Industrial chain risk prediction method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention provides an industrial chain risk prediction method and device, electronic equipment and a storage medium, and belongs to the technical field of artificial intelligence. The method comprises the steps that product supply relation data of an industrial chain is acquired, an initial industrial chain relation graph is constructed, the relation graph comprises initial nodes, and the initial nodes are simplex; grouping the simplex in the initial industry chain relation graph according to the dimensions of the simplex to obtain simplex combinations, and then constructing an inter-dimension dependency matrix according to the simplex combination of each dimension and the adjacent simplex combinations; and obtaining a target industry chain relation graph according to a preset sample risk event node, the initial node and the inter-dimension dependency matrix. And training a preset initial risk prediction model according to the target industry chain relation graph to obtain a target risk prediction model. And finally performing risk prediction on the target risk event through the target risk prediction model. According to the embodiment of the invention, the accuracy of risk prediction of an industrial chain can be improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, is applicable to the financial field, and in particular, relates to a risk prediction method and device, electronic equipment and storage medium for an industrial chain. Background Art

[0002] In related technologies, graph convolutional networks (GCN) are usually used to learn multi-node network graphs that represent industrial chain relationships, and a risk prediction model is obtained after training to evaluate the potential impact of risk events on different objects in the industrial chain (such as raw material suppliers, manufacturers, distributors, and customers, etc.).

[0003] In fact, there are often many-to-many or many-to-one complex dimensional interactive relationships between nodes in a complex industrial chain, and the multi-node network diagram is mainly manifested as a simple binary relationship between nodes and edges, which cannot effectively represent the complex interactive relationships in the industrial chain, thereby affecting the accuracy of risk prediction.

[0004] Therefore, how to improve the accuracy of risk prediction in the industrial chain has become a technical problem that needs to be solved urgently. Summary of the invention

[0005] The main purpose of the embodiments of the present application is to propose a risk prediction method and device, electronic device and storage medium for an industrial chain, aiming to improve the accuracy of risk prediction for the industrial chain.

[0006] To achieve the above-mentioned purpose, a first aspect of an embodiment of the present application proposes a risk prediction method for an industrial chain, the method comprising:

[0007] Obtain product supply relationship data of at least two objects in the target industrial chain;

[0008] An initial industrial chain relationship graph is constructed based on the product supply relationship data of each object; wherein the initial industrial chain relationship graph includes at least two initial nodes, and the at least two initial nodes form at least two simplexes;

[0009] Grouping the simplexes in the initial industrial chain relationship diagram according to the dimensions of the simplexes to obtain at least two simplex combinations;

[0010] Constructing diffusion relationships based on the simplex combination and adjacent simplex combinations in each dimension to obtain an inter-dimensional dependency matrix; wherein the adjacent simplex combination is a combination adjacent to the dimension of the simplex combination;

[0011] According to the preset sample risk event node, the initial node and the inter-dimensional dependency matrix, the sample risk event node is embedded in the initial industrial chain relationship diagram to obtain a target industrial chain relationship diagram;

[0012] Performing model training on a preset initial risk prediction model according to the target industrial chain relationship diagram to obtain a target risk prediction model;

[0013] Obtain a target risk event, and perform risk prediction on the target risk event using the target risk prediction model.

[0014] In some embodiments, the adjacent simplex combination includes a first adjacent simplex combination and a second adjacent simplex combination, and the dimension of the first adjacent simplex combination is smaller than the dimension of the second adjacent simplex combination; the step of constructing a diffusion relationship based on the simplex combination and the adjacent simplex combination in each dimension to obtain an inter-dimensional dependency matrix includes:

[0015] Performing matrix calculation according to the diffusion relationship between the simplex combination and the first adjacent simplex combination, and the diffusion relationship between the simplex combination and the second adjacent simplex combination, to obtain an inter-dimensional dependency submatrix of the simplex combination;

[0016] The inter-dimensional dependency matrix is ​​obtained by fusing the inter-dimensional dependency sub-matrices of the various simplex combinations.

[0017] In some embodiments, performing matrix calculation according to the diffusion relationship between the simplex combination and the first adjacent simplex combination, and the diffusion relationship between the simplex combination and the second adjacent simplex combination to obtain the inter-dimensional dependency submatrix of the simplex combination includes:

[0018] Construct a relationship matrix of the diffusion of the simplex combination to the first adjacent simplex combination to obtain a first boundary matrix;

[0019] Construct a relationship matrix from which the second adjacent simplex combination diffuses to the simplex combination to obtain a second boundary matrix;

[0020] Performing a matrix transposition on the first boundary matrix to obtain a first co-boundary matrix;

[0021] Performing a matrix transposition on the second boundary matrix to obtain a second co-boundary matrix;

[0022] An inter-dimensional dependency submatrix of the simplex combination is obtained according to the first boundary matrix, the first co-boundary matrix, the second boundary matrix and the second co-boundary matrix.

[0023] In some embodiments, embedding the sample risk event node into the initial industrial chain relationship diagram according to the preset sample risk event node, the initial node and the inter-dimensional dependency matrix to obtain the target industrial chain relationship diagram includes:

[0024] Calculating the similarity between the sample risk event node and each of the initial nodes to obtain initial node similarity data;

[0025] Selecting candidate nodes from the initial nodes according to the initial node similarity data, and connecting the sample risk event node with the candidate node to obtain a candidate connection edge;

[0026] Determine an initial propagation node sequence according to the sample risk event node, the candidate node and the candidate connection edge;

[0027] The initial industrial chain relationship diagram is updated according to the initial propagation node sequence and the inter-dimensional dependency matrix to obtain the target industrial chain relationship diagram.

[0028] In some embodiments, updating the initial industrial chain relationship graph according to the initial propagation node sequence and the inter-dimensional dependency matrix to obtain the target industrial chain relationship graph includes:

[0029] Selecting a starting propagation node from the initial propagation nodes of the initial propagation node sequence;

[0030] Determine the neighbor nodes of the initial propagation node according to the inter-dimensional dependency matrix, calculate the similarity between the neighbor nodes and each of the initial propagation nodes, and obtain candidate node similarity data;

[0031] Selecting a target node from the initial propagation nodes according to the candidate node similarity data, connecting the target node with the neighbor node, and updating the initial propagation node sequence according to the connection edge between the target node and the neighbor node to obtain a target propagation node sequence;

[0032] The initial industrial chain relationship graph is updated according to the target propagation node sequence to obtain the target industrial chain relationship graph.

[0033] In some embodiments, the model training of a preset initial risk prediction model according to the target industrial chain relationship diagram to obtain a target risk prediction model includes:

[0034] The inter-dimensional dependency matrix is ​​updated according to the target industrial chain relationship diagram to obtain a target dependency matrix;

[0035] Integrate node relationships according to the target dependency matrix and each node in the target industry chain relationship graph to obtain sample graph structural features;

[0036] The parameters of the initial risk prediction model are adjusted according to the structural characteristics of the sample graph to obtain the target risk prediction model.

