A method for analyzing indirect interaction relationships among trading entities based on intermediary subgraphs

Abstract

The present invention discloses a method for analyzing indirect interactive relationships between trading entities based on media subgraphs. The method combines a trade database and logistics transport records to construct a weighted directed graph of trade logistics, converts it into a weighted undirected graph or unidirectional graph of trade logistics, extracts media subgraphs from the weighted undirected graph or unidirectional graph of trade logistics, obtains a total set of media subgraphs, traverses and calculates the generalized competition and homogeneous functional relationship strength of each media subgraph in the total set of media subgraphs, constructs a strength threshold, compares the generalized competition and homogeneous functional relationship strength of each media subgraph after normalization, and obtains a coexistence association pattern. Based on the pattern and strength, two visualization tools, node-to-node pair space and generalized competition homogeneous functional space, are constructed. The present invention can effectively reveal the implicit indirect relationships between trading entities, quantitatively analyze the relationship strength, identify coexistence association patterns, and provide support for structural analysis and strategy optimization of trade logistics networks.

Description

Technical Field

[0001] The present invention relates to the field of trade logistics network analysis, and in particular to a method for analyzing indirect interaction relationships between trade entities based on a media subgraph. Background Art

[0002] In trade logistics network (TLNS) analysis, uncovering the indirect interactions between trading entities is crucial for understanding network structure and optimizing resource allocation. Traditional research has focused on direct relationships between nodes, such as bilateral trade flows, but lacks systematic analysis of indirect relationships formed through intermediary nodes. Existing methods for extracting indirect relationships based on network subgraphs have yet to be developed, making it difficult to fully characterize the broad competitive relationships between trading entities arising from shared intermediary nodes.

[0003] Furthermore, existing methods fail to consider the relative importance of interactions between nodes, lack a means to quantify the strength of indirect relationships, and lack a framework for identifying coexistence patterns. This hinders in-depth analysis of the implicit interaction mechanisms between trading entities in the network. To address these shortcomings, an analytical method is urgently needed that can effectively extract indirect relationships, quantify relationship strength, and identify node coexistence patterns. This method can fill the technical gap in the study of indirect relationships in trade logistics networks and provide theoretical support for trade strategy formulation and network optimization. To this end, we propose a method for analyzing indirect interactions between trading entities based on intermediary subgraphs. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for analyzing indirect interaction relationships among trading entities based on a media subgraph, so as to solve the problems raised in the above-mentioned background technology.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] A method for analyzing indirect interaction relationships among trading entities based on a media subgraph includes the following steps:

[0007] S1. Combine the trade database and logistics transportation records to construct a weighted directed trade logistics graph. Based on the research strategy, transform the weighted directed trade logistics graph into a weighted undirected trade logistics graph or a weighted unidirectional trade logistics graph.

[0008] S2. Extracting media subgraphs from the weighted undirected trade logistics graph or the weighted unidirectional trade logistics graph to obtain a total set of media subgraphs;

[0009] S3. traverse each media subgraph in the total set of media subgraphs, and calculate the generalized competitive relationship strength and homogeneous functional relationship strength corresponding to each media subgraph;

[0010] S4. Construct thresholds for the strength of generalized competitive relationships and homogeneous functional relationships, normalize the strength of generalized competitive relationships and homogeneous functional relationships corresponding to each media subgraph, and compare the normalized strength of generalized competitive relationships and homogeneous functional relationships corresponding to each media subgraph with their respective strength thresholds to obtain the coexistence association pattern corresponding to each media subgraph.

[0011] S5. Based on the coexistence association pattern, generalized competition relationship strength, and homogeneous functional relationship strength corresponding to each media subgraph, a node-node pair space and a generalized competition homogeneous functional space are constructed for the total set of media subgraphs.

[0012] Preferably, the method for constructing a weighted directed graph of trade logistics by combining the trade database and logistics transport records is:

[0013] Extract the bilateral trade flow and volume between various trading entities in the trade database and logistics transport records, and use each trading entity including regions, enterprises, cities and ports as nodes , Node number. Each trading entity has a unique node number. , according to the bilateral trade flow between each trading entity, build a slave node To Node The directed edges connect the trading entities as nodes. The directed edges include and , Representation node and nodes The trade flow is node To Node , Representation node and nodes The trade flow is node To Node The direction of the directed edge strictly follows the trade flow, and the bilateral trade volume between each trading entity is used as the edge weight. The edge weight includes and , Representation node To Node The size of trade volume, Representation node To Node The trade volume is large, and the directed edges are marked with corresponding edge weights to obtain the weighted directed graph of trade logistics. , The node set is composed of all nodes in the weighted directed graph of trade logistics , including various trading entities, It is the edge set consisting of all directed edges in the weighted directed graph of trade logistics, including the trade flow and connections between various trading entities. for , contains the edge weights corresponding to all directed edges in the trade logistics weighted directed graph, where For directed edges The corresponding edge weight.

