ESG evaluation data management system based on multi-source data fusion and dynamic verification

Abstract

The invention discloses an ESG evaluation data management system based on multi-source data fusion and dynamic verification, and relates to the technical field of enterprise ESG evaluation. Comprising a multi-source data access module, an intelligent standardized processing module, an ESG index accounting engine, a hierarchical data verification and processing module, a data quality dynamic scoring module, an intelligent updating and tracing module and a standardized ESG data packet generation module. The system realizes automatic closed-loop management through hierarchical verification, priority acquisition and dynamic scoring, outputs a standardized data packet with quality proof, and solves the problems of insufficient data authenticity and non-closed-loop management in the prior art.

Description

Technical Field

[0001] This invention relates to the field of enterprise ESG assessment technology, specifically to an ESG assessment data governance system based on multi-source data fusion and dynamic verification. Background Technology

[0002] Environmental, social, and governance (ESG) assessments have become a core tool for measuring a company's sustainable development capabilities. However, current corporate ESG data suffers from three major pain points: First, diverse data sources (including self-disclosure by companies, government regulatory data, and data from third-party institutions) lead to inconsistent data formats and definitions, creating "data silos." Second, data credibility is poor; voluntarily disclosed data carries the risk of "greenwashing" (referring to companies falsely promoting environmental protection to mislead stakeholders), and there is a lack of efficient verification mechanisms. Third, data verification is costly; current technologies mainly rely on manual verification, which is inefficient and prone to errors. These problems result in low-quality underlying data in the existing ESG assessment system, severely impacting the accuracy and credibility of the final rating results.

[0003] In recent years, although various ESG evaluation technologies have emerged, such as deep learning models based on weight allocation (CN114943458A), indicator systems based on the analytic hierarchy process (CN119313471A), conflict detection methods based on data fusion (CN119477059A), and full-scenario evaluation systems (CN120764965A), these solutions still have significant shortcomings in areas such as data authenticity verification, intelligent fusion of multi-source data, and automated governance loops. These shortcomings can be summarized into four core dimensions: (1) strong human intervention, resulting in low efficiency in verification and conflict resolution; (2) weak domain verification, lacking dedicated cross-validation logic for ESG business, making it difficult to prevent "greenwashing"; (3) non-closed-loop governance, with breaks in the data verification, decision-making, and update processes, failing to form an automated governance flow; and (4) non-standard output, with results mostly being ratings or pending data, unable to provide standardized data products with quality certification that can be directly accepted by downstream systems. These systemic defects prevent the provision of a credible data foundation for ESG evaluation from the source. Therefore, there is an urgent need for a system and method that can automatically integrate, standardize, and intelligently verify multi-source ESG data to ensure data quality from the source, prevent "greenwashing" behavior, and provide directly credible standard data products for downstream rating applications. Summary of the Invention

[0004] To address the aforementioned shortcomings in existing technologies, this invention provides an ESG evaluation data governance system based on multi-source data fusion and dynamic verification, which solves the problems of insufficient data authenticity and non-closed-loop governance in existing technologies.

[0005] To achieve the aforementioned objectives, the technical solution adopted by this invention is: an ESG evaluation data governance system based on multi-source data fusion and dynamic verification, comprising: The multi-source data access module is used to obtain the enterprise's original ESG-related data from multiple heterogeneous data sources; The intelligent standardization processing module is connected to the multi-source data access module and has a built-in ESG data standardization dictionary. The ESG data standardization dictionary defines the concepts, units of measurement and statistical standards of core ESG indicators and is used to clean, format and standardize the units of the acquired raw data. The ESG indicator calculation engine is connected to the intelligent standardization processing module and is used to automatically calculate derived ESG indicators, including carbon emissions, green income ratio, and environmental risk index, according to the preset formulas in the ESG data standardization dictionary. The hierarchical data verification and processing module is connected to the ESG indicator calculation engine and includes a logical verification unit, a range verification unit, and a cross-verification unit, which is used to perform automated hierarchical verification and processing on the processed and calculated data. The dynamic data quality scoring module is connected to the hierarchical data verification and processing module and is used to dynamically calculate and output the enterprise's ESG data quality score based on the data verification results. The intelligent update and traceability module is connected to the hierarchical data verification and processing module and the data quality dynamic scoring module. It is used to automatically update the corresponding data in the enterprise unified data pool after determining the final acceptance value according to the priority acceptance rules, and record complete verification and acceptance logs to establish a data traceability chain. The standardized ESG data packet generation module, connected to the intelligent update and traceability module, is used to generate standardized, independently verifiable data packets for direct use by upper-level ESG rating models or financial institution risk control systems.

