Cross-border e-commerce supply chain data collaborative pushing method and platform based on graph calculation

Abstract

The invention discloses a graph calculation-based cross-border e-commerce supply chain data collaborative pushing method and platform, and relates to the field of data pushing, and the method comprises the steps: obtaining historical supply data of a plurality of supply nodes in a supply chain in a recent historical time range, and obtaining a historical supply data chain set; constructing a supply node graph according to the historical supply data chain set, updating the supply node graph in real time, and constructing the supply node graph according to supply data of adjacent supply nodes; obtaining product characteristics of a current target product, and obtaining multiple pieces of supply state information sent by multiple supply nodes; and carrying out supply data prediction according to the product features and the plurality of pieces of supply state information to obtain supply data of a plurality of nodes, carrying out fusion processing to obtain collaborative supply data, and pushing the collaborative supply data. According to the collaborative pushing method and platform for the supply chain data, the problem that logistics pushing information is inaccurate due to the fact that the number of supply chain nodes is large is solved.

Description

TECHNICAL FIELD

[0001] The application relates to the field of data pushing, in particular to a cross-border e-commerce supply chain data collaborative pushing method and platform based on graph computing. BACKGROUND

[0002] Supply chain data collaborative pushing refers to realizing information circulation and cooperation between supply chain links through data sharing and real-time pushing technology, so as to improve overall operation efficiency and response speed.

[0003] The existing supply chain data pushing technology relies on manual operation, and has poor data pushing flexibility. Due to the large number of supply chain nodes, the supply chain data pushing accuracy is poor. SUMMARY

[0004] The application provides a cross-border e-commerce supply chain data collaborative pushing method and platform based on graph computing, which solves the problem of inaccurate supply chain data pushing caused by the large number of supply chain nodes in the prior art, and has a high error rate of pushed information.

[0005] In view of the above problems, the application provides a cross-border e-commerce supply chain data collaborative pushing method and platform based on graph computing.

[0006] In a first aspect, the application provides a cross-border e-commerce supply chain data collaborative pushing method based on graph computing, which comprises the following steps: acquiring a supply chain of a target product in cross-border e-commerce, and acquiring historical supply data of a plurality of supply nodes in the supply chain within a recent historical time range to obtain a historical supply data chain set; constructing a supply node graph according to the historical supply data chain set, wherein the supply node graph comprises nodes and edges, and is updated in real time, a supply stability coefficient is analyzed according to supply data of adjacent supply nodes, the scale of the nodes is configured, and the construction of the supply node graph is performed; acquiring product features of a current target product, and acquiring a plurality of supply state information sent by a plurality of supply nodes; configuring a plurality of supply data prediction resources according to the scales of a plurality of nodes in the supply node graph and the product features, performing supply data prediction according to the product features and the plurality of supply state information respectively to obtain a plurality of node supply data, fusing and processing to obtain collaborative supply data, and performing pushing, wherein the supply data comprises supply time.

[0007] In a second aspect, the application provides a cross-border e-commerce supply chain data collaborative pushing platform based on graph computing, which comprises: a data acquisition module, configured to acquire a supply chain of a target product in cross-border e-commerce, and acquire historical supply data of a plurality of supply nodes in the supply chain within a recent historical time range to obtain a historical supply data chain set; A node graph construction module is configured to construct a supply node graph according to the set of historical supply data chains, wherein the supply node graph comprises nodes and edges and is updated in real time, analyze supply stability coefficients according to supply data of adjacent supply nodes, configure scales of the nodes, and construct the supply node graph; An information acquisition module is configured to acquire product features of a current target product and acquire a plurality of supply state information sent by a plurality of supply nodes; A data prediction module is configured to configure a plurality of supply data prediction resources according to the scales of the plurality of nodes in the supply node graph in combination with the product features, perform supply data prediction according to the product features and the plurality of supply state information respectively, acquire a plurality of node supply data, fuse and process the node supply data to obtain collaborative supply data, and perform pushing, wherein the supply data comprises supply time.

[0008] The one or more technical solutions provided in the present application have at least the following technical effects or advantages: The present application provides a cross-border e-commerce supply chain data collaborative pushing method and platform based on graph computing. The method comprises the following steps: constructing a prediction branch model to obtain a prediction branch group, performing prediction according to the prediction branch group, obtaining a supply prediction coefficient, and finally performing weighted calculation on node supply data according to the supply prediction coefficient to obtain collaborative supply data. The prediction improves the accuracy of supply data. Compared with the previous single node data supply method, the cross-border e-commerce supply chain data collaborative pushing platform based on graph computing provides systematic prediction analysis support, solves the problem of inaccurate data supply and high false positive rate in cross-border e-commerce, avoids the problem of being unable to balance supply accuracy and multi-node data supply due to the large number of supply chain nodes, and improves the effect of multi-node supply of the supply chain, the accuracy of supply chain data supply, the timeliness of multi-node data supply, and reduces the problem of inaccurate supply. BRIEF DESCRIPTION OF DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0010] Figure 1 It is a flowchart of a cross-border e-commerce supply chain data collaborative pushing method based on graph computing. Figure 2 It is a structural schematic diagram of a cross-border e-commerce supply chain data collaborative pushing platform based on graph computing.

