the cross-domain feature evaluation coefficient of goods available for sale based on logistics information; ig represents the number of orders for goods available for sale that have been sold but not yet completed; j1 represents the number of logistics links in BLs; r1 represents the number of pre-defined keyword types in BLc; N (BLs，i，j) This represents the percentage of element values ​​in BLs corresponding to the j-th logistics link that are less than or equal to the element values ​​in Bsi corresponding to the j-th logistics link; NV (BLc，r) NV represents the ratio of the frequency of the r-th preset keyword in BLC to the sum of the frequencies of all preset keywords; (BLc，i，r) This represents the quotient of the frequency of the r-th pre-defined keyword in Bci divided by the frequency of the r-th pre-defined keyword in Blc;

[0025] The predicted result of the order load analysis value of the currently sold goods in S3 is equal to the sum of the inventory of the currently sold goods and the predicted value of the order after-sales pressure of the currently sold goods.

[0026] Furthermore, S4 includes:

[0027] S41. Obtain pending orders for goods on sale and user information for each pending order, and construct a user profile for each pending order. The user profile includes historical order rollback rate, order receiving location, and the number of goods on sale in the corresponding pending order.

[0028] S42. Use the historical order rollback rate in the user profile corresponding to the m-th pending order as the user anomaly risk assessment coefficient for the corresponding pending order, denoted as US. m ;

[0029] S43. Obtain the order transfer fit between each sold but not yet completed order of the goods on sale and each pending order. The order transfer fit between the i-th sold but not yet completed order of the goods on sale and the m-th pending order is denoted as FD. (i，m) ,

[0030]

[0031] LC (i，m) LD represents the length of the overlapping path between the planned route from the shipping location of the product in sale to the receiving location of the corresponding order for the m-th pending order and the planned route from the shipping location of the product in sale to the receiving location of the i-th sold but not yet completed order for the product in sale; m This represents the planned route length from the shipping location of the goods currently for sale to the receiving location of the corresponding order for the m-th pending order; LY i The planned route length from the shipping location of the product to the receiving location of the i-th order that has been sold but not yet completed; ε represents the conversion coefficient; when COU i ≥COUD m When the condition is met, ε = 1; otherwise, ε = 0. m COU represents the number of available products in the user profile corresponding to the m-th order to be executed; i This represents the number of items available for sale in the i-th order that has been sold but not yet completed; min{} represents the operation to find the minimum value;

[0032] S44. Obtain the execution risk perception coefficient of the m-th pending order of the goods for sale, denoted as G. m ,

[0033] G m =(1-US) m )·max{FD (i,m) |i∈[0,i1],m∈[0,m1]}

[0034] Where m1 represents the number of pending orders for goods currently on sale; max{} represents the operation to find the maximum value.

[0035] Furthermore, in S5, the order pending evaluation sequence of the goods for sale is the sorting result of each pending order of the goods for sale in descending order of execution risk perception coefficient;

[0036] The calculation formula for dynamically adjusting the execution risk perception threshold of pending orders in S5 is as follows:

[0037] ED = EY·LFD / LFV

[0038] Wherein, ED represents the dynamically adjusted execution risk perception threshold for orders to be executed; EY represents the preset execution risk perception reference threshold in the database; and LFD represents the order sales trend curve of the products on sale over time within the time interval [T]. now TS+T now The integral value within ]; T now This indicates the current time point; TS indicates the preset statistical duration; LFV indicates the predicted result of the order load analysis value of the currently available products;

[0039] The order risk warning information includes the order evaluation sequence of goods for sale and the execution risk perception threshold of the orders to be executed, and marks the elements in the order risk warning information whose execution risk perception coefficient is less than or equal to the execution risk perception threshold of the orders to be executed within the order evaluation sequence of goods for sale.

[0040] A dynamic analysis and management system for order information based on multi-source data, the system comprising the following modules:

[0041] The sales trend analysis module acquires historical order data of products on sale and analyzes the order sales trend curve of products on sale over time.

