A Method for Mining the Maximum Propagation Point Set of a Large-Scale Express Logistics Network

Abstract

The present invention discloses a method for mining the maximum propagation point set of a large-scale express logistics network, comprising the following steps: obtaining express logistics network information, extracting the directed information in the express logistics network information, and constructing a directed weighted network according to the directed information; associating the original KPP-Pos model with the directed weighted network, and injecting direction and weight to construct a mining model; optimizing the mining model by using a heuristic algorithm until the mining model outputs the maximum distance proximity, and generating a maximum propagation node set according to the maximum distance proximity. The beneficial effects of the present invention are: by associating the original KPP-Pos model with the directed weighted network, injecting direction and weight to construct a mining model, and finally optimizing the mining model by using a heuristic algorithm, the operation efficiency of the model can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of complex network applications, and particularly to a method for mining the maximum propagation point set of a large-scale express logistics network. Background Art

[0002] Mining the node cities with the maximum propagation effect in the express logistics network can improve the operation efficiency and network accessibility of express logistics, and optimize the layout of logistics infrastructure. At present, the evaluation of the propagation ability of nodes in the network is mainly based on four perspectives: the local attributes, global attributes, location, and random walk of the network. The typical index based on the local attributes of the network is the node degree, which mainly considers the direct connection relationship between the node and its neighboring nodes. Nodes with a high degree value can directly spread more neighboring points. The typical index based on the global attributes of the network is the betweenness centrality, which measures the importance of a node in the propagation process by the number of the shortest paths passing through a certain node in the network. The indicators based on location and random walk, such as the K-core decomposition algorithm and the PageRank node ranking algorithm, etc., are mostly based on the node degree, and the measurement of the node propagation ability is respectively based on whether the node is in the core position of the network and the mutual influence relationship between the node and its neighboring points.

[0003] The above indicators lay a method foundation for mining the key nodes of the logistics network, but there are still the following problems: ① Most of them focus on the comparison of the propagation ability of single nodes in the network, ignoring the role of nodes in realizing the overall maximum propagation of the network. In a complex network, the node scale is large, and the connection paths and directions between nodes show diversified characteristics, which means that the realization of the overall maximum propagation effect of the network often cannot rely on a single node but a set of nodes. ② Even if the node set is concerned, traditional research mostly forms a point set by sequentially picking the optimal nodes based on the quantization results such as the node degree. This is a "single individual optimal" selection criterion, and the obvious problem brought is the "rich club" effect, that is, nodes with higher individual centrality form a specific community, making the most compact connection paths limited within the community and unable to achieve the maximum propagation.

[0004] To address the problems brought about by the above metrics, social science expert Stephen P. Borgatti proposed the Key Player Problem-Positive (KPP-Pos) model to identify a set of key nodes (Nodes-set) that can maximize connections to other nodes. This is an integrated goal-oriented combinatorial optimization model that has been applied in areas such as traffic congestion management, precise web information retrieval, and funding for key patent technologies. Given that the original KPP-Pos model was proposed based on an undirected unweighted network, to meet the requirements of complex weighted network analysis, relevant scholars have optimized issues such as node selection and solution efficiency, node weights, and edge weights in the KPP method. There are still some problems in existing research: ① There are few node mining models for directed networks, ignoring the two-way characteristics of node connections; ② The node mining model for weighted networks is still under exploration, lacking relevant empirical research to evaluate the effectiveness of the model; ③ The algorithm complexity is relatively high, especially in large-scale node networks, lacking practical feasibility due to computational magnitude limitations.

[0005] In summary, based on the characteristics of the KPP-Pos model, it can theoretically identify the "maximum propagation point set" in the express logistics network, thereby providing technical support for the layout of express logistics infrastructure and becoming one of the important means for building logistics infrastructure. However, the algorithm of this model is complex and difficult to implement in large-scale node networks. Summary of the Invention

[0006] In view of the above problems, the present invention proposes a method for mining the maximum propagation point set in a large-scale express logistics network, aiming to solve the problem that the algorithm of the original KPP-Pos model is complex and difficult to implement in large-scale node networks.

