Intelligent single-ticket profit analysis method and system in logistics field based on knowledge graph

Abstract

The invention belongs to the technical field of knowledge maps, and discloses a logistics field intelligent single-ticket profit analysis method and system based on a knowledge map, and the method comprises the steps: obtaining the feature impact factor of each type on the single-ticket profit; based on a single-ticket profit model, obtaining relevance between each feature influence factor and other feature influence factors and a positive and negative influence relation of each feature influence factor on the single-ticket profit; constructing a logistics field single-ticket profit characteristic graph based on a relationship and a change condition between the knowledge graph representation and the single-ticket profit characteristic influence factor; and answering the questions of the user based on the subgraph triple knowledge to generate answers in a natural language form. According to the logistics field single-ticket profit characteristic spectrum and subgraph triple knowledge, the interaction and the change trend among the factors are clearly expressed, and targeted decision support is provided for enterprises, so that the overall operation efficiency and the response speed are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of knowledge graphs, and relates to an intelligent single-ticket profit analysis method and system for the logistics field based on a knowledge graph. Background Art

[0002] In the logistics field, the core indicators that franchise enterprises focus on are business volume, profit, and cost. These three are interrelated. Single-ticket profit = (profit - cost) / business volume. The single-ticket profit is the most comprehensive indicator reflecting the enterprise's profitability. Therefore, improving the single-ticket profit of each subordinate institution of the enterprise is one of the core business objectives. Large branches of logistics enterprises will also conduct single-ticket profit analysis and modeling within their jurisdiction. For the headquarters, it is necessary to take overall consideration, not only understand the single-ticket profit situation of each branch, but also explore the core factors affecting its single-ticket profit from multiple dimensions, and then formulate targeted policies or give decision-making suggestions. However, in the process of pursuing the improvement of single-ticket profit, logistics enterprises face many problems such as high difficulty in data collection and integration, complex and inaccurate profit calculation, lack of depth and breadth in data analysis, and lack of flexibility and coordination in information systems. Summary of the Invention

[0003] The purpose of the present invention is to solve the problems in the prior art such as large calculation amount of single-ticket profit, lack of depth and breadth in data analysis, and lack of flexibility and coordination in information systems, and provide an intelligent single-ticket profit analysis method and system for the logistics field based on a knowledge graph.

[0004] To achieve the above object, the present invention adopts the following technical solutions:

[0005] An intelligent single-ticket profit analysis method for the logistics field based on a knowledge graph includes:

[0006] Classify the collected logistics data information to obtain the characteristic influence factors of each category on the single-ticket profit;

[0007] Based on the single-ticket profit model, construct a characteristic correlation covariance matrix to obtain the correlation between each characteristic influence factor and other characteristic influence factors and the positive and negative influence relationships of each characteristic influence factor on the single-ticket profit;

[0008] Based on the relationship and change situation between the knowledge graph representation and the characteristic influence factors of the single-ticket profit, construct a single-ticket profit characteristic graph for the logistics field;

[0009] Based on the single-ticket profit characteristic graph of the logistics field and the graph traversal algorithm, recall sub- Figure 3 tuple knowledge;

[0010] Based on the sub- Figure 3Tuple knowledge answers the user's questions and generates answers in natural language form.

[0011] A further improvement of the present invention lies in:

[0012] Furthermore, the logistics data information includes: cost classification, revenue classification, weight segment, channel shipment volume, and delivery timeliness; the logistics data information is subdivided, and the ratio of each sub-classification to the total of its corresponding sub-classification is used as the characteristic influence factor;

[0013] The characteristic influence factors of each category on the profit per order include:

[0014] The logistics shipment volume is divided according to the delivery timeliness to obtain the same-day delivery volume, next-day delivery volume, delivery volume the day after the next day, 3-day delivery volume, 4-day delivery volume, and other timeliness delivery volumes;

[0015] The logistics shipment weight is divided according to the delivery timeliness to obtain the same-day delivery weight ratio, next-day delivery weight ratio, delivery weight ratio the day after the next day, 3-day delivery weight ratio, 4-day delivery weight ratio, and other timeliness weight ratios;

[0016] The business volume is divided according to the weight to obtain the business volume ratio of weight segment 1, business volume ratio of weight segment 2, business volume ratio of weight segment 3, business volume ratio of weight segment 4, business volume ratio of weight segment 5, business volume ratio of weight segment 6, and business volume ratio of weight segment 7;

[0017] The business volume is divided according to the channel shipment volume to obtain the Douyin shipment volume ratio, JD.com shipment volume ratio, Kuaishou shipment volume ratio, Pinduoduo shipment volume ratio, Taobao system shipment volume ratio, total business volume - branch company business volume ratio, total business volume - marketing department business volume ratio, total business volume - community business volume ratio, and other channel shipment volume ratios;

