Risk identification method based on supplier association relationship, medium and electronic equipment

Abstract

The invention provides a risk identification method based on a supplier association relationship, a medium and electronic equipment. The method comprises the steps of obtaining supplier information of a project to be subjected to risk analysis, and matching the supplier information with data of a preset association relationship risk index library to obtain supplier holding association information; generating a supplier stock equity relation graph and a natural person stock equity relation graph based on the supplier holding association information; obtaining supplier feature data based on the supplier equity relation graph and the natural person equity relation graph, inputting the supplier feature data into a pre-trained equity penetration model, and predicting an equity control relation of the supplier and control right ownership of the supplier based on the equity penetration model; and determining a risk level based on the predicted stock right control relationship of the supplier, the degree of influence of the control right attribution of the supplier on the project and the possibility of the influence. According to the invention, efficient, accurate and real-time risk identification of the supplier association relationship is realized.

Description

Technical Field

[0001] This application belongs to the technical field of big data processing, and particularly relates to the technical field of risk identification through data analysis. Background Art

[0002] In today's digital age, procurement transactions, as a key link in the operation of state-owned assets and state-owned enterprises, the importance of its supervision work has become increasingly prominent. In key areas such as procurement transactions, the use of digital means and big data for joint penetration supervision has already taken the lead in the supervision work of state-owned assets and state-owned enterprises. However, the existing procurement supervision systems still face many problems that need to be solved urgently in terms of risk identification, which to a certain extent affect the effectiveness and accuracy of supervision.

[0003] First of all, a high degree of manual dependence is a major shortcoming in risk identification of existing procurement supervision systems. Under the current supervision mode, the number of risk supervision and management personnel is relatively limited. Facing a large amount of complex procurement data and intricate enterprise association relationships, it is difficult to conduct a comprehensive and detailed risk investigation only by manual review. Manual review is not only inefficient but also easily affected by subjective factors, resulting in some potential risks being ignored or misjudged. For example, when facing a large amount of supplier information and transaction records, it is difficult for manual work to sort out the association relationships between enterprises in a short time, thus unable to accurately identify risks such as potential benefit transfers and bid rigging.

[0004] Secondly, limited data sources seriously restrict the coverage and accuracy of risk identification. At present, procurement supervision systems often only rely on internal procurement data of enterprises, which makes the perspective of risk identification relatively narrow. Although internal procurement data of enterprises can reflect part of the transaction situation, they lack understanding of external association information of enterprises, such as the cooperation relationships between suppliers and other enterprises, market dynamics, etc. This limitation leads to risk identification being limited to known internal information, unable to comprehensively grasp risk factors that may affect procurement transactions, thereby affecting the accuracy and comprehensiveness of risk identification.

[0005] Furthermore, poor real-time performance is also a prominent problem existing in the current system. In a rapidly changing market environment, procurement transaction activities are frequent and complex, and the supervision system is required to be able to respond and process various risk information in a timely manner. However, the existing methods usually require a long time for data processing and analysis, unable to meet the real-time or near-real-time risk identification requirements. When a risk occurs, it cannot be detected and measures taken in a timely manner, which may lead to the further expansion of the risk and cause greater losses to the enterprise.

[0006] Finally, the weak relevance makes it difficult for existing methods to cope with complex risk situations. Procurement transactions involve multiple links and multiple parties, and there are intricate connections between different data sources. However, existing methods are difficult to handle the correlation queries of multi-source data and cannot comprehensively identify potential risks. For example, it is impossible to effectively correlate and analyze the financial status and credit records of suppliers with procurement transaction data, making it difficult to discover the hidden risk hazards.

[0007] Therefore, to improve the efficiency of procurement transaction supervision, it is necessary to increase technological investment, optimize the risk identification mechanism, and achieve more comprehensive, accurate, and real-time supervision in response to the problems existing in the existing system. Summary of the Invention

[0008] This application provides a risk identification method, medium, and electronic device based on supplier association relationships to solve the technical problems of limited data volume, poor real-time performance, and weak relevance in the existing risk identification of supplier association relationships.

[0009] In a first aspect, an embodiment of this application provides a risk identification method based on supplier association relationships, including: obtaining supplier information of a project to be analyzed for risks, and matching the supplier information with the data in a preset association relationship risk index library to obtain supplier holding association information associated with the supplier information; generating a supplier equity relationship diagram and a natural person equity relationship diagram based on the supplier holding association information; wherein, each node in the supplier equity relationship diagram and the natural person equity relationship diagram represents an enterprise or an individual, the edge represents an equity relationship, and the weight of the edge represents the shareholding ratio; obtaining supplier feature data based on the supplier equity relationship diagram and the natural person equity relationship diagram, inputting the supplier feature data into a pre-trained equity penetration model, and predicting the equity control relationship of the supplier and the ownership of the supplier's control right based on the equity penetration model; determining the risk level based on the influence degree and the possibility of occurrence of the predicted equity control relationship of the supplier and the ownership of the supplier's control right on the project.

[0010] In an implementation manner of the first aspect, it further includes: training the equity penetration model: preprocessing the data in the association relationship risk index library to form a training data set; generating a supplier equity relationship diagram and a natural person equity relationship diagram for each supplier based on the training data set; wherein, each node in the supplier equity relationship diagram and the natural person equity relationship diagram represents an enterprise or an individual, the edge represents an equity relationship, and the weight of the edge represents the shareholding ratio; obtaining corresponding supplier feature data for each supplier based on each supplier equity relationship diagram and each natural person equity relationship diagram, and repeatedly training a random forest model based on each supplier feature data to obtain an equity penetration model whose output is the predicted equity control relationship of the supplier and the ownership of the supplier's control right.