[0007] To solve the above technical problems, the first aspect of the present invention proposes a method for mining the maximum propagation point set in a large-scale express logistics network, including the following steps:

[0008] Obtain express logistics network information, extract the directed information in the express logistics network information, and construct a directed weighted network based on the directed information.

[0009] Associate the original KPP-Pos model with the directed weighted network and inject direction and weight to construct a mining model.

[0010] Optimize the mining model using a heuristic algorithm until the mining model outputs the maximum distance proximity, and generate a maximum propagation node set based on the maximum distance proximity.

[0011] In some embodiments, the directed information at least includes the starting city, the ending city, and the number of transportation paths between the starting city and the ending city.

[0012] In some embodiments, an n×n directed weighted network is constructed with the starting cities as columns and the ending cities as rows:

[0013]

[0014] where N is the directed weighted network, and N (i-1)i is the number of transportation paths from the (i - 1)-th starting city node to the i-th ending city node.

[0015] In some embodiments, the mining model includes:

[0016]

[0017] where K is a key node set composed of k nodes selected from the directed weighted network, j is a remaining node set composed of all nodes other than the key node set in the directed weighted network, n is the total number of nodes in the directed weighted network, and w Kj is the distance proximity between the nodes in the key node set and node j, and r Kj is the distance from the nodes in the key node set to node j, and r jK is the distance from node j to the nodes in the key node set, and w is the weight.

[0018] In some embodiments, optimizing the mining model by using a heuristic algorithm includes: selecting the node with the largest node importance value in the directed weighted network as the first member of the key node set, sequentially selecting nodes from the remaining node set to join the key node set, calculating the distance proximity, and filling the key node set with the node having the highest distance proximity until the mining model outputs the maximum distance proximity, and generating the maximum propagation node set according to the maximum distance proximity.

[0019] A second aspect of the present invention proposes a system for mining the maximum propagation point set of a large-scale express logistics network, including:

[0020] A directed weighted network generation module, configured to obtain express logistics network information, extract the directed information in the express logistics network information, and construct a directed weighted network according to the directed information;

[0021] A mining model generation module, configured to associate the original KPP-Pos model with the directed weighted network and inject directions and weights to construct a mining model;

[0022] A maximum propagation node set generation module, configured to optimize the mining model by using a heuristic algorithm until the mining model outputs the maximum distance proximity, and generate a maximum propagation node set according to the maximum distance proximity

[0023] In a third aspect of the present invention, a device for mining the maximum propagation point set of a large-scale express logistics network is proposed. The device includes a memory, a processor, and a communication module, where,

[0024] The memory is used to store executable program codes;

[0025] The processor is coupled to the memory;

[0026] The processor calls the executable program codes stored in the memory and executes the above-mentioned method for mining the maximum propagation point set of a large-scale express logistics network.

[0027] In a fourth aspect of the present invention, a computer-readable storage medium is proposed. The computer-readable storage medium stores computer instructions, which are used to execute the above-mentioned method for mining the maximum propagation point set of a large-scale express logistics network when the computer instructions are called.

[0028] The beneficial effects of the present invention are as follows: By associating the original KPP-Pos model with a directed weighted network, injecting directions and weights to construct a mining model, and finally optimizing the mining model using a heuristic algorithm, the operation efficiency of the model can be effectively improved. Description of the Drawings

[0029] Figure 1 It is a schematic diagram of the express logistics route between the place of dispatch and the place of receipt;

[0030] Figure 2 It is a schematic diagram of the mining steps of the "maximum propagation node set" in the key node set;

[0031] Figure 3 It is a flow chart of the heuristic algorithm for key node mining;

[0032] Figure 4 It is a numerical schematic diagram of DW-KPP-Pos of K-set members and their node combinations;

[0033] Figure 5 It is a schematic diagram of the propagation efficiency of point sets Kdeg, Kpag, and Kbet;

[0034] Figure 6 It is a schematic diagram of the structure of the device for mining the maximum propagation point set of a large-scale express logistics network disclosed in Embodiment 3. Detailed Embodiments

[0035] To make the objectives, technical solutions and advantages of the present invention clearer and more explicit, the content of the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It can be understood that the specific embodiments described herein are only used to explain the present invention, rather than limiting the present invention. Additionally, it should be noted that for the sake of description, only the parts related to the present invention are shown in the drawings rather than all the content.