[0006] Furthermore, the calculation formulas for carbon emissions, green income ratio, and environmental risk index in the ESG indicator calculation engine are as follows: carbon emissions :

[0007]

[0008]

[0009] in, , and The figures represent emissions for ranges 1, 2, and 3, respectively. Range 1 represents direct emissions, range 2 represents indirect emissions from purchased electricity and heat, and range 3 represents other indirect emissions. fuel consumption, fuel Emission factors For purchased electricity, For power grid emission factors, For externally purchased calories, Thermal emission factor; Green income share :

[0010] in, For green business revenue that meets the standards, This refers to the company's total operating revenue; Environmental Risk Index :

[0011] in, , , and These are the number of environmental penalties, the number of records of exceeding emission standards, the amount of high-risk waste generated, and the environmental sensitivity of the location. , , and For weights.

[0012] Furthermore, the logic verification unit is used to perform hard logic rule checks, including numerical non-negativity, consistency between items and totals, and continuity of time series; The range verification unit is used to check whether the data fluctuation is within a preset reasonable range based on historical data and industry benchmarks, mark data that exceeds the range as needing verification, and automatically send an inquiry letter. The cross-validation unit is used to automatically compare the calculation results from external authoritative data sources with the enterprise's internal declaration values, including environmental data validation, compliance data validation, and financial data validation. Data with differences exceeding a preset threshold is marked as seriously inconsistent and triggers priority acceptance rules.

[0013] Furthermore, the enterprise's ESG data quality score The calculation formula is:

[0014] in, , , and These are the data integrity score, data consistency score, data timeliness score, and verification pass rate score. , , and The weights assigned to each dimension;

[0015] in, The number of indicators that have been reported. This represents the total number of core indicators;

[0016] in, The number of consistency rules that passed. This represents the total number of consistency rules.

[0017] in, For the number of days of delay, This is the deduction coefficient;

[0018] in, The number of indicators that have passed verification. This represents the total number of indicators that were verified.

[0019] Furthermore, the priority acceptance rule determines the final value based on the authority of the data source and historical consistency.

[0020] Furthermore, the data packets generated by the standardized ESG data packet generation module include all verified core ESG indicator data, data source information and acceptance criteria, data verification status records, final data quality scores, and data update timestamps.

[0021] An ESG assessment data governance method based on multi-source data fusion and dynamic verification includes the following steps: S1: Obtain the enterprise's raw ESG-related data in parallel from multiple heterogeneous data sources; S2: The raw data is cleaned, formatted and standardized using an ESG data standardization dictionary. The ESG data standardization dictionary defines the concepts, units of measurement and statistical scope of core ESG indicators. S3: Based on the accounting formula in the ESG data standardization dictionary, automatically calculate derived ESG indicators including carbon emissions, green income ratio, and environmental risk index; S4: Perform automated hierarchical verification and processing on the processed and calculated data; S4 includes: S41: Perform hard logic rule checks, including numerical non-negativity, consistency between items and totals, and time series continuity; S42: Execution range verification, based on historical data and industry benchmarks, checks whether data fluctuations are within a preset reasonable range, marks data that exceeds the range as pending verification, and automatically sends an inquiry letter; S43: Perform cross-validation, automatically compare the calculation results from external authoritative data sources with the enterprise's internal reported values, including environmental data validation, compliance data validation, and financial data validation. Data with differences exceeding a preset threshold is marked as seriously inconsistent and triggers priority acceptance rules. S5: Dynamically calculate and output the enterprise's ESG data quality score based on the data verification results; S6: After determining the final acceptance value based on the priority acceptance rules, automatically update the corresponding data in the enterprise unified data pool, record complete verification and acceptance logs, and establish a data traceability chain; S7: Generates standardized, independently verifiable data packages for direct use by upper-level ESG rating models or financial institution risk control systems.