[0011] In an implementation manner of the first aspect, it further includes: constructing the associated relationship risk index library, including: obtaining historical bidding and tendering data from the e-bidding system, and at least extracting supplier participation information, bidding information, and winning bid information from the historical bidding and tendering data; obtaining external public data of the supplier from an external public data source, where the external public data of the supplier includes multiple combinations of the following: actual controller of the enterprise, superior-subordinate information of the enterprise, inter-enterprise associated relationship, enterprise associated risk, enterprise suppliers, main member information of the enterprise, member appointment information, and partner information; and forming the associated relationship risk index library from the historical bidding and tendering data and the external public data of the supplier.

[0012] In an implementation manner of the first aspect, the generation of the supplier equity relationship diagram and the natural person equity relationship diagram for each supplier includes: extracting enterprises, individuals, and shareholding ratios from the obtained data, and using the enterprises and individuals as nodes respectively, and the shareholding ratio as an edge; determining the root node, and calculating the hierarchy and path from the root node to each of the nodes based on the obtained data, to form the corresponding supplier equity relationship diagram and natural person equity relationship diagram.

[0013] In an implementation manner of the first aspect, the obtaining of supplier feature data based on the supplier equity relationship diagram and the natural person equity relationship diagram includes: mapping the nodes in the supplier equity relationship diagram and the natural person equity relationship diagram to low-dimensional vectors; starting from a certain node, randomly selecting neighbor nodes to perform a walk, generating a series of node sequences, and capturing the hierarchy, path, and indirect shareholding information in the supplier equity relationship diagram and the natural person equity relationship diagram; using the Word2Vec model to learn the low-dimensional vectors of each node in the node sequence, obtaining node embeddings, and using the node embeddings as supplier feature data; where the node embeddings include an adjacency matrix representing the connection relationship between nodes, and a feature matrix representing the shareholding ratio, equity hierarchy, and shareholder type of the nodes.

[0014] In an implementation manner of the first aspect, the prediction of the equity control relationship of the supplier and the ownership of the control right of the supplier based on the equity penetration model includes: the equity penetration model calculates the direct shareholding ratio, indirect shareholding ratio, and total shareholding ratio of each node based on the supplier feature data; the equity penetration model outputs the predicted equity control relationship of the supplier and the ownership of the control right of the supplier based on the direct shareholding ratio, the indirect shareholding ratio, and the total shareholding ratio.

[0015] In one implementation of the first aspect, determining the risk level based on the predicted equity control relationship of the supplier and the degree of influence and the likelihood of the impact of the ownership of the control right of the supplier on the project includes: determining the category of the label according to the predicted equity control relationship of the supplier and the ownership of the control right of the supplier; determining the probability scores of the degree of influence on the project, the likelihood of the impact, and the type of risk based on the label category; obtaining the label risk score based on the degree of influence and the likelihood of the impact; and determining the risk level based on the label risk score and the probability score of the risk type.

[0016] In one implementation of the first aspect, it further includes: generating a warning notification message based on the determined risk level, and sending the warning notification message to the user terminal by means of an in-system message, an email, or a short message.

[0017] In a second aspect, an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the risk identification method based on the supplier association relationship described in any item of the first aspect of the present application.

[0018] In a third aspect, an embodiment of the present application provides an electronic device, the electronic device includes a processor and a memory; the memory stores program instructions; the processor is used to run the program instructions to execute the risk identification method based on the supplier association relationship described in any item of the first aspect of the present application.

[0019] The risk identification method based on the supplier association relationship provided by the embodiment of the present application has the following beneficial effects:

[0020] The present application overcomes the limitation of the limited amount of data in the prior art. By integrating multi-source data, deeply analyzing various aspects of information of the supplier, accurately identifying potential risk points, effectively avoiding risk hidden dangers caused by misjudgment or missed judgment, enhancing the relevance of risk identification by deeply mining the supplier association relationship, and significantly shortening the time required for risk identification through optimized algorithms and powerful computing capabilities, greatly improving the efficiency of risk management work. Therefore, the present application realizes efficient, accurate, and real-time risk identification of the supplier association relationship, and solves the problems of limited data volume, poor real-time performance, and weak relevance in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It shows a flowchart of the risk identification method based on the supplier association relationship according to an embodiment of the present application.

[0022] Figure 2 It shows a flowchart of training an equity penetration model in the risk identification method based on the supplier association relationship according to an embodiment of the present application.

[0023] Figure 3 It shows a schematic diagram of the associated relationship risk index library in the risk identification method based on the supplier associated relationship according to an embodiment of the present application.

[0024] Figure 4 It shows a flowchart of constructing a supplier equity relationship diagram and a natural person equity relationship diagram in the risk identification method based on the supplier associated relationship according to an embodiment of the present application.

[0025] Figure 5 It shows an example diagram of the supplier equity relationship generated in the risk identification method based on the supplier associated relationship according to an embodiment of the present application.

[0026] Figure 6 It shows a flowchart of obtaining supplier characteristic data in the risk identification method based on the supplier associated relationship according to an embodiment of the present application.

[0027] Figure 7 It shows a schematic diagram of the principle of predicting the equity control relationship of a supplier and the attribution of the control right of the supplier in the risk identification method based on the supplier associated relationship according to an embodiment of the present application.

[0028] Figure 8 It shows a schematic diagram of the structure of an electronic device according to an embodiment of the present application.