[0018] The business volume is divided according to the channel shipment weight to obtain the Douyin weight ratio, JD.com weight ratio, Kuaishou weight ratio, loading rate, Pinduoduo weight ratio, and Taobao system weight ratio;

[0019] The business types are divided according to the cost to obtain the proportion of outbound cost subtotal - outbound operation cost, proportion of outbound cost subtotal - outbound dispatch fee cost, proportion of outbound cost subtotal - other outbound costs, proportion of outbound cost subtotal - provincial management fees, proportion of outbound cost subtotal - provincial sales fees, proportion of outbound cost subtotal - outbound transport capacity cost, and proportion of outbound cost subtotal - incremental incentives;

[0020] The business types are divided according to the revenue to obtain the proportion of outbound revenue - outbound transfer revenue, proportion of outbound revenue - other outbound revenues, proportion of outbound revenue - outbound waybill revenue, and proportion of outbound revenue - outbound dispatch fee revenue.

[0021] Further, weight segment 1 represents 0 - 0.5 kg, weight segment 2 represents 0.5 - 1 kg, weight segment 3 represents 1 - 1.5 kg, weight segment 4 represents 1.5 - 2 kg, weight segment 5 represents 2 - 3 kg, weight segment 6 represents 3 - 5 kg, and weight segment 7 represents over 5 kg.

[0022] Further, based on the single - ticket profit model, a characteristic correlation covariance matrix is constructed to obtain the correlation between each characteristic influencing factor and other characteristic influencing factors, as well as the positive and negative influence relationships of each characteristic influencing factor on the single - ticket profit. Specifically:

[0023] For characteristic influencing factors X i and X j , their covariance calculation formula is:

[0024]

[0025] where n is the number of samples, X ik and X jk are the values of characteristic X i and X j on the k - th sample respectively, and are the means of characteristic X i and X j respectively;

[0026] The covariances between all characteristics are formed into a matrix, namely the covariance matrix; for m characteristics, the covariance matrix is an m×m matrix, where the element (i, j) is the covariance between characteristic influencing factors X i and X j ; the element values in the covariance matrix reflect the linear correlation between characteristics; positive values indicate positive correlation, negative values indicate negative correlation, and the larger the absolute value, the stronger the correlation.

[0027] Further, the core factors affecting the single - ticket profit are: the proportion of same - day delivery volume, the proportion of out - bound delivery fee cost, the proportion of out - bound delivery fee income, the proportion of branch business volume, the proportion of next - day delivery volume, the proportion of other - time - limit weight, and the proportion of JD.com orders; among the costs, out - bound delivery fees, among the incomes, out - bound delivery fee income, among the time - limits, same - day delivery and next - day delivery, and among the weights, other - time - limit have the greatest influence weights on the single - ticket profit.

[0028] Further, based on the relationship and change situation between the knowledge - graph representation and the single - ticket profit characteristic influencing factors, a single - ticket profit characteristic graph in the logistics field is constructed. Specifically:

[0029] Construct index entity nodes, institutional entity nodes, and time entity nodes; meanwhile, assign a unique ID to each index entity node and use the index name as an attribute; assign a unique ID to each institutional entity node and use the institutional name, type, and establishment time as attributes; assign a unique ID to each time entity node and use the time value as an attribute;

[0030] Establish directed edges between feature factors and between feature factors and single-ticket profit to represent the influence relationships between them; each edge has positive and reverse attributes indicating the direction of influence, either positive or reverse; write the normalized influence degree into the attribute of the edge to represent the intensity of influence; the weight value ranges from 0 to 1, where 1 represents the maximum influence and 0 represents no influence;

[0031] Establish the relationship between the index and the institutional entity node, indicating that the index is responsible for or generated by the institution in the institutional entity node; establish the relationship between the index and the corresponding time entity node, indicating that the index occurs during the time period in the time entity node; write the above entities and relationships into the graph database to form a single-ticket profit feature graph in the logistics field.

[0032] Furthermore, recall sub Figure 3 tuple knowledge based on the single-ticket profit feature graph in the logistics field and the graph traversal algorithm, and answer the user's question based on the sub Figure 3 tuple knowledge to generate an answer in natural language form, specifically:

[0033] According to the user's question, determine the starting node in the knowledge graph;

[0034] Starting from the starting node, traverse the knowledge graph using the depth-first search DFS or breadth-first search BFS algorithm to find all triple knowledge related to the starting node; during the traversal process, set the depth or breadth of the traversal as needed to control the quantity and quality of the recalled triple knowledge;

[0035] From the triple knowledge obtained by traversal, filter out the sub Figure 3 tuple knowledge directly related to the user's question;

[0036] Organize the extracted key information into an answer in natural language form and present the organized natural language answer to the user.