[0022] A computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described ESG evaluation data governance method based on multi-source data fusion and dynamic verification.

[0023] The beneficial effects of this invention are: (1) This invention establishes a fully automatic verification-decision-update closed loop, replacing its conflict resolution mode that relies on manual intervention, and realizes the leap from "statistical conflict detection" to "business logic verification". Through the cross-validation rules specific to the ESG field, it solves the core defect of insufficient verification of data authenticity.

[0024] (2) This invention focuses on “ESG data authenticity governance” and for the first time forms a three-layer automatic verification and priority acceptance mechanism in the ESG scenario, and outputs a standardized and credible data package with quality score and acceptance basis, providing a data foundation that can be directly accepted for downstream systems such as rating and risk control.

[0025] (3) This invention provides a reliable data foundation for ESG rating agencies: the generated standardized data packages can be directly used as rating inputs, significantly improving the accuracy and credibility of rating results. It provides decision support for financial institutions' risk control: high-quality, verifiable ESG data provides a reliable basis for risk assessment in green lending, responsible investment, and other businesses. It provides monitoring tools for regulatory agencies: through automated data verification and quality assessment, it assists regulatory authorities in identifying corporate ESG data fraud. It promotes the standardization of ESG data: through a unified standardized dictionary and quality assessment system, it promotes the standardization and comparability of ESG data across the entire industry. Attached Figure Description

[0026] Figure 1 This is a flowchart of an ESG evaluation data governance method based on multi-source data fusion and dynamic verification according to the present invention. Detailed Implementation

[0027] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0028] An ESG assessment data governance system based on multi-source data fusion and dynamic verification includes: The multi-source data access module is used to obtain the enterprise's original ESG-related data from multiple heterogeneous data sources; Data sources include: authoritative official data sources: publicly available data interfaces of government agencies such as the ecological and environmental protection departments, power regulatory departments, and the China Securities Regulatory Commission; data sources declared by enterprises themselves: enterprise ESG reports, sustainable development reports, annual reports, etc.; data sources from third-party institutions: data from ESG rating agencies such as MSCI and FTSE Russell, as well as alternative data such as news media and social media.

[0029] The intelligent standardization processing module is connected to the multi-source data access module and has a built-in ESG data standardization dictionary. The ESG data standardization dictionary defines the concepts, units of measurement and statistical standards of core ESG indicators and is used to clean, format and standardize the units of the acquired raw data. The ESG indicator calculation engine is connected to the intelligent standardization processing module and is used to automatically calculate derived ESG indicators, including carbon emissions, green income ratio, and environmental risk index, according to the preset formulas in the ESG data standardization dictionary. The formulas for calculating carbon emissions, green income ratio, and environmental risk index in the ESG indicator calculation engine are as follows: According to the internationally accepted Greenhouse Gas Accounting System, carbon emissions are divided into three categories and then summed up.

[0030] carbon emissions :

[0031]

[0032]

[0033] in, , and The figures represent emissions for ranges 1, 2, and 3, respectively. Range 1 represents direct emissions, range 2 represents indirect emissions from purchased electricity and heat, and range 3 represents other indirect emissions. fuel consumption, fuel Emission factors For purchased electricity, For power grid emission factors, For externally purchased calories, Thermal emission factor; Scope 1 (Direct Emissions): Emissions directly from business activities, such as fuel combustion and company vehicles. This is calculated by summing the consumption of natural gas, diesel, gasoline, etc., declared by the company and multiplying each by its corresponding emission factor.

