A multi-level correlation relationship penetration enterprise risk transmission processing method and system

Abstract

This invention discloses a method and system for processing enterprise risk transmission data through multi-level relationship penetration, belonging to the field of data information processing technology. It involves standardizing multi-source related data to obtain standardized related data; based on this standardized related data, a multi-level relationship graph is constructed with the target enterprise as the core node, including equity relationships, personnel relationships, transaction relationships, contact information relationships, and implicit relationships, penetrating down to the actual controller; the comprehensive relationship weight of each relationship in the multi-level relationship graph is determined; a risk transmission model is constructed, and a risk transmission coefficient is determined based on the model; enterprise risk is assessed based on the risk transmission coefficient and a preset risk threshold, and an alarm is triggered when the risk transmission coefficient exceeds the preset risk threshold. This invention provides more accurate risk transmission calculations and allows for traceability of risk sources.

Description

Technical Field

[0001] This invention relates to a method and system for processing enterprise risk transmission data that penetrates multi-level relationships, belonging to the field of data information processing technology. Background Technology

[0002] In the current economic environment, large corporate groups are formed through equity control and cross-appointment of personnel. When any company within a group experiences defaults, irregularities, or financial difficulties, these risks can easily spread to other companies within the group through related relationships, potentially triggering regional or systemic risks. This poses serious challenges to financial institutions' credit lines, supply chain finance security, and legitimate government regulation. Therefore, to improve risk identification and prevention capabilities, multi-dimensional analysis and hierarchical penetration of the relationships between these corporate entities are necessary, along with quantitative calculations and path localization of the risk transmission process. Current technologies primarily identify relationships between companies using knowledge graphs and the Delphi method, but lack quantitative assessment of risk transmission between related companies and cannot distinguish whether the risk originates from external transmission or is generated by the companies themselves. Summary of the Invention

[0003] Purpose of the invention: In order to solve the problems of incomplete identification of enterprise relationships, inaccurate quantification of risk transmission, inability to determine the source of risk, and difficulty in locating critical paths, the present invention provides a method for handling enterprise risk transmission through multi-level relationship penetration.

[0004] Technical solution: To achieve the above objectives, the technical solution adopted by this invention is as follows: A method for handling enterprise risk transmission through multi-level relationships includes the following steps: Step S1: Collect multi-source related data of the enterprise, and perform standardization processing on the multi-source related data to obtain standardized related data.

[0005] Step S2: Based on standardized related data, with the target company as the core node, construct a multi-level relationship graph of equity relationship, personnel relationship, transaction relationship, contact information relationship, and implicit relationship, until it penetrates to the actual controller.

[0006] Step S3: Based on the risk transmission theory, determine the comprehensive association weight of a single association in the multi-level association graph according to the association type and association level.

[0007] Step S4: Based on the inherent risk value of the related enterprises and the comprehensive association weight, construct a risk transmission model according to the multi-level association relationship graph, and determine the risk transmission coefficient according to the risk transmission model.

[0008] Step S5: Assess the enterprise risk based on the risk transmission coefficient and the preset risk threshold. When the risk transmission coefficient exceeds the preset risk threshold, trigger an alarm.

[0009] Preferably, the risk transmission model is as follows:

[0010] in, Indicates core node enterprises With enterprises The risk transmission coefficient, This represents the risk transmission weighting coefficient of core node enterprises. This represents the risk weighting coefficient of the related enterprise itself. Indicates the weight of judicial risk factors. Indicates the weight of financial risk factors. Indicates the weight of public opinion risk factors. Indicates judicial risk factors, Indicates financial risk factor, Indicates public opinion risk factors, Indicates enterprise With enterprises The overall correlation weight between them This indicates the inherent risk value of the related enterprise itself.

[0011] The preferred formula for calculating the comprehensive correlation weight is as follows:

[0012] in, Indicates enterprise With enterprises The overall correlation weight between them Indicates the weight of the association type. This represents a generally accepted neutral and stable value in the industry. For related levels, Indicates the strength of the association.

[0013] Preferably, when determining the overall association weight of a single association, if cross-shareholding, cyclical appointment, or closed association path loop transmission is detected, only the one with the largest overall association weight is retained.