[0037] In some embodiments, the obtaining of a target risk event and performing risk prediction on the target risk event using the target risk prediction model include:

[0038] Constructing an original risk node based on the target risk event;

[0039] The initial industrial chain relationship graph is updated according to the original risk nodes to obtain the target graph structure characteristics;

[0040] Perform risk prediction on the target graph structure characteristics according to the target risk prediction model to obtain a reference risk characteristic value of each node in the target industry chain relationship graph;

[0041] Selecting a target risk node according to the reference risk characteristic value;

[0042] According to the degree of dependence between the target risk node and the associated risk nodes, the connection edges associated with the target risk node are screened to obtain target risk edges;

[0043] Risk prediction information is determined based on the target risk node and the target risk edge.

[0044] To achieve the above-mentioned purpose, a second aspect of an embodiment of the present application proposes a risk prediction device for an industrial chain, the device comprising:

[0045] A data acquisition module, used to acquire product supply relationship data of at least two objects in the target industrial chain;

[0046] A relationship graph construction module, used to construct an initial industry chain relationship graph based on the product supply relationship data of each object; wherein the initial industry chain relationship graph includes at least two initial nodes, and the at least two initial nodes form at least two simplexes;

[0047] A grouping module, used for grouping the simplexes in the initial industrial chain relationship diagram according to the dimensions of the simplexes to obtain at least two simplex combinations;

[0048] A matrix construction module, used for constructing diffusion relationships according to the simplex combination and adjacent simplex combination of each dimension to obtain an inter-dimensional dependency matrix; wherein the adjacent simplex combination is a combination adjacent to the dimension of the simplex combination;

[0049] A relationship graph updating module, used to embed the sample risk event node into the initial industrial chain relationship graph according to the preset sample risk event node, the initial node and the inter-dimensional dependency matrix, so as to obtain a target industrial chain relationship graph;

[0050] A prediction model training module is used to perform model training on a preset initial risk prediction model according to the target industrial chain relationship diagram to obtain a target risk prediction model;

[0051] The risk prediction module is used to obtain a target risk event and perform risk prediction on the target risk event through the target risk prediction model.

[0052] To achieve the above-mentioned purpose, the third aspect of an embodiment of the present application proposes an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the method described in the first aspect is implemented.

[0053] To achieve the above-mentioned purpose, the fourth aspect of an embodiment of the present application proposes a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method described in the first aspect is implemented.

[0054] The risk prediction method and device, electronic device and storage medium of the industrial chain proposed in this application first construct an initial industrial chain relationship diagram including a simplex structure by acquiring product supply relationship data of different objects in the target industrial chain. Then, the simplex of the initial industrial chain relationship diagram is grouped according to the dimension of the simplex, and the diffusion relationship between the simplex combinations of different dimensions is analyzed to generate an inter-dimensional dependency matrix, so that the inter-dimensional dependency matrix can characterize the high-dimensional interaction characteristics, thereby overcoming the defect that the multi-node network diagram in the background technology cannot effectively reflect the complex dimensional interaction in the complex industrial chain. Subsequently, the preset sample risk event node is embedded in the graph structure to simulate the propagation characteristics of the actual risk event, and the target industrial chain relationship diagram is dynamically updated. Then the initial risk prediction model is trained in combination with the target industrial chain relationship diagram to obtain the target risk prediction model. Finally, the target risk prediction model is used to predict the target risk event, and the risk characteristic value and potential propagation path of each node in the industrial chain are accurately evaluated. Compared with the traditional scheme of risk prediction based on the multi-node network diagram, the method of this embodiment introduces a simplex structure when constructing the industrial chain relationship, which can reflect the complex dimensional interaction relationship between each node in the complex industrial chain. By combining the inter-dimensional dependency matrix to update the structure of the initial industrial chain relationship diagram, the target industrial chain relationship diagram obtained can reflect the impact of the sample risk event node on the target industrial chain. The target risk prediction model trained based on the target industrial chain relationship diagram can accurately capture the characteristics of risk propagation, thereby improving the accuracy of risk prediction for the industrial chain. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 is a flow chart of a risk prediction method for an industrial chain provided in an embodiment of the present application;

[0056] Figure 2 yes Figure 1 Flow chart of step S104 in FIG.

[0057] Figure 3 yes Figure 2 Flow chart of step S201 in FIG.

[0058] Figure 4 yes Figure 1 Flow chart of step S105 in FIG.

[0059] Figure 5 yes Figure 4 Flow chart of step S404 in FIG.

[0060] Figure 6 It is an industrial chain relationship diagram of the risk prediction method of the industrial chain provided in the embodiment of the present application;

[0061] Figure 7 yes Figure 1 Flow chart of step S106 in FIG.

[0062] Figure 8 yes Figure 1 Flow chart of step S107 in FIG.

[0063] Fig. 9 It is a schematic diagram of the structure of a risk prediction device for an industrial chain provided in an embodiment of the present application;

[0064] Fig.10 It is a schematic diagram of the hardware structure of the electronic device provided in the embodiment of the present application. DETAILED DESCRIPTION

[0065] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0066] It should be noted that, although the functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first", "second", etc. in the specification, claims and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0067] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0068] First, some nouns involved in this application are analyzed:

[0069] Artificial intelligence (AI) is a new technical science that studies and develops theories, methods, technologies and application systems for simulating, extending and expanding human intelligence. AI is a branch of computer science that attempts to understand the essence of intelligence and produce a new type of intelligent machine that can respond in a similar way to human intelligence. Research in this field includes robotics, speech recognition, image recognition, natural language processing and expert systems. AI can simulate the information process of human consciousness and thinking. AI is also a theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0070] Natural language processing (NLP): NLP uses computers to process, understand and apply human languages ​​(such as Chinese, English, etc.). NLP is a branch of artificial intelligence and an interdisciplinary subject between computer science and linguistics. It is often referred to as computational linguistics. Natural language processing includes grammatical analysis, semantic analysis, and text understanding. Natural language processing is often used in technical fields such as machine translation, handwritten and printed character recognition, speech recognition and text-to-speech conversion, information intent recognition, information extraction and filtering, text classification and clustering, public opinion analysis and opinion mining. It involves data mining related to language processing, machine learning, knowledge acquisition, knowledge engineering, artificial intelligence research, and linguistic research related to language computing.

[0071] Simplex: In geometry, a simplex or n-simplex is an n-dimensional geometric body similar to a triangle. To be precise, a simplex is the convex hull of a set of (n+1) affine-independent points in some Euclidean space with more than n dimensions. For example, a 0-simplex is a point, a 1-simplex is a line segment, and a 2-simplex is a triangle. A 3-simplex is a tetrahedron, and each face of the tetrahedron is a triangle. A 4-simplex is a pentagon.

[0072] Simplicial Complex: A concept in topology, which refers to a topological object obtained by "gluing" simplexes such as points, line segments, and triangles.