[0036] Embodiment 1

[0037] This embodiment proposes a method for mining the maximum propagation point set of a large-scale express logistics network. By associating the original KPP-Pos model with a directed weighted network, injecting directions and weights to construct a mining model, and finally using a heuristic algorithm to optimize the mining model, the operation efficiency of the model can be effectively improved.

[0038] This method includes the following steps S1 - S3:

[0039] S1. Obtain the express logistics network information, extract the directed information in the express logistics network information, and construct a directed weighted network according to the directed information.

[0040] The directed information includes at least the starting city, the ending city, and the number of transportation paths between the starting city and the ending city. In this embodiment, the "relationship data" for constructing the directed weighted network is sourced from the China Smart Logistics Network (CSN). The China Smart Logistics Network is a logistics network mainly serving the online retail industry, an important infrastructure for the e-commerce industry, and the largest socialized express logistics platform in China. The CSN network contains express logistics transportation line information such as the starting and ending cities and the number. In this solution, at the shipping line query entry on the official website merchant workbench of CSN, the directed logistics line numbers between various cities in China are retrieved and obtained. For example Figure 1 , there are different transportation paths a - c from the shipping city O to the receiving city D, and the logistics line number is counted as 3. This line can accurately represent the express logistics line connection between cities. Considering the periodic variation of the lines, to avoid contingency, five batches of data are collected between July 5th and 9th, 2020, and their average value is taken. The preprocessing of the data includes removing some statistically repeated lines, same-city lines, and lines without site tracking. Finally, 111,366 pairs of starting and ending points (O - D) combinations between cities are actually obtained, totaling 21,430,731 directed lines.

[0041] The network matrix can directly reflect the flow direction of express logistics and the weight characteristics of the flow lines, which is a necessary link for node mining and network structure analysis, and also the basis for network drawing and the analysis of the spatial pattern of logistics elements. This solution constructs a directed weighted express logistics network based on the principle of graph theory. In this network structure, each city in China is a node, the directed express logistics line between cities is an edge, and the number of lines is the weight. Specifically, an n×n directed weighted network is constructed with the starting city as the column and the ending city as the row:

[0042]

[0043] In the formula, N is the directed weighted network, N (i-1)i is the number of transportation paths from the (i - 1)-th starting city node to the i-th ending city node. For example, N 1i is the number of express logistics lines starting from the first city to the i-th city; N i1 is the number of express logistics lines starting from the i-th city to the first city, and so on.

[0044] S2. Associate the original KPP-Pos model with the directed weighted network and inject direction and weight to construct a mining model.

[0045] The original KPP-Pos model is used to mine a set of node sets that can connect all nodes with the minimum distance. That is, given an unweighted and undirected network G, find a set of k nodes to form a node set K, which connects the remaining nodes with the minimum distance, or it can also be understood as connecting the remaining nodes along the maximum propagation path. The original formula of the classic KPP-Pos is based on an unweighted and undirected network for measurement, and the formula is as follows (Equation 3):

[0046]