[0037] The intelligent single-ticket profit analysis system in the logistics field based on the knowledge graph includes:

[0038] A classification module that classifies the collected logistics data information to obtain the characteristic influence factors of each category on the single-ticket profit;

[0039] An acquisition module, which constructs a feature correlation covariance matrix based on a single-ticket profit model to obtain the correlation between each feature influence factor and other feature influence factors, and the positive and negative influence relationships of each feature influence factor on the single-ticket profit;

[0040] A construction module, which constructs a single-ticket profit feature map for the logistics field based on the relationship and change situation between the knowledge graph representation and the single-ticket profit feature influence factors;

[0041] A recall module, which recalls sub- Figure 3 tuple knowledge based on the single-ticket profit feature map for the logistics field and the graph traversal algorithm;

[0042] A generation module, which answers the user's question based on the sub- Figure 3 tuple knowledge and generates an answer in the form of natural language.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] The present invention systematically classifies the collected logistics data information, obtains the feature influence factors of each category on the single-ticket profit, and through an automated and intelligent data processing process, can greatly improve the data integration efficiency, ensure the accuracy and integrity of the data, and provide a solid foundation for subsequent analysis. And according to the single-ticket profit feature map for the logistics field and the sub- Figure 3 tuple knowledge, it can deepen the depth and breadth of data analysis, clearly express the interaction and change trend between various factors, provide more scientific and targeted decision-making support for enterprises, thereby improving the single-ticket profit of each subordinate institution, realizing overall planning, collaborative optimization, information sharing and collaborative work between different departments, and thus improving the overall operation efficiency and response speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present invention, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0046] Figure 1 It is a schematic flow chart of the intelligent single-ticket profit analysis method for the logistics field based on the knowledge graph of the present invention;

[0047] Figure 2 It is a schematic structural diagram of the intelligent single-ticket profit analysis system for the logistics field based on the knowledge graph of the present invention;

[0048] Figure 3It is a probability distribution curve graph of the profit per order;

[0049] Figure 4 It is a schematic diagram of the key influencing factors affecting the profit per order;

[0050] Figure 5 It is a schematic diagram of the positive and negative qualitative analysis of the influencing factors;

[0051] Figure 6 It is a fitting curve graph of the top influencing factors and the profit per order. Detailed implementation manners

[0052] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Generally, the components of the embodiments of the present invention described and illustrated in the accompanying drawings here can be arranged and designed in various different configurations.

[0053] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed present invention, but merely represents selected embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without making creative efforts fall within the scope of protection of the present invention.

[0054] It should be noted that: similar reference numerals and letters denote similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0055] In the description of the embodiments of the present invention, it should be noted that if terms such as "upper", "lower", "horizontal", "inner", etc. are used to indicate the orientation or positional relationship, it is based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship when the product of the present invention is normally placed. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus cannot be construed as a limitation of the present invention. In addition, terms such as "first", "second", etc. are only used for distinguishing descriptions and cannot be construed as indicating or implying relative importance.

[0056] In addition, if the term "horizontal" appears, it does not mean that the component is required to be absolutely horizontal, but it can be slightly inclined. For example, "horizontal" only means that its direction is more horizontal relative to "vertical", and does not mean that the structure must be completely horizontal, but it can be slightly inclined.

[0057] In the description of the embodiments of the present invention, it is also necessary to explain that, unless otherwise clearly specified and limited, the terms "set", "install", "connect", and "connect" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal connection of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0058] The present invention is further described in detail below in conjunction with the accompanying drawings:

[0059] See also Figure 1 The present invention discloses an intelligent single-ticket profit analysis method in the logistics field based on a knowledge graph, comprising:

[0060] S101, classify the collected logistics data information to obtain the characteristic influencing factors of each category on the single ticket profit;

[0061] Logistics data information includes: cost classification, income classification, weight range, channel shipment volume and delivery time; the logistics data information is subdivided, and the ratio of each subdivision to the corresponding subdivision sum is used as the characteristic influencing factor;

[0062] The characteristic influencing factors of each category on the single ticket profit include:

[0063] The logistics delivery volume is divided according to the delivery time, and the delivery volume on the same day, the delivery volume on the next day, the delivery volume on the next day, the delivery volume on the third day, the delivery volume on the fourth day, and the delivery volume on other delivery time are obtained;

[0064] The weight of logistics delivery is divided according to the delivery time, and the weight proportion of same-day delivery, next-day delivery, next-day delivery, 3-day delivery, 4-day delivery and other time-limited weight proportions are obtained;