[0014] Preferably, the method for converting the trade logistics weighted directed graph into the trade logistics weighted undirected graph or the trade logistics weighted unidirectional graph according to the research strategy is:

[0015] If the research strategy is an undirected strategy, then the trade logistics weighted directed graph Transformed into a weighted undirected graph of trade logistics , traverse the weighted directed graph of trade logistics Each node pair in ,node and satisfy and , for each node pair Two nodes in and Directed edges that exist and The corresponding edge weight and , by combining the two-way edge weight formula, we can get The corresponding undirected edge weight , each node in the trade logistics weighted directed graph is Two nodes in and Directed edges that exist and Delete and replace with an undirected edge , Corresponding node pairs Bidirectional interaction, and annotate the corresponding undirected edge weights , For node pairs The total interaction volume of trade logistics is obtained based on this weighted undirected graph , Weighted directed graph for trade logistics in , remain unchanged, is the set of all undirected edges in the trade logistics weighted undirected graph, is the set of undirected edge weights corresponding to all undirected edges in the trade logistics weighted undirected graph;

[0016] The formula for merging bidirectional edge weights is:

[0017]

[0018] in, is an undirected edge The weight of and are slave nodes To Node The trade volume and slave nodes To Node the size of trade volume;

[0019] If the research strategy is a unidirectional strategy, then the trade logistics weighted directed graph Transformed into a weighted one-way graph of trade logistics , traverse the weighted directed graph of trade logistics Each node pair in ,node and satisfy and , for each node pair Two nodes in and Directed edges that exist and The corresponding edge weight and The traffic asymmetry index formula is used to calculate the traffic asymmetry index of each node. The normalized difference in bidirectional flow symmetry of ;

[0020] The result interval is , the closer the value is to 1, the and The more trade volume between same direction;

[0021] The closer , it indicates that the node and The more trade volume between same direction;

[0022] The closer to 0, the better the node and The trade volume between Direction and The closer the size in the direction, the closer the distance;

[0023] Filter out node pairs that do not have a single directionality and whose normalized difference in the symmetry of each bidirectional flow is less than 0.5 and greater than -0.5 , delete each node pair that does not have a single direction from the trade logistics weighted directed graph Two nodes in and Directed edges that exist and At the same time, filter out the node pairs with a single direction whose normalized difference of bidirectional traffic symmetry is less than or equal to 0.5 or greater than or equal to -0.5. , for each node pair with a single direction Two nodes in and Directed edges that exist and The corresponding edge weight and The flow difference formula between node pairs is used to calculate the flow difference between each node pair. The corresponding one-way edge weight , reflecting the two nodes and The net trade volume in the dominant direction is calculated by pairing each node with a single direction in the weighted directed graph of trade logistics. Two nodes in and Directed edges that exist and Delete and replace with a one-way edge or , and mark the corresponding one-way edge weight , we get the weighted one-way graph of trade logistics , Weighted directed graph for trade logistics in , remain unchanged, is the set of all unidirectional edges in the trade logistics weighted unidirectional graph, The set of one-way edge weights corresponding to all one-way edges in the trade logistics weighted one-way graph;

[0024] The flow asymmetry index formula is:

[0025] ;

[0026] in, is the normalized difference in bidirectional traffic symmetry, and are slave nodes To Node The trade volume and slave nodes To Node the size of trade volume;

[0027] The formula for the traffic difference between node pairs is:

[0028] ;

[0029] in, and are slave nodes To Node The trade volume and slave nodes To Node The size of trade volume, and are slave nodes To Node The net trade volume and slave nodes To Node the size of net trade;

[0030] The undirected strategy does not distinguish the direction of the edge and emphasizes the total scale of interaction between nodes;

[0031] The unidirectional strategy is to analyze the inflow and outflow conditions of the node.

[0032] Preferably, the method of extracting media subgraphs from the trade logistics weighted undirected graph to obtain a total set of media subgraphs comprises the following steps:

[0033] S011. From the weighted undirected graph of trade logistics Filter node degree Nodes greater than or equal to 2 , that is, the mediating point;

[0034] S012. From the weighted undirected graph of trade logistics Extract the neighbor node set of the median point ,in express The jth neighbor node of the neighbor node set is combined into pairs to obtain the media-related node pair , the number of media-related node pairs is Based on this, a set of media-related node pairs is constructed. ;

[0035] S013. Each media-associated node pair in the media-associated node pair set , respectively with the media point Combine to form a set of media point subgraph nodes At the same time, according to the set of media point subgraph nodes , from the weighted undirected graph of trade logistics Extract the two undirected edges corresponding to the mediating point subgraph node set and and the corresponding edge weights and , get the edge set of the media point subgraph and the edge weight set of the media point subgraph , construct the media subgraph , is the lth undirected mediator subgraph of the ith mediator point, is the lth subgraph node set of the ith intermediate node, is the lth subgraph edge set of the ith intermediate point, is the set of edge weights of the lth subgraph of the ith intermediate node;

[0036] S014. Traversing the weighted undirected graph of trade logistics Each media point , for each intermediate point Repeat S011, S012, and S013 in sequence to obtain all media subgraphs corresponding to each media point: , Represents the subgraph set of the i-th media point, and merges all the media subgraphs corresponding to all media points to obtain the total set of media subgraphs ;

[0037] The media subgraph includes a media point subgraph node set, a media point subgraph edge set and a media point subgraph edge weight set, and the media point subgraph node set includes media points and media-related node pairs.