[0042] The order feature analysis module obtains cross-domain feature evaluation coefficients for the products based on historical order after-sales records of the products on sale, using logistics information and social communication information respectively; and obtains the order after-sales pressure prediction value of the products on sale based on the extraction results of logistics information and social communication information corresponding to each sold but not completed order of the products on sale.

[0043] The order load analysis module obtains the current inventory of goods on sale and the predicted value of the order after-sales pressure of goods on sale, and predicts the order load analysis value of the goods on sale at present.

[0044] The execution risk perception module obtains the pending orders of the goods for sale and the user information of each pending order, constructs the user profile corresponding to each pending order, and generates the user abnormal risk assessment coefficient for each pending order; it also obtains the order transfer fit between the information of each sold but not completed order of the goods for sale and each pending order, and obtains the execution risk perception coefficient for each pending order of the goods for sale.

[0045] The order risk warning module generates an order execution assessment sequence for each pending order of the goods on sale based on the execution risk perception coefficient of each pending order. It dynamically adjusts the execution risk perception threshold of the pending orders by combining the order sales trend curve of the goods on sale over time and the prediction results of the order load analysis value of the current goods on sale, and constructs order risk warning information by combining the order execution assessment sequence of the goods on sale and the execution risk perception threshold of the pending orders.

[0046] Furthermore, the order risk warning module includes an execution evaluation sequence generation unit, a risk perception threshold dynamic adjustment unit, and a warning information construction unit.

[0047] The execution evaluation sequence generation unit generates an execution evaluation sequence for orders of goods for sale based on the execution risk perception coefficient of each order to be executed for the goods for sale.

[0048] The risk perception threshold dynamic adjustment unit combines the order sales trend curve of the products on sale over time with the prediction results of the order load analysis value of the current products on sale to dynamically adjust the execution risk perception threshold of the orders to be executed.

[0049] The early warning information construction unit combines the order pending evaluation sequence of goods on sale and the execution risk perception threshold of pending orders to construct order risk early warning information.

[0050] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0051] (1) This invention analyzes the trend adaptation deviation between different sales trend curve segments of sold orders by analyzing the periodic fluctuations of order sales in historical data, thereby enabling the screening of order sales trend curve segments after the current time. By adjusting and splicing the screening results, the accuracy of the order sales trend based on the current time is ensured in the obtained order sales trend curve of the sold goods changing over time; providing data support for the construction of order risk warning information in the subsequent process.

[0052] (2) This invention combines the logistics information of orders with the social communication information during the order execution process to accurately predict the order after-sales pressure of goods on sale, and combines the current inventory status of goods to predict the number of goods available in subsequent orders (the order load analysis value of goods currently on sale), providing a basis for the subsequent construction of order risk warning information;

[0053] (3) This invention not only considers the influence relationship between after-sales order information and pending order information and the user profile of pending orders, but also effectively combines the inventory status of goods on sale to achieve order control, completes the dynamic adjustment of the execution risk perception threshold of pending orders, assists the goods supplier in making execution priority decisions for pending orders, improves the execution efficiency of order goods, and effectively reduces the impact of after-sales return status on inventory goods. Attached Figure Description

[0054] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0055] Figure 1 This is a schematic diagram of the structure of an order information dynamic analysis and management system based on multi-source data according to the present invention;

[0056] Figure 2 This is a flowchart illustrating a dynamic analysis and management method for order information based on multi-source data, according to the present invention. Detailed Implementation

[0057] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0058] Please see Figure 1 The present invention provides a technical solution: a dynamic analysis and management system for order information based on multi-source data, the system comprising the following modules:

[0059] The sales trend analysis module acquires historical order data of products on sale and analyzes the order sales trend curve of products on sale over time.

[0060] The order feature analysis module obtains cross-domain feature evaluation coefficients for the products based on historical order after-sales records of the products on sale, using logistics information and social communication information respectively; and obtains the order after-sales pressure prediction value of the products on sale based on the extraction results of logistics information and social communication information corresponding to each sold but not completed order of the products on sale.

[0061] The order load analysis module obtains the current inventory of goods on sale and the predicted value of the order after-sales pressure of goods on sale, and predicts the order load analysis value of the goods on sale at present.