[0014] Preferably, when an alarm is triggered, the core transmission path is selected based on the risk transmission coefficient and a preset risk threshold, and key nodes are marked, and the risk level is classified according to the risk value range.

[0015] Preferably, it also includes real-time collection and updating of related data, periodic recalculation of maps and risk indicators, and incremental updates.

[0016] Preferably, the multi-source associated data includes entity identifiers, association types, initial values ​​of association strength, data timestamps, business registration information, legal litigation information, equity holding information, senior management appointments, related-party transactions, financial and guarantee information, and public opinion data.

[0017] Preferably, the contact information association includes at least one of the following: same registered address, office address, contact number, email address, domain name, and IP address. Implicit association includes at least one of the following: same personnel, similar equity structure, or suspected actual controller.

[0018] Preferably, the implicit association is quantitatively identified using a feature similarity scoring method, including the following steps: Step S211: Extract the identity information of the legal representatives, supervisors, financial officers, and managers of suspected related entities, and calculate the personnel overlap score.

[0019] Step S212: Extract the registration address and office address information, and calculate the address similarity score.

[0020] Step S213: Extract contact phone number and email address information, and calculate the overlap score of contact information.

[0021] Step S214: Extract business scope and registered capital structure information, and calculate business similarity score.

[0022] Step S215: Based on the weighted summation of the personnel overlap score, address similarity score, contact information overlap score, and business similarity score, the implicit association strength score is obtained.

[0023] Step S216: When the latent association strength score exceeds the preset latent association threshold, it is determined that a latent association exists.

[0024] Another objective of this invention is to provide a multi-level relationship penetration enterprise risk transmission processing system, used to realize a multi-level relationship penetration enterprise risk transmission processing method, including a data acquisition and preprocessing module, a relationship graph construction module, a weight calculation module, a risk transmission calculation module, and a risk identification module, wherein: The data acquisition and preprocessing module is used to collect multi-source related data of the enterprise and to perform standardization processing on the multi-source related data to obtain standardized related data.

[0025] The association graph construction module is used to construct a multi-level association graph based on standardized association data, with the target enterprise as the core node, including equity association, personnel association, transaction association, contact information association, and implicit association, down to the actual controller.

[0026] The weight calculation module is used to determine the comprehensive association weight of a single association in a multi-level association graph based on the association type and association level according to the risk transmission theory.

[0027] The risk transmission calculation module is used to determine the risk transmission coefficient based on the risk transmission model.

[0028] The risk identification module is used to assess enterprise risks based on the risk transmission coefficient and a preset risk threshold. When the risk transmission coefficient exceeds the preset risk threshold, an alarm is triggered.

[0029] Compared with the prior art, the present invention has the following advantages: 1. The scope of content covered by the association identification has been expanded to include equity, personnel, transactions, contact information, and implicit association information.

[0030] 2. More accurate risk transmission calculation. A hierarchical attenuation method is adopted to address the overestimation of risks at distant levels and reduce interference. Decoupling calculations reduce misjudgments.

[0031] 3. Results are explainable and attributable. Effective separation of transmitted risks and enterprise-specific risks, traceability of risk sources, and facilitating regulatory, risk control approval, and business decision-making.

[0032] 4. High execution efficiency. Supports incremental updates, reducing computational costs. It can support both large-scale enterprise group analysis and precise calculations for single groups of related enterprises. Computational complexity is controllable, making it suitable for system deployment. Attached Figure Description

[0033] Figure 1 This is a flowchart of the present invention.

[0034] Figure 2 This is a schematic diagram of the risk transmission calculation module of the present invention.

[0035] Figure 3 This is a schematic diagram of the five-category association recognition system architecture. Detailed Implementation

[0036] The present invention will be further illustrated below with reference to the accompanying drawings and specific embodiments. It should be understood that these examples are for illustrative purposes only and are not intended to limit the scope of the invention. After reading this invention, any modifications of the invention in various equivalent forms by those skilled in the art will fall within the scope defined by the appended claims.