[0073] In related technologies, graph convolutional networks (GCN) are usually used to learn multi-node network graphs that represent industrial chain relationships, and a risk prediction model is obtained after training to evaluate the potential impact of risk events on different objects in the industrial chain (such as raw material suppliers, manufacturers, distributors, and customers, etc.).

[0074] In fact, in complex industrial chains, risk propagation is often not just carried out through simple nodes and edges (such as the relationship between suppliers and manufacturers), because there are often many-to-many or many-to-one complex dimensional interactive relationships between nodes in complex industrial chains. For example, multiple suppliers may affect a manufacturer at the same time, and multiple manufacturers may jointly affect a distributor. The multi-node network graph is mainly expressed as a simple binary relationship between nodes and edges, which cannot effectively represent the complex interactive relationships in the industrial chain. As a result, the graph convolutional network trained based on this graph structure is more inclined to capture the one-to-one connection or overall layout characteristics between nodes, and thus cannot effectively reflect the complex coordination and dependence between multiple nodes, which in turn affects the accuracy of risk prediction.

[0075] Therefore, how to improve the accuracy of risk prediction in the industrial chain has become a technical problem that needs to be solved urgently.

[0076] Based on this, the embodiments of the present application provide a risk prediction method and device for an industrial chain, an electronic device and a storage medium, aiming to improve the accuracy of risk prediction for the industrial chain.

[0077] The risk prediction method and device, electronic device and storage medium for the industrial chain provided in the embodiments of the present application are specifically illustrated through the following embodiments. First, the risk prediction method for the industrial chain in the embodiments of the present application is described.

[0078] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0079] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0080] The risk prediction method for the industrial chain provided in the embodiment of the present application relates to the field of artificial intelligence technology. The risk prediction method for the industrial chain provided in the embodiment of the present application can be applied to the terminal, can also be applied to the server side, and can also be software running in the terminal or the server side. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc.; the server side can be configured as an independent physical server, or it can be configured as a server cluster or a distributed system composed of multiple physical servers, and can also be configured as a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the risk prediction method for the industrial chain, etc., but is not limited to the above forms.

[0081] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0082] Figure 1 is an optional flow chart of the risk prediction method for the industrial chain provided in the embodiment of the present application. Figure 1 The method may include but is not limited to steps S101 to S107.

[0083] Step S101, obtaining product supply relationship data of at least two objects in a target industrial chain.

[0084] Step S102, constructing an initial industry chain relationship diagram based on the product supply relationship data of each object.

[0085] Step S103, grouping the simplexes in the initial industrial chain relationship diagram according to the simplex dimensions to obtain at least two simplex combinations.

[0086] Step S104, constructing diffusion relationships based on simplex combinations and adjacent simplex combinations in each dimension to obtain an inter-dimensional dependency matrix.

[0087] Step S105, embedding the sample risk event node into the initial industrial chain relationship diagram according to the preset sample risk event node, initial node and inter-dimensional dependency matrix to obtain the target industrial chain relationship diagram.

[0088] Step S106, training the preset initial risk prediction model according to the target industry chain relationship diagram to obtain a target risk prediction model.

[0089] Step S107, obtaining a target risk event, and performing risk prediction on the target risk event using a target risk prediction model.

[0090] Steps S101 to S107 shown in the embodiment of the present application first construct an initial industrial chain relationship diagram including a simplex structure by acquiring product supply relationship data of different objects in the target industrial chain. Then, the simplex of the initial industrial chain relationship diagram is grouped according to the dimensions of the simplex, and the diffusion relationship between simplex combinations of different dimensions is analyzed to generate an inter-dimensional dependency matrix, so that the inter-dimensional dependency matrix can characterize high-dimensional interaction characteristics, thereby overcoming the defect that the multi-node network diagram in the background technology cannot effectively reflect the complex dimensional interaction in the complex industrial chain. Subsequently, the preset sample risk event node is embedded in the graph structure to simulate the propagation characteristics of the actual risk event, and the target industrial chain relationship diagram is dynamically updated. Then the initial risk prediction model is trained in combination with the target industrial chain relationship diagram to obtain the target risk prediction model. Finally, the target risk prediction model is used to predict the target risk event, and the risk characteristic value and potential propagation path of each node in the industrial chain are accurately evaluated. Compared with the traditional scheme of risk prediction based on the multi-node network diagram, the method of this embodiment introduces a simplex structure when constructing the industrial chain relationship, which can reflect the complex dimensional interaction relationship between each node in the complex industrial chain. By combining the inter-dimensional dependency matrix to update the structure of the initial industrial chain relationship diagram, the target industrial chain relationship diagram obtained can reflect the impact of the sample risk event node on the target industrial chain. The target risk prediction model trained based on the target industrial chain relationship diagram can accurately capture the characteristics of risk propagation, thereby improving the accuracy of risk prediction for the industrial chain.

[0091] In step S101 of some embodiments, the target industrial chain refers to the industrial chain that currently needs to be risk analyzed. The object refers to the basic subject involved in production, transportation or sales in the industrial chain, such as raw material suppliers, manufacturers, distributors and customers, but not limited to this. Product supply relationship data is data used to clarify the upstream and downstream relationships between various objects related to product supply in the target industrial chain, which can be obtained from data such as contract data, logistics data and transaction records between enterprises.

[0092] In step S102 of some embodiments, the initial industrial chain relationship diagram is a graph structure constructed based on the product supply relationship data of each object in the target industrial chain. In this embodiment, the initial industrial chain relationship diagram is a simple complex graph, including at least two initial nodes, and at least two initial nodes form at least two simplexes. It can be understood that each initial node represents an object in the target industrial chain. If there is a connecting edge between two objects, the connecting edge has a direction, which can represent the product supply or logistics path between different objects. In some embodiments, the weight data of the connecting edge in the initial industrial chain relationship diagram defaults to 1.

[0093] In step S103 of some embodiments, simplexes (such as nodes, edges, faces, etc.) with different dimensions are classified and grouped according to their dimensions to obtain simplex combinations of different dimensions. It should be noted that the simplex elements in the initial industrial chain relationship diagram are not mutually exclusive in the grouping process, that is to say, all initial nodes in the initial industrial chain relationship diagram can be classified as the combination of 0-dimensional simplexes, and all edges can be classified as the combination of 1-dimensional simplexes, even if the two end nodes constituting the edge have been classified as the combination of 0-dimensional simplexes, the connection edges between them are not excluded from being grouped into the combination of 1-dimensional simplexes. Similarly, even if each edge constituting the triangle is grouped into the 1-dimensional simplex combination, all triangles in the initial industrial chain relationship diagram can still be classified into the combination of 2-dimensional simplexes. Exemplarily, the initial industrial chain relationship diagram is a structure in which three nodes of node A, node B and node C are connected end to end in sequence, and the grouping result is to obtain a simplex combination of three dimensions. Wherein, the 0-dimensional simplex combination includes node A, node B and node C. The 1-dimensional simplex combination includes edge AB, edge BC and edge CA. The 2-dimensional simplex combination includes face ABC.