[0038] Preferably, the method of extracting media subgraphs from the weighted one-way graph of trade logistics to obtain a total set of media subgraphs comprises the following steps:

[0039] S021. From the weighted one-way graph of trade logistics Filter the nodes that connect two nodes as the starting point As a medium point, that is, existence Make and ; or as the end point connected by two nodes as the intermediate point, that is, there is Make and ;

[0040] S022. From the weighted one-way graph of trade logistics Extract the neighbor node set of the median point ,in 'express The jth neighbor node of the neighbor node set is combined into pairs to obtain the media-related node pair , the number of media-related node pairs is indivual, For nodes The out-degree or in-degree of ;

[0041] S023. Each media-associated node pair in the media-associated node pair set , respectively with the media point Combine to form a set of media point subgraph nodes At the same time, according to the set of media point subgraph nodes , from the weighted one-way graph of trade logistics Extract the two undirected edges corresponding to the mediating point subgraph node set and and the corresponding edge weights and , or two undirected edges and and the corresponding edge weights and , get the edge set of the media point subgraph and the edge weight set of the media point subgraph ,or and the edge weight set of the media point subgraph , construct the media subgraph , is the lth undirected mediator subgraph of the ith mediator point, is the lth subgraph node set of the ith intermediate node, is the lth subgraph edge set of the ith intermediate point, is the set of edge weights of the lth subgraph of the ith intermediate node;

[0042] S024. Traversing the weighted one-way graph of trade logistics Each media point , for each intermediate point Repeat S021, S022, and S023 in sequence to obtain all media subgraphs corresponding to each media point: , Represents the subgraph set of the i-th media point, and merges all the media subgraphs corresponding to all media points to obtain the total set of media subgraphs .

[0043] Preferably, the method for calculating the strength of the generalized competitive relationship and the strength of the homogeneous functional relationship corresponding to each media subgraph is:

[0044] The calculation of the generalized competitive relationship strength corresponding to each media subgraph is used to quantify the competitive strength of the media-related nodes for the media point resources, based on the relative importance and similarity of the media point to the two nodes. The calculation of the homogeneous functional relationship strength corresponding to each media subgraph is used to quantify the evaluation of the functional similarity of the media point to the two related nodes, based on the relative importance and similarity of the related nodes to the media point.

[0045] For each media subgraph, the comprehensive importance CI formula and the proportional similarity CS formula are used to calculate respectively, and the comprehensive importance degree of the generalized competitive relationship and the proportional similarity degree of the generalized competitive relationship in the media subgraph are obtained. The comprehensive importance degree of the generalized competitive relationship and the proportional similarity degree of the generalized competitive relationship in the media subgraph are calculated by the GCR intensity formula to obtain the generalized competitive relationship intensity;

[0046] The comprehensive importance CI formula is:

[0047] ;

[0048] in, is the comprehensive importance of the generalized competitive relationship in the media subgraph, and They are respectively and The edge weight of an arbitrary connecting edge, i is the media point number, j and k are the associated node numbers of the media point, l is the media subgraph number, and They are respectively and The total weight of all possible connecting edges, It represents the media subgraph type;

[0049] The proportional similarity CS formula is:

[0050] ;

[0051] in, is the similarity degree of the generalized competitive relationship ratio, and They are respectively and The edge weight of any connecting edge, i is the media point number, j and k are the associated node numbers of the media point, l is the media subgraph number, and They are respectively and The total weight of all possible connecting edges, It represents the medium subgraph type;

[0052] The GCR intensity formula is:

[0053] ;

[0054] in, is the generalized competitive relationship strength of the media subgraph, is the similarity degree of the generalized competitive relationship ratio, is the comprehensive importance of the generalized competitive relationship in the media subgraph;

[0055] For each media subgraph, the comprehensive importance HI formula and the proportional similarity HS formula are used to calculate respectively, and the comprehensive importance degree of homogeneous functional relationships and the proportional similarity degree of homogeneous functional relationships in the media subgraph are obtained. The comprehensive importance degree of homogeneous functional relationships and the proportional similarity degree of homogeneous functional relationships in the media subgraph are calculated by the HFR intensity formula to obtain the homogeneous functional relationship strength;

[0056] The comprehensive importance HI formula is:

[0057] ;

[0058] in, is the comprehensive importance of homogeneous functional relationships in the media subgraph, and They are respectively and The edge weight of any connecting edge, i is the media point number, j and k are the associated node numbers of the media point, l is the media subgraph number, Indicates the media point The total weight of all possible connecting edges, It represents the medium subgraph type;

[0059] The proportional similarity HS formula is:

[0060] ;

[0061] in, is the degree of similarity of the proportion of homogeneous functional relationships, and They are respectively and The edge weight of any connecting edge, i is the media point number, j and k are the associated node numbers of the media point, l is the media subgraph number, It represents the medium subgraph type;

[0062] The HFR intensity formula is:

[0063] ;

[0064] in, is the strength of the homogeneous functional relationship representing the media subgraph, is the degree of similarity of the proportion of homogeneous functional relationships, It is the comprehensive importance of homogeneous functional relationships in the media subgraph.