[0065] Divide the business volume according to weight to obtain the business volume proportion of weight segment 1, the business volume proportion of weight segment 2, the business volume proportion of weight segment 3, the business volume proportion of weight segment 4, the business volume proportion of weight segment 5, the business volume proportion of weight segment 6, and the business volume proportion of weight segment 7;

[0066] The business volume is divided according to the channel shipment volume, and the proportion of Douyin, JD.com, Kuaishou, Pinduoduo, Taobao, total business volume-branch business volume, total business volume-marketing department business volume, total business volume-community business volume and other channel volume are obtained;

[0067] Divide the business volume according to the shipping weight of each channel to obtain the weight ratios of Douyin, JD.com, Kuaishou, loading rate, Pinduoduo, and Taobao systems;

[0068] Divide the business types according to the cost to obtain the proportion of outbound operation cost in the total outbound cost, the proportion of outbound dispatch fee cost in the total outbound cost, the proportion of other outbound costs in the total outbound cost, the proportion of provincial management fees in the total outbound cost, the proportion of provincial sales expenses in the total outbound cost, the proportion of outbound transport capacity cost in the total outbound cost, and the proportion of incremental incentives in the total outbound cost;

[0069] Divide the business types according to the revenue to obtain the proportion of outbound transfer revenue in the total outbound revenue, the proportion of other outbound revenues in the total outbound revenue, the proportion of outbound waybill revenue in the total outbound revenue, and the proportion of outbound dispatch fee revenue in the total outbound revenue.

[0070] The weight segment 1 represents 0 - 0.5 kg, the weight segment 2 represents 0.5 - 1 kg, the weight segment 3 represents 1 - 1.5 kg, the weight segment 4 represents 1.5 - 2 kg, the weight segment 5 represents 2 - 3 kg, the weight segment 6 represents 3 - 5 kg, and the weight segment 7 represents over 5 kg.

[0071] S102. Based on the single - ticket profit model, construct a characteristic correlation covariance matrix to obtain the correlation between each characteristic influencing factor and other characteristic influencing factors, and the positive - negative influence relationship of each characteristic influencing factor on the single - ticket profit;

[0072] For the characteristic influencing factor X i and X j , its covariance calculation formula is:

[0073]

[0074] where n is the number of samples, X ik and X jk are the values of the characteristics X i and X j on the k - th sample respectively, and are the means of the characteristics X i and X j respectively;

[0075] Form a matrix with the covariances between all characteristics, that is, the covariance matrix; for m characteristics, the covariance matrix is an m×m matrix, where the element (i, j) is the covariance between the characteristic influencing factors X i and X j ; the element values in the covariance matrix reflect the linear correlation between characteristics; positive values indicate positive correlation, negative values indicate negative correlation, and the larger the absolute value, the stronger the correlation.

[0076] The core factors affecting the profit per order are: the proportion of same-day delivery volume, the proportion of outbound dispatch cost, the proportion of outbound dispatch revenue, the proportion of branch business volume, the proportion of next-day delivery volume, the proportion of other time-sensitive weight, and the proportion of JD.com orders; among the costs, the outbound dispatch fee, the inbound outbound dispatch revenue, the same-day delivery in the timeliness, the next-day delivery, and the other time-sensitive weight in the weight have the greatest impact on the profit per order.

[0077] S103. Based on the relationship and change situation between the knowledge graph representation and the influencing factors of the profit per order characteristics, construct a profit per order characteristics graph in the logistics field;

[0078] Construct index entity nodes, institutional entity nodes, and time entity nodes; at the same time, assign a unique ID to each index entity node and use the index name as an attribute; assign a unique ID to each institutional entity node and use the institutional name, type, and establishment time as attributes; assign a unique ID to each time entity node and use the time value as an attribute;

[0079] Establish directed edges between the characteristic factors and between the characteristic factors and the profit per order to represent the influencing relationships between them; each edge has positive and reverse attributes to represent the direction of influence, forward or reverse; write the normalized influence degree into the attribute of the edge to represent the intensity of influence; the weight value is between 0 and 1, where 1 represents the greatest influence and 0 represents no influence;

[0080] Establish the relationship between the index and the institutional entity node, indicating that the index is responsible for or generated by the institution in the institutional entity node; establish the relationship between the index and the corresponding time entity node, indicating that the index occurs during the time period in the time entity node; write the above entities and relationships into the graph database to form a profit per order characteristics graph in the logistics field.

[0081] S104. Recall sub Figure 3 tuple knowledge based on the profit per order characteristics graph in the logistics field and the graph traversal algorithm;

[0082] S105. Answer the user's question based on the sub Figure 3 tuple knowledge and generate an answer in natural language form.