[0094] In the step S104 of some embodiments, it is understandable that adjacent simplex combination is the combination adjacent to the dimension of simplex combination.Particularly, adjacent simplex combination comprises the first adjacent simplex combination and the second adjacent simplex combination, and the dimension of the first adjacent simplex combination is less than the dimension of the second adjacent simplex combination.Inter-dimensional dependency matrix refers to the boundary relationship between different simplex combinations used to describe the industrial chain relationship diagram, and captures conversion, contact and the degree of dependence between the different dimensional simplexes, thereby can be used to describe the interactive relationship between multiple nodes, this matrix can be Laplace matrix, normalized Laplace matrix and Hodge Laplace matrix etc., is not limited to this.

[0095] See also Figure 2 In some embodiments, step S104 may include but is not limited to steps S201 to S202:

[0096] Step S201, performing matrix calculation based on the diffusion relationship between the simplex combination and the first adjacent simplex combination, and the diffusion relationship between the simplex combination and the second adjacent simplex combination, to obtain the inter-dimensional dependency submatrix of the simplex combination.

[0097] Step S202, fusing the inter-dimensional dependency sub-matrices of each simplex combination to obtain an inter-dimensional dependency matrix.

[0098] In step S201 of some embodiments, for each simplex combination of dimension, it is necessary to construct a diffusion relationship between it and two simplex combinations of adjacent dimensions. The diffusion relationship between a simplex combination and a first adjacent simplex combination refers to the diffusion relationship between the current simplex combination and a low-dimensional simplex combination, and the diffusion relationship between a simplex combination and a second adjacent simplex combination refers to the diffusion relationship between the current simplex combination and a high-dimensional simplex combination. The dimensions of the first adjacent simplex combination, the simplex combination, and the second adjacent simplex combination increase successively, and the increment of the dimensions may be equal or different.

[0099] Next, the following embodiments will be used to illustrate the construction process of the inter-dimensional dependency matrix when the characteristics of the Hodge Laplace matrix are used to describe the diffusion relationship between simplex combinations of different dimensions. Figure 3 In this embodiment, the dimensions of the first adjacent simplex combination, the simplex combination, and the second adjacent simplex combination are increased in sequence, and the increment is 1. Step S201 may include but is not limited to steps S301 to S305:

[0100] Step S301, constructing a relationship matrix of a simplex combination diffusing to a first adjacent simplex combination to obtain a first boundary matrix.

[0101] Step S302, constructing a relationship matrix of the second adjacent simplex combination diffused to the simplex combination to obtain a second boundary matrix.

[0102] Step S303: performing matrix transposition on the first boundary matrix to obtain a first common boundary matrix.

[0103] Step S304: performing matrix transposition on the second boundary matrix to obtain a second common boundary matrix.

[0104] Step S305 , obtaining the inter-dimensional dependency submatrix of the simplex combination according to the first boundary matrix, the first co-boundary matrix, the second boundary matrix and the second co-boundary matrix.

[0105] In step S301 of some embodiments, illustratively, the symbol k is used to describe the dimension of the current simplex combination, the dimension of the first adjacent simplex combination is (k-1), and the first boundary matrix is ​​used to describe the mapping relationship from the k-dimensional simplex to the (k-1)-dimensional simplex.

[0106] In step S302 of some embodiments, illustratively, the symbol k is used to describe the dimension of the current simplex combination, then the dimension of the second adjacent simplex combination is (k+1), and the second boundary matrix is ​​used to describe the mapping relationship from the (k+1)-dimensional simplex to the k-dimensional simplex.

[0107] In step S303 of some embodiments, the first co-boundary matrix is ​​a transposed matrix of the first boundary matrix, which is used to describe the mapping relationship from the (k-1)-dimensional simplex to the k-dimensional simplex.

[0108] In step S304 of some embodiments, the second co-boundary matrix is ​​a transposed matrix of the second boundary matrix, which is used to describe the mapping relationship from the k-dimensional simplex to the (k+1)-dimensional simplex.

[0109] In step S305 of some embodiments, the first boundary matrix, the first co-boundary matrix, the second boundary matrix, the second co-boundary matrix, and the inter-dimensional dependency sub-matrix may satisfy the following analytical expression:

[0110]

[0111] Among them, the dimension of the simplex combination is denoted as k, and the first boundary matrix is ​​denoted as B k , the first common boundary matrix is ​​expressed as The second boundary matrix is ​​denoted as B k+1 , the second common boundary matrix is ​​expressed as The inter-dimensional dependency submatrix is ​​represented as L k .

[0112] It should be noted that in the same simple complex graph, when dimension k is the lowest dimension and the highest dimension of the simplex, the specific structure of the corresponding inter-dimensional dependency submatrix will discard the boundary matrix corresponding to the non-existent dimension. Exemplarily, the initial industrial chain relationship graph has 0-dimensional simplex, 1-dimensional simplex and 2-dimensional simplex. For the 0-dimensional simplex combination, there is no dimension lower than 0 dimension, and k is 0 at this time, so the corresponding inter-dimensional dependency submatrix For a 1-dimensional simplex combination, k is 1, so the corresponding inter-dimensional dependency submatrix is For a 2-dimensional simplex combination, there is no dimension higher than 2, in which case k is 2, so the corresponding inter-dimensional dependency submatrix is In addition, when calculating L1 and L2, the boundary matrix B2 that maps the 2D simplex combination to the 1D simplex combination is used, but and The meaning is different. It works on 1-dimensional simplex and is used to describe the connection between edges through triangles, such as how many triangles each edge contains. It works on 2D simplexes and is used to describe the connection relationship between triangles through edges, such as how many edges each triangle shares with itself, or how many common edges two triangles are connected by.

[0113] Steps S301 to S305 shown in the embodiment of the present application accurately describe the topological associations of multidimensional structures in the simplicial complex graph by precisely constructing a relationship matrix between simplices of different dimensions. This is particularly suitable for risk analysis scenarios in complex industrial chains. It can capture dependency paths and propagation effects that cannot be described by single-dimensional analysis in multidimensional data structures, and provides an efficient tool for analyzing the diffusion behavior and dependency relationships of complex networks, thereby significantly improving the accuracy of the model.

[0114] In step S202 of some embodiments, the inter-dimensional dependency sub-matrices corresponding to each dimension are merged to obtain an inter-dimensional dependency matrix representing the entire initial industrial chain relationship diagram. In other embodiments, each inter-dimensional dependency sub-matrix is ​​independent of each other, and the set of all inter-dimensional dependency sub-matrices can be used as the inter-dimensional dependency matrix.