[0011] In the drawings, the following signs represent the following explanations: The data acquisition module 11; the node graph construction module 12; the information acquisition module 13; the data prediction module 14. DETAILED DESCRIPTION

[0012] The application provides a cross-border e-commerce supply chain data collaborative pushing method and platform, which solves the problem of inaccurate logistics information data pushing caused by inaccurate supply chain data pushing and too many supply chain nodes.

[0013] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the application.

[0014] It should be noted that the terms "comprising" and "having" are intended to cover non-exclusive inclusion, for example, a process, method, system, product or server including a series of steps or units does not have to be limited to those clearly listed steps or units, but can include other steps or modules that are not clearly listed or inherent to the process, method, product or device.

[0015] The application will be described in detail below with reference to the drawings.

[0016] Embodiment one, as shown in the application provides a cross-border e-commerce supply chain data collaborative pushing method based on graph computing, the method comprises: Figure 1 S10: obtaining the supply chain of the target product in the cross-border e-commerce, and obtaining the historical supply data of the plurality of supply nodes in the supply chain in the recent historical time range to obtain a historical supply data chain set; First, the historical supply data of the plurality of supply nodes is obtained, preferably the supply chain of the customized product. The supply chain includes a plurality of supply nodes such as raw material suppliers, material processing suppliers, assembly suppliers, multi-level logistics suppliers, and end users. The supply chain is a chain structure including a plurality of nodes, and the overall functional network chain structure connected by a plurality of supply nodes. Because there are a plurality of supply nodes, the collaboration between the plurality of supply nodes is insufficient, thereby causing the problem of inaccurate pushing effect of traditional cross-border e-commerce supply chain data. The method provided in the embodiments of the application comprises the following steps:

[0017] ​Obtain all supply data of the plurality of supply nodes and the next supply node in the recent historical time range respectively, and obtain a historical supply data chain set, wherein the last supply node has no supply data, each supply data includes supply characteristics, expected supply time and actual supply time.

[0018] Specifically, all supply data of all supply nodes in the recent historical time range (such as the recent one year) is obtained through the supply data table in the historical supply process, such as the supply record table. The supply data without table can be constructed into the supply data table of the corresponding node, and the historical data chain set is obtained through the historical supply data table construction.

[0019] The all supply data includes: supply characteristics, expected supply time and actual supply time of raw material suppliers; supply characteristics, expected supply time and actual supply time of material processing suppliers; supply characteristics, expected supply time and actual supply time of assembly suppliers; supply characteristics, expected supply time and actual supply time of multi-level logistics supply, etc.

[0020] For example, the supply characteristics of the raw material supplier are: general service quality, better product quality, and better price competitiveness; the expected supply time is 90 days; and the actual supply time is 75 days.

[0021] In summary, compared with the prior art, in the embodiment of the application, all supply data in the recent historical event range is obtained to construct a historical data chain set, thereby providing a data basis for calculating the collaborative supply data, ensuring the completeness of the data, and providing sufficient data basis for subsequent analysis.

[0022] S20: constructing a supply node graph according to the historical supply data chain set, wherein the supply node graph includes nodes and edges, and is updated in real time, analyzing a supply stability coefficient according to the supply data of adjacent supply nodes, configuring the size of the node, and constructing the supply node graph; The on-time proportion of each node to the next level node is analyzed as the supply stability coefficient. The size of the node is set, the larger the supply stability coefficient is, the larger the node is, and the more visual it is. The supply node graph is constructed through the supply stability coefficient, and the visualization degree is improved. The supply node graph includes nodes and edges, and the node is the supply node and the edge is the connection line. The nodes can be connected by the connection line to form the supply node graph.

[0023] The method provided in the embodiment of the application, step S20, the historical supply data chain set, constructs a supply node graph, including: According to the historical supply data chain set, a plurality of historical supply data sets of the plurality of supply nodes and the next supply node are obtained; In each historical supply data set, a proportion of supply data with actual supply time less than or equal to expected supply time is calculated as a plurality of supply stability coefficients. According to the plurality of supply stability coefficients, the scale of the node is configured, and a supply node graph is constructed.

[0024] Specifically, according to the historical supply data chain set, a plurality of historical supply data sets of a supply node and a next supply node are divided. The historical supply data chain set is divided according to raw material suppliers, material processing suppliers, assembly suppliers, and multi-level logistics supply to obtain historical supply data sets corresponding to the node and the next node. Each historical supply data set includes historical supply data of the supply node and the next supply node.