[0034] Scope 2 (Indirect emissions from purchased electricity and heat): Calculated using the total electricity consumption of enterprises and the average emission factor of the regional power grid obtained from the power sector through the data cross-validation unit. This is the key to demonstrating the cross-validation capability of this invention.

[0035] Scope 3 (Other Indirect Emissions): This may include employee travel, purchased goods and services, etc. The system can first calculate the more mature parts, such as employee commuting emissions, by estimating them based on the commuting distance, mode of transportation, and corresponding emission factors declared by the company.

[0036] Green revenue share: This measures the proportion of a company's main business that meets green and environmental protection standards.

[0037] Green income share :

[0038] in, For green business revenue that meets the standards, This refers to the company's total operating revenue; Numerical component (Green business revenue): Enterprises report the revenue of each business line in the declaration data according to standards such as the "Green Industry Guidance Catalogue". The accounting engine will identify and summarize the revenue based on a pre-set list of green industry codes.

[0039] Denominator (Total Revenue): Obtained directly from the company's financial data.

[0040] Verification: The system can incorporate third-party certification reports or use natural language processing technology to analyze corporate annual reports to assist in verifying the rationality of the definition of "green business".

[0041] Environmental Risk Index: This is a comprehensive composite indicator designed to quantify the environmental compliance and physical risks faced by enterprises.

[0042] Environmental Risk Index :

[0043] in, , , and These are the number of environmental penalties, the number of records of exceeding emission standards, the amount of high-risk waste generated, and the environmental sensitivity of the location. , , and For weights.

[0044] Number of environmental penalties: obtained through access to the official interface of the ecological and environmental protection department to ensure data authority.

[0045] Excessive emission records: Comparison results of data from the company's environmental self-monitoring platform with national standard limits.

[0046] High-risk waste generation: based on enterprise declarations and undergoes a reasonableness check through range verification (comparing with historical values ​​and industry averages).

[0047] Location environmental sensitivity: Incorporate external geographic information data, such as whether it is located within a water source protection area or an ecological red line.

[0048] Weighting: Weight , , and It is determined by domain experts through pre-set criteria or by analyzing historical environmental event data through machine learning models.

[0049] The hierarchical data verification and processing module is connected to the ESG indicator calculation engine and includes a logical verification unit, a range verification unit, and a cross-verification unit, which is used to perform automated hierarchical verification and processing on the processed and calculated data. The logic verification unit is used to perform hard logic rule checks, including numerical non-negativity, consistency between items and totals, and time series continuity. The range verification unit is used to check whether the data fluctuation is within a preset reasonable range based on historical data and industry benchmarks, mark data that exceeds the range as needing verification, and automatically send an inquiry letter. The cross-validation unit is used to automatically compare the calculation results from external authoritative data sources with the enterprise's internally declared values. This includes environmental data validation, compliance data validation, and financial data validation. Data with discrepancies exceeding a preset threshold is marked as seriously inconsistent and triggers priority acceptance rules. Specific implementation includes: Environmental data validation: The estimated indirect carbon emissions of the enterprise are calculated by multiplying the enterprise's electricity consumption obtained from the power sector by the regional power grid emission factor, and then compared with the enterprise's self-declared carbon emissions in Scope 2. Compliance data validation: The enterprise's environmental credit rating or administrative penalty records are obtained through access to the official data interface of the ecological and environmental protection department and compared with the enterprise's self-declared statement of no major environmental violations. Financial data validation: The consistency between the green tax incentives declared by the enterprise and the actual situation is verified by accessing data from the tax department.