[0083] Determine the starting node in the knowledge graph according to the user's question;

[0084] Start from the starting node and use the depth-first search DFS or breadth-first search BFS algorithm to traverse the knowledge graph to find all triple knowledge related to the starting node; during the traversal process, set the depth or breadth of the traversal as needed to control the quantity and quality of the recalled triple knowledge;

[0085] From the triple knowledge obtained by traversal, filter out the sub-tuple knowledge directly related to the user's question Figure 3 tuple knowledge;

[0086] Organize the extracted key information into an answer in natural language form and present the organized natural language answer to the user.

[0087] See Figure 2 , the present invention discloses an intelligent single-ticket profit analysis system in the logistics field based on a knowledge graph, including:

[0088] A classification module, which classifies the collected logistics data information to obtain the characteristic influence factors of each category on the single-ticket profit;

[0089] An acquisition module, which constructs a characteristic correlation covariance matrix based on the single-ticket profit model to obtain the correlation between each characteristic influence factor and other characteristic influence factors and the positive and negative influence relationships of each characteristic influence factor on the single-ticket profit;

[0090] A construction module, which constructs a single-ticket profit characteristic graph in the logistics field based on the relationship and change situation between the knowledge graph representation and the single-ticket profit characteristic influence factors;

[0091] A recall module, which recalls sub-tuple knowledge based on the single-ticket profit characteristic graph in the logistics field and the graph traversal algorithm Figure 3 tuple knowledge;

[0092] A generation module, which answers the user's question based on the sub-tuple knowledge and generates an answer in natural language form. Figure 3 tuple knowledge

[0093] Embodiment:

[0094] This research mainly considers the influence of several major categories on the single-ticket profit, including cost classification, revenue classification, weight segment, channel, timeliness. The single-ticket profit is a ratio index. The above categories are further divided, and the ratio is calculated as the characteristic influence factor.

[0095] The characteristic influence factors are shown in Table 1:

[0096] Table 1

[0097]

[0098]

[0099] Note: Both costs and revenues have different components. This study explores the main influencing factors, and other factors with very small proportions are not considered. Weight segment 1 represents 0 - 0.5 kg, weight segment 2 represents 0.5 - 1 kg, weight segment 3 represents 1 - 1.5 kg, weight segment 4 represents 1.5 - 2 kg, weight segment 5 represents 2 - 3 kg, weight segment 6 represents 3 - 5 kg, and weight segment 7 represents over 5 kg.

[0100] Data range: The data of the characteristic factors taken in this invention are from January 2022 to July 2024, statistically monthly.

[0101] The single - ticket profit model is a machine - learning regression model; based on the regression model and the characteristic correlation covariance matrix, we obtained the following analysis results for Anhui Province. The probability distribution curve of the single - ticket profit is as Figure 3 , and it can be found that: the distribution of the single - ticket profit over the time line conforms to the normal distribution. The single - ticket profit in Anhui is relatively low, with small profits but quick turnover, and sometimes there may be losses.

[0102] Based on the single - ticket profit model, a characteristic correlation covariance matrix is constructed to obtain the correlation between each characteristic influencing factor and other characteristic influencing factors, as well as the positive and negative influence relationships of each characteristic influencing factor on the single - ticket profit;

[0103] For the characteristic influencing factors X i and X j , its covariance calculation formula is:

[0104]

[0105] where n is the number of samples, X ik and X jk are the values of the characteristics X i and X j on the k - th sample respectively, and are the means of the characteristics X i and X j respectively;

[0106] The covariances between all characteristics are formed into a matrix, that is, the covariance matrix; for m characteristics, the covariance matrix is an m×m matrix, where the element (i, j) is the covariance between the characteristic influencing factors X i and X j . The element values in the covariance matrix reflect the linear correlation between the characteristics; positive values indicate positive correlation, negative values indicate negative correlation, and the larger the absolute value, the stronger the correlation.

[0107] The core factors affecting the profit per order are: the proportion of same-day shipments, the proportion of outbound delivery costs, the proportion of outbound delivery revenue, the proportion of the business volume of branch companies, the proportion of next-day shipments, the proportion of weights for other delivery timings, and the proportion of JD.com shipments; among the costs, the outbound delivery fees, among the revenues, the outbound delivery revenue, among the delivery timings, same-day and next-day deliveries, and among the weights, other delivery timings have the greatest impact on the profit per order in terms of weight.

[0108] The key influencing factors affecting the profit per order are as Figure 4 , and the analysis conclusion is: The core factors affecting the profit per order are: the proportion of same-day shipments, the proportion of outbound delivery costs, the proportion of outbound delivery revenue, the proportion of the business volume of branch companies, the proportion of next-day shipments, the proportion of weights for other delivery timings, and the proportion of JD.com shipments; among the costs, the outbound delivery fees, among the revenues, the outbound delivery revenue, among the delivery timings (delivery distance), same-day and next-day deliveries, and among the weights, other delivery timings (long-time deliveries) have the greatest impact on the profit per order in terms of weight.