[0047] Among them, K is the key node set composed of k selected nodes; j is any node in the remaining node set; d Kj is the minimum distance from any member of K to node j, and the reciprocal is taken to standardize the measurement index, so that D R ∈[0, 1]; D R is the target value, which can be regarded as the weighted proportion of the set reaching all nodes. The larger the D R value, the smaller the distance from the current key node set to all the remaining nodes and the maximum propagation efficiency under the condition that the number of nodes is restricted to k. The distance between the key node of the original KPP-Pos formula and itself is 1 (i.e., d ii = 1, i ∈ K), which is inconsistent with the fact that the distance from a node to itself in graph theory is equal to 0. And the construction of the network in this application is based on the principle of graph theory, taking d ii= 0, that is, the connection of a node itself in the key node set is removed, and the denominator of the above formula changes from n to n - k (n > k). In addition, considering that the Chinese express logistics network is a directed weighted network, the directionality of the connection between nodes i and j is considered based on the above original KPP-Pos formula, and the weight w is added.

[0048] Therefore, to solve the problem of huge computational complexity of complex nodes, this embodiment optimizes the original KPP-Pos model through the following two steps.

[0049] The first step: Consider the directionality between nodes, that is, the express logistics starting from a node in the key node set and sent to node j is different from the express logistics starting from node j and sent to the nodes in the key node set, thus forming a bidirectional symmetric matrix. In addition, the improvement of this solution to the KPP-Pos model will be based on one-step connectivity. When there is two-step or more connectivity between cities, then d Kj = ∞, and the principle of taking the reciprocal of the distance to standardize the measurement in formula (3) can be simplified here, and the formula is as follows:

[0050]

[0051] where r Kj is the distance between the node in the key node set and node j; r jK is the distance between node j and the node in the key node set.

[0052] The second step: Add the weight w to scientifically measure the tightness between the node in the key node set and node j, which can also be understood as the distance proximity, to replace the simple setting of 0 / 1 indicating whether the node connection exists or not. The weighted operator WD R is as follows: The finally obtained mining model includes:

[0053]

[0054] where K is the key node set composed of k nodes selected in the directed weighted network, j is the remaining node set composed of all nodes except the key node set in the directed weighted network, n is the total number of nodes in the directed weighted network, w Kj is the distance proximity between the node in the key node set and node j, represented by the number of express logistics lines between two points after standardization, to ensure that WD R ∈ [0, 1]; w jK Similarly. r Kj is the distance between the node in the key node set and node j, r jK is the distance between node j and the node in the key node set, and w is the weight.

[0055] The above formula (4) is defined as the DW-KPP-Pos (Directed Weighted-Key Players Problem-Positive) model, that is, the mining model referred to in this embodiment.

[0056] The principle of the above DW-KPP-Pos node mining model is relatively simple, that is, to find a set of nodes that are connected to all the remaining nodes through the shortest distance path, so as to achieve the maximum propagation effect. However, in the actual operation process, due to the need to repeatedly reorganize nodes and reconstruct the relationships between nodes, the calculation difficulty is relatively high, and the feasibility is relatively low when mining the key point set of a large-scale complex network. Therefore, this embodiment also designs a heuristic algorithm to reduce the complexity of WDR solution, as follows in step S3.

[0057] S3. Optimize the mining model using a heuristic algorithm until the mining model outputs the maximum distance proximity, and generate the maximum propagation node set according to the maximum distance proximity.

[0058] Optimizing the mining model using a heuristic algorithm includes: selecting the node with the largest point importance value in the directed weighted network as the first member of the key node set, sequentially selecting nodes from the remaining node set to join the key node set, and calculating the distance proximity. Select the node with the highest distance proximity to fill the key node set until the mining model outputs the maximum distance proximity, and generate the maximum propagation node set according to the maximum distance proximity. The specific idea is as Figure 2 follows: ① First, calculate the point degrees of all n nodes in the network, and select the node with the largest point degree as the first member of the key node set. Assume that the node with the largest point degree is node 1 in the following figure. ② The idea of selecting the second node is: sequentially select nodes from the remaining (n-1) nodes to join the key node set, calculate the WD R value, and select the node corresponding to the highest value to fill the K set. That is, select the key node set binary groups (1, 2), (1, 3), (1, 4)... (1, 9) formed by node 1 and the remaining nodes in the following figure. If the WD R value is the largest when the binary combination is (1, 3), then node 3 becomes the second member of the key node set. ③ The mining method of other nodes in the key node set is the same, until a "maximum propagation node set" that can connect all nodes with the shortest distance is formed. The flow of the heuristic algorithm is as Figure 3 shown.