[0050] The dynamic data quality scoring module is connected to the hierarchical data verification and processing module and is used to dynamically calculate and output the enterprise's ESG data quality score based on the data verification results. The enterprise's ESG data quality score The calculation formula is:

[0051] in, , , and These are the data integrity score, data consistency score, data timeliness score, and verification pass rate score. , , and The weights of each dimension Weight , , , It is preset by experts or dynamically adjusted by machine learning.

[0052] The weighting of each dimension in the aforementioned data quality scoring module is based on, but is not limited to, the following: Downstream application orientation: Different weight configuration schemes are preset according to the main application scenarios of ESG data packages (such as investment decision-making, internal management, and compliance disclosure). For example, for the investment decision-making scenario, due to the extremely high requirements for data authenticity, the weight of the verification pass rate score is set to the highest (e.g., 0.5) to significantly reflect the contribution of external verification to data credibility.

[0053] Quantification of expert experience: Through mathematical models such as the analytic hierarchy process, the judgments of domain experts on the relative importance of each dimension of data quality are transformed into quantitative weights, ensuring the objectivity and scientific nature of weight allocation.

[0054] Configurability and Adaptability: The weighting coefficients can be manually configured by system administrators according to actual needs. In another preferred embodiment, the system can also dynamically fine-tune the weights based on the evolution of the data source and feedback from downstream users.

[0055] The assessment can be adjusted according to the assessment objectives. For example, if credibility is emphasized, the verification pass rate can be given higher weight.

[0056] The role and calculation process of scoring elements (example): Suppose we want to evaluate the quality of a company's "environmental" data. This dimension has 10 core indicators (such as carbon emissions, energy consumption, water consumption, and waste volume).

[0057] a) Data integrity score Purpose: To measure the degree of missing required ESG data.

[0058] Calculation process: The system checks several of the 10 core indicators and finds that data was successfully obtained for each indicator (regardless of the data source). Assume that data is missing for two indicators.

[0059]

[0060] in, The number of indicators that have been reported. This represents the total number of core indicators; b) Data consistency score Purpose: To measure the internal consistency of data and its logical consistency with historical data.

[0061] Calculation process: The system executes the rules of the logic verification unit. For example, rule 1: "The sum of carbon emissions from all branches ≈ total carbon emissions"; rule 2: "The ratio of this year's water consumption to the previous year is within a reasonable range [0.8, 1.2]". Assume that a total of 5 consistency rules are executed, and 1 rule fails (e.g., the sum of carbon emissions from branches differs too much from the total).

[0062]

[0063] in, The number of consistency rules that passed. This represents the total number of consistency rules. c) Data timeliness score Purpose: To measure whether the frequency of data updates meets the requirements.

[0064] Calculation Process: The system defaults to requiring "annual data to be disclosed within four months of the end of the fiscal year." Assume the current date is June 1st, and the company's fiscal year ends on December 31st. The data should have been updated by April 30th. The company's data update timestamp is May 15th, a delay of 15 days. A decay function is used to calculate the score, for example:

[0065] Assuming the deduction coefficient is 2, the score = 100 - 15 × 2 = 70 points.

[0066] in, For the number of days of delay, This is the deduction coefficient; d) Verification pass rate score Function: This is the core scoring dimension of this invention, which directly measures the credibility of the data after it has been verified by the system.

[0067] Calculation process: The system summarizes the results of range verification and cross-validation. Assume that of the 8 reported indicators, the system is capable of verifying 6 of them (the other 2 may not be verifiable due to a lack of external data sources). Of these 6 verified indicators, 5 passed verification successfully, and 1 failed cross-validation (e.g., the reported carbon emissions differed significantly from the estimated electricity consumption).

[0068]

[0069] in, The number of indicators that have passed verification. This represents the total number of indicators that were verified.

[0070] Assume the weights for each dimension are set as follows: =0.2, =0.2, =0.1, =0.5 (emphasizing the credibility of the verification).