[0037] Example Large corporate groups are often formed through equity control and cross-appointment of personnel. When a risk event occurs in any company within the group, it can easily spread to other companies through related relationships. To address the problems of incomplete identification of corporate relationships, inaccurate quantification of risk transmission, inability to determine the source of risk, and difficulty in locating critical paths, this embodiment provides a multi-level, interconnected method for handling corporate risk transmission. Figure 1-3 As shown, it includes the following steps: Step S1: Collect multi-source related data of the enterprise, and perform standardization processing on the multi-source related data to obtain standardized related data. The multi-source related data is cleaned, deduplicated, and standardized to obtain standardized related data.

[0038] In some examples, the multi-source associated data includes subject identifiers, association types, initial values ​​of association strength, data timestamps, business registrations, legal litigation, equity holdings, senior management appointments, related-party transactions, financial and guarantee data, and public opinion data.

[0039] Step S2: Based on standardized related data, with the target company as the core node, construct a multi-level relationship graph of equity relationship, personnel relationship, transaction relationship, contact information relationship, and implicit relationship, until it penetrates to the actual controller.

[0040] In some examples, the contact information association includes at least one of the following: the same registered address, office address, contact number, email address, domain name, and IP address; the implicit association includes at least one of the following: the same team, similar equity structure, and suspected actual controller.

[0041] In some examples, the implicit associations are quantitatively identified using a feature similarity scoring method, including the following steps: Step S211: Extract the identity information of the legal representatives, supervisors, financial officers, and managers of suspected related entities, and calculate the personnel overlap score.

[0042] Step S212: Extract the registration address and office address information, and calculate the address similarity score.

[0043] Step S213: Extract contact phone number and email address information, and calculate the overlap score of contact information.

[0044] Step S214: Extract business scope and registered capital structure information, and calculate business similarity score.

[0045] Step S215: Based on the weighted summation of the personnel overlap score, address similarity score, contact information overlap score, and business similarity score, the implicit association strength score is obtained.

[0046]

[0047] in, Indicates the score of the strength of the implicit association. The score indicates the degree of overlap among personnel. This represents the address similarity score. The score indicates the degree of overlap in contact information. This represents the score for business similarity.

[0048] Step S216: When the latent association strength score exceeds the preset latent association threshold, it is determined that a latent association exists.

[0049] In some examples, when the latent association strength score When the value is greater than or equal to 0.6, it is determined that there is a hidden relationship between the subjects.

[0050] Step S3: Based on the risk transmission theory, determine the comprehensive association weight of a single association in the multi-level association graph according to the association type and association level.

[0051] In some examples, the association type refers to equity association, personnel association, transaction association, contact information association, and implicit association. The association type weights satisfy the following order: equity association > personnel association > transaction association > contact information association > implicit association. The weight range for equity association is 0.30–0.50, for personnel association it is 0.20–0.30, for transaction association it is 0.10–0.20, for contact information association it is 0.05–0.15, and for implicit association it is 0.02–0.10. The sum of the weights for the five types is 1.

[0052] In some examples, the weight for equity association is 0.40, the weight for personnel association is 0.25, the weight for transaction association is 0.15, the weight for contact information association is 0.12, and the weight for implicit association is 0.08.

[0053] The formula for hierarchy decay of related levels is as follows:

[0054] in, Indicates hierarchical weight. This represents a generally accepted neutral and stable value in the industry. ∈[0.7,0.9], For related levels, ;when At that time, the weights were 1.0 for level 1, 0.8 for level 2, 0.64 for level 3, 0.512 for level 4, and 0.41 for level 5. According to the theory of financial risk contagion, risk diminishes by approximately 20% with each layer of transmission. It is a generally accepted neutral and stable value in the industry, and conforms to the decreasing exposure law of the correlation hierarchy in the KMV model.

[0055] The comprehensive weight is calculated based on the association type weight, hierarchy weight, and association strength. When the same entity has multiple types of associations, the comprehensive weight is the sum of the comprehensive weights of all associations. The formula for calculating the comprehensive association weight is as follows:

[0056] but:

[0057] in, Indicates enterprise With enterprises The overall correlation weight between them Indicates the weight of the association type. Indicates hierarchical weight. This represents a generally accepted neutral and stable value in the industry. For related levels, Indicates the strength of the association.