[0025] For example, a historical supply data set from a raw material supplier to a material processing supplier, a historical supply data set from a material processing supplier to an assembly supplier, and a historical supply data set from an assembly supplier to multi-level logistics supply.

[0026] Further, in each historical supply data set, a proportion of supply data with actual supply time less than or equal to expected supply time is calculated as a plurality of supply stability coefficients. The supply stability coefficient is actually a just-in-time delivery rate, and the calculation formula is: supply stability coefficient = number of supply data with actual supply time less than or equal to expected supply time / total number of historical supply data. The larger the ratio, the stronger the supply stability and the higher the just-in-time delivery rate. The expected supply time is the originally planned supply time, which can be determined according to the original planned supply time recorded by the plurality of supply nodes.

[0027] For example, in a historical supply data set of a certain supply node, the number of times that the actual supply time is less than the expected supply time is 83, and there are a total of 100 historical supply data. Therefore, the proportion of supply data with actual supply time less than or equal to expected supply time is 83 / 100 = 83%.

[0028] Further, according to the plurality of supply stability coefficients, the scale of the node is configured, and a supply node graph is constructed. The node scale is, for example, the size of the node.

[0029] In the method provided by the embodiments of the present application, step S20 includes: obtaining a preset node scale; calculating a ratio of each supply stability coefficient to a mean value of the plurality of supply stability coefficients, and calculating a plurality of node scale adjustment coefficients; using the plurality of node scale adjustment coefficients to respectively adjust and calculate the preset node scale to obtain a plurality of adjusted node scales; Based on the plurality of adjusted node scales, a plurality of nodes are constructed, and edges are used to connect the plurality of nodes according to the supply relationship of the plurality of supply nodes in the supply chain to obtain a supply node graph.

[0030] Specifically, by reading, a preset node size is obtained, for example, the preset node size is 4cm².

[0031] Further, the ratio of each supply stability coefficient to the average of the plurality of supply stability coefficients is calculated, and a plurality of node scale adjustment coefficients are calculated and configured. The calculation formula is node scale adjustment coefficient = supply stability coefficient / average of the plurality of supply stability coefficients. Wherein, the larger the node scale adjustment coefficient, the smaller the degree of adjustment required.

[0032] Illustratively, the supply stability coefficient is 0.83, and the average of the plurality of supply stability coefficients is 0.8, so the node scale adjustment coefficient is 0.83 / 0.8 = 1.0375.

[0033] Further, the plurality of node scale adjustment coefficients are used to respectively adjust and calculate the preset node scale to obtain a plurality of adjusted node scales. The obtained node scale adjustment coefficient is multiplied by the preset node scale to adjust and calculate the adjusted node scale. The calculation formula is: node scale adjustment coefficient x preset node scale = adjusted node scale.

[0034] Illustratively, the adjusted node scale is 1.0375 x 4 = 4.15cm².

[0035] Further, based on the plurality of adjusted node scales, a plurality of nodes are constructed, wherein the node size of each node is different. After the nodes are constructed, edges are used to connect the plurality of nodes according to the supply relationship of the plurality of supply nodes in the supply chain to obtain a supply node graph.

[0036] Illustratively, the adjusted node scales such as 4.15cm² and 3.8cm² are obtained, a plurality of nodes such as circles with different area sizes are constructed, and then edges are used to connect them in turn to obtain a supply node graph.

[0037] In summary, compared with the prior art, in the embodiments of the present application, the node scale adjustment coefficient is obtained through the supply stability coefficient, and the node scale of each supply chain is adjusted, so that the node size corresponding to the supply data is obtained, avoiding the problem of inaccurate supply data caused by too many nodes, and improving the accuracy of the supply data.

[0038] S30: Obtain the product characteristics of the current target product, and obtain a plurality of supply state information sent by a plurality of supply nodes; Obtaining the quantity of the current product, the degree of customization, and obtaining a plurality of supply state information sent by a plurality of supply nodes, and each supply state information includes the processing progress reported by each node to the e-commerce platform according to the supply progress of the upstream node or the node itself. For example, two nodes A and B, and A node produces 50%. The progress reported by A node is 25%, and the progress reported by B node is 50%.

[0039] In the embodiment of the application, step S30 comprises: Obtaining product characteristics of the current target product, wherein the product characteristics include product quantity and product customization coefficient; Obtaining a plurality of supply state information sent by a plurality of supply nodes, wherein each supply state information includes supply progress information.

[0040] For example, the product characteristics of the current target product are obtained, the product characteristics include product quantity and product customization coefficient, the product customization coefficient is the difference amplitude such as the difference amplitude of 20%, the product customization coefficient is 20%, and the product quantity is, for example, 100 pieces.