[0029] Description of component numbers

[0030] 100 Electronic device

[0031] 101 Memory

[0032] 102 Processor

[0033] 103 Display

[0034] Steps S100 - S400

[0035] Steps S10 - S30

[0036] Steps S210 - S220

[0037] Steps S310 - S330 Detailed implementation manners

[0038] The following describes the implementation manners of the present application through specific examples. Those skilled in the art can easily understand the other advantages and effects of the present application from the content disclosed in this specification. The present application can also be implemented or applied through other different specific implementation manners. Various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present application. It should be noted that, without conflict, the following embodiments and the features in the embodiments can be combined with each other.

[0039] This embodiment provides a risk identification method, medium, and electronic device based on supplier association relationships, which can be applied to the procurement field and are used to identify and warn of potential risks of supplier association relationships in real time or near real time, and solve the technical problems of limited data volume, poor real-time performance, and weak relevance in the existing risk identification of supplier association relationships.

[0040] This embodiment provides a risk identification method, medium, and electronic device based on supplier association relationships. Massive data is collected in real time from multiple data sources. Among them, the internal data source includes historical procurement project information, and the external data sources include public information such as enterprise basic operation information, historical risk information, and public bidding information. Then, the collected data is cleaned, de-noised, and format-converted to ensure data quality. Mark the association query conditions for each type of data (such as fields like section number, contract number, enterprise unified credit code, etc.), associate them with data indicators, and obtain key information (such as list of shareholders and shareholding ratios, list of parent companies of shareholders and shareholding ratios). After that, data mining and machine learning algorithms are used to analyze the association relationships between suppliers and identify potential risk nodes. For example: according to the obtained structured information, an equity relationship graph is constructed, where each node in the graph represents an enterprise or an individual, the edge represents the equity relationship, and the weight of the edge represents the shareholding ratio. Calculate the direct shareholding ratio and indirect shareholding ratio of enterprises and natural persons, and obtain the total association degree analysis result. Then, according to the association relationship analysis result, combined with the risk assessment model, the risk level of each supplier is evaluated. And the enterprises that jointly participate in the project within the internal units of the enterprise group are excluded, and the risk level is defined according to the impact degree of the risk on the project. Finally, according to the risk assessment result, warning information is generated in real time or near real time and sent to relevant personnel or systems. Through the combination of the above technical features, this embodiment realizes the functions of efficient, accurate, and real-time risk identification, and solves the problems of limited data volume, poor real-time performance, and weak relevance in the prior art.

[0041] The following will combine the appended Figure 1 to the appended Figure 8 to describe the technical solutions in the embodiments of the present application in detail. Those skilled in the art can understand and implement the risk identification method based on supplier association relationships in this embodiment without creative labor.

[0042] Figure 1 It is shown as the flowchart of the risk identification method based on supplier association relationships in the embodiments of the present application. As Figure 1 shown, the risk identification method based on supplier association relationships provided by the embodiments of the present application includes the following steps S100 to S400.

[0043] Step S100: Obtain the supplier information of the project to be analyzed for risks, and match the supplier information with the data in the preset associated relationship risk index library to obtain the supplier holding associated information associated with the supplier information;

[0044] Step S200: Generate a supplier equity relationship graph and a natural person equity relationship graph based on the supplier holding associated information; wherein, each node in the supplier equity relationship graph and the natural person equity relationship graph represents an enterprise or an individual, the edge represents an equity relationship, and the weight of the edge represents the shareholding ratio;

[0045] Step S300: Obtain supplier characteristic data based on the supplier equity relationship graph and the natural person equity relationship graph, input the supplier characteristic data into a pre-trained equity penetration model, and predict the equity control relationship of the supplier and the ownership of the supplier's control right based on the equity penetration model;

[0046] Step S400: Determine the risk level based on the degree of influence and the possibility of influence of the predicted equity control relationship of the supplier and the ownership of the supplier's control right on the project.

[0047] The risk identification method based on supplier association relationships in this embodiment first forms data indicators by cleaning, denoising, and format conversion of the collected data according to internal (historical project information) and external real-time data (public information such as enterprise superior-subordinate information, enterprise main member information, personal employment information, and partner information), extracts key structured data such as the supplier name and credit code in the project to be analyzed for risks, forms a supplier association relationship graph, calculates the direct shareholding and indirect shareholding ratios between suppliers, obtains the analysis result of the supplier association degree, and issues a notice to relevant project personnel or procurement supervision and management personnel through early warning.

[0048] The above steps S100 to S400 of the risk identification method based on supplier association relationships in this embodiment are described in detail below.

[0049] In this embodiment, first, an equity penetration model needs to be trained, and the equity control relationship of the supplier and the ownership of the supplier's control right are predicted based on the trained equity penetration model.

[0050] Therefore, in one implementation of this embodiment, it further includes: training the equity penetration model. Figure 2 It shows a flowchart of training the equity penetration model in the risk identification method based on supplier association relationships according to an embodiment of the present application. As Figure 2 shown, training the equity penetration model includes the following steps:

[0051] Step S10: Preprocess the data in the association relationship risk index library to form a training data set.

[0052] In one implementation of this embodiment, it further includes: constructing the association relationship risk index library, and forming the association relationship risk index library by collecting internal and external information:

[0053] 1) Collect internal data: Obtain historical bidding and tendering data in the e-bidding system, and extract supplier participation information, bidding information, winning bid information, etc. from it.

[0054] 2) Collect external data: Obtain enterprise superior-subordinate information, enterprise main member information, personal appointment information, partner information, etc. in public information.