[0065] Preferably, the method for constructing the generalized competitive relationship strength threshold and the homogeneous functional relationship strength threshold is:

[0066] The generalized competitive relationship strength corresponding to each media subgraph in the total set of media subgraphs and homogeneous functional relationship strength Arrange in descending order respectively, represents the generalized competitive relationship strength of the nth intermediary subgraph of the lth intermediary point, Representing the homogeneous functional relationship strength of the nth media subgraph of the lth media point, we get the descending series of the generalized competitive relationship strength and the descending series of the homogeneous functional relationship strength;

[0067] If the number of elements n in the descending sequence of the strength of generalized competition relationship and homogeneous functional relationship is an odd number, then the threshold value of the strength of generalized competition relationship is is the mid-th generalized competitive relationship strength in the middle position in the descending sequence of generalized competitive relationship strength, and the homogeneous functional relationship strength threshold The midth homogeneous functional relationship strength in the middle position of the descending sequence of homogeneous functional relationship strengths, the value of mid is (n+1) / 2 and is rounded down;

[0068] If the number of elements n in the descending sequence of the strength of generalized competition relationship and homogeneous functional relationship is an even number, then the threshold value of the strength of generalized competition relationship is is the average value of the mid1 and mid2 generalized competitive relationship strengths in the middle of the descending series of generalized competitive relationship strengths. The homogeneous functional relationship strength threshold It is the average value of the homogeneous functional relationship strengths of the two middle positions mid1 and mid2 in the descending sequence of homogeneous functional relationship strengths, where the value of mid1 is n / 2 and the value of mid2 is n / 2+1.

[0069] Preferably, the method for obtaining the coexistence association pattern corresponding to each media subgraph is:

[0070] The generalized competitive relationship strength and homogeneous functional relationship strength corresponding to each media subgraph are calculated using the normalization formula respectively, so that the generalized competitive relationship strength and homogeneous functional relationship strength corresponding to each media subgraph are scaled to the [0,1] interval, and the normalized generalized competitive relationship strength corresponding to each media subgraph is obtained. and homogeneous functional relationship strength ;

[0071] The normalized generalized competitive relationship strength corresponding to each media subgraph and homogeneous functional relationship strength Compare with their respective intensity thresholds;

[0072] like and , then the corresponding coexistence association mode is a stable bidirectional dependence mode;

[0073] like and , then the corresponding coexistence association mode is the association node dominant mode;

[0074] like and , then the corresponding coexistence association mode is the bidirectional independent edge mode;

[0075] like and , then the corresponding coexistence association mode is the media node monopoly mode;

[0076] The normalization formula is:

[0077] ;

[0078] in, is the normalized value, is the maximum value, is the minimum value, is the value to be normalized.

[0079] Preferably, the method for constructing the node-node pair space and the generalized competitive homogeneous function space for the total set of medium subgraphs is:

[0080] For each media subgraph in the total set of media subgraphs, the media points in the media point subgraph node set of each media subgraph are used as the horizontal coordinates, and the media-related node pairs in the media point subgraph node set of each media subgraph are used as the vertical coordinates. Each media subgraph is mapped to a unique corresponding coordinate position on the rectangular coordinate system, and the corresponding coexistence association pattern, normalized generalized competition relationship intensity and homogeneous functional relationship intensity are marked on the coordinate position. Based on this, a node and node pair space NNPS is constructed. The constructed node and node pair space is a two-dimensional table. The rows represent media-related node pairs, and the columns represent media points. Each cell corresponds to a media subgraph, which contains its relationship strength and pattern information. It intuitively displays: the node pairs connected by media points, the difference in relationship strength between different node pairs at the same media point, and the distribution characteristics of the coexistence pattern in the media point to node pair dimension;

[0081] For each media subgraph in the total set of media subgraphs, the normalized generalized competition relationship intensity corresponding to each media subgraph is used as the horizontal coordinate, and the normalized homogeneous function relationship intensity corresponding to each media subgraph is used as the vertical coordinate. Each media subgraph is mapped to a unique corresponding coordinate position on the rectangular coordinate system, and a generalized competition homogeneous function space CHS is constructed accordingly. The constructed generalized competition homogeneous function space is a two-dimensional scatter plot. The generalized competition homogeneous function space can be divided into four subspaces of high-high, high-low, low-high and low-low based on the generalized competition relationship intensity threshold and the homogeneous function relationship intensity threshold, representing four types of node coexistence association patterns. At the same time, the kernel density estimation method can be used to create a probability distribution map for the generalized competition homogeneous function space in order to analyze the overall distribution characteristics of the media subgraph in the CHS.