[0041] In summary, compared with the prior art, in the embodiment of the application, the product characteristics of the current target product are obtained, and a plurality of supply state information sent by a plurality of supply nodes is obtained to obtain supply state data of a plurality of nodes, thereby ensuring the integrity of the data.

[0042] S40: According to the scale of the plurality of nodes in the supply node graph and the product characteristics, a plurality of supply data prediction resources are configured, supply data prediction is performed according to the product characteristics and the plurality of supply state information respectively, a plurality of node supply data are obtained, collaborative supply data are obtained through fusion processing, and the collaborative supply data are pushed, wherein the supply data include supply time.

[0043] The scale of the node is combined with the product characteristics to set the number of branches of integrated prediction. The smaller the scale, the worse the node supply stability, the larger the product quantity, the higher the degree of customization, and the more branches should be used for prediction to improve the accuracy. According to the product characteristics and the scale of the node, the number of prediction resources used for predicting the data of the node is determined. This reflects an intelligent resource scheduling strategy, which allocates more computing resources to unstable nodes to improve the prediction accuracy and ensure the effect of supply prediction. The integrated learning technology is used to predict the actual remaining supply time, such as 12 days.

[0044] In the embodiment of the application, S40 comprises: obtaining the maximum product quantity and the maximum product customization coefficient supplied in the historical time of the target product; Calculating the ratio of the product quantity and the product customization coefficient in the product characteristics to the maximum product quantity and the maximum product customization coefficient to obtain a first supply prediction coefficient; The ratio of the size of each node in the supply node graph to the maximum size is calculated to obtain a plurality of second supply prediction coefficients. According to the first supply prediction coefficient and the plurality of second supply prediction coefficients, a plurality of supply prediction coefficients are respectively calculated. Based on the plurality of supply prediction branch groups pre-configured for the plurality of supply nodes, the plurality of supply prediction branch numbers are respectively calculated according to the plurality of supply prediction coefficients combined with the total branch number in each supply prediction branch group. A plurality of supply prediction branches of the plurality of supply prediction branch numbers are respectively randomly selected, and product features and a plurality of supply state information are respectively input and output to obtain a plurality of predicted node supply data sets, and the mean value is calculated to obtain a plurality of node supply data. According to the plurality of supply prediction coefficients, the plurality of node supply data are weighted and calculated to obtain collaborative supply data, which are pushed.

[0045] Specifically, first, the maximum product quantity and the maximum product customization coefficient of the target product in the historical time are obtained.

[0046] The ratio of the product quantity and the product customization coefficient in the product feature to the maximum product quantity and the maximum product customization coefficient is calculated to obtain the first supply prediction coefficient. The first supply prediction coefficient is the average of the ratio of the product quantity to the maximum product quantity and the ratio of the product customization coefficient to the maximum product customization coefficient. Comparing the current product quantity and product customization coefficient with the maximum product quantity and maximum product customization coefficient, it can be seen that the larger the first supply prediction coefficient, the more difficult the prediction.

[0047] For example, the product quantity is 100 pieces, the product customization coefficient is 0.2, the maximum product quantity and the maximum product customization coefficient are 120 pieces and 0.3 respectively, [(100 / 120)]≈0.83, [(0.2 / 0.3)]≈0.66, and the first supply prediction coefficient is [(0.83+0.66) / 2]≈0.746. The first supply prediction coefficient is large, which proves that the prediction is difficult.

[0048] Further, the ratio of the size of each node in the supply node graph to the maximum size is calculated to obtain a plurality of second supply prediction coefficients.

[0049] The size of each node is compared with the maximum size in the graph. The smaller the node size, the worse the node supply stability, and the larger the second supply prediction coefficient. The calculation formula is: the second supply prediction coefficient=1 minus the ratio of the size of each node to the maximum size. The smaller the size of the node, the smaller the supply stability coefficient, the larger the second supply prediction coefficient should be configured, the more branch numbers should be configured, and the prediction accuracy of the supply should be improved.

[0050] For example, the size of the node is 3 square centimeters, the maximum size is 8 square centimeters, and the second supply prediction coefficient is 1-(3 / 8)=0.625. The size of the node is smaller, and the configured second supply prediction coefficient is larger, so more branch numbers can be configured for prediction, and the accuracy of prediction is improved.

[0051] Further, according to the first supply prediction coefficient and the plurality of second supply prediction coefficients, a plurality of supply prediction coefficients are respectively calculated and obtained. The larger the supply prediction coefficient, the more supply prediction branches are needed.

[0052] The first supply prediction coefficient and the plurality of second supply prediction coefficients are respectively weighted and fused.