[0062] The execution risk perception module obtains the pending orders of the goods for sale and the user information of each pending order, constructs the user profile corresponding to each pending order, and generates the user abnormal risk assessment coefficient for each pending order; it also obtains the order transfer fit between the information of each sold but not completed order of the goods for sale and each pending order, and obtains the execution risk perception coefficient for each pending order of the goods for sale.

[0063] The order risk warning module includes an execution evaluation sequence generation unit, a risk perception threshold dynamic adjustment unit, and a warning information construction unit.

[0064] The execution evaluation sequence generation unit generates an execution evaluation sequence for orders of goods for sale based on the execution risk perception coefficient of each order to be executed for the goods for sale.

[0065] The risk perception threshold dynamic adjustment unit combines the order sales trend curve of the products on sale over time with the prediction results of the order load analysis value of the current products on sale to dynamically adjust the execution risk perception threshold of the orders to be executed.

[0066] The early warning information construction unit combines the order pending evaluation sequence of goods on sale and the execution risk perception threshold of pending orders to construct order risk early warning information.

[0067] like Figure 2 As shown, a method for dynamic analysis and management of order information based on multi-source data is described, the method comprising the following steps:

[0068] S1. Obtain historical order data for products currently on sale and analyze the order sales trend curve of these products over time;

[0069] S1 includes:

[0070] S11. Obtain historical order data of products for sale. The historical order data includes each order data statistical time interval and the sum of the number of products for sale corresponding to each order within each order data statistical time interval. The interval lengths corresponding to each order data statistical time interval are the same and are preset values.

[0071] S12. Obtain the data pairs corresponding to the statistical time intervals of each order data in the historical data. The first value in the data pair represents the time point corresponding to the center point within the statistical time interval of the corresponding order data. The second value in the data pair represents the quotient of the sum of the quantities of goods for sale corresponding to each order within the statistical time interval of the corresponding order data divided by the duration of the corresponding interval. Obtain the mapping coordinates of each data pair corresponding to the historical data in the coordinate system of order quantity and time relationship, and connect each adjacent mapping coordinate point in the coordinate system of order quantity and time relationship in chronological order to construct the sales trend curve of sold orders changing over time.

[0072] S13. Based on the preset period duration in the database, segment the sales trend curve of sold orders for the products on sale over time in ascending order of the time interval from the current time, generating a set of sales trend segmentation fragments of sold orders for the products on sale, denoted as {A1, A2, A3, ..., A...}. n A n+1}, where A n This represents the nth segment of the sales trend curve of sold orders for products on sale over time, based on a preset period in the database; the number of segments is denoted as n+1; the result is the set of sales trend curve segments of sold orders for products on sale excluding A1 and A... n+1 The trend fit deviation between each element other than A1 is calculated using the following formula:

[0073] Q n =|K (T,1) -K (T,n) |+μ·P n

[0074] Q n A represents n Trend adaptation deviation between A1 and A1; T represents the time interval between the current time and the nearest extreme point in A1 (the extreme points include the maximum and minimum values ​​in the segments of the corresponding sold order sales trend change curves, and the maximum and minimum values ​​are obtained based on the change trends in the corresponding segments); K (T，1)K represents the slope between the coordinate point in A1 with an interval T from the current time and the starting coordinate point in A1; (T，n) A represents n The coordinates of the starting coordinates at a time interval of T and A n The slope between the initial coordinate points; μ is the preset weighting coefficient; P n Indicates A n The average deviation between the functions corresponding to A1 when the horizontal translation is performed so that the corresponding time interval is the same as the time interval of A1.

[0075]

[0076] F1(t) represents the function of A1 within the corresponding time interval; F n (t) represents A n The function within the corresponding time interval when the horizontal shift is the same as the time interval of A1; TY represents the preset period duration in the database;

[0077] In this embodiment, the selection of the extreme point closest to the current time in A1 is obtained by judging the changing trend in the corresponding segment. The extreme point includes the maximum value coordinate point and the minimum value coordinate point. The function value of the maximum value coordinate point is the largest compared with the function value of its neighboring points, and the intersection of the continuous function that is first increasing and then decreasing is the maximum value coordinate point. The function value of the minimum value coordinate point is the smallest compared with the function value of its neighboring points, and the intersection of the continuous function that is first decreasing and then increasing is the minimum value coordinate point.