[0055] Figure 3 It shows a schematic diagram of the association relationship risk index library in the risk identification method based on supplier association relationships according to an embodiment of the present application. As Figure 3 shown, constructing the association relationship risk index library includes:

[0056] Obtain historical bidding and tendering data from the e-bidding system, and at least extract supplier participation information, bidding information, and winning bid information from the historical bidding and tendering data; obtain external public data of the supplier from external public data sources, and the external public data of the supplier includes multiple combinations of the following: actual controller of the enterprise, enterprise superior-subordinate information, inter-enterprise association relationship, enterprise association risk, enterprise supplier, enterprise main member information, member appointment information, and partner information; the historical bidding and tendering data and the external public data of the supplier constitute the association relationship risk index library.

[0057] Among them, the details of the external data indicators are shown in Table 1 below.

[0058] Table 1

[0059]

[0060]

[0061] In this embodiment, preprocessing the data in the association relationship risk index library includes, but is not limited to, cleaning, denoising, and format conversion of the collected data, etc.

[0062] Among them, the purpose of data denoising is to eliminate inaccurate or irrelevant data points in the data. For example, if the data is significantly out of range, the IsolationForest algorithm is used to remove the dirty data.

[0063] Unify the data format. For example: Date format: Unify the date format, such as YYYYMMDD; Numeric format: Unify the numeric format, such as retaining 2 digits after the decimal point; Text format: Unify the text format, such as capitalization, spaces, full-width and half-width symbols, special characters, etc.

[0064] Data type conversion, such as converting strings to numbers, dates, etc. The methods include: String to number: Use methods such as astype() or pd.to_numeric(); String to date: Use methods such as pd.to_datetime().

[0065] Data deduplication: Delete duplicate data records to ensure the uniqueness of the data. Use methods such as drop_duplicates().

[0066] Data encoding: For categorical variables, use encoding methods to convert them into numerical data for machine learning models to process. The methods include: OneHotEncoding: Convert categorical variables into binary vectors; LabelEncoding: Convert categorical variables into integers.

[0067] In addition, in this embodiment, the preprocessing of the data in the associated relationship risk index library includes marking the associated query conditions for each type of data. For example, fields such as bid section number, contract number, and unified social credit code of the enterprise are used to associate the preprocessed data with the data indicators to obtain key information, such as: list of shareholders and shareholding ratios, list of parent companies of shareholders and shareholding ratios. In this embodiment, the data indicator association values are preconfigured, and the association order of the association values is configured. For example, in the enterprise dimension, the unified social credit code of the supplier is used as the first association value. If this field is missing, the supplier name is used as the sequential association value.

[0068] Step S20, generate a supplier equity relationship graph and a natural person equity relationship graph for each supplier based on the training data set; among them, each node in the supplier equity relationship graph and the natural person equity relationship graph represents an enterprise or an individual, the edge represents the equity relationship, and the weight of the edge represents the shareholding ratio.

[0069] According to the collected data, use the graph data structure (NetworkX) to construct a supplier equity relationship graph and a natural person equity relationship graph involving suppliers' ancient shareholders, legal persons, directors, supervisors, etc. Each node in the graph represents an enterprise or an individual, the edge represents the equity relationship, and the weight of the edge represents the shareholding ratio.

[0070] Step S30: Obtain the corresponding supplier feature data based on each of the supplier equity relationship diagrams and each of the natural person equity relationship diagrams, and repeatedly train the random forest model based on each of the supplier feature data to obtain an equity penetration model with the output being the predicted equity control relationship of the supplier and the attribution of the supplier's control right.

[0071] In this embodiment, features related to the association relationship are extracted, including: 1) Shareholding ratio: direct shareholding ratio, indirect shareholding ratio; 2) Equity level: the number of levels of the equity relationship; 3) Control right: board seat, shareholder agreement, etc.; 4) Historical changes: equity change records, historical shareholding ratio changes, etc.

[0072] In this embodiment, each of the supplier equity relationship diagrams and each of the natural person equity relationship diagrams are converted into a feature representation that can be processed by a machine learning model, i.e., supplier feature data.

[0073] Among them, the supplier feature data includes:

[0074] Graph Embedding: Use DeepWalk to map the nodes in the graph to a low-dimensional vector space.

[0075] Adjacency matrix: Represents the connection relationship between nodes.

[0076] Feature matrix: Represents the features of nodes (such as shareholding ratio, shareholder type, etc.).

[0077] During training, a graph neural network (GNN) and a random forest model can be selected for training. Among them, the graph neural network (GNN) is suitable for processing graph-structured data and can capture the complex relationships between nodes. The random forest model is suitable for processing high-dimensional features and can handle non-linear relationships. The trained random forest model can make predictions based on the feature matrix (such as shareholding ratio, equity level, board seat, etc.) and output classification results (such as enterprise type) or regression results (such as shareholding ratio prediction).

[0078] Use the training set composed of historical data to train the model, and the goal is to predict and analyze the equity relationship. During the training process, cross-validation and hyperparameter tuning can be used to improve the model performance.

[0079] Divide the data set into a training set and a test set. Usually, 70% of the data is used for training and 30% of the data is used for testing. The specific training process includes:

[0080] Perform preliminary training using a random forest model because random forests can handle high-dimensional data and have a certain degree of robustness to noisy data. Initialize parameters: Set the parameters of the random forest model, such as the number of trees, maximum depth, minimum sample split number, etc. Train the model: Use the training set data to train the random forest model. Cross-validation: Use k-fold cross-validation to evaluate the performance of the model and prevent overfitting. Hyperparameter tuning: Adjust the model parameters according to the results of cross-validation to improve the model performance.

[0081] In this embodiment, after obtaining the trained equity penetration model, directly apply this equity penetration model in the risk identification process of the supplier's associated relationship. That is, execute steps S100 to S400.

[0082] Step S100, obtain the supplier information of the project to be analyzed for risks, and match the supplier information with the data in the preset associated relationship risk index library to obtain the supplier holding associated information associated with the supplier information.