[0053] The weights of the first supply prediction coefficient and the second supply prediction coefficient are set according to the product customization degree and the influence degree of the node supply stability on the supply time. The greater the influence degree, the greater the weight, which can be set by those skilled in the art. For example, the weights are 0.5 and 0.5, and the supply prediction coefficient is (0.746*0.5)+(0.625*0.5)≈0.68.

[0054] Further, based on the plurality of supply prediction branch groups pre-configured for the plurality of supply nodes, the plurality of supply prediction branch numbers are respectively calculated and determined according to the plurality of supply prediction coefficients combined with the total branch numbers in each supply prediction branch group. The plurality of supply prediction branch numbers are obtained by multiplying the plurality of supply prediction coefficients and the total branch numbers in each supply prediction branch group.

[0055] For example, the supply prediction coefficient is 0.68, the total branch number in each supply prediction branch group is 5, and the supply prediction branch number=0.68*5≈3.4, which is rounded up to 4.

[0056] Further, a plurality of supply prediction branches of the plurality of supply prediction branch numbers are respectively randomly selected, and product features and a plurality of supply state information are respectively input, and a plurality of predicted node supply data sets such as actual delivery time are output and obtained, and a plurality of node supply data are calculated and obtained.

[0057] For example, the supply time of the predicted node supply data of A, B and C is 10 days, 8 days and 12 days respectively.

[0058] Based on multiple supply forecast coefficients, weighted calculations are performed on supply data from multiple nodes to obtain collaborative supply data, which is then pushed out. The weights represent the degree of influence on the node supply data; the greater the influence, the greater the weight. The collaborative supply data is the sum of the products of all node supply data and their respective weights. The calculation formula is: (1st node supply data × 1st supply forecast coefficient weight + ... + nth node supply data × nth supply forecast coefficient weight). The weight of each supply forecast coefficient is calculated and configured by the ratio of each supply forecast coefficient to the sum of all supply forecast coefficients. Collaborative supply data provides coordinated supply of data from multiple nodes, making the timing and analysis of collaborative supply more accurate, and improving the prediction accuracy of collaborative data supply.

[0059] For example, the supply forecast coefficients for A, B, and C are 0.68, 0.5, and 0.75, respectively, with n being 3. Based on the supply forecast coefficients, their weights are 0.35, 0.25, and 0.4, respectively. Their collaborative supply data is (10×0.35+8×0.25+12×0.4)=10.3 days, which is the remaining supply time of the target product in the collaborative forecast.

[0060] The method provided in this application includes a pre-configuration step for multiple supply forecasting branch groups, comprising: Based on the historical supply record data of the first supply node within multiple supply nodes, the sample first product feature set, the sample first supply status information set, and the final actual product supply time are collected and labeled to obtain the sample first node supply data set; The first product feature set, the first supply status information set, and the first node supply data set of the sample are divided multiple times to obtain multiple first supply prediction training datasets. Based on machine learning, multiple first supply prediction branches are constructed, and each branch is trained under supervision until convergence using multiple first supply prediction training datasets to obtain a group of first supply prediction branches. Continue training to obtain multiple supply prediction branches for other supply nodes.

[0061] Specifically, based on the historical supply record data of the first supply node among multiple supply nodes, the system collects the first product feature set, the first supply status information set, and the final actual product supply time of the sample, thereby labeling and constructing the first node supply data set of the sample. The first supply node is the most upstream supply node among multiple supply nodes in the supply chain, such as the node corresponding to the raw material supplier.

[0062] Furthermore, the mean squared error loss function (MSE) is used as the partitioning criterion, and the optimal bisection point is selected to partition the first product feature set, the first supply status information set, and the first node supply data set of the sample multiple times to obtain multiple first supply prediction training datasets.

[0063] Furthermore, a decision tree model, CART (Classification and Regression Tree), using machine learning algorithms, is employed to construct multiple first supply prediction branches. CART is a supervised learning algorithm with a binary tree structure that can classify or regress data using a series of rules. K first supply prediction branches are constructed, where K is a positive integer. Each branch represents an output judgment result, and each node in the decision tree represents a classification feature. The number of nodes in each prediction branch on the decision tree is K-1.

[0064] For example, if K is 4, the number of predicted branches is 4, and the number of nodes for the predicted branches on the decision tree is 3.

[0065] Furthermore, multiple first supply prediction training datasets are used, and the process is repeated iteratively and supervised until convergence to obtain the first supply prediction branch group. That is, the first product features and first supply status information in the historical supply data of each node are used as input, and the actual supply time of the first node is used as the label to train multiple prediction models, i.e., branches, forming a model group, which is the first supply prediction branch group.

[0066] By continuing training and inputting historical supply record data from multiple other supply nodes into the model, multiple supply prediction branches from those other supply nodes can be obtained.