[0058] When determining the overall association weight of a single relationship, if cross-shareholding, cyclical appointment, or closed-loop transmission of association path (A→B→C→A) is detected, only the one with the largest overall association weight is retained to avoid repeated accumulation of high risk values.

[0059] Step S4, as follows Figure 2 As shown, a risk transmission model is constructed based on the inherent risk value of related enterprises and the comprehensive association weight, according to the multi-level association relationship graph, and the risk transmission coefficient is determined based on the risk transmission model.

[0060] The risk transmission model is as follows:

[0061] in, Indicates core node enterprises With enterprises The risk transmission coefficient, The risk transmission coefficient represents the intensity of risk transmission; the larger the value, the greater the risk impact of the core node enterprise on related enterprises. The closer the value is to 1, the more likely the risk is caused by external transmission; the closer it is to 0, the more likely the risk is caused by internal factors. This represents the risk transmission weighting coefficient of core node enterprises. , This represents the risk weighting coefficient of the related enterprise itself. , , Indicates the weight of judicial risk factors. Indicates the weight of financial risk factors. Indicates the weight of public opinion risk factors, satisfying In some examples, the value is... , , , Indicates judicial risk factors, Indicates financial risk factor, Indicates public opinion risk factors, , , The value range is [0,1]. Indicates enterprise With enterprises The overall correlation weight between them This represents the inherent risk value of the related enterprise, which is assessed based on the financial indicators and legal records of the related entity.

[0062] In some examples, the total risk value of related enterprises is determined based on their inherent risk value. The formula for calculating the total risk value of related enterprises is as follows:

[0063]

[0064]

[0065] in, This represents the total risk value of related enterprises. This represents the risk transmission weighting coefficient of core node enterprises. This represents the risk weighting coefficient of the related enterprise itself. Indicates the weight of judicial risk factors. Indicates the weight of financial risk factors. Indicates the weight of public opinion risk factors. Indicates judicial risk factors, Indicates financial risk factor, Indicates public opinion risk factors, Indicates enterprise With enterprises The overall correlation weight between them This indicates the inherent risk value of the related enterprise itself. This indicates the risk value of the core node enterprise.

[0066] Step S5: Based on the risk transmission coefficient, core transmission paths are selected according to preset risk thresholds, key nodes are marked, risk level classification is completed according to risk value ranges, structured risk transmission data is output and visualized.

[0067] In some examples, the preset risk threshold is 0.1. If the risk transmission coefficient is greater than or equal to the preset risk threshold, it is considered to be risky. The risk value range is divided into: 0–0.2 low risk, 0.2–0.5 medium risk, 0.5–0.8 relatively high risk, and 0.8–1 extremely high risk.

[0068] Step S6: Collect and update related data in real time, recalculate the map and risk indicators periodically to achieve incremental updates, and trigger an alarm and push it to the terminal when the main risk transmission coefficient exceeds the preset threshold.

[0069] In some examples, a multi-level relationship penetration enterprise risk transmission processing system is provided to implement a multi-level relationship penetration enterprise risk transmission processing method, such as... Figure 1 As shown, it includes a data acquisition and preprocessing module, a correlation graph construction module, a weight calculation module, a risk transmission calculation module, a risk identification module, and a dynamic update and early warning module, wherein: The data acquisition and preprocessing module is used to collect multi-source related data of the enterprise and to perform standardization processing on the multi-source related data to obtain standardized related data.

[0070] The association graph construction module is used to construct a multi-level association graph based on standardized association data, with the target enterprise as the core node, including equity association, personnel association, transaction association, contact information association, and implicit association, down to the actual controller.

[0071] The weight calculation module is used to determine the comprehensive association weight of a single association in a multi-level association graph based on the association type and association level according to the risk transmission theory.

[0072] The risk transmission calculation module is used to determine the risk transmission coefficient based on the risk transmission model.

[0073] The risk identification module is used to assess enterprise risks based on the risk transmission coefficient and a preset risk threshold. When the risk transmission coefficient exceeds the preset risk threshold, an alarm is triggered.

[0074] The dynamic update and early warning module is used to collect and update related data in real time, recalculate the map and risk indicators periodically, and trigger an alarm and push it to the terminal when the main risk transmission coefficient exceeds the preset risk threshold.