[0083] That is, obtain the supplier information of the project that needs to be analyzed for risks, including unified social credit code, supplier name, participation stage, etc., compare and calculate with the data in the associated relationship risk index library, and obtain the list of upper and lower level enterprises of the suppliers participating in the project, information on natural persons' shareholding or controlling suppliers, etc.

[0084] Step S200, generate a supplier equity relationship diagram and a natural person equity relationship diagram based on the supplier holding associated information; wherein, each node in the supplier equity relationship diagram and the natural person equity relationship diagram represents an enterprise or an individual, the edge represents the equity relationship, and the weight of the edge represents the shareholding ratio.

[0085] Figure 4 Shown is a flowchart of constructing a supplier equity relationship diagram and a natural person equity relationship diagram in the risk identification method based on supplier associated relationships according to an embodiment of the present application. As Figure 4 shown, in one implementation manner of this embodiment, the generation of the supplier equity relationship diagram and the natural person equity relationship diagram for each supplier includes:

[0086] Step S210, extract enterprises, individuals, and shareholding ratios from the obtained data, and use the enterprises and individuals as nodes respectively, and the shareholding ratio as the edge;

[0087] Step S220, determine the root node, and calculate the levels and paths from the root node to each of the nodes based on the obtained data to form the corresponding supplier equity relationship diagram and natural person equity relationship diagram of the supplier.

[0088] Figure 5It shows an example diagram of the supplier equity relationship generated in the risk identification method based on supplier association relationships according to an embodiment of the present application. As Figure 5 shown, the example process of generating the supplier equity relationship diagram includes:

[0089] 1) Create nodes and edges based on the collected data:

[0090] Enterprise A, Enterprise B, weight = 0.3

[0091] Enterprise B, Enterprise C, weight = 0.8

[0092] Enterprise A, Enterprise C, weight = 0.1

[0093] Enterprise C, Enterprise D, weight = 0.7

[0094] Individual X, Enterprise B, weight = 0.2

[0095] 2) Define the root node: Enterprise A;

[0096] 3) Calculate the level and path of each node, and output: The level from Enterprise A to Enterprise B: 1, path: ['Enterprise A', 'Enterprise B']; The level from Enterprise A to Enterprise C: 1, path 1: ['Enterprise A', 'Enterprise C'], path 2: ['Enterprise A', 'Enterprise B', 'Enterprise C']; The level from Enterprise A to Enterprise D: 2, path 1: ['Enterprise A', 'Enterprise C', 'Enterprise D'], path 2: ['Enterprise A', "Enterprise B", Enterprise C', 'Enterprise D']; There is no path from Enterprise A to Individual X.

[0097] Step S300, obtain supplier feature data based on the supplier equity relationship diagram and the natural person equity relationship diagram, input the supplier feature data into a pre-trained equity penetration model, and predict the equity control relationship of the supplier and the ownership of the supplier's control right based on the equity penetration model.

[0098] Figure 6 It shows a flowchart of obtaining supplier feature data in the risk identification method based on supplier association relationships according to an embodiment of the present application. As Figure 6 shown, in one implementation manner of this embodiment, obtaining the supplier feature data based on the supplier equity relationship diagram and the natural person equity relationship diagram includes:

[0099] Step S310, map the nodes in the supplier equity relationship diagram and the natural person equity relationship diagram to low-dimensional vectors;

[0100] Step S320: Starting from a certain node, randomly select neighbor nodes for a walk to generate a series of node sequences, and capture the hierarchy, paths, and indirect shareholding information in the supplier equity relationship graph and the natural person equity relationship graph;

[0101] Step S330: Use the Word2Vec model to learn the low-dimensional vectors of each node in the node sequence to obtain node embeddings, and use the node embeddings as supplier feature data; among them, a node embedding is a representation that maps each node to a low-dimensional vector space. This mapping can preserve the similarity of nodes in the network, so that the node relationships in the embedding space can approximately reflect the structure and properties of the original network. In this embodiment, the node embedding includes an adjacency matrix representing the connection relationships between nodes and a feature matrix representing the shareholding ratio, equity hierarchy, and shareholder type of nodes.

[0102] In this embodiment, the equity relationship graph is converted into a feature representation that can be processed by a machine learning model. That is, extract features related to the association relationship to obtain supplier feature data, including:

[0103] Shareholding ratio: direct shareholding ratio, indirect shareholding ratio, calculated using the paths and weights in the equity relationship graph.

[0104] Equity hierarchy: the number of levels of the equity relationship.

[0105] Control right: board seats, shareholder agreements, etc., obtained from the number of levels from the root node to the target node.

[0106] In this embodiment, Graph Embedding: Use DeepWalk to map the nodes in the graph to a low-dimensional vector space. Specifically includes:

[0107] 1) Map nodes to low-dimensional vectors:

[0108] Map enterprise and individual nodes to low-dimensional vectors for easy processing by machine learning models.

[0109] For example: Enterprise A → [0.1, 0.3, -0.2], Enterprise B → [0.4, -0.1, 0.5].

[0110] 2) Random Walk:

[0111] Starting from a certain node, randomly select neighbor nodes for a walk to generate a series of node sequences, and capture the hierarchy, paths, and indirect shareholding information in the equity relationship graph.

[0112] For example: Starting from "Enterprise A", the random walk may generate a sequence ["Enterprise A", "Enterprise B", "Enterprise C"].

[0113] 3) Language Model (Word2Vec):

[0114] Regard the node sequence generated by random walk as "sentences" and the nodes as "words". Use the Word2Vec model to learn the low-dimensional vector representation of nodes and obtain node embeddings.