[0071] The overall quality score for the enterprise's environmental dimension data is: Total score = (80 × 0.2) + (80 × 0.2) + (70 × 0.1) + (83.3 × 0.5) = 16 + 16 + 7 + 41.65 = 80.65 If the data missing rate is greater than 50%, the system will automatically mark it as "low quality" and trigger manual intervention.

[0072] This score and its dimensional breakdown (completeness 80, consistency 80, timeliness 70, verification pass rate 83.3%) are written into a standardized data package, so downstream users can clearly see that although the company's data updates are slightly delayed, most of the key data is complete and has passed the system's authenticity verification, making the overall quality reliable.

[0073] The intelligent update and traceability module is connected to the hierarchical data verification and processing module and the data quality dynamic scoring module. It is used to automatically update the corresponding data in the enterprise unified data pool after determining the final acceptance value according to the priority acceptance rules, and record complete verification and acceptance logs to establish a data traceability chain. The priority acceptance rules determine the final value based on the authority of the data source and historical consistency.

[0074] The standardized ESG data packet generation module, connected to the intelligent update and traceability module, is used to generate standardized, independently verifiable data packets (such as JSON / XML format) for direct use by upper-level ESG rating models or financial institution risk control systems.

[0075] The data packets generated by the standardized ESG data packet generation module include all verified core ESG indicator data, data source information and acceptance criteria, data verification status records, final data quality scores, and data update timestamps.

[0076] This data packet serves as the sole trusted data source for direct access by upper-level ESG rating models or financial institution risk control systems.

[0077] The standardized ESG data packets generated by this invention contain the following key fields: Validation Status; Source of Trust; Quality Score; Update Timestamp. This structured field is designed to facilitate direct access by downstream rating, risk control, and auditing systems, enabling reliable data traceability and differentiated acceptance.

[0078] like Figure 1 As shown, an ESG evaluation data governance method based on multi-source data fusion and dynamic verification includes the following steps: S1: Obtain the enterprise's raw ESG-related data in parallel from multiple heterogeneous data sources; S2: The raw data is cleaned, formatted and standardized using an ESG data standardization dictionary. The ESG data standardization dictionary defines the concepts, units of measurement and statistical scope of core ESG indicators. S3: Based on the accounting formula in the ESG data standardization dictionary, automatically calculate derived ESG indicators including carbon emissions, green income ratio, and environmental risk index; S4: Perform automated hierarchical verification and processing on the processed and calculated data; S4 includes: S41: Perform hard logic rule checks, including numerical non-negativity, consistency between items and totals, and time series continuity; S42: Execution range verification, based on historical data and industry benchmarks, checks whether data fluctuations are within a preset reasonable range, marks data that exceeds the range as pending verification, and automatically sends an inquiry letter; S43: Perform cross-validation, automatically compare the calculation results from external authoritative data sources with the enterprise's internal reported values, including environmental data validation, compliance data validation, and financial data validation. Data with differences exceeding a preset threshold is marked as seriously inconsistent and triggers priority acceptance rules. S5: Dynamically calculate and output the enterprise's ESG data quality score based on the data verification results; S6: After determining the final acceptance value based on the priority acceptance rules, automatically update the corresponding data in the enterprise unified data pool, record complete verification and acceptance logs, and establish a data traceability chain; S7: Generates standardized, independently verifiable data packages for direct use by upper-level ESG rating models or financial institution risk control systems.

[0079] A computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described ESG evaluation data governance method based on multi-source data fusion and dynamic verification.