[0082] The kernel density estimation is a non-parametric estimation method used to estimate the probability density function of a random variable;

[0083] The node-node pair space and the generalized competitive homogeneous functional space are visualization tools for analyzing and displaying the indirect relationship between nodes in the trade logistics network TLNS. The visualization tools refer to the probability distribution diagram, the two-dimensional scatter plot and the two-dimensional table.

[0084] Compared with the prior art, the present invention has the following beneficial effects:

[0085] 1. The present invention constructs an indirect relationship analysis framework, filling a technological gap. Existing technologies mostly focus on direct relationships between trading entities, and lack a systematic analysis of indirect relationships formed through third-party intermediary points. The present invention introduces the intermediary subgraph as the smallest structural unit, processes the original network through an undirected or unidirectional strategy, and systematically extracts the triple structure mediated by intermediary points. For the first time, it constructs an extraction and analysis framework for indirect relationships in trade logistics networks, filling a technological gap in indirect relationship research. It can fully reveal the broad competitive relationships and homogeneous functional relationships between trading entities due to shared intermediary points, and deepen the understanding of the implicit interaction mechanism of the network.

[0086] 2. This paper proposes a relationship strength measurement method based on relative importance. Unlike existing technologies that rely on the limitation of absolute values, this method accurately quantifies the strength of indirect relationships through a joint evaluation of comprehensive importance and proportional similarity, combined with natural logarithm correction. This method takes into account both local interactions and the overall network structure, considers the product of the relative importance of media points to associated nodes in generalized competitive relationships, and evaluates the similarity of the functional contributions of associated nodes to media points in homogeneous functional relationships. It can more accurately reflect the actual degree of dependence and functional association between nodes, providing a more scientific quantitative tool for complex network analysis.

[0087] 3. The present invention can identify coexistence association patterns and reveal deep interaction mechanisms. Based on the strong and weak combination of generalized competition and homogeneous functional relationships, it defines four coexistence association patterns for the first time, namely stable two-way dependence, association node dominance, two-way independent edge, and intermediary node monopoly. The pattern boundaries are divided by the median threshold, realizing the classification and identification of dependence, dominance, edge, and monopoly relationships between nodes. This model framework goes beyond simple binary relationship analysis and can reveal the role positioning of trading entities in the network, providing a new perspective for understanding the hierarchical structure and dynamic interaction of trade logistics networks.

[0088] 4. This invention provides a multidimensional visualization tool that supports intuitive analysis and decision-making, and constructs two visualization strategies: NNPS and CHS. NNPS uses a two-dimensional table to accurately locate the node composition, relationship strength, and pattern of each media subgraph, suitable for detailed analysis of small-scale networks. CHS uses scatter plots and kernel density estimation to display relationship strength distribution and pattern clustering, supporting macro-pattern analysis of large-scale networks. The two tools complement each other, converting abstract indirect relationships into intuitive visualization results, facilitating researchers and decision makers to quickly identify key nodes, core relationships, and potential risks, providing an efficient analytical tool for trade strategy formulation and resource allocation optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0089] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0090] Figure 1 It is a flow chart of the steps of the present invention. DETAILED DESCRIPTION

[0091] To make the objectives, technical solutions, and advantages of the present invention more apparent, the technical solutions of the present invention will be described in detail below. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other implementations obtained by those of ordinary skill in the art without inventive effort are within the scope of protection of the present invention.

[0092] Examples, such as Figure 1 As shown, a method for analyzing indirect interaction relationships between trading entities based on a media subgraph includes the following steps:

[0093] S001. Combine the trade database and logistics transportation records to construct a weighted directed trade logistics graph. Based on the research strategy, transform the weighted directed trade logistics graph into a weighted undirected trade logistics graph or a weighted unidirectional trade logistics graph.

[0094] S002. Extract media subgraphs from the weighted undirected trade logistics graph or the weighted unidirectional trade logistics graph to obtain a total set of media subgraphs;

[0095] S003. Traverse each media subgraph in the total set of media subgraphs and calculate the generalized competitive relationship strength and homogeneous functional relationship strength corresponding to each media subgraph;

[0096] S004. Construct a threshold for the strength of a generalized competitive relationship and a threshold for the strength of a homogeneous functional relationship. Normalize the strength of the generalized competitive relationship and the strength of the homogeneous functional relationship corresponding to each media subgraph, and compare the normalized strength of the generalized competitive relationship and the strength of the homogeneous functional relationship corresponding to each media subgraph with the respective strength thresholds to obtain the coexistence association pattern corresponding to each media subgraph.

[0097] S5. Based on the coexistence association pattern, generalized competition relationship strength, and homogeneous functional relationship strength corresponding to each media subgraph, a node-node pair space and a generalized competition homogeneous functional space are constructed for the total set of media subgraphs.