[0115] Steps S201 to S202 shown in the embodiment of the present application effectively integrate the diffusion characteristics and dependencies of simplexes of different dimensions in the initial industrial chain relationship diagram by constructing and fusing the inter-dimensional dependency submatrices. The method of this embodiment provides higher expressive power in modeling topological structures, and also lays a solid foundation for risk propagation analysis, feature extraction, and model optimization, greatly improving the system's prediction accuracy and reliability for complex industrial chain risks.

[0116] In step S105 of some embodiments, according to the preset sample risk event nodes, initial nodes and inter-dimensional dependency matrix, the sample risk event nodes are embedded in the initial industrial chain relationship diagram to obtain the target industrial chain relationship diagram. Specifically, the sample risk event node is the starting point of risk propagation constructed according to the risk event. Risk events refer to events that have an impact or chain reaction on the product supply relationship of the target industrial chain, illustratively, trade strategy adjustments between different regions, logistics delays, price fluctuations and natural disasters, etc., but are not limited to this.

[0117] See also Figure 4 In some embodiments, step S105 may include but is not limited to steps S401 to S404:

[0118] Step S401, calculate the similarity between the sample risk event node and each initial node to obtain initial node similarity data.

[0119] Step S402 , selecting candidate nodes from the initial nodes according to the initial node similarity data, and connecting the sample risk event nodes with the candidate nodes to obtain candidate connection edges.

[0120] Step S403: determining an initial propagation node sequence according to the sample risk event nodes, candidate nodes and candidate connection edges.

[0121] Step S404, updating the initial industrial chain relationship diagram according to the initial propagation node sequence and the inter-dimensional dependency matrix to obtain the target industrial chain relationship diagram.

[0122] In step S401 of some embodiments, it should be noted that when constructing the initial industrial chain relationship diagram, the initial node is assigned initial node features, and each initial node feature includes the inventory level, production capacity, geographical location of the corresponding object, and the degree of dependence with other objects, and is stored in the form of an array. In other embodiments, the degree of dependence between different objects can be set as the weight data of the connecting edges between different nodes. The sample risk event node also has a preset sample node feature. Since the initial industrial chain relationship diagram includes at least two initial nodes, at least two similarity data can be obtained by calculating the similarity between the sample node feature and each initial node feature. The similarity data can be cosine similarity, Euclidean distance or dot product, etc., but is not limited to this.

[0123] In step S402 of some embodiments, a similarity threshold is set for the initial industrial chain relationship graph that is connected to the sample risk event node, and the corresponding initial nodes whose similarity data is greater than the similarity threshold are used as candidate nodes. It can be understood that the candidate node can represent the initial node that is directly affected after the sample risk event node is added to the initial industrial chain relationship graph, and the candidate connection edge used to connect the sample risk event node and the candidate node represents the direction of risk propagation.

[0124] In step S403 of some embodiments, the initial propagation node sequence represents the direct propagation path of the risk event in the target industrial chain. For the node sequence, a word embedding method (such as Word2Vec) in natural language processing is used to learn the representation of the node, so that each node in the initial industrial chain relationship diagram is mapped to a low-dimensional vector space, and these vectors represent the structural characteristics of the node in the diagram.

[0125] In step S404 of some embodiments, after the initial propagation node sequence is determined, the node features in the sequence need to be updated, and the initial nodes adjacent to the candidate nodes in the initial industrial chain relationship diagram are determined according to the inter-dimensional dependency matrix. Then, based on the updated node features, it is determined whether to add new connection edges. The above steps are repeated, and after risk propagation simulation is performed on all nodes that may be affected by risks, the obtained industrial chain relationship diagram is the target industrial chain relationship diagram.

[0126] Steps S401 to S404 shown in the embodiment of the present application, by gradually screening, constructing and updating risk propagation relationships, accurately and dynamically simulate the diffusion path of risk events in the industrial chain, adjust and expand the initial industrial chain relationship diagram, and finally generate a target industrial chain relationship diagram, which can clearly depict the propagation path of risk events in the industrial chain and its impact on upstream and downstream nodes.

[0127] In some embodiments, the initial propagation node sequence is updated by random walk. Figure 5 Step S404 may also include but is not limited to steps S501 to S504:

[0128] Step S501 : selecting a starting propagation node from the initial propagation nodes in the initial propagation node sequence.

[0129] Step S502, determining the neighbor nodes of the initial propagation node according to the inter-dimensional dependency matrix, calculating the similarity between the neighbor nodes and each initial propagation node, and obtaining candidate node similarity data.

[0130] Step S503 , selecting a target node from the initial propagation nodes according to the candidate node similarity data, connecting the target node with the neighboring nodes, and updating the initial propagation node sequence according to the connection edges between the target node and the neighboring nodes to obtain a target propagation node sequence.

[0131] Step S504, updating the initial industrial chain relationship diagram according to the target propagation node sequence to obtain the target industrial chain relationship diagram.

[0132] In step S501 of some embodiments, the nodes in the initial propagation nodes are all initial propagation nodes. In this embodiment, the method of selecting the initial propagation node is random selection. For example, see Figure 6 , Figure 6 This is a feasible industrial chain relationship diagram in the embodiment of this application. Figure 6In the example, the sample risk event node is represented as X, and nodes A, B, C, E, D, and F are the structure of the initial industrial chain relationship graph. When executing steps S401 to S403 in the above embodiment, it is determined that in the initial industrial chain relationship graph, the similarity data between the sample risk event node X and nodes A and B respectively exceeds the similarity threshold, and it is determined to connect the sample risk event node X to nodes A and B respectively, and the candidate connection edges are as follows: Figure 6 It should be noted that for each newly added connection edge, the features of the two nodes connected by the newly added connection edge are updated, and the two nodes will add their own upstream and downstream relationships. For example, the node feature of node A will add "is downstream of node X", and the node feature of the sample risk event node X will add "is upstream of node A".

[0133] At this time, the initial propagation nodes of the initial propagation node sequence include the sample risk event node X, node A and node B. When executing step S501, a node is randomly selected from the sample risk event node X, node A and node B as the starting propagation node.

[0134] In step S502 of some embodiments, in combination Figure 6 In step S501, node A is randomly selected as the starting propagation node. Through the inter-dimensional dependency matrix, the neighbor nodes can be determined to be node D, node E, node F, and sample risk event node X. For node D, the candidate node similarity data between node D and each propagation node in the propagation node sequence is calculated, that is, the similarity between node D and node A, the similarity between node D and node B, and the similarity between node D and sample risk event node X are calculated respectively. For node E and node F, the calculation logic of candidate node similarity is the same as that of node D, which will not be repeated here.

[0135] In step S503 of some embodiments, when the candidate node similarity data exceeds a preset similarity threshold, the corresponding node is used as the target node. Figure 6 As shown, if the similarity between node D and node B exceeds the similarity threshold, node B is taken as the target node, node B is connected to node D, and the initial propagation node sequence is updated, and finally the target propagation node sequence is obtained: [sample risk event node X, node A, node D], and [sample risk event node X, node B, node D].