[0059] The DW-KPP-Pos model provided by this application improves the model operation efficiency by designing a heuristic algorithm. Taking the express logistics network of Chinese cities as an example for empirical analysis, by comparing the node analysis results of the DW-KPP-Pos model with those of centrality, PageRank algorithm, and betweenness centrality, the effectiveness of the model for mining logistics network nodes is tested. The experimental results show that: (1) The DW-KPP-Pos model based on the heuristic algorithm can improve the efficiency of mining complex network nodes. As Figure 4 shown, the "maximum propagation point set" of the Chinese express logistics network includes Shanghai, Chongqing, Guangzhou, Beijing, Jinhua City, Zhejiang Province, and the Hong Kong Special Administrative Region. The point set K composed of the above 6 cities connects 100% of the external nodes with the minimum distance and the most complete range. In addition, Chinese express logistics has an obvious hierarchical diffusion effect. Most of the nodes in the K set are cities with a large population and large economic scale in China, with strong logistics radiation power, attracting cities outside the K set to overcome distance friction to form the maximum propagation link. (2) The point set K mined by the DW-KPP-Pos model based on the "collective optimal principle" has the highest propagation efficiency. Under the same node quantity constraint, the key node set is relative to the point degree point set K deg selected sequentially based on the "individual optimal principle", the PageRank point set K pag and the betweenness centrality point set K bet , and the propagation efficiencies are 0.59%, 0.88%, and 6.19% higher respectively, as Figure 5 shown. In summary, the proposed DW-KPP-Pos model is applicable to the mining of the maximum propagation point set in large-scale, directed weighted logistics networks, and can provide a method reference for the layout of express logistics infrastructure, etc.

[0060] Example Two

[0061] This example proposes a system for mining the maximum propagation point set of a large-scale express logistics network, including:

[0062] A directed weighted network generation module, which is used to obtain express logistics network information, extract the directed information in the express logistics network information, and construct a directed weighted network according to the directed information;

[0063] A mining model generation module, which is used to associate the original KPP-Pos model with the directed weighted network, and inject direction and weight to construct a mining model;

[0064] A maximum propagation node set generation module, which is used to optimize the mining model by using a heuristic algorithm until the mining model outputs the maximum distance proximity, and generate a maximum propagation node set according to the maximum distance proximity.

[0065] For the data processing methods of the directed weighted network generation module, mining model generation module, and maximum propagation node set generation module in this embodiment, please refer to S1-S3 in Embodiment 1, which will not be elaborated here.

[0066] Embodiment 3

[0067] Please refer to Figure 6 , Figure 6 which is a schematic structural diagram of another large-scale express logistics network maximum propagation point set mining device disclosed in the embodiments of the present invention. As Figure 6 shown, the control platform may include:

[0068] A memory 301 storing executable program code;

[0069] A processor 302 coupled to the memory 301;

[0070] The processor 302 calls the executable program code stored in the memory 301 to execute the steps in the large-scale express logistics network maximum propagation point set mining method described in Embodiment 1.

[0071] Embodiment 4

[0072] The embodiments of the present invention disclose a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to execute the steps in the large-scale express logistics network maximum propagation point set mining method described in Embodiment 1.

[0073] The device embodiments described above are only illustrative. The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed to multiple network modules. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative labor.

[0074] Through the specific descriptions of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, and the storage medium includes read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc memories, magnetic disk memories, tape memories, or any other computer-readable medium that can be used to carry or store data.

[0075] Finally, it should be noted that: What is disclosed in a method for mining the maximum propagation point set of a large-scale express logistics network disclosed in the embodiments of the present invention is only the preferred embodiments of the present invention, which are only used to illustrate the technical solutions of the present invention, rather than to limit it; Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: They can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; And these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the foregoing embodiments.