[0078] S14. Extract the element preceding the element with the smallest trend fit deviation between the sold order sales trend change segment set and A1, denoted as AX; obtain the symmetrical result of AX about the vertical axis, denoted as AXD; translate AXD so that the minimum value coordinate in the time interval coincides with the coordinate point of the current time in the sold order sales trend change curve of the sold goods over time, and obtain the order sales trend curve of the sold goods over time.

[0079] S2. Based on the historical order after-sales records of the products on sale, obtain the cross-domain feature evaluation coefficients of the products on sale based on logistics information and social communication information respectively; and based on the extraction results of logistics information and social communication information corresponding to each order of the products on sale that has been sold but not completed, obtain the order after-sales pressure prediction value of the products on sale.

[0080] In step S2, during the process of obtaining the cross-domain feature evaluation coefficients of the products for sale based on logistics information and social communication information, the number of returned orders in the historical order after-sales records of the products for sale due to logistics reasons and social communication reasons is obtained. The cross-domain feature evaluation coefficient of the products for sale based on logistics information represents the ratio of the number of returned orders due to logistics reasons to the total number of returned orders in the historical order after-sales records of the products for sale; the cross-domain feature evaluation coefficient of the products for sale based on social communication information represents the ratio of the number of returned orders due to social communication reasons to the total number of returned orders in the historical order after-sales records of the products for sale.

[0081] For each sold but not yet completed order of goods in sale, feature extraction is performed on the logistics information and social communication information corresponding to them. The feature extraction result of the logistics information corresponding to the i-th sold but not yet completed order of goods in sale is denoted as Bsi, and the feature extraction result of the social communication information corresponding to the i-th sold but not yet completed order of goods in sale is denoted as Bci. The feature extraction result of the logistics information includes the difference between the actual execution time of each logistics link and the preset execution time reference value of the corresponding logistics link. The feature extraction result of the social communication information includes each preset keyword in the social communication information that belongs to the preset form of the database and the frequency of each preset keyword.

[0082] For returned orders due to logistics reasons in the historical order after-sales records of products on sale, obtain the summary set of differences between the actual execution time of each logistics link and the preset execution time reference value of the corresponding logistics link in the logistics information of each returned order, denoted as BLs; obtain the summary set of each preset keyword in the database preset form and the frequency of occurrence of each preset keyword in the social communication information of each returned order in the historical order after-sales records of products on sale, denoted as Blc;

[0083] The formula for calculating the predicted post-sales pressure of orders for goods on sale in S2 is as follows:

[0084]

[0085] OFP represents the projected post-order pressure for products currently on sale; COU i Rs represents the number of goods available for sale in the i-th order that has been sold but not yet completed; Rc represents the cross-domain feature evaluation coefficient of goods available for sale based on logistics information; ig represents the number of orders for goods available for sale that have been sold but not yet completed; j1 represents the number of logistics links in BLs; r1 represents the number of pre-defined keyword types in BLc; N (BLs，i，j)This represents the percentage of element values ​​in BLs corresponding to the j-th logistics link that are less than or equal to the element values ​​in Bsi corresponding to the j-th logistics link; NV (BLc，r) NV represents the ratio of the frequency of the r-th preset keyword in BLC to the sum of the frequencies of all preset keywords; (BLc，i，r) This represents the quotient of the frequency of the r-th pre-defined keyword in Bci divided by the frequency of the r-th pre-defined keyword in Blc;

[0086] S3. Obtain the current inventory of goods on sale and the predicted value of order after-sales pressure for goods on sale, and predict the order load analysis value of goods on sale. The predicted result of the order load analysis value of goods on sale in S3 is equal to the sum of the current inventory of goods on sale and the predicted value of order after-sales pressure for goods on sale.