[0136] In step S504 of some embodiments, the node characteristics of the relevant nodes are updated according to the target propagation node sequence, and steps S501 to S504 are repeated, and all nodes in the initial industrial chain relationship diagram that may be affected by risks are subjected to risk propagation simulation, and the final industrial chain relationship diagram is determined as the target industrial chain relationship diagram.

[0137] Steps S501 to S504 shown in the embodiment of the present application, by gradually screening and constructing the propagation path, determine the starting propagation node from the initial propagation node and expand its influence range, and finally dynamically update the initial industrial chain relationship diagram to generate a target industrial chain relationship diagram that represents the risk propagation. This process screens out nodes that may be affected by the risk based on the inter-dimensional dependency matrix, and then accurately screens out nodes affected by the chain reaction of the risk event through similarity calculation, and then constructs the propagation path, realizing the gradual expansion simulation of the risk impact.

[0138] See also Figure 7 In some embodiments, step S106 includes but is not limited to steps S701 to S703:

[0139] Step S701, updating the inter-dimensional dependency matrix according to the target industry chain relationship diagram to obtain a target dependency matrix.

[0140] Step S702, integrating node relationships according to the target dependency matrix and each node in the target industry chain relationship graph to obtain sample graph structural features.

[0141] Step S703, adjusting the parameters of the initial risk prediction model according to the sample graph structural characteristics to obtain a target risk prediction model.

[0142] In step S701 of some embodiments, each time a connection edge is added, the structure of the industrial chain relationship diagram will change, and the corresponding combination of simplexes will also be adjusted accordingly, so it is necessary to update the inter-dimensional dependency matrix, and the updated inter-dimensional dependency matrix is ​​the target dependency matrix. The process of constructing the target dependency matrix is ​​the same as the process of constructing the inter-dimensional dependency matrix, which will not be repeated here.

[0143] In step S702 of some embodiments, different from the node feature, the sample graph structure feature of this embodiment is a vector. The sample graph structure feature satisfies the following analytical expression:

[0144]

[0145] Among them, i and j represent node sequences, f i represents the node feature of node i, f j represents the node feature of node j, A ijRepresents the elements of the adjacency matrix, which is used to describe whether there is a connection edge between node i and node j. If there is a connection edge, A ij is 1 if there is no connecting edge, A ij is 0. L represents the target dependency matrix, f represents the sample graph structural features, and f Τ is the transposed matrix of f.

[0146] In step S703 of some embodiments, the risk characteristic values ​​of all nodes in the target industrial chain relationship graph are initialized, and the risk characteristic values ​​of the sample risk event nodes are set to a higher value. For example, the risk characteristic values ​​of the sample risk event nodes are set to be greater than 0, and the risk characteristic values ​​of all nodes except the sample risk event nodes are set to 0. The initial risk prediction model propagates the high risk characteristic values ​​of the sample risk event nodes in the target industrial chain relationship graph according to the structural characteristics of the sample graph, and updates the risk characteristic values ​​of other nodes.

[0147] A risk characteristic threshold is set. In this embodiment, the risk characteristic threshold may be 0. If the risk characteristic value of a node is greater than the risk characteristic threshold, the node is determined to be a high-risk node. If the risk characteristic value of a node is less than or equal to the risk characteristic threshold, the node is determined to be a low-risk node.

[0148] The risk characteristic value of the high-risk node obtained by the initial risk prediction model is compared with the actual risk characteristic value of the corresponding node through a preset loss function to obtain a loss value, and the model parameters of the initial risk prediction model are adjusted according to the loss value to reduce the loss value. In this embodiment, the initial risk prediction model can be a graph convolutional network, and the loss value is reduced by adjusting the weight matrix of each layer of the network during the training process. Repeat the steps of predicting risk nodes, calculating loss values, and adjusting parameters until the loss value is reduced to a preset value, or the number of repetitions reaches the preset maximum number of iterations, then end the model training and obtain the target risk prediction model.

[0149] Steps S701 to S703 shown in the embodiment of the present application update the dependencies between dimensions in a timely manner so that the target dependency matrix can accurately reflect the latest topological changes of the industrial chain, and then construct the sample graph structure features, fully integrate the node features and adjacency information, and quantize the complex graph structure characteristics. By initializing the node risk feature value and propagating the high-risk feature value of the sample risk event node, the diffusion path of the risk in the industrial chain is gradually simulated, and the model parameters are continuously adjusted to reduce the loss value, so that the model training can dynamically adapt to the structural changes of the industrial chain and improve the accuracy of the model prediction.

[0150] See also Figure 8 In some embodiments, step S107 may include but is not limited to steps S801 to S806:

[0151] Step S801, constructing an original risk node based on the target risk event.

[0152] Step S802, updating the initial industrial chain relationship graph according to the original risk nodes to obtain the target graph structure characteristics.

[0153] Step S803, performing risk prediction on the target graph structure characteristics according to the target risk prediction model to obtain a reference risk characteristic value of each node in the target industry chain relationship graph.

[0154] Step S804: Select a target risk node according to the reference risk characteristic value.

[0155] Step S805 , screening the connection edges associated with the target risk node according to the degree of dependency between the target risk node and the associated risk nodes to obtain the target risk edges.

[0156] Step S806, determining risk prediction information according to the target risk node and the target risk edge.

[0157] In step S801 of some embodiments, the target risk event refers to a risk event that requires analysis of the impact that the target industrial chain may suffer in the actual risk prediction application scenario. The target risk event may include natural disasters, market fluctuations, supply strategies, technical failures, etc. In some embodiments, based on the nature of the target risk event, an object node with the same nature is constructed as the original risk node. For example, the target risk event is that an object in the target industrial chain is unable to purchase or produce a certain raw material, and the event ultimately leads to a supply interruption of the corresponding raw materials. In this case, a node representing a bankrupt supplier is constructed as the original risk node. Assuming that the characteristic value interval is [0, 1], the node characteristics of the original risk node can be [inventory level: 0, production capacity: 0, downstream node dependence on this node: 1].

[0158] In step S802 of some embodiments, the original risk node is embedded into the initial industrial chain relationship graph to obtain a new industrial chain relationship graph. The line of sight process of embedding the node into the industrial chain relationship graph has been explained in detail in the embodiment of step S105, and will not be repeated here. Then, the target graph structure feature is calculated according to analytical formula (2).

[0159] In step S803 of some embodiments, the original risk node is preset with an original risk characteristic value, and the target risk prediction model simulates the propagation of the original risk characteristic value in the industrial chain to obtain the risk characteristic value of each node after the target risk event occurs. The risk characteristic value is the reference risk characteristic value.

[0160] In step S804 of some embodiments, the target risk node refers to a node with a higher risk after prediction. A risk characteristic threshold is set, and if the reference risk characteristic value of the node is greater than the risk characteristic threshold, the node is determined to be a target risk node. If the reference risk characteristic value of the node is less than or equal to the risk characteristic threshold, the node is determined to be a low risk node.