[0115] For example: "Enterprise A": [0.1, 0.3, -0.2,...],

[0116] "Enterprise B": [0.4, -0.1, 0.5,...],

[0117] "Enterprise C": [-0.2, 0.6, 0.1,...],

[0118] "Individual X": [0.3, 0.2, -0.4,...].

[0119] In this embodiment, the generated random walk sequence is input into the Word2Vec model for training. The model will automatically learn the low-dimensional vector representation of each node according to the co-occurrence relationship between nodes. During the training process, the model will continuously adjust the values of node vectors to maximize the co-occurrence probability of node contexts.

[0120] In this embodiment, the adjacency matrix represents the connection relationship between nodes. The feature matrix represents the features of nodes, such as shareholding ratio, shareholder type, etc.

[0121] For example, generate the vector representations generated by nodes (individuals, enterprises) and weights (shareholding ratios) into a matrix, as shown in Table 2.

[0122] Table 2

[0123]

[0124] In this embodiment, the obtained supplier feature data is input into a pre-trained equity penetration model, and based on the equity penetration model, predict the equity control relationship of the supplier and the ownership of the supplier's control right. Among them, the results output by the equity penetration model include:

[0125] 1) Predicted equity relationship: Such as whether there is a control relationship, shareholding ratio, etc., judge whether an enterprise has control over another enterprise, and calculate the shareholding ratio between enterprises.

[0126] 2) Ownership of enterprise control: Such as whether an enterprise is controlled by a certain group, analyze the control structure of the enterprise group, identify potential actual controllers, and analyze the relationship between individuals and enterprises.

[0127] Figure 7It is a schematic diagram showing the principle of predicting the equity control relationship of a supplier and the ownership of the control right of the supplier in the risk identification method based on supplier association relationships according to an embodiment of the present application. As Figure 7 shown, in one implementation manner of this embodiment, predicting the equity control relationship of a supplier and the ownership of the control right of the supplier based on the equity penetration model includes: the equity penetration model calculates the direct shareholding ratio, indirect shareholding ratio, and total shareholding ratio of each node based on supplier characteristic data; the equity penetration model outputs the predicted equity control relationship of the supplier and the ownership of the control right of the supplier based on the direct shareholding ratio, the indirect shareholding ratio, and the total shareholding ratio.

[0128] In this embodiment, the equity penetration model calculates the direct shareholding ratio of each shareholder in the target enterprise. For example, if Company A holds 30% of the shares of Company B, then the direct shareholding ratio of Company A in Company B is 30%. For indirect shareholding, it needs to be calculated by penetrating layer by layer.

[0129] Through the following formula:

[0130] Indirect shareholding ratio = ∑i (shareholding ratio of shareholder A in company B × shareholding ratio of company B in the target company)

[0131] For example, if Company A holds 30% of the shares of Company B, and Company B holds 40% of the shares of Company C, then the indirect shareholding ratio of Company A in Company C through Company B is:

[0132] Indirect shareholding ratio = 30% × 40% = 12% Indirect shareholding ratio = 30% × 40% = 12%.

[0133] Adding the direct shareholding ratio and the indirect shareholding ratio gives the total shareholding ratio. Suppose Company A holds 60% of the shares of Company B, Company B holds 55% of the shares of Company C, and Company C holds 70% of the shares of the target company D.

[0134] Direct shareholding ratio of Company A in Company B: 60%

[0135] Direct shareholding ratio of Company B in Company C: 55%

[0136] Direct shareholding ratio of Company C in the target company D: 70%

[0137] Indirect shareholding ratio of Company A in the target company D:

[0138] Indirect shareholding ratio = 60% × 55% × 70% = 0.6 × 0.55 × 0.7 = 0.231 = 23.1% Indirect shareholding ratio = 60% × 55% × 70% = 0.6 × 0.55 × 0.7 = 0.231 = 23.1%

[0139] Through model training and parameter tuning, the above algorithm can obtain the calculation results of the association information within 5 layers among the top 10 enterprises in a relatively short time.

[0140] Through the equity penetration model, the ultimate controller of the target enterprise can be determined. The ultimate controller usually refers to an individual or enterprise that holds the largest proportion of shares in the target enterprise.

[0141] Step S400, determine the risk level based on the predicted equity control relationship of the supplier and the degree of influence and the likelihood of influence of the ownership attribution of the supplier on the project.

[0142] In an implementation manner of this embodiment, the determining the risk level based on the predicted equity control relationship of the supplier and the degree of influence and the likelihood of influence of the ownership attribution of the supplier on the project includes:

[0143] Determine the label category according to the predicted equity control relationship of the supplier and the ownership attribution of the supplier;

[0144] Based on the label category, determine the degree of influence on the project, the likelihood of influence, and the probability score of the risk type;

[0145] Obtain the label risk score based on the degree of influence and the likelihood of influence; determine the risk level based on the label risk score and the probability score of the risk type.

[0146] For example, according to the degree of influence of the risk on the project, the risk is rated into five levels: high, medium - high, medium, medium - low, and low, and the following risk items are generated: as shown in Table 3.