[0109] The positive and negative qualitative analysis of the influencing factors is as Figure 5 , and the analysis finds that:

[0110] 1. According to the analysis of the multi-dimensional covariance matrix: greater than 0 indicates a positive correlation (increasing or decreasing together); less than 0 indicates a negative correlation (one increasing and the other decreasing).

[0111] 2. Typical positive correlation factors include: the proportion of outbound transportation capacity costs (that is, excluding other major influencing factors, the higher the proportion of outbound transportation capacity costs in the total cost, the higher the profit per order), and the proportion of the volume (weight) of Pinduoduo shipments, that is, the more Pinduoduo in the business volume, the higher the profit per order.

[0112] 3. Typical negative correlation factors include: the proportion of shipments from other channels (that is, excluding Pinduoduo, JD.com, Taobao and other platforms, the higher the proportion of other shipments, the lower the profit per order). The analysis finds that non-typical positive or negative correlations mean that under normal circumstances, it is a positive or negative correlation, and the influencing factors affect each other and are complex and diverse, and there are special cases, and it is not excluded that a positive correlation shows a negative correlation.

[0113] The fitting curve of the top influencing factors and the profit per order is as Figure 6 :

[0114] (1). The pink one is the distribution of the profit per order. From the perspective of the curve fitting degree, the proportion of same-day shipments has the highest fitting degree, which also reflects the highest correlation.

[0115] (2). At point 1, the profit per order has a significant increase. Looking at each influencing factor, the main reason is that the proportion of same-day shipments increases and the outbound delivery costs rise. Although other influencing factors decrease, it still causes a significant increase in the profit per order.

[0116] (3) At point 3, the profit per order reaches its maximum value. It can be seen that the proportion of same-day delivery volume is also the highest on that day, while the proportion of other delivery time weights (heavy goods), outgoing dispatch fee income, and the business volume of the branch company are at relatively low levels. As the proportion of same-day delivery volume decreases, the proportion of outgoing dispatch fee income and the business volume of the branch company increases, and the profit per order begins to decline.

[0117] (4) At point 5, the profit per order fluctuates greatly. The main reason is the impact of the large fluctuation in outgoing dispatch fee income (negatively correlated).

[0118] (5) At point 6, there should be relatively large changes in the company's orders. The increase in long-distance goods (the proportion of other delivery time weights jumps, negatively correlated) leads to a significant decrease in the proportion of next-day delivery and also a decrease in the proportion of same-day delivery. The proportion of outgoing dispatch fee cost decreases significantly, and the outgoing dispatch fee income increases significantly. Overall, this results in a significant decrease in the profit per order, and then all indicators return to normal.

[0119] (6) Similar analyses can be conducted at other points. Regarding how the proportion of other delivery time weights affects the proportion of next-day delivery, outgoing dispatch fee income, and costs. As for the specific ways in which these costs and incomes affect, it is still necessary to supplement business and data, increase logic and dimensions, further explore and analyze, and use a constructed graph to represent their impact context for convenient implementation and application.

[0120] Construct a graph of profit per order

[0121] Entities are established for the profit per order and its characteristic factors. The characteristic of the profit per order is named bill, and other characteristic factors are named as shown in the English names in Table 1. The graph database uses NebulaGraph, which has distributed functions and relatively comprehensive performance. The process of constructing the graph is as follows:

[0122] (1) Establish entity nodes with characteristics for different institutions, incorporating the institution names into the entity attributes; establish time and institution entity nodes;

[0123] (2) Establish directed edges between the characteristic factors and between them and the profit per order. Normalize the influence degree of the positive and negative edges and write it into the edge attributes; establish directed edges between the characteristic influencing factors and time and institutions;

[0124] (3) Write the entities and relationships into the graph database.

[0125] Based on the graph of profit per order characteristics in the logistics field and the graph traversal algorithm, recall sub- Figure 3 tuple knowledge;

[0126] Based on the sub- Figure 3 tuple knowledge, answer the user's questions and generate answers in natural language form.

[0127] Use Qwen 72B to build a large model, and determine the starting node in the knowledge graph according to the user's question;

[0128] Starting from the starting node, use the depth-first search DFS or breadth-first search BFS algorithm to traverse the knowledge graph to find all triple knowledge related to the starting node; during the traversal process, set the depth or breadth of the traversal as needed to control the quantity and quality of the recalled triple knowledge;

[0129] From the triple knowledge obtained by traversal, filter out the sub- Figure 3 tuple knowledge directly related to the user's question;

[0130] Organize the extracted key information into an answer in natural language form, and present the organized natural language answer to the user.