[0087] In this embodiment, the order cancellation situation during the after-sales process of the goods on sale is taken into account. Therefore, the actual available quantity of goods on sale by the supplier is greater than the current inventory of goods on sale, specifically the sum of the quantity of cancelled goods on sale and the actual inventory. In this embodiment, the predicted value of the order after-sales pressure of goods on sale is the predicted quantity of cancelled goods on sale.

[0088] S4. Obtain the pending orders for the products on sale and the user information for each pending order, construct the user profile for each pending order, and generate the user anomaly risk assessment coefficient for each pending order; obtain the order transfer fit between the information of each sold but not completed order for the products on sale and each pending order, and obtain the execution risk perception coefficient for each pending order for the products on sale.

[0089] S4 includes:

[0090] S41. Obtain pending orders for goods on sale and user information for each pending order, and construct a user profile for each pending order. The user profile includes historical order rollback rate, order receiving location, and the number of goods on sale in the corresponding pending order.

[0091] S42. Use the historical order rollback rate in the user profile corresponding to the m-th pending order as the user anomaly risk assessment coefficient for the corresponding pending order, denoted as US. m ;

[0092] S43. Obtain the order transfer fit between each sold but not yet completed order of the goods on sale and each pending order. The order transfer fit between the i-th sold but not yet completed order of the goods on sale and the m-th pending order is denoted as FD. (i，m) ,

[0093]

[0094] LC (i，m) LD represents the length of the overlapping path between the planned route from the shipping location of the product in sale to the receiving location of the corresponding order for the m-th pending order and the planned route from the shipping location of the product in sale to the receiving location of the i-th sold but not yet completed order for the product in sale; m This represents the planned route length from the shipping location of the goods currently for sale to the receiving location of the corresponding order for the m-th pending order; LY i σ represents the planned route length from the shipping location of the product to the receiving location of the i-th order that has been sold but not yet completed; σ represents the conversion coefficient; when COU i ≥COUD m When σ = 1, then σ = 1; otherwise, ε = 0. m COU represents the number of available products in the user profile corresponding to the m-th order to be executed; i This represents the number of items available for sale in the i-th order that has been sold but not yet completed; min{} represents the operation to find the minimum value;

[0095] S44. Obtain the execution risk perception coefficient of the m-th pending order of the goods for sale, denoted as G. m ,

[0096] G m =(1-US) m )·max{FD (i,m) |i∈[0,i1],m∈[0,m1]}

[0097] Where m1 represents the number of pending orders for goods currently on sale; max{} represents the operation to find the maximum value.

[0098] S5. Based on the execution risk perception coefficient of each pending order of the goods on sale, generate the order pending evaluation sequence of the goods on sale; combine the order sales trend curve of the goods on sale over time and the prediction results of the order load analysis value of the current goods on sale, dynamically adjust the execution risk perception threshold of the pending orders, and construct order risk early warning information by combining the order pending evaluation sequence of the goods on sale and the execution risk perception threshold of the pending orders.

[0099] The order evaluation sequence for goods on sale in S5 is the sorting result of each order to be executed for goods on sale in descending order of execution risk perception coefficient;

[0100] The calculation formula for dynamically adjusting the execution risk perception threshold of pending orders in S5 is as follows:

[0101] ED = EY·LFD / LFV

[0102] Wherein, ED represents the dynamically adjusted execution risk perception threshold for orders to be executed; EY represents the preset execution risk perception reference threshold in the database; and LFD represents the order sales trend curve of the products on sale over time within the time interval [T]. now TS+T now The integral value within ]; T now This indicates the current time point; TS indicates the preset statistical duration; LFV indicates the predicted result of the order load analysis value of the currently available products;

[0103] The order risk warning information includes the order evaluation sequence of goods for sale and the execution risk perception threshold of the orders to be executed, and marks the elements in the order risk warning information whose execution risk perception coefficient is less than or equal to the execution risk perception threshold of the orders to be executed within the order evaluation sequence of goods for sale.