[0067] In summary, compared to existing technologies, this application's embodiments construct a supply forecasting branch group to obtain multiple supply forecasting branches. Then, by configuring multiple supply data forecasting resources based on the scale of multiple nodes within the node graph, supply data forecasting is performed within the supply forecasting branch group based on product characteristics and multiple supply status information. The resulting supply forecasting coefficient initially reflects the accuracy of the supply forecast. Collaborative supply data is obtained through weighted fusion processing of supply data from multiple nodes. The resulting collaborative supply coefficient is then weighted according to the influence of the node supply data to obtain collaborative supply data that can coordinate multi-node data supply. This avoids the inaccuracy problem caused by numerous nodes, solves the multi-node data supply problem in cross-border e-commerce, and improves the data supply effect of the supply chain. It achieves efficient and accurate delivery of supply chain data.

[0068] The embodiments of this application, through the specific implementation methods described above, achieve the following technical effects: In this embodiment, firstly, a historical data chain set is constructed by acquiring all supply data within the range of recent historical events. This provides a data foundation for calculating collaborative supply data, ensuring data completeness and providing sufficient data basis for subsequent analysis. Secondly, a node scale adjustment coefficient is obtained through the supply stability coefficient, and the node scale of each node in the supply chain is adjusted accordingly. This yields the node size corresponding to the supply data, avoiding the problem of inaccurate supply data caused by a large number of nodes and improving the accuracy of the supply data. Furthermore, by acquiring the product characteristics of the current target product and obtaining multiple supply status information sent by multiple supply nodes, supply status data for multiple nodes is obtained, ensuring data integrity.

[0069] Finally, this embodiment of the application constructs a supply forecasting branch group to obtain multiple supply forecasting branches. Then, by configuring multiple supply data forecasting resources based on the scale of multiple nodes within the node graph, a forecasting branch model is constructed, resulting in a first supply forecasting branch group and several other supply forecasting branch groups. This construction of the supply forecasting branch model avoids the problem of inaccurate supply data push effects across multiple nodes, improves the ability to push supply data information across nodes, and enhances the effectiveness of supply data push. Supply data is predicted within the supply forecasting branch group based on product characteristics and multiple supply status information. The resulting supply forecasting coefficient initially reflects the accuracy of supply forecasting. Collaborative supply data is obtained through weighted fusion processing of supply data from multiple nodes. The resulting collaborative supply coefficient is weighted according to the influence of node supply data to obtain collaborative supply data that can coordinate multi-node data supply. This avoids the problem of inaccurate data caused by multiple nodes, solves the problem of multi-node data supply in cross-border e-commerce, and improves the effectiveness of supply chain data supply. It achieves efficient and accurate push of supply chain data supply.

[0070] Example 2, as Figure 2 As shown, based on the same inventive concept as the graph computing-based cross-border e-commerce supply chain data collaborative push method provided in Embodiment 1, this embodiment of the invention also provides a graph computing-based cross-border e-commerce supply chain data collaborative push platform, including: The data acquisition module 11 is used to acquire the supply chain of the target product in cross-border e-commerce, and to acquire historical supply data of multiple supply nodes within the supply chain within the most recent historical time range, thereby obtaining a set of historical supply data chains. The node graph construction module 12 is used to construct a supply node graph based on the historical supply data chain set. The supply node graph includes nodes and edges and is updated in real time. The supply stability coefficient is analyzed based on the supply data of adjacent supply nodes, the node scale is configured, and the supply node graph is constructed. The information acquisition module 13 is used to acquire the product characteristics of the current target product and acquire multiple supply status information sent by multiple supply nodes; The data prediction module 14 is used to configure multiple supply data prediction resources based on the scale of multiple nodes in the supply node diagram and product characteristics, and to perform supply data prediction based on product characteristics and multiple supply status information respectively, to obtain supply data of multiple nodes, to merge and process to obtain collaborative supply data, and to push it. The supply data includes supply time.

[0071] In one embodiment, the data acquisition module 11 is used for: Acquire the supply chain of the target product in cross-border e-commerce, which includes multiple supply nodes; Obtain all supply data from multiple supply nodes and the next supply node within the most recent historical time range to obtain a historical supply data chain set. The last supply node has no supply data. Each supply data includes supply characteristics, expected supply time, and actual supply time.

[0072] In one embodiment, the node graph construction module 12 is used for: Based on the historical supply data chain set, multiple historical supply datasets are obtained for multiple supply nodes and the next supply node; Within each historical supply dataset, the proportion of supply data where the actual supply time is less than or equal to the expected supply time is calculated and used as multiple supply stability coefficients. Based on multiple supply stability coefficients, configure the node scale and construct a supply node graph.