[0098] Furthermore, the working principle of the present invention is described below by way of examples:

[0099] Assuming that the crude oil trade data of major regions in the world in 2023 is taken as an example, the specific implementation process of the method of the present invention is described, involving nodes including region A 、Region B 、Region C 、Region D and Region E .

[0100] Extract bilateral trade flows and volumes from the UN Comtrade database and construct a weighted directed graph of trade logistics ,in ,

[0101] Include 、 、 etc., W contains 、 For example, if region A exports 1 million tons of crude oil to region C, then the weighted directed graph of trade logistics is There is an edge ,and is 100, and region C exports 00,000 tons of crude oil to region A. Then the weighted directed graph of trade logistics is There is an edge ,and is 0; Region B exports 500,000 tons of crude oil to Region C, then the trade logistics weighted directed graph There is an edge ,and is 50, and region C imports 100,000 tons of crude oil from region B. Then the weighted directed graph of trade logistics is There is an edge ,and =10; Taking the unidirectional strategy as an example, the weighted directed graph of trade logistics is transformed into a weighted unidirectional graph of trade logistics, and the flow asymmetry index is calculated. , For example, =(100-0)\(100+0), which is 1, because 1 is greater than 0.5, retaining one-way edges , The weight is 1 million tons, =(50-10)\(50+10), which is about 0.67, because 0.67 is greater than 0.5, retaining one-way edges , The weight is 400,000 tons, based on which the trade logistics weighted one-way graph , Include 、 Equal one-way edges;

[0102] Taking the undirected network as an example, assuming the trade logistics weighted unidirectional graph Node in Region C The neighbor nodes of 、 and , node degree If 3 is greater than 2, then confirm As the medium point, , the node pair obtained by combination is , and , forming a set of media-related node pairs , with node pairs For example, the node set , edge set , weight set , assuming The weight is 600,000 tons, based on which the media subgraph is constructed.

[0103] Calculate the relationship strength of the media subgraph and the generalized competitive relation GCR of the media subgraph. The total trade volume is 1.5 million tons. Under the condition that the total trade volume is 1.1 million tons, is about 0.36, is about 0.82, for , about -1.22; for the homogeneous functional relationship HFR of the media subgraph, under the assumption Under the condition that the total trade volume is 2 million tons, is 0.15, is 0.75, for , about -2.10.

[0104] Assuming the media point The GCR intensity range of all media subgraphs is [-2, 0], and the HFR intensity range is [-3, -1]. After normalization, is 0.45, is 0.39. Assuming that the threshold of the strength of the generalized competitive relationship is 0.3 and the threshold of the strength of the homogeneous functional relationship is 0.4, then Greater than 0.4 and If it is greater than 0.3, it is judged to be a stable two-way dependence model, that is, region A and region B form resource competition and functional dependence through region C.

[0105] Construct NNPS, with the horizontal axis representing the median point , the vertical axis is the node pair , for coordinates The GCR is marked as 0.39 and the HFR is marked as 0.45, indicating a stable two-way dependence pattern. The CHS is constructed with the GCR value of 0.39 on the horizontal axis and the HFR value of 0.45 on the vertical axis, falling into the first quadrant of high GCR and high HFR, corresponding to a stable two-way dependence pattern. The probability density map shows that the data points in this area are dense, indicating the importance of oil transactions between Region C and Region A, and between Region B and Region C, to the crude oil market of Region C.

[0106] The above embodiments strictly follow the logic of the claims, fully demonstrating the entire process from data modeling to pattern recognition, and reflecting the specific application and technical advantages of the present invention in actual trade networks. The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Modifications to the technical solutions described in the above embodiments, or equivalent replacements of some of the technical features therein, do not deviate from the essence of the corresponding technical solutions within the scope of the technical solutions of the embodiments of this application, and should be included within the scope of protection of this application.

Claims

1. A method for analyzing indirect interaction relationships between trading entities based on a media subgraph, characterized by: The following steps are involved: S1. Combine the trade database and logistics transportation records to construct a weighted directed trade logistics graph. Based on the research strategy, transform the weighted directed trade logistics graph into a weighted undirected trade logistics graph or a weighted unidirectional trade logistics graph. S2. Extracting media subgraphs from the weighted undirected trade logistics graph or the weighted unidirectional trade logistics graph to obtain a total set of media subgraphs; S3. traverse each media subgraph in the total set of media subgraphs, and calculate the generalized competitive relationship strength and homogeneous functional relationship strength corresponding to each media subgraph; S4. Construct thresholds for the strength of generalized competitive relationships and homogeneous functional relationships, normalize the strength of generalized competitive relationships and homogeneous functional relationships corresponding to each media subgraph, and compare the normalized strength of generalized competitive relationships and homogeneous functional relationships corresponding to each media subgraph with their respective strength thresholds to obtain the coexistence association pattern corresponding to each media subgraph. S5. Based on the coexistence association pattern, generalized competition relationship strength, and homogeneous functional relationship strength corresponding to each media subgraph, a node-node pair space and a generalized competition homogeneous functional space are constructed for the total set of media subgraphs.