[0104] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0105] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for dynamic analysis and management of order information based on multi-source data, characterized in that, The method includes the following steps: S1. Obtain historical order data for products currently on sale and analyze the order sales trend curve of these products over time; S2. Based on the historical order after-sales records of the products on sale, obtain the cross-domain feature evaluation coefficients of the products on sale based on logistics information and social communication information respectively; and based on the extraction results of logistics information and social communication information corresponding to each order of the products on sale that has been sold but not completed, obtain the order after-sales pressure prediction value of the products on sale. S3. Obtain the current inventory of goods on sale, the predicted order after-sales pressure of goods on sale, and the predicted order load analysis value of goods on sale. S4. Obtain the pending orders for the products on sale and the user information for each pending order, construct the user profile for each pending order, and generate the user anomaly risk assessment coefficient for each pending order; obtain the order transfer fit between the information of each sold but not completed order for the products on sale and each pending order, and obtain the execution risk perception coefficient for each pending order for the products on sale. S5. Based on the execution risk perception coefficient of each pending order of the goods on sale, generate the order pending evaluation sequence of the goods on sale; combine the order sales trend curve of the goods on sale over time and the prediction results of the order load analysis value of the current goods on sale, dynamically adjust the execution risk perception threshold of the pending orders, and construct order risk early warning information by combining the order pending evaluation sequence of the goods on sale and the execution risk perception threshold of the pending orders. For each sold but not yet completed order of goods for sale, feature extraction is performed on the corresponding logistics and social communication information. i The logistics information feature extraction results corresponding to each order are denoted as follows: Bsi The first day after the sale of goods that have been sold but not yet completed. i The social communication information feature extraction results corresponding to each order are denoted as follows: Bci The feature extraction results of the logistics information include the difference between the actual execution time of each logistics link and the preset execution time reference value of the corresponding logistics link; the feature extraction results of the social communication information include each preset keyword in the social communication information that belongs to the preset form of the database and the frequency of each preset keyword. The summary set of differences between the actual execution time of each logistics step and the preset execution time reference value of the corresponding logistics step in the after-sales records of historical orders of goods for sale due to logistics reasons is denoted as: BLs ; Retrieve the historical order after-sales records of products currently on sale, specifically the returned orders due to social communication reasons. For each returned order, collect the social communication information containing each pre-defined keyword from a pre-defined form in the database, along with the frequency of each keyword. This collection is denoted as: BLc ; The formula for calculating the predicted post-sales pressure of orders for goods on sale in S2 is as follows: ； in, OFP This represents the projected post-sales pressure of orders for products currently on sale. COU i This indicates the number of items sold but not yet completed in the transaction process. i The number of items available for sale within each order; Rs This represents the cross-domain characteristic evaluation coefficient of products for sale based on logistics information; Rc This represents the cross-domain feature evaluation coefficient of products for sale based on social communication information; ig This indicates the number of orders for goods that have been sold but not yet completed. j1 express BLs The number of logistics links in the process; r1 express BLc The number of preset keyword types in the data; N （BLs，i，j） express BLs The Middle j Within each element value corresponding to each logistics link, less than or equal to Bsi The Middle j The percentage of element values ​​corresponding to each logistics link; NV （BLc，r） express BLc The Middle r The ratio of the frequency of occurrence of a certain preset keyword to the sum of the frequencies of occurrence of all preset keywords; NV （BLc，i，r） express Bci The Middle r The frequency of occurrence of a preset keyword divided by BLc The Middle r The frequency of occurrence of pre-set keywords in the business; The predicted result of the order load analysis value of the currently sold goods in S3 is equal to the sum of the inventory of the currently sold goods and the predicted value of the order after-sales pressure of the currently sold goods. S4 includes: S41. Obtain pending orders for goods on sale and user information for each pending order, and construct a user profile for each pending order. The user profile includes historical order rollback rate, order receiving location, and the number of goods on sale in the corresponding pending order. S42, the first m The historical order rollback rate in the user profile corresponding to each pending order is used as the user anomaly risk assessment coefficient for that pending order, denoted as . US m ; S43. Obtain the order transfer fit between each sold but incomplete transaction of the goods for sale and each pending order, and determine the order transfer fit between each sold but incomplete transaction of the goods for sale. i The order and the first m The order transfer fit between pending orders is denoted as: FD （i，m） , ； LC （i，m） This indicates the shipping location of the goods for sale up to the specified number. m The planned route from the order receiving location to the shipment location of the pending orders, and the route from the shipment location of the goods for sale to the first sold but not yet completed transaction of the goods for sale. i The length of overlapping paths in the planned routes for each order receiving location; LD m This indicates the shipping location of the goods for sale up to the specified number. m The planned route length for the receiving location of each pending order; LY i This indicates the shipping location of the product for sale up to the number of sold but not yet completed transactions. i The planned route length for each order receiving location; ℇ Indicates the conversion factor; when COU i ≥COUD m When ℇ = 1, then ℇ = 1; otherwise, ℇ = 0. COUD m Indicates the first m The number of products available for sale in the user profile corresponding to each pending order; COU i This indicates the number of items sold but not yet completed in the transaction process. i The number of items available for sale in each order; min{} This represents the operation of finding the minimum value; S44, Obtain the first item of the products for sale m The perceived risk coefficient for the execution of each pending order is denoted as... G m , ； in, m1 This indicates the number of pending orders for products currently on sale. max{} This represents the operation of finding the maximum value.