[0080] Without changing the overall objective and technical effect of the present invention, the following alternative solutions can also achieve the purpose of the invention: (1) Replacement of rules and models: The three-layer verification of logic / range / cross can be replaced by a scheme combining statistics and machine learning, such as replacing some fixed thresholds with anomaly detection (IQR, ARIMA residual threshold) based on distribution assumptions and seasonal decomposition; or introducing an anomaly interpretation model based on feature importance to improve the sensitivity of anomaly identification. (2) Replacement of acceptance decision mechanism: The priority acceptance of multi-source data can be replaced by a weighted credibility model or Bayesian posterior update, and the acceptance value can be dynamically calculated according to the source weight and historical consistency when conflicts occur; the weight can be automatically optimized according to industry, year or enterprise size. (3) Replacement of threshold and benchmark: The range verification threshold can be replaced by an adaptive control chart (EWMA / MCUSUM) or by the industry quantile bandwidth (P5–P95); the benchmark parameters in cross verification (such as power grid emission factor, industry energy consumption intensity, compliance penalty timeliness) can be automatically updated according to region / time period. (4) Alternatives to modularization and deployment: The system can be deployed in the cloud as a standalone application, or microservices can be used to make the layered verification and processing modules independent services that can be called via API, or edge verification proxies can be deployed on the data source side to improve real-time performance. (5) Alternatives to data packet formats and exchange protocols: In addition to JSON / XML, standardized data packets can use formats such as Avro / Protobuf / Parquet to improve compression ratio and cross-language portability. (6) Alternatives to privacy and compliance: To protect data privacy, the cross-verification process can use federated learning or secure multi-party computation technology, so that each data source does not need to share the original data, but only needs to interact or upload the intermediate results or verification summaries obtained from the computation; in addition, for the final output standardized data packets, sensitive indicators can be processed using data desensitization technologies such as k-anonymity or differential privacy to prevent information leakage. (7) Alternatives to auditing and traceability: The audit chain used to record quality scores and acceptance decisions can be implemented using evidence storage technology with anti-tampering characteristics (such as hash on-chain based on blockchain or verifiable logs); or a timestamp service provided by a third-party timestamp service provider can be used as an alternative. The above alternatives do not change the technical approach of this invention, which is "layered automatic verification—priority / weighted acceptance—dynamic quality scoring—standardized trusted data packets." They only involve equivalent substitutions and engineering optimizations to the implementation details. These substitutions do not depart from the scope of protection of this invention.

[0081] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of the invention.

Claims

1. An ESG evaluation data governance system based on multi-source data fusion and dynamic verification, characterized in that, include: The multi-source data access module is used to obtain the enterprise's original ESG-related data from multiple heterogeneous data sources; The intelligent standardization processing module is connected to the multi-source data access module and has a built-in ESG data standardization dictionary. The ESG data standardization dictionary defines the concepts, units of measurement and statistical standards of core ESG indicators and is used to clean, format and standardize the units of the acquired raw data. The ESG indicator calculation engine is connected to the intelligent standardization processing module and is used to automatically calculate derived ESG indicators, including carbon emissions, green income ratio, and environmental risk index, according to the preset formulas in the ESG data standardization dictionary. The hierarchical data verification and processing module is connected to the ESG indicator calculation engine and includes a logical verification unit, a range verification unit, and a cross-verification unit, which is used to perform automated hierarchical verification and processing on the processed and calculated data. The dynamic data quality scoring module is connected to the hierarchical data verification and processing module and is used to dynamically calculate and output the enterprise's ESG data quality score based on the data verification results. The intelligent update and traceability module is connected to the hierarchical data verification and processing module and the data quality dynamic scoring module. It is used to automatically update the corresponding data in the enterprise unified data pool after determining the final acceptance value according to the priority acceptance rules, and record complete verification and acceptance logs to establish a data traceability chain. The standardized ESG data packet generation module, connected to the intelligent update and traceability module, is used to generate standardized, independently verifiable data packets for direct use by upper-level ESG rating models or financial institution risk control systems.