[0073] In one embodiment, the node graph construction module 12 is further configured to: Obtain the preset node scale; Calculate the ratio of each supply stability coefficient to the average of multiple supply stability coefficients, and calculate the adjustment coefficients for configuring multiple node scales; Multiple node scale adjustment coefficients are used to adjust the preset node scales respectively, thereby obtaining multiple adjusted node scales; Based on multiple adjustment node scales, multiple nodes are constructed, and according to the supply relationships of multiple supply nodes within the supply chain, multiple nodes are connected by edges to obtain a supply node graph.

[0074] In one embodiment, the information acquisition module 13 is used for: Obtain the product characteristics of the current target product, including the product quantity and product customization coefficient; Obtain multiple supply status information sent by multiple supply nodes, where each supply status information includes supply progress information.

[0075] In one embodiment, the data prediction module 14 is used for: Calculate the ratio of the number of products within the product characteristics and the product customization coefficient to the maximum number of products and the maximum product customization coefficient to obtain the first supply forecast coefficient; Calculate the ratio of the scale of multiple nodes in the supply node diagram to the largest scale, and process them to obtain multiple second supply prediction coefficients; Multiple supply forecast coefficients are calculated based on the first supply forecast coefficient and multiple second supply forecast coefficients. Based on multiple supply forecasting branch groups pre-configured for multiple supply nodes, the number of multiple supply forecasting branches is calculated and determined according to multiple supply forecasting coefficients and the total number of branches in each supply forecasting branch group. Randomly select multiple supply forecast branches, input product characteristics and multiple supply status information respectively, and output multiple forecast node supply datasets. Calculate the mean to obtain multiple node supply data. Based on multiple supply forecasting coefficients, weighted calculations are performed on supply data from multiple nodes to obtain collaborative supply data, which is then pushed out.

[0076] The pre-configuration of multiple supply forecasting subgroups includes: Based on the historical supply record data of the first supply node within multiple supply nodes, the sample first product feature set, the sample first supply status information set, and the final actual product supply time are collected and labeled to obtain the sample first node supply data set; The first product feature set, the first supply status information set, and the first node supply data set of the sample are divided multiple times to obtain multiple first supply prediction training datasets. Based on machine learning, multiple first supply prediction branches are constructed, and each branch is trained under supervision until convergence using multiple first supply prediction training datasets to obtain a group of first supply prediction branches. Continue training to obtain multiple supply prediction branches for other supply nodes.

[0077] The graph computing-based collaborative push method for cross-border e-commerce supply chain data first utilizes a data acquisition module to collect comprehensive and sufficient data, ensuring data availability and providing ample data support for subsequent predictive analysis of supply data. Second, a node graph construction module constructs a node graph, using a supply stability coefficient to obtain node scale adjustment coefficients. This adjusts the node scale for each supply chain node, resulting in node sizes corresponding to the supply data, avoiding inaccuracies caused by numerous nodes and improving data accuracy. Third, an information acquisition module acquires the product characteristics of the current target product and multiple supply status information sent by various supply nodes, obtaining supply status data for multiple nodes and ensuring data integrity. Finally, a data prediction module constructs a prediction branch model to predict and analyze node supply data, obtaining accurate predictions and improving the precision of node supply, reducing inaccuracies in supply information caused by supply chain data nodes. This method effectively improves the information supply effect in cross-border e-commerce, increases the timeliness of supply chain data supply, and ensures the effectiveness of data supply.

[0078] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0079] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for collaborative data push in cross-border e-commerce supply chain using graph computing, characterized in that, The method includes: Obtain the supply chain of the target product in cross-border e-commerce, and obtain historical supply data of multiple supply nodes within the supply chain within the most recent historical time range to obtain a set of historical supply data chains; Based on the historical supply data chain set, a supply node graph is constructed, which includes nodes and edges and is updated in real time. The supply stability coefficient is analyzed based on the supply data of adjacent supply nodes, the node scale is configured, and the supply node graph is constructed. Obtain the product characteristics of the current target product and obtain multiple supply status information sent by multiple supply nodes; Based on the scale of multiple nodes in the supply node graph and the product characteristics, multiple supply data prediction resources are configured. Supply data prediction is performed according to the product characteristics and multiple supply status information to obtain supply data from multiple nodes. The data is then fused and processed to obtain collaborative supply data, which is then pushed out. The supply data includes supply time.

2. The cross-border e-commerce supply chain data collaborative push method based on graph computing according to claim 1, characterized in that, Obtain the supply chain of the target product in cross-border e-commerce, and acquire historical supply data from multiple supply nodes within the supply chain within the most recent historical time range to obtain a historical supply data chain set, including: Acquire the supply chain of the target product in cross-border e-commerce, which includes multiple supply nodes; Obtain all supply data from multiple supply nodes and the next supply node within the most recent historical time range to obtain a historical supply data chain set. The last supply node has no supply data. Each supply data includes supply characteristics, expected supply time, and actual supply time.