2. The method for analyzing indirect interaction relationships between trading entities based on a media subgraph according to claim 1, characterized in that: The method of combining trade database and logistics transportation records to construct a trade logistics weighted directed graph: Extract the bilateral trade flow and bilateral trade volume between each trading entity in the trade database and logistics transportation records, take each trading entity as a node, and construct directed edges to connect each trading entity as a node according to the bilateral trade flow between each trading entity. And use the bilateral trade volume between each trading entity as the edge weight, mark the directed edges with the corresponding edge weight, and obtain a trade logistics weighted directed graph. .

3. The method for analyzing indirect interaction relationships between trading entities based on a media subgraph according to claim 2, characterized in that: According to the research strategy, the method of converting the trade logistics weighted directed graph into the trade logistics weighted undirected graph or the trade logistics weighted unidirectional graph is as follows: If the research strategy is an undirected strategy, the trade logistics weighted directed graph is converted into a trade logistics weighted undirected graph. Each node pair in the trade logistics weighted directed graph is traversed, and the edge weights corresponding to the directed edges between the two nodes in each node pair are calculated by combining the bidirectional edge weight formula to obtain the undirected edge weights corresponding to each node pair. The directed edges between the two nodes in each node pair in the trade logistics weighted directed graph are replaced with undirected edges, and the corresponding undirected edge weights are marked to obtain the trade logistics weighted undirected graph. ; If the research strategy is a unidirectional strategy, the trade logistics weighted directed graph is converted into a trade logistics weighted unidirectional graph. Each node pair in the trade logistics weighted directed graph is traversed, and the edge weights corresponding to the directed edges between the two nodes in each node pair are calculated using the flow asymmetry index formula to obtain the normalized difference in the bidirectional flow symmetry of each node pair. The node pairs without unidirectionality whose normalized difference in bidirectional flow symmetry is less than 0.5 and greater than -0.5 are screened, and the two nodes in each node pair without unidirectionality are deleted from the trade logistics weighted directed graph. At the same time, the node pairs with single directionality whose normalized difference of bidirectional flow symmetry is less than or equal to 0.5 or greater than or equal to -0.5 are screened. The edge weights corresponding to the directed edges between the two nodes in each node pair with single directionality are calculated by the flow difference formula between the node pairs to obtain the unidirectional edge weights corresponding to each node pair with single directionality. The directed edges between the two nodes in each node pair with single directionality in the trade logistics weighted directed graph are replaced with unidirectional edges, and the corresponding unidirectional edge weights are marked to obtain the trade logistics weighted unidirectional graph. ; The undirected strategy does not distinguish the direction of the edge and emphasizes the total scale of interaction between nodes; The unidirectional strategy is to analyze the inflow and outflow conditions of the node.

4. The method for analyzing indirect interaction relationships between trading entities based on a media subgraph according to claim 3 is characterized in that: The method for extracting media subgraphs from a weighted undirected trade logistics graph to obtain a total set of media subgraphs comprises the following steps: S011. From the weighted undirected graph of trade logistics Filter nodes with a degree greater than or equal to 2 , that is, the media point ; S012. From the weighted undirected graph of trade logistics Extracting media points The set of neighbor nodes , for the set of neighbor nodes The neighbor nodes in the pair are combined to obtain the media-related node pairs, and the media-related node pair set is constructed based on this. ; S013. Associate the media node pair set Each media-associated node pair in Combine to form a set of media point subgraph nodes At the same time, according to the set of media point subgraph nodes , from the weighted undirected graph of trade logistics Extracting the media point subgraph node set The corresponding media point subgraph edge set and the edge weight set of the media point subgraph , construct the media subgraph ; S014. Traversing the weighted undirected graph of trade logistics For each media point, repeat S011, S012 and S013 in sequence to obtain all media subgraphs corresponding to each media point , merge all media subgraphs corresponding to all media points to obtain the total set of media subgraphs ; The media subgraph includes a media point subgraph node set, a media point subgraph edge set and a media point subgraph edge weight set, and the media point subgraph node set includes media points and media-related node pairs.

5. The method for analyzing indirect interaction relationships between trading entities based on a media subgraph according to claim 3 is characterized in that: The method for extracting media subgraphs from the weighted one-way trade logistics graph to obtain a total set of media subgraphs includes the following steps: S021. From the weighted one-way graph of trade logistics Filter nodes that connect two nodes as the starting point or are connected by two nodes as the end point , that is, the media point ; S022. From the weighted one-way graph of trade logistics Extracting media points The set of neighbor nodes , the neighbor nodes in the neighbor node set are combined into pairs to obtain the media-related node pairs, and the media-related node pair set is constructed accordingly ; S023. Associate the media node pair set Each media-related node pair in is combined with the media point to form a media point subgraph node set At the same time, according to the set of media point subgraph nodes , from the weighted one-way graph of trade logistics Extracting the media point subgraph node set The corresponding media point subgraph edge set and the edge weight set of the media point subgraph , construct the media subgraph ; S024. Traversing the weighted one-way graph of trade logistics For each media point, repeat S021, S022 and S023 in sequence to obtain all media subgraphs corresponding to each media point , merge all media subgraphs corresponding to all media points to obtain the total set of media subgraphs .