[0131] Among them, the sub- Figure 3 tuple includes the following situations:

[0132] (1) The sub-graph structure of the core impact factor of the current institution at the current time, used to describe the current situation of the institution's single-ticket profit;

[0133] (2) The sub-graph structure of the core impact factor of the current institution at the current time and the previous period, used to describe and prompt the change of the institution's single-ticket profit;

[0134] (3) The sub-graph structure of the impact factor with a large change between the current institution at the current time and the previous period, used to describe and prompt the impact of the change of this impact factor;

[0135] (4) The sub-graph structure of the core impact factor of the current institution at the current time and similar institutions, used to describe the comparison of the institution's single-ticket profit.

[0136] The terminal device provided by the embodiment of the present invention. The terminal device of this embodiment includes: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the above-mentioned method embodiments are implemented. Or, when the processor executes the computer program, the functions of each module / unit in the above-mentioned device embodiments are implemented.

[0137] The computer program can be divided into one or more modules / units, and the one or more modules / units are stored in the memory and executed by the processor to complete the present invention.

[0138] The terminal device can be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The terminal device may include, but is not limited to, a processor and a memory.

[0139] The processor may be a Central Processing Unit (CPU), or may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0140] The memory can be used to store the computer program and / or module. By running or executing the computer program and / or module stored in the memory, and by invoking the data stored in the memory, the processor implements various functions of the terminal device.

[0141] If the modules / units integrated in the terminal device are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, to implement all or part of the processes in the above-described embodiment methods of the present invention, it can also be completed by a computer program instructing relevant hardware. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-described various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, Read-Only Memory (ROM), Random Access Memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content included in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0142] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An intelligent single-ticket profit analysis method in the logistics field based on knowledge graph, characterized by: include: Classify the collected logistics data information and obtain the characteristic influencing factors of each category on the profit of a single ticket; Based on the single-ticket profit model, the feature correlation covariance matrix is ​​constructed to obtain the correlation between each feature influencing factor and other feature influencing factors and the positive and negative impact relationship of each feature influencing factor on the single-ticket profit; Based on the relationship and changes between knowledge graph representation and single-ticket profit characteristic influencing factors, a single-ticket profit characteristic graph in the logistics field is constructed; Recall subgraph triples knowledge based on single-ticket profit feature graph and graph traversal algorithm in logistics field; The user's questions are answered based on the subgraph triple knowledge, and the answers in natural language are generated.

2. The intelligent single-ticket profit analysis method in the logistics field based on knowledge graph according to claim 1 is characterized in that: The logistics data information includes: cost classification, income classification, weight range, channel shipment volume and delivery time; the logistics data information is subdivided, and the ratio of each subdivision to the sum of its corresponding subdivisions is used as a characteristic influencing factor; The characteristic influencing factors of each category on the single ticket profit include: The logistics delivery volume is divided according to the delivery time, and the delivery volume on the same day, the delivery volume on the next day, the delivery volume on the next day, the delivery volume on the third day, the delivery volume on the fourth day, and the delivery volume on other delivery time are obtained; The weight of logistics delivery is divided according to the delivery time, and the weight proportion of same-day delivery, next-day delivery, next-day delivery, 3-day delivery, 4-day delivery and other time-limited weight proportions are obtained; Divide the business volume according to weight to obtain the business volume proportion of weight segment 1, the business volume proportion of weight segment 2, the business volume proportion of weight segment 3, the business volume proportion of weight segment 4, the business volume proportion of weight segment 5, the business volume proportion of weight segment 6, and the business volume proportion of weight segment 7; The business volume is divided according to the channel shipment volume, and the proportion of Douyin, JD.com, Kuaishou, Pinduoduo, Taobao, total business volume-branch business volume, total business volume-marketing department business volume, total business volume-community business volume and other channel volume are obtained; The business volume is divided according to the channel shipment weight, and the weight share of Douyin, JD.com, Kuaishou, loading rate, Pinduoduo and Taobao are obtained; Divide the business types according to the cost, and obtain the proportion of outbound operation cost subtotal - outbound operation cost, the proportion of outbound dispatch cost subtotal - outbound dispatch fee cost, the proportion of other outbound costs subtotal - outbound management cost, the proportion of provincial sales cost, the proportion of outbound transportation capacity cost and the proportion of incremental incentives; By dividing the business types according to revenue, we obtain the proportion of outbound revenue - outbound transit revenue, the proportion of outbound revenue - other outbound revenue, the proportion of outbound revenue - outbound waybill revenue and the proportion of outbound revenue - outbound delivery fee revenue.