2. The method for dynamic analysis and management of order information based on multi-source data according to claim 1, characterized in that: S1 includes: S11. Obtain historical order data of products for sale. The historical order data includes each order data statistical time interval and the sum of the number of products for sale corresponding to each order within each order data statistical time interval. The interval lengths corresponding to each order data statistical time interval are the same and are preset values. S12. Obtain the data pairs corresponding to the statistical time intervals of each order data in the historical data. The first value in the data pair represents the time point corresponding to the center point within the statistical time interval of the corresponding order data. The second value in the data pair represents the quotient of the sum of the quantities of goods for sale corresponding to each order within the statistical time interval of the corresponding order data divided by the duration of the corresponding interval. Obtain the mapping coordinates of each data pair corresponding to the historical data in the coordinate system of order quantity and time relationship, and connect each adjacent mapping coordinate point in the coordinate system of order quantity and time relationship in chronological order to construct the sales trend curve of sold orders changing over time. S13. Based on the preset period duration in the database, segment the sales trend curve of sold orders for the products on sale over time in ascending order of the time interval from the current time, generating a set of sales trend segmentation data for sold orders of the products on sale, denoted as { A 1 ，A 2 ，A 3 ，...，A n ，A n+1 },in, A n This represents the first digit of the sales trend curve of sold orders for products on sale over time, based on a preset period in the database. n The number of segments is recorded as follows: The number of segments representing the sales trend curve of sold orders for products on sale over time, based on a preset period in the database. n+1 ; obtain fragments of sales trend changes in sold orders for products currently on sale, excluding A 1 and A n+1 The other elements are respectively with A 1 The trend adaptation deviation is calculated using the following formula: ； Q n express A n and A 1 Trend mismatch between them; T express A 1 The time interval between the time point in the coordinates of the nearest extreme point to the current time; K （T，1） express A 1 The coordinates of the point with a time interval T from the current time and A 1 The slope between the initial coordinate points; K （T，n） express A n The time interval between the starting coordinates at the mid-distance point is T coordinates and A n The slope between the initial coordinate points; μ Preset weighting coefficients; P n Indicates to A n Perform a horizontal translation so that its corresponding time interval is... A 1 When the time intervals are the same, the average deviation between the functions corresponding to the two are the same. ； F 1 （t） express A 1 The function within the corresponding time interval; F n （t） express A n Horizontally shift to the corresponding time interval and A 1 When the time intervals are the same, the functions within the corresponding time intervals; TY This indicates the preset period duration in the database; S14. Extract and aggregate segments of sales trend changes from sold orders. A 1 The element preceding the element with the smallest trend fit deviation is denoted as . AX ; obtain AX The symmetric result about the ordinate is denoted as AXD ;right AXD The coordinates of the minimum value in the time interval are shifted to coincide with the coordinates of the current time in the sales trend curve of the sold orders of the goods over time, thus obtaining the sales trend curve of the orders of the goods over time.