[0161] In step S805 of some embodiments, the associated risk node refers to a node that has a connection relationship with the target risk node. Since the characteristics of each node record its own upstream and downstream relationships and the degree of dependence between the upstream and downstream nodes, the degree of dependence is compared with a preset dependency feature threshold, and the connection edge with a degree of dependence greater than the dependency feature threshold is taken as the target risk edge. Figure 6 For example, node A has been determined to be the target risk node, then the associated risk nodes are node D, node E, node F and risk event node X, the node characteristics of node A are [is the downstream of node X, the degree of dependence on node X, is the upstream of node D, the degree of dependence on node D, is the upstream of node E, the degree of dependence on node E, is the upstream of node F, the degree of dependence on node F], and the degree of dependence on node D is greater than the preset dependency feature threshold, and the connecting edge AD is determined to be the target risk edge.

[0162] In step S806 of some embodiments, risk prediction information is generated based on the object information represented by the target risk node and the event information represented by the target risk edge for reference by risk analysts. For example, when the target risk event is a restriction on the export of a certain type of chip in a remote place, the local supplier cannot provide the manufacturer with the chip of this model, and some manufacturers are unable to produce the corresponding electronic products due to the lack of the chip of this model and the lack of other types of chips that can replace the chip of this model in the market, which in turn causes the corresponding electronic products to be removed from the shelves, and ultimately causes customers to be unable to purchase the electronic products.

[0163] Steps S801 to S806 shown in the embodiment of the present application, by constructing the original risk node, updating the initial industrial chain relationship diagram and obtaining the target graph structure characteristics, and using the target risk prediction model, the reference risk characteristic value of each node in the industrial chain can be predicted. Then, the target risk node is selected according to the reference risk characteristic value, and the degree of dependence between these nodes and other nodes is evaluated to screen out the target risk edge, which helps to identify the key paths and vulnerable links in the industrial chain. Finally, the information of the target risk node and the target risk edge is integrated to determine the risk prediction information, providing decision makers with a clear view of the industrial chain risk, improving the accuracy of risk assessment, and enhancing the understanding of the risk propagation path in the complex industrial chain, thereby helping to protect the industrial chain from potential risks.

[0164] See also Fig. 9The embodiment of the present application further provides a risk prediction device for an industrial chain, which can implement the risk prediction method for the above industrial chain, and the device includes:

[0165] The data acquisition module 901 is used to acquire product supply relationship data of at least two objects in the target industrial chain;

[0166] The relationship graph construction module 902 is used to construct an initial industry chain relationship graph based on the product supply relationship data of each object; wherein the initial industry chain relationship graph includes at least two initial nodes, and the at least two initial nodes form at least two simplexes.

[0167] The grouping module 903 is used to group the simplexes in the initial industrial chain relationship diagram according to the dimensions of the simplexes to obtain at least two simplex combinations.

[0168] The matrix construction module 904 is used to construct diffusion relationships according to the simplex combination and the adjacent simplex combination of each dimension to obtain the inter-dimensional dependency matrix; wherein the adjacent simplex combination is a combination adjacent to the dimension of the simplex combination.

[0169] The relationship diagram updating module 905 is used to embed the sample risk event node into the initial industrial chain relationship diagram according to the preset sample risk event node, initial node and inter-dimensional dependency matrix to obtain the target industrial chain relationship diagram.

[0170] The prediction model training module 906 is used to perform model training on a preset initial risk prediction model according to the target industrial chain relationship diagram to obtain a target risk prediction model.

[0171] The risk prediction module 907 is used to obtain target risk events and perform risk prediction on the target risk events through a target risk prediction model.

[0172] The specific implementation of the risk prediction device for the industrial chain is basically the same as the specific implementation of the risk prediction method for the above-mentioned industrial chain, and will not be repeated here.

[0173] The embodiment of the present application also provides an electronic device, the electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the risk prediction method of the above industrial chain when executing the computer program. The electronic device can be any smart terminal including a tablet computer, a car computer, etc.

[0174] See also Fig.10 , Fig.10 The hardware structure of an electronic device of another embodiment is illustrated, and the electronic device includes:

[0175] The processor 1001 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application;

[0176] The memory 1002 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 1002 can store an operating system and other applications. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 1002, and the processor 1001 calls and executes the risk prediction method of the industrial chain in the embodiment of this application;

[0177] Input / output interface 1003, used to implement information input and output;

[0178] The communication interface 1004 is used to realize the communication interaction between the device and other devices. The communication can be realized through a wired manner (such as USB, network cable, etc.) or a wireless manner (such as mobile network, WIFI, Bluetooth, etc.);

[0179] A bus 1005 , which transmits information between various components of the device (e.g., the processor 1001 , the memory 1002 , the input / output interface 1003 , and the communication interface 1004 );

[0180] The processor 1001 , the memory 1002 , the input / output interface 1003 and the communication interface 1004 are connected to each other in communication within the device via the bus 1005 .

[0181] An embodiment of the present application also provides a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, the risk prediction method for the above-mentioned industrial chain is implemented.

[0182] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0183] The risk prediction method, risk prediction device, electronic device and storage medium of the industrial chain provided by the embodiment of the present application first construct an initial industrial chain relationship diagram including a simplex structure by acquiring product supply relationship data of different objects in the target industrial chain. Then, the simplex of the initial industrial chain relationship diagram is grouped according to the dimension of the simplex, and the diffusion relationship between the simplex combinations of different dimensions is analyzed, so as to generate an inter-dimensional dependency matrix, so that the inter-dimensional dependency matrix can characterize the high-dimensional interaction characteristics, thereby overcoming the defect that the multi-node network diagram in the background technology cannot effectively reflect the complex dimensional interaction in the complex industrial chain. Subsequently, the preset sample risk event node is embedded in the graph structure to simulate the propagation characteristics of the actual risk event, and the target industrial chain relationship diagram is dynamically updated. Then the initial risk prediction model is trained in combination with the target industrial chain relationship diagram to obtain the target risk prediction model. Finally, the target risk prediction model is used to predict the target risk event, and the risk characteristic value and potential propagation path of each node in the industrial chain are accurately evaluated. Compared with the traditional scheme of risk prediction based on the multi-node network diagram, the method of this embodiment introduces a simplex structure when constructing the industrial chain relationship, which can reflect the complex dimensional interaction relationship between each node in the complex industrial chain. By combining the inter-dimensional dependency matrix to update the structure of the initial industrial chain relationship diagram, the target industrial chain relationship diagram obtained can reflect the impact of the sample risk event node on the target industrial chain. The target risk prediction model trained based on the target industrial chain relationship diagram can accurately capture the characteristics of risk propagation, thereby improving the accuracy of risk prediction for the industrial chain.

[0184] The embodiments described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0185] Those skilled in the art will appreciate that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0186] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0187] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.

[0188] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0189] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0190] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the above units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0191] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0192] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0193] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes multiple instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store programs.