2. The ESG evaluation data governance system based on multi-source data fusion and dynamic verification according to claim 1, characterized in that, The formulas for calculating carbon emissions, green income ratio, and environmental risk index in the ESG indicator calculation engine are as follows: carbon emissions : in, , and The figures represent emissions for ranges 1, 2, and 3, respectively. Range 1 represents direct emissions, range 2 represents indirect emissions from purchased electricity and heat, and range 3 represents other indirect emissions. fuel consumption, fuel Emission factors For purchased electricity, For power grid emission factors, For externally purchased calories, Thermal emission factor; Green income share : in, For green business revenue that meets the standards, Total operating revenue of the enterprise; Environmental Risk Index : in, , , and These are the number of environmental penalties, the number of records of exceeding emission standards, the amount of high-risk waste generated, and the environmental sensitivity of the location. , , and As weight.

3. The ESG evaluation data governance system based on multi-source data fusion and dynamic verification according to claim 1, characterized in that, The logic verification unit is used to perform hard logic rule checks, including numerical non-negativity, consistency between items and totals, and time series continuity. The range verification unit is used to check whether the data fluctuation is within a preset reasonable range based on historical data and industry benchmarks, mark data that exceeds the range as needing verification, and automatically send an inquiry letter. The cross-validation unit is used to automatically compare the calculation results from external authoritative data sources with the enterprise's internal declaration values, including environmental data validation, compliance data validation, and financial data validation. Data with differences exceeding a preset threshold is marked as seriously inconsistent and triggers priority acceptance rules.

4. The ESG evaluation data governance system based on multi-source data fusion and dynamic verification according to claim 1, characterized in that, The enterprise's ESG data quality score The calculation formula is: in, , , and These are the data integrity score, data consistency score, data timeliness score, and verification pass rate score. , , and The weights assigned to each dimension; in, The number of indicators that have been reported. This represents the total number of core indicators; in, The number of consistency rules that passed. This represents the total number of consistency rules. in, For the number of days of delay, This is the deduction coefficient; in, The number of indicators that have passed verification. This represents the total number of indicators that were verified.

5. The ESG evaluation data governance system based on multi-source data fusion and dynamic verification according to claim 1, characterized in that, The priority acceptance rules determine the final value based on the authority of the data source and historical consistency.

6. The ESG evaluation data governance system based on multi-source data fusion and dynamic verification according to claim 1, characterized in that, The data packets generated by the standardized ESG data packet generation module include all verified core ESG indicator data, data source information and acceptance criteria, data verification status records, final data quality scores, and data update timestamps.

7. A method for ESG evaluation data governance based on multi-source data fusion and dynamic verification, characterized in that, Includes the following steps: S1: Obtain the enterprise's raw ESG-related data in parallel from multiple heterogeneous data sources; S2: The raw data is cleaned, formatted and standardized using an ESG data standardization dictionary. The ESG data standardization dictionary defines the concepts, units of measurement and statistical scope of core ESG indicators. S3: Based on the accounting formula in the ESG data standardization dictionary, automatically calculate derived ESG indicators including carbon emissions, green income ratio, and environmental risk index; S4: Perform automated hierarchical verification and processing on the processed and calculated data; S4 includes: S41: Perform hard logic rule checks, including numerical non-negativity, consistency between items and totals, and time series continuity; S42: Execution range verification, based on historical data and industry benchmarks, checks whether data fluctuations are within a preset reasonable range, marks data that exceeds the range as pending verification, and automatically sends an inquiry letter; S43: Perform cross-validation, automatically compare the calculation results from external authoritative data sources with the enterprise's internal reported values, including environmental data validation, compliance data validation, and financial data validation. Data with differences exceeding a preset threshold is marked as seriously inconsistent and triggers priority acceptance rules. S5: Dynamically calculate and output the enterprise's ESG data quality score based on the data verification results; S6: After determining the final acceptance value based on the priority acceptance rules, automatically update the corresponding data in the enterprise unified data pool, record complete verification and acceptance logs, and establish a data traceability chain; S7: Generates standardized, independently verifiable data packages for direct use by upper-level ESG rating models or financial institution risk control systems.

8. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the steps of the ESG evaluation data governance method based on multi-source data fusion and dynamic verification as described in claim 7.

Images