6. The method for analyzing indirect interaction relationships between trading entities based on a media subgraph according to claim 1, characterized in that: The method for calculating the strength of the generalized competitive relationship and the strength of the homogeneous functional relationship corresponding to each media subgraph is: For each media subgraph, the comprehensive importance CI formula and the proportional similarity CS formula are used to calculate respectively, and the comprehensive importance degree of the generalized competitive relationship and the proportional similarity degree of the generalized competitive relationship in the media subgraph are obtained. The comprehensive importance degree of the generalized competitive relationship and the proportional similarity degree of the generalized competitive relationship in the media subgraph are calculated by the GCR intensity formula to obtain the generalized competitive relationship intensity; The comprehensive importance HI formula and the proportional similarity HS formula are used to calculate each media subgraph respectively to obtain the comprehensive importance degree of homogeneous functional relationships and the proportional similarity degree of homogeneous functional relationships in the media subgraph. The comprehensive importance degree of homogeneous functional relationships and the proportional similarity degree of homogeneous functional relationships in the media subgraph are calculated using the HFR intensity formula to obtain the homogeneous functional relationship strength.

7. The method for analyzing indirect interaction relationships between trading entities based on a media subgraph according to claim 6, characterized in that: The method for constructing the generalized competitive relationship strength threshold and the homogeneous functional relationship strength threshold: Arrange all the generalized competition relationship intensities and all the homogeneous functional relationship intensities corresponding to each media subgraph in the total set of media subgraphs in descending order, and obtain a descending sequence of generalized competition relationship intensities and a descending sequence of homogeneous functional relationship intensities; If the number of elements in the descending order of the strength of generalized competition relationship and homogeneous functional relationship is odd, then the threshold value of the strength of generalized competition relationship is is the generalized competitive relationship strength in the middle position in the descending series of generalized competitive relationship strength, and the homogeneous functional relationship strength threshold The strength of the homogeneous functional relationship in the middle position in the descending sequence of the strength of the homogeneous functional relationship; If the number of elements in the descending order of the strength of generalized competition relationship and homogeneous functional relationship is even, then the threshold value of the strength of generalized competition relationship is is the average value of the generalized competitive relationship strength in the two middle positions in the descending series of generalized competitive relationship strength, and the homogeneous functional relationship strength threshold It is the average value of the homogeneous functional relationship strength in the middle two positions in the descending sequence of homogeneous functional relationship strength.

8. The method for analyzing indirect interaction relationships between trading entities based on a media subgraph according to claim 7 is characterized in that: The method for obtaining the coexistence association pattern corresponding to each media subgraph: The generalized competitive relationship strength and homogeneous functional relationship strength corresponding to each media subgraph are calculated using the normalization formula respectively, so that the generalized competitive relationship strength and homogeneous functional relationship strength corresponding to each media subgraph are scaled to the interval [0, 1], and the normalized generalized competitive relationship strength corresponding to each media subgraph is obtained. and homogeneous functional relationship strength ; The normalized generalized competitive relationship strength corresponding to each media subgraph and homogeneous functional relationship strength Compare with their respective intensity thresholds; like and , then the corresponding coexistence association mode is a stable bidirectional dependence mode; like and , then the corresponding coexistence association mode is the association node dominant mode; like and , then the corresponding coexistence association mode is the bidirectional independent edge mode; like and , then the corresponding coexistence association mode is the media node monopoly mode.

9. The method for analyzing indirect interaction relationships between trading entities based on a media subgraph according to claim 8, characterized in that: The method for constructing the node-node pair space and the generalized competitive homogeneous function space for the total set of medium subgraphs is as follows: For each media subgraph in the total set of media subgraphs, use the media points in the media point subgraph node set of each media subgraph as the horizontal coordinate and the media-related node pairs in the media point subgraph node set of each media subgraph as the vertical coordinate. Map each media subgraph to a unique corresponding coordinate position on the rectangular coordinate system, and annotate the corresponding coexistence association pattern, generalized competitive relationship strength, and homogeneous functional relationship strength on the coordinate position, and construct a node and node pair space based on this. For each media subgraph in the total set of media subgraphs, the strength of the generalized competitive relationship corresponding to each media subgraph is used as the horizontal coordinate, and the strength of the homogeneous functional relationship corresponding to each media subgraph is used as the vertical coordinate. Each media subgraph is mapped to a unique corresponding coordinate position on the rectangular coordinate system, and a generalized competitive homogeneous functional space is constructed accordingly. The node-node pair space and the generalized competitive homogeneous functional space are visualization tools for analyzing and displaying the indirect relationships between nodes in the trade logistics network TLNS.

Images