[0194] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but the scope of the rights of the present invention is not limited thereto. Any modification, equivalent substitution and improvement made by a person skilled in the art without departing from the scope and essence of the present invention should be within the scope of the rights of the present invention.

Claims

1. A risk prediction method for an industrial chain, characterized in that: The method comprises: Obtain product supply relationship data of at least two objects in the target industrial chain; An initial industrial chain relationship graph is constructed based on the product supply relationship data of each object; wherein the initial industrial chain relationship graph includes at least two initial nodes, and the at least two initial nodes form at least two simplexes; Grouping the simplexes in the initial industrial chain relationship diagram according to the dimensions of the simplexes to obtain at least two simplex combinations; Constructing diffusion relationships based on the simplex combination and adjacent simplex combinations in each dimension to obtain an inter-dimensional dependency matrix; wherein the adjacent simplex combination is a combination adjacent to the dimension of the simplex combination; According to the preset sample risk event node, the initial node and the inter-dimensional dependency matrix, the sample risk event node is embedded in the initial industrial chain relationship diagram to obtain a target industrial chain relationship diagram; Performing model training on a preset initial risk prediction model according to the target industrial chain relationship diagram to obtain a target risk prediction model; Obtain a target risk event, and perform risk prediction on the target risk event using the target risk prediction model.

2. The method according to claim 1, characterized in that The adjacent simplex combination includes a first adjacent simplex combination and a second adjacent simplex combination, and the dimension of the first adjacent simplex combination is smaller than the dimension of the second adjacent simplex combination; the diffusion relationship is constructed according to the simplex combination and the adjacent simplex combination of each dimension to obtain the inter-dimensional dependency matrix, including: Performing matrix calculation according to the diffusion relationship between the simplex combination and the first adjacent simplex combination, and the diffusion relationship between the simplex combination and the second adjacent simplex combination, to obtain an inter-dimensional dependency submatrix of the simplex combination; The inter-dimensional dependency matrix is ​​obtained by fusing the inter-dimensional dependency sub-matrices of the various simplex combinations.

3. The method according to claim 2, characterized in that The matrix calculation is performed according to the diffusion relationship between the simplex combination and the first adjacent simplex combination, and the diffusion relationship between the simplex combination and the second adjacent simplex combination to obtain the inter-dimensional dependency submatrix of the simplex combination, including: Construct a relationship matrix of the diffusion of the simplex combination to the first adjacent simplex combination to obtain a first boundary matrix; Construct a relationship matrix from the second adjacent simplex combination to the simplex combination to obtain a second boundary matrix; Performing a matrix transposition on the first boundary matrix to obtain a first co-boundary matrix; Performing a matrix transposition on the second boundary matrix to obtain a second co-boundary matrix; An inter-dimensional dependency submatrix of the simplex combination is obtained according to the first boundary matrix, the first co-boundary matrix, the second boundary matrix and the second co-boundary matrix.

4. The method according to claim 1, characterized in that The method of embedding the sample risk event node into the initial industrial chain relationship diagram according to the preset sample risk event node, the initial node and the inter-dimensional dependency matrix to obtain the target industrial chain relationship diagram includes: Calculating the similarity between the sample risk event node and each of the initial nodes to obtain initial node similarity data; Selecting candidate nodes from the initial nodes according to the initial node similarity data, and connecting the sample risk event node with the candidate node to obtain a candidate connection edge; Determine an initial propagation node sequence according to the sample risk event node, the candidate node and the candidate connection edge; The initial industrial chain relationship diagram is updated according to the initial propagation node sequence and the inter-dimensional dependency matrix to obtain the target industrial chain relationship diagram.

5. The method according to claim 4, characterized in that The updating of the initial industrial chain relationship graph according to the initial propagation node sequence and the inter-dimensional dependency matrix to obtain the target industrial chain relationship graph includes: Selecting a starting propagation node from the initial propagation nodes of the initial propagation node sequence; Determine the neighbor nodes of the initial propagation node according to the inter-dimensional dependency matrix, calculate the similarity between the neighbor nodes and each of the initial propagation nodes, and obtain candidate node similarity data; Selecting a target node from the initial propagation nodes according to the candidate node similarity data, connecting the target node with the neighbor node, and updating the initial propagation node sequence according to the connection edge between the target node and the neighbor node to obtain a target propagation node sequence; The initial industrial chain relationship graph is updated according to the target propagation node sequence to obtain the target industrial chain relationship graph.

6. The method according to claim 1, characterized in that The method of training a preset initial risk prediction model according to the target industrial chain relationship diagram to obtain a target risk prediction model includes: The inter-dimensional dependency matrix is ​​updated according to the target industrial chain relationship diagram to obtain a target dependency matrix; Integrate node relationships according to the target dependency matrix and each node in the target industry chain relationship graph to obtain sample graph structural features; The parameters of the initial risk prediction model are adjusted according to the structural characteristics of the sample graph to obtain the target risk prediction model.

7. The method according to any one of claims 1 to 6, characterized in that: The obtaining of a target risk event and performing risk prediction on the target risk event by using the target risk prediction model includes: Constructing an original risk node based on the target risk event; The initial industrial chain relationship graph is updated according to the original risk nodes to obtain the target graph structure characteristics; Perform risk prediction on the target graph structure characteristics according to the target risk prediction model to obtain a reference risk characteristic value of each node in the target industry chain relationship graph; Selecting a target risk node according to the reference risk characteristic value; According to the degree of dependence between the target risk node and the associated risk nodes, the connection edges associated with the target risk node are screened to obtain target risk edges; Risk prediction information is determined based on the target risk node and the target risk edge.

8. A risk prediction device for an industrial chain, characterized in that: The device comprises: A data acquisition module, used to acquire product supply relationship data of at least two objects in the target industrial chain; A relationship graph construction module, used to construct an initial industry chain relationship graph based on the product supply relationship data of each object; wherein the initial industry chain relationship graph includes at least two initial nodes, and the at least two initial nodes form at least two simplexes; A grouping module, used for grouping the simplexes in the initial industrial chain relationship diagram according to the dimensions of the simplexes to obtain at least two simplex combinations; A matrix construction module, used for constructing diffusion relationships according to the simplex combination and adjacent simplex combination of each dimension to obtain an inter-dimensional dependency matrix; wherein the adjacent simplex combination is a combination adjacent to the dimension of the simplex combination; A relationship graph updating module, used to embed the sample risk event node into the initial industrial chain relationship graph according to the preset sample risk event node, the initial node and the inter-dimensional dependency matrix, so as to obtain a target industrial chain relationship graph; A prediction model training module is used to perform model training on a preset initial risk prediction model according to the target industrial chain relationship diagram to obtain a target risk prediction model; The risk prediction module is used to obtain a target risk event and perform risk prediction on the target risk event through the target risk prediction model.

9. An electronic device, characterized in that: The electronic device comprises a memory and a processor, the memory stores a computer program, and the processor implements the method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images