3. The intelligent single-ticket profit analysis method in the logistics field based on knowledge graph according to claim 2 is characterized in that: The weight segment 1 represents 0-0.5kg, the weight segment 2 represents 0.5-1kg, the weight segment 3 represents 1-1.5kg, the weight segment 4 represents 1.5-2kg, the weight segment 5 represents 2-3kg, the weight segment 6 represents 3-5kg, and the weight segment 7 represents more than 5kg.

4. The intelligent single-ticket profit analysis method in the field of logistics based on knowledge graph according to claim 3 is characterized in that: Based on the single-ticket profit model, a feature correlation covariance matrix is ​​constructed to obtain the correlation between each feature influencing factor and other feature influencing factors and the positive and negative impact relationship of each feature influencing factor on the single-ticket profit, specifically: For the characteristic influence factor X i and X j , and its covariance calculation formula is: Where n is the number of samples, X ik and X jk They are feature X i and X j The value at the kth sample, and They are feature X i and X j The mean of The covariances between all features are combined into a matrix, namely the covariance matrix; for m features, the covariance matrix is ​​an m×m matrix, where the element (i, j) is the feature influence factor X i and X j The covariance between them; the element values ​​in the covariance matrix reflect the linear correlation between the features; positive values ​​indicate positive correlation, negative values ​​indicate negative correlation, and the larger the absolute value, the stronger the correlation.

5. The intelligent single-ticket profit analysis method in the logistics field based on knowledge graph according to claim 4 is characterized in that: The core factors affecting the profit of a single ticket are: the proportion of same-day delivery volume, the proportion of outbound delivery fee costs, the proportion of outbound delivery fee income, the proportion of branch business volume, the proportion of next-day delivery volume, the proportion of other time-limited weights, and the proportion of JD.com volume; among the costs, outbound delivery fees, outbound delivery fee income, among the time limits, same-day delivery and next-day delivery, and among the weight, other time limits have the greatest impact on the profit of a single ticket.

6. The intelligent single-ticket profit analysis method in the field of logistics based on knowledge graph according to claim 5 is characterized in that: The relationship and changes between the knowledge graph representation and the influencing factors of the single-ticket profit characteristics are used to construct a single-ticket profit characteristic graph in the logistics field, specifically: Construct indicator entity nodes, organization entity nodes, and time entity nodes; at the same time, assign a unique ID to each indicator entity node and use the indicator name as an attribute; Assign a unique ID to each organization entity node, and use the organization name, type, and establishment time as attributes; assign a unique ID to each time entity node, and use the time value as an attribute; Directed edges are established between characteristic factors and between characteristic factors and single-ticket profits to indicate the influence relationship between them; each edge has positive and negative attributes to indicate the direction of influence, positive or negative; the influence degree is normalized and written into the edge attribute to indicate the intensity of influence; the weight value is between 0 and 1, where 1 indicates the maximum influence and 0 indicates no influence; Establish a relationship between the indicator and the organization entity node, indicating that the indicator is responsible for or generated by the organization in the organization entity node; establish a relationship between the indicator and the time entity node, indicating that the indicator occurs between the time periods in the time entity node; write the above entities and relationships into the graph database to form a single ticket profit characteristic map in the logistics field.

7. The intelligent single-ticket profit analysis method in the field of logistics based on knowledge graph according to claim 6 is characterized in that: The single-ticket profit characteristic graph and graph traversal algorithm in the logistics field recall subgraph triple knowledge, answer the user's question based on the subgraph triple knowledge, and generate answers in natural language form, specifically: Determine the starting node in the knowledge graph based on the user's question; Starting from the starting node, use the depth-first traversal DFS or breadth-first traversal BFS algorithm to traverse the knowledge graph and find all triples of knowledge related to the starting node; during the traversal process, set the depth or breadth of the traversal as needed to control the quantity and quality of the recalled triples of knowledge; From the traversed triple knowledge, filter out the subgraph triple knowledge directly related to the user's question; The extracted key information is organized into answers in natural language form, and the organized natural language answers are presented to the user.

8. Intelligent single-ticket profit analysis system in the logistics field based on knowledge graph, characterized by: include: A classification module, which classifies the collected logistics data information to obtain characteristic influencing factors of each category on the profit of a single ticket; An acquisition module, which constructs a feature correlation covariance matrix based on a single-ticket profit model to obtain the correlation between each feature influencing factor and other feature influencing factors and the positive and negative influence relationship of each feature influencing factor on the single-ticket profit; A construction module, wherein the construction module constructs a single-ticket profit characteristic graph in the logistics field based on the relationship and changes between the knowledge graph representation and the single-ticket profit characteristic influencing factors; A recall module, wherein the recall module recalls subgraph triple knowledge based on a single-ticket profit characteristic graph in the logistics field and a graph traversal algorithm; A generation module is used to answer the user's questions based on the subgraph triple knowledge and generate answers in natural language form.

Images