[0147] Table 3

[0148]

[0149] In this embodiment, among the degree of influence on the project and the likelihood of influence determined based on the label category:

[0150] 1) The degree of influence includes:

[0151] ① Very serious influence: serious influences such as project cancellation, tender invalidation, bid winning invalidation, contract invalidation, criminal punishment, safety and quality accidents, contract termination, contract revocation, etc.;

[0152] ② Relatively serious influence: project suspension, affecting the fairness, impartiality, competitiveness of tendering and procurement, increasing unreasonable costs, being revoked of business license, qualification certificate, administrative detention, being ordered to suspend production and business, affecting contract performance, etc.;

[0153] ③General impacts: invalid bids, quality and safety issues, resource waste, warnings, fines, confiscation of illegal income and property, and other administrative penalties stipulated by laws, which do not conform to industry rules and common sense and have a certain substantial impact on all parties involved in the procurement;

[0154] ④Minor impacts: acts that violate laws and regulations, contract agreements, common sense, etc., but with minor impacts (except for the above impacts);

[0155] ⑤Almost no impact: there are defects that do not cause substantial impacts;

[0156] 2) The possibilities of impacts are divided into: ①Definitely have an impact; ②Relatively high possibility of having an impact; ③General possibility of having an impact; ④Relatively low possibility of having an impact.

[0157] In this embodiment, the label risk score (R) = impact (A) × possibility (B). Table 4 shows the judgment criteria and label risk scores.

[0158] Table 4

[0159]

[0160] For risks of bid rigging and collusive bidding, the probability scores are as follows:

[0161] Probability: direct holding 4, indirect holding 3, direct shareholding 2, indirect shareholding 1.

[0162] The scores for determining the risk levels are, for example: high risk: 15 - 20; medium - high risk: 9 - 14; medium risk: 6 - 8; medium - low risk:

[0163] 4 - 5; low risk: 1 - 3.

[0164] In this embodiment, it is necessary to conduct a risk assessment on the procurement activity based on the calculation results of the risk indicators, generate a warning message, and notify the project leader or procurement supervision and management personnel in the form of system internal messages, emails, text messages, etc.

[0165] In one implementation manner of this embodiment, it further includes: generating a warning notification message based on the determined risk level, and sending the warning notification message to the user terminal by means of system internal messages, emails, or text messages.

[0166] In this embodiment, it is necessary to conduct a risk assessment on the procurement activity based on the calculation results of the risk indicators, generate a warning message, and notify the project leader or procurement supervision and management personnel in the form of system internal messages, emails, text messages, etc.

[0167] In summary, the risk identification method based on supplier association relationships described in the embodiments of the present application overcomes the limitation of limited data volume in the prior art. By integrating multi-source data, deeply analyzing various aspects of supplier information, accurately identifying potential risk points, effectively avoiding risk hidden dangers caused by misjudgment or missed judgment, enhancing the relevance of risk identification by deeply mining supplier association relationships, and significantly shortening the time required for risk identification through optimized algorithms and powerful computing capabilities, greatly improving the efficiency of risk management work. Therefore, the present application realizes efficient, accurate, and real-time risk identification of supplier association relationships, and solves problems such as limited data volume, poor real-time performance, and weak relevance in the prior art.

[0168] The protection scope of the risk identification method based on supplier association relationships described in the embodiments of the present application is not limited to the execution order of the steps listed in this embodiment. Any solution achieved by adding or subtracting steps of the prior art and replacing steps according to the principle of the present application is included in the protection scope of the present application.

[0169] The embodiments of the present application also provide a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the risk identification method based on supplier association relationships provided in any embodiment of the present application.

[0170] In the embodiments of the present application, any combination of one or more storage media can be used. The storage medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, RAM, ROM, erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0171] The embodiments of the present application also provide an electronic device. Figure 8It shows a schematic structural diagram of the electronic device 100 provided by an embodiment of the present application. In some embodiments, the electronic device may be a mobile phone, a tablet computer, a wearable device, a vehicle-mounted device, an Augmented Reality (AR) / Virtual Reality (VR) device, a laptop computer, an Ultra-Mobile Personal Computer (UMPC), a netbook, a Personal Digital Assistant (PDA), or other terminal devices. In addition, the risk identification method based on supplier association relationships provided by the present application can also be applied to databases, servers, and service response systems based on terminal artificial intelligence. The embodiments of the present application do not impose any restrictions on the specific application scenarios of the risk identification method based on supplier association relationships.

[0172] As Figure 8 shown, the electronic device 100 provided by an embodiment of the present application includes a memory 101 and a processor 102.

[0173] The memory 101 is used to store computer programs; preferably, the memory 101 includes various media that can store program codes, such as ROM, RAM, magnetic disks, USB flash drives, memory cards, or optical discs.

[0174] Specifically, the memory 101 may include a computer system readable medium in the form of volatile memory, such as random access memory (RAM) and / or cache memory. The electronic device 100 may further include other removable / non-removable, volatile / non-volatile computer system storage media. The memory 101 may include at least one program product, and this program product has a set (for example, at least one) of program modules, and these program modules are configured to execute the functions of the embodiments of the present application.

[0175] The processor 102 is connected to the memory 101 and is used to execute the computer programs stored in the memory 101, so that the electronic device 100 executes the risk identification method based on supplier association relationships provided in any embodiment of the present application.

[0176] Optionally, the processor 102 may be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0177] Optionally, in this embodiment, the electronic device 100 may further include a display 103. The display 103 is communicatively connected to the memory 101 and the processor 102, and is configured to display a relevant GUI interaction interface of the risk identification method based on the supplier association relationship.

[0178] In summary, this application overcomes the limitation of limited data volume in the prior art. By integrating multi-source data, deeply analyzing various aspects of information of suppliers, accurately identifying potential risk points, effectively avoiding risk hazards caused by misjudgment or missed judgment, enhancing the relevance of risk identification by deeply mining supplier association relationships, and greatly shortening the time required for risk identification through optimized algorithms and powerful computing capabilities, the efficiency of risk management work is greatly improved. Therefore, this application realizes efficient, accurate, and real-time risk identification of supplier association relationships, and solves problems such as limited data volume, poor real-time performance, and weak relevance in the prior art. Thus, this application effectively overcomes various disadvantages in the prior art and has high industrial utilization value.