[0075] Example To better illustrate the enterprise risk transmission processing method of multi-level relationship penetration according to the present invention, we provide the following examples.

[0076] 1. Scenario and Data Preparation A core enterprise A within a certain group is selected as the source node of risk, and its related enterprises are B and C. Data on business registration, equity, senior management, transactions, legal matters, finance, and public opinion of these enterprises are collected and then cleaned, deduplicated, and standardized to form standardized related data.

[0077] 2. Relationship Identification and Graph Construction Centered on core enterprise A, identify related entities B and C layer by layer, clarify the relationship between A and B, and between A and C, and label the relationship type and level to construct a relationship graph.

[0078] 1) A and B: There are two types of relationships. Equity relationship: A directly and wholly owns B. Contact information relationship: They share the same registered address and contact number.

[0079] 2) A and C: Both types of relationships exist. Equity relationship: A holds 5% of C's shares, a small minority stake. Transaction relationship: A and C have frequent transactions.

[0080] 3) All are first-level associations, hierarchy It penetrates to the directly related layer.

[0081] 4) The detection process eliminates cross-shareholding and circular transmission, thus eliminating the need for path elimination.

[0082] 3. Weight Calculation 1) Weights of association types (preferred values): Equity 0.40, Personnel 0.25, Transactions 0.15, Contact Information 0.12, Implicit 0.08; 2) Hierarchical decay algorithm: , The preferred value is 0.8. , .

[0083] 3) Initial value of association strength: Quantified into a range of 0-1 based on factors such as shareholding ratio, personnel overlap, and transaction proportion; 3.1), AB equity: wholly owned. .

[0084] 3.2), A and B contact information: their addresses and phone numbers overlap. .

[0085] 3.3), AC equity: 5% shareholding, .

[0086] 3.4), AC trading: regular trading, .

[0087] 4) Single-class association weight: 4.1), Equity = 0.4 × 1.0 × 1.0 = 0.4.

[0088] 4.2), Contact information = 0.12 × 1.0 × 0.8 = 0.096.

[0089] 4.3), Equity = 0.4 × 1.0 × 0.05 = 0.02.

[0090] 4.4), Transaction = 0.15 × 1.0 × 0.6 = 0.09.

[0091] 5) Overall weighting:

[0092]

[0093] 4. Core Node Risk Calculation The risk value of core enterprise A is obtained by weighting three factors: legal, financial, and public opinion.

[0094] in, .

[0095] Substitute: .

[0096] .

[0097] 5. Decoupled Risk Transmission Calculation Pick ; B's own risks C's own risks .

[0098] 1) Risk value of related entities:

[0099]

[0100]

[0101] 2) Risk transmission coefficient

[0102]

[0103]

[0104] 6. Path identification and risk level determination Transmission coefficient threshold:

[0105] Risk levels: 0-0.2 - Low risk; 0.2-0.5 - Medium risk; 0.5-0.8 - Higher risk; 0.8-1 - Very high risk; Judgment result: A→B: This is a key transmission path with higher risks.

[0106] A→C: This is a key transmission path for medium-risk.

[0107] 7. Dynamic updates and early warnings The system collects data in real time, updates the data incrementally on a monthly basis, and recalculates the map and risk indicators. When the risk value of the main body exceeds the preset risk threshold (0.7), an early warning is automatically triggered and pushed to the terminal.

[0108] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for handling enterprise risk transmission through multi-level relationship penetration, characterized in that, Includes the following steps: Step S1: Collect multi-source related data of the enterprise, and perform standardization processing on the multi-source related data to obtain standardized related data; Step S2: Based on standardized related data, with the target company as the core node, construct a multi-level relationship graph of equity relationship, personnel relationship, transaction relationship, contact information relationship, and implicit relationship, until it penetrates to the actual controller; Step S3: Based on the risk transmission theory, determine the comprehensive association weight of a single association in the multi-level association graph according to the association type and association level; Step S4: Based on the inherent risk value of the related enterprises and the comprehensive association weight, construct a risk transmission model according to the multi-level association relationship map, and determine the risk transmission coefficient according to the risk transmission model; Step S5: Assess the enterprise risk based on the risk transmission coefficient and the preset risk threshold. When the risk transmission coefficient exceeds the preset risk threshold, trigger an alarm.