3. The method for dynamic analysis and management of order information based on multi-source data according to claim 1, characterized in that: In step S2, during the process of obtaining the cross-domain feature evaluation coefficients of the products for sale based on logistics information and social communication information, the number of returned orders in the historical order after-sales records of the products for sale due to logistics reasons and social communication reasons is obtained. The cross-domain feature evaluation coefficient of the products for sale based on logistics information represents the ratio of the number of returned orders due to logistics reasons to the total number of returned orders in the historical order after-sales records of the products for sale; the cross-domain feature evaluation coefficient of the products for sale based on social communication information represents the ratio of the number of returned orders due to social communication reasons to the total number of returned orders in the historical order after-sales records of the products for sale.

4. The method for dynamic analysis and management of order information based on multi-source data according to claim 1, characterized in that: The order evaluation sequence for goods on sale in S5 is the sorting result of each order to be executed for goods on sale in descending order of execution risk perception coefficient; The calculation formula for dynamically adjusting the execution risk perception threshold of pending orders in S5 is as follows: ； in, ED This represents the dynamically adjusted threshold for perceived execution risk of pending orders. EY This represents the preset execution risk perception reference threshold in the database; LFD The order sales trend curve representing the changes in available products over time is located within the time interval []. T now , TS + T now The integral value within ]; T now Indicates the current time point; TS Indicates the preset statistical duration; LFV This represents the predicted order load analysis value of currently available products. The order risk warning information includes the order evaluation sequence of goods for sale and the execution risk perception threshold of the orders to be executed, and marks the elements in the order risk warning information whose execution risk perception coefficient is less than or equal to the execution risk perception threshold of the orders to be executed within the order evaluation sequence of goods for sale.

5. A dynamic analysis and management system for order information based on multi-source data, wherein the dynamic analysis and management system for order information is applied to the dynamic analysis and management method for order information based on multi-source data as described in claim 1, characterized in that, The system includes the following modules: The sales trend analysis module acquires historical order data of products on sale and analyzes the order sales trend curve of products on sale over time. The order feature analysis module obtains cross-domain feature evaluation coefficients for the products on sale based on historical order after-sales records of the products on sale, respectively, based on logistics information and social communication information. Based on the extraction results of logistics information and social communication information corresponding to each order of sold but not completed goods, the post-sales pressure of the goods is predicted. The order load analysis module obtains the current inventory of goods on sale and the predicted value of the order after-sales pressure of goods on sale, and predicts the order load analysis value of the goods on sale at present. The execution risk perception module obtains the pending orders of the goods for sale and the user information of each pending order, constructs the user profile corresponding to each pending order, and generates the user abnormal risk assessment coefficient for each pending order; it also obtains the order transfer fit between the information of each sold but not completed order of the goods for sale and each pending order, and obtains the execution risk perception coefficient for each pending order of the goods for sale. The order risk warning module generates an order execution assessment sequence for each pending order of the goods on sale based on the execution risk perception coefficient of each pending order. It dynamically adjusts the execution risk perception threshold of the pending orders by combining the order sales trend curve of the goods on sale over time and the prediction results of the order load analysis value of the current goods on sale, and constructs order risk warning information by combining the order execution assessment sequence of the goods on sale and the execution risk perception threshold of the pending orders.

6. The order information dynamic analysis and management system based on multi-source data according to claim 5, characterized in that: The order risk warning module includes an execution evaluation sequence generation unit, a risk perception threshold dynamic adjustment unit, and a warning information construction unit. The execution evaluation sequence generation unit generates an execution evaluation sequence for orders of goods for sale based on the execution risk perception coefficient of each order to be executed for the goods for sale. The risk perception threshold dynamic adjustment unit combines the order sales trend curve of the products on sale over time with the prediction results of the order load analysis value of the current products on sale to dynamically adjust the execution risk perception threshold of the orders to be executed. The early warning information construction unit combines the order pending evaluation sequence of goods on sale and the execution risk perception threshold of pending orders to construct order risk early warning information.

Images