[0076] The above embodiments are only for explaining the technical concept and features of the present invention, and their purpose is to enable ordinary technical personnel in the art to understand the content of the present invention and implement it accordingly, and cannot be used to limit the protection scope of the present invention. Any equivalent changes or modifications made according to the essence of the content of the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for mining the maximum propagation point set of a large-scale express logistics network, characterized in that, Including the following steps: Obtain express logistics network information, extract the directed information in the express logistics network information, and construct a directed weighted network according to the directed information; Associate the original KPP-Pos model with the directed weighted network, and inject direction and weight to construct a mining model; Optimize the mining model using a heuristic algorithm until the mining model outputs the maximum distance proximity, and generate a maximum propagation node set according to the maximum distance proximity; The mining model includes: Among them, K is the key node set composed of k nodes selected from the directed weighted network, j is the remaining node set composed of all nodes in the directed weighted network except the key node set, n is the total number of nodes in the directed weighted network, w Kj is the distance proximity between the nodes in the key node set and node j, r Kj is the distance from the nodes in the key node set to node j, r jK is the distance from node j to the nodes in the key node set, and w is the weight; The optimizing the mining model using a heuristic algorithm includes: selecting the node with the largest point importance value in the directed weighted network as the first member of the key node set, sequentially selecting nodes from the remaining node set to join the key node set, calculating the distance proximity, and filling the key node set with the node having the highest distance proximity until the mining model outputs the maximum distance proximity, and generating the maximum propagation node set according to the maximum distance proximity.

2. The method for mining the maximum propagation point set of a large-scale express logistics network according to claim 1, characterized in that, The directed information at least includes the starting city, the ending city, and the number of transportation paths between the starting city and the ending city.

3. The method for mining the maximum propagation point set of a large-scale express logistics network according to claim 2, wherein, Construct an n×n directed weighted network with the starting city as the column and the ending city as the row: In the formula, N is a directed weighted network, N (i-1)i is the number of transportation paths from the (i - 1)-th starting city node to the i-th ending city node.

4. A system for mining the maximum propagation point set of a large-scale express logistics network, characterized in that, Including: A directed weighted network generation module, configured to obtain express logistics network information, extract the directed information in the express logistics network information, and construct a directed weighted network according to the directed information; A mining model generation module, configured to associate the original KPP-Pos model with the directed weighted network, and inject direction and weight to construct a mining model; A maximum propagation node set generation module, configured to optimize the mining model using a heuristic algorithm until the mining model outputs the maximum distance proximity, and generate a maximum propagation node set according to the maximum distance proximity; The mining model includes: Among them, K is the key node set composed of k nodes selected from the directed weighted network, j is the remaining node set composed of all nodes in the directed weighted network except the key node set, n is the total number of nodes in the directed weighted network, w Kj is the distance proximity between the nodes in the key node set and node j, r Kj is the distance from the nodes in the key node set to node j, r jK is the distance from node j to the nodes in the key node set, and w is the weight; The optimizing the mining model using a heuristic algorithm includes: selecting the node with the largest point importance value in the directed weighted network as the first member of the key node set, sequentially selecting nodes from the remaining node set to join the key node set, calculating the distance proximity, and filling the key node set with the node having the highest distance proximity until the mining model outputs the maximum distance proximity, and generating the maximum propagation node set according to the maximum distance proximity.

5. A device for mining the maximum propagation point set of a large-scale express logistics network, characterized in that, The device includes a memory, a processor, and a communication module, wherein The memory is used to store executable program code; The processor is coupled to the memory; The processor calls the executable program code stored in the memory to execute the method for mining the maximum propagation point set of a large-scale express logistics network according to any one of claims 1-3.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions, which when called, are used to execute the method for mining the maximum propagation point set of a large-scale express logistics network according to any one of claims 1-3.

Images