3. The cross-border e-commerce supply chain data collaborative push method based on graph computing according to claim 1, characterized in that, Based on the historical supply data chain set, a supply node graph is constructed, including: Based on the historical supply data chain set, multiple historical supply datasets for multiple supply nodes and the next supply node are obtained; Within each historical supply dataset, the proportion of supply data where the actual supply time is less than or equal to the expected supply time is calculated and used as multiple supply stability coefficients. The supply node graph is constructed by configuring the node scale based on multiple supply stability coefficients.

4. The cross-border e-commerce supply chain data collaborative push method based on graph computing according to claim 3, characterized in that, Based on multiple supply stability coefficients, the scale of the nodes is configured, and the supply node graph is constructed, including: Obtain the preset node scale; Calculate the ratio of each supply stability coefficient to the average of multiple supply stability coefficients, and calculate the adjustment coefficients for configuring multiple node scales; The preset node scale is adjusted using the multiple node scale adjustment coefficients to obtain multiple adjusted node scales. Based on multiple adjustment node scales, multiple nodes are constructed, and according to the supply relationships of multiple supply nodes within the supply chain, the multiple nodes are connected by edges to obtain a supply node graph.

5. The cross-border e-commerce supply chain data collaborative push method based on graph computing according to claim 1, characterized in that, Obtain the product characteristics of the current target product and acquire multiple supply status information sent by multiple supply nodes, including: Obtain the product characteristics of the current target product, including the product quantity and product customization coefficient; Obtain multiple supply status information sent by multiple supply nodes, where each supply status information includes supply progress information.

6. The cross-border e-commerce supply chain data collaborative push method based on graph computing according to claim 1, characterized in that, Based on the scale of multiple nodes within the supply node graph and the product characteristics, multiple supply data prediction resources are configured. Supply data prediction is performed based on the product characteristics and multiple supply status information to obtain supply data from multiple nodes. This data is then fused and processed to obtain collaborative supply data, which is then pushed to relevant departments. Obtain the maximum product quantity and maximum product customization coefficient of the target product within a historical period; Calculate the ratio of the product quantity and product customization coefficient within the product characteristics to the maximum product quantity and maximum product customization coefficient to obtain the first supply forecast coefficient; Calculate the ratio of the scale of multiple nodes in the supply node diagram to the largest scale, and process them to obtain multiple second supply prediction coefficients; Multiple supply forecast coefficients are calculated based on the first supply forecast coefficient and multiple second supply forecast coefficients. Based on multiple supply forecasting branch groups pre-configured for multiple supply nodes, the number of multiple supply forecasting branches is calculated and determined according to the multiple supply forecasting coefficients and the total number of branches in each supply forecasting branch group. Multiple supply forecast branches are randomly selected, and the product characteristics and multiple supply status information are input respectively. Multiple forecast node supply datasets are output, and the average value is calculated to obtain multiple node supply data. Based on multiple supply forecasting coefficients, the supply data of the multiple nodes are weighted and calculated to obtain collaborative supply data, which is then pushed out.

7. The cross-border e-commerce supply chain data collaborative push method based on graph computing according to claim 6, characterized in that, The pre-configuration steps for multiple supply forecasting branch clusters include: Based on the historical supply record data of the first supply node among multiple supply nodes, the sample first product feature set, the sample first supply status information set, and the final actual product supply time are collected and labeled to obtain the sample first node supply data set; The first product feature set, the first supply status information set, and the first node supply data set of the sample are divided multiple times to obtain multiple first supply prediction training datasets. Based on machine learning, multiple first supply prediction branches are constructed, and each branch is trained under supervision until convergence using the multiple first supply prediction training datasets to obtain a group of first supply prediction branches. Continue training to obtain multiple supply prediction branches for other supply nodes.

8. A graph computing-based cross-border e-commerce supply chain data collaborative push platform, characterized in that, The platform for implementing the method according to any one of claims 1-7 comprises: The data acquisition module is used to acquire the supply chain of the target product in cross-border e-commerce, and to acquire historical supply data of multiple supply nodes within the supply chain within the most recent historical time range, thereby obtaining a set of historical supply data chains. The node graph construction module is used to construct a supply node graph based on the historical supply data chain set. The supply node graph includes nodes and edges and is updated in real time. The supply stability coefficient is analyzed based on the supply data of adjacent supply nodes, the node scale is configured, and the supply node graph is constructed. The information acquisition module is used to acquire the product characteristics of the current target product and to acquire multiple supply status information sent by multiple supply nodes; The data prediction module is used to configure multiple supply data prediction resources based on the scale of multiple nodes in the supply node graph and the product characteristics, respectively perform supply data prediction based on the product characteristics and multiple supply status information, obtain supply data of multiple nodes, fuse and process to obtain collaborative supply data, and push it, wherein the supply data includes supply time.

Images