[0179] The above embodiments are only illustrative of the principles and effects of this application, and are not used to limit this application. Any person familiar with this technology can modify or change the above embodiments without departing from the spirit and scope of this application. Therefore, all equivalent modifications or changes made by those with ordinary knowledge in the technical field without departing from the spirit and technical idea disclosed in this application should still be covered by the claims of this application.

Claims

1. A risk identification method based on supplier association relationships, characterized in that, Including: Obtain the supplier information of the project to be risk-analyzed, match the supplier information with the data in the preset associated relationship risk index library, and obtain the supplier holding associated information associated with the supplier information; Generate a supplier equity relationship diagram and a natural person equity relationship diagram based on the supplier holding associated information; wherein, each node in the supplier equity relationship diagram and the natural person equity relationship diagram represents an enterprise or an individual, the edge represents the equity relationship, and the weight of the edge represents the shareholding ratio; Obtain supplier characteristic data based on the supplier equity relationship diagram and the natural person equity relationship diagram, input the supplier characteristic data into a pre-trained equity penetration model, and predict the equity control relationship of the supplier and the ownership of the supplier's control right based on the equity penetration model; Determine the risk level based on the degree of influence and the possibility of influence of the predicted equity control relationship of the supplier and the ownership of the supplier's control right on the project.

2. The risk identification method based on supplier association relationship according to claim 1, wherein Also including: Train the equity penetration model: Preprocess the data in the associated relationship risk index library to form a training data set; Generate a supplier equity relationship diagram and a natural person equity relationship diagram for each supplier based on the training data set; wherein, each node in the supplier equity relationship diagram and the natural person equity relationship diagram represents an enterprise or an individual, the edge represents the equity relationship, and the weight of the edge represents the shareholding ratio; Obtain the corresponding supplier characteristic data for each supplier based on each supplier equity relationship diagram and each natural person equity relationship diagram, and repeatedly train a random forest model based on each supplier characteristic data to obtain an equity penetration model whose output is the predicted equity control relationship of the supplier and the ownership of the supplier's control right.

3. The risk identification method based on supplier association relationships according to claim 1 or 2, wherein Also including: Construct the associated relationship risk index library, including: Obtain historical bidding data from the e-bidding system, and extract at least supplier participation information, bidding information, and winning bid information from the historical bidding data; Obtain the external public data of the supplier from external public data sources, and the external public data of the supplier includes multiple combinations of the following: actual controller of the enterprise, superior and subordinate information of the enterprise, inter-enterprise associated relationship, enterprise associated risk, enterprise suppliers, main member information of the enterprise, member appointment information, and partner information; The associated relationship risk index library is composed of the historical bidding data and the external public data of the supplier.

4. The risk identification method based on supplier association relationships according to claim 1 or 2, characterized in that The generation of the supplier equity relationship diagram and the natural person equity relationship diagram for each supplier includes: Extract enterprises, individuals, and shareholding ratios from the obtained data, use the enterprises and individuals as nodes respectively, and the shareholding ratio as the edge; Determine the root node, and calculate the hierarchy and path from the root node to each node based on the obtained data to form the corresponding supplier equity relationship diagram and natural person equity relationship diagram.

5. The risk identification method based on supplier association relationship according to claim 1 or 2, characterized in that, The obtaining of the supplier characteristic data based on the supplier equity relationship diagram and the natural person equity relationship diagram includes: Map the nodes in the supplier equity relationship diagram and the natural person equity relationship diagram to low-dimensional vectors; Starting from a certain node, randomly select neighbor nodes for a walk to generate a series of node sequences, and capture the hierarchy, paths, and indirect shareholding information in the supplier equity relationship graph and the natural person equity relationship graph; Use the Word2Vec model to learn the low-dimensional vectors of each node in the node sequence, obtain node embeddings, and use the node embeddings as supplier feature data; among them, the node embeddings include an adjacency matrix representing the connection relationship between nodes and a feature matrix representing the shareholding ratio, equity hierarchy, and shareholder type of the nodes.

6. The risk identification method based on supplier association relationship according to claim 1, wherein The predicting the equity control relationship of the supplier and the attribution of the supplier's control right based on the equity penetration model includes: The equity penetration model calculates the direct shareholding ratio, indirect shareholding ratio, and total shareholding ratio of each node based on the supplier feature data; The equity penetration model outputs the predicted equity control relationship of the supplier and the attribution of the supplier's control right based on the direct shareholding ratio, the indirect shareholding ratio, and the total shareholding ratio.

7. The risk identification method based on supplier association relationship according to claim 1, wherein The determining the risk level based on the degree of influence and the likelihood of influence of the predicted equity control relationship of the supplier and the attribution of the supplier's control right on the project includes: Determine the label category according to the predicted equity control relationship of the supplier and the attribution of the supplier's control right; Based on the label category, determine the probability scores of the degree of influence, the likelihood of influence, and the risk type on the project; Obtain the label risk score based on the degree of influence and the likelihood of influence; Determine the risk level based on the label risk score and the probability score of the risk type.

8. The risk identification method based on supplier association relationship according to claim 1, wherein Further includes: Generate a warning notification message based on the determined risk level, and send the warning notification message to the user terminal by means of a system internal message, email, or text message.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the risk identification method based on supplier association relationships described in any one of claims 1 to 8.

10. An electronic device, characterized in that, The electronic device includes: A processor and a memory; The memory stores program instructions; The processor is used to run the program instructions to execute the risk identification method based on supplier association relationships described in any one of claims 1 to 8.

Images