2. The enterprise risk transmission handling method based on multi-level relationship penetration according to claim 1, characterized in that: The risk transmission model is as follows: in, Indicates core node enterprises With enterprises The risk transmission coefficient, This represents the risk transmission weighting coefficient of core node enterprises. This represents the risk weighting coefficient of the related enterprise itself. Indicates the weight of judicial risk factors. Indicates the weight of financial risk factors. Indicates the weight of public opinion risk factors. Indicates judicial risk factors, Indicates financial risk factor, Indicates public opinion risk factors, Indicates enterprise With enterprises The overall correlation weight between them This indicates the inherent risk value of the related enterprise itself.

3. The enterprise risk transmission handling method based on multi-level relationship penetration according to claim 2, characterized in that: The formula for calculating the overall correlation weight is as follows: in, Indicates enterprise With enterprises The overall correlation weight between them Indicates the weight of the association type. This represents a generally accepted neutral and stable value in the industry. For related levels, Indicates the strength of the association.

4. The enterprise risk transmission handling method based on multi-level relationship penetration according to claim 3, characterized in that: When determining the overall association weight of a single relationship, if cross-shareholding, cyclical appointment, or closed-loop transmission of related paths are detected, only the one with the largest overall association weight is retained.

5. The enterprise risk transmission handling method based on multi-level relationship penetration according to claim 4, characterized in that: When an alarm is triggered, the core transmission path is selected based on the risk transmission coefficient and the preset risk threshold, and key nodes are marked. The risk level is classified according to the risk value range.

6. The enterprise risk transmission handling method based on multi-level relationship penetration according to claim 5, characterized in that: It also includes real-time collection and updating of related data, periodic recalculation of maps and risk indicators, and incremental updates.

7. The enterprise risk transmission handling method based on multi-level relationship penetration according to claim 6, characterized in that: The multi-source associated data includes entity identifiers, association types, initial values ​​of association strength, data timestamps, business registration information, legal litigation information, equity holdings, senior management appointments, related-party transactions, financial and guarantee information, and public opinion data.

8. The enterprise risk transmission handling method based on multi-level relationship penetration according to claim 7, characterized in that: Contact information association includes at least one of the following: same registered address, office address, contact number, email address, domain name, and IP address; implicit association includes at least one of the following: same personnel, similar equity structure, and suspected actual controller.

9. The enterprise risk transmission handling method based on multi-level relationship penetration according to claim 8, characterized in that: The implicit associations are quantitatively identified using a feature similarity scoring method, including the following steps: Step S211: Extract the identity information of the legal representatives, supervisors, financial officers, and managers of suspected related entities, and calculate the personnel overlap score; Step S212: Extract the registration address and office address information, and calculate the address similarity score; Step S213: Extract contact phone number and email address information, and calculate the overlap score of contact information; Step S214: Extract business scope and registered capital structure information, and calculate business similarity score; Step S215: Based on the weighted sum of the personnel overlap score, address similarity score, contact information overlap score, and business similarity score, the implicit association strength score is obtained. Step S216: When the latent association strength score exceeds the preset latent association threshold, it is determined that a latent association exists.

10. A system for handling enterprise risk transmission based on the multi-level relationship penetration method described in claim 1, characterized in that: It includes a data acquisition and preprocessing module, a correlation graph construction module, a weight calculation module, a risk transmission calculation module, and a risk identification module, among which: The data acquisition and preprocessing module is used to collect multi-source related data of the enterprise and perform standardization processing on the multi-source related data to obtain standardized related data. The association graph construction module is used to construct a multi-level association graph based on standardized association data, with the target enterprise as the core node, including equity association, personnel association, transaction association, contact information association, and implicit association, down to the actual controller. The weight calculation module is used to determine the comprehensive association weight of a single association in a multi-level association graph based on the association type and association level according to the risk transmission theory. The risk transmission calculation module is used to determine the risk transmission coefficient based on the risk transmission model. The risk identification module is used to assess enterprise risks based on the risk transmission coefficient and a preset risk threshold. When the risk transmission coefficient exceeds the preset risk threshold, an alarm is triggered.

Images