Enterprise ESG information disclosure optimization method and rating maintenance system

Through the collection, fusion, sentiment analysis and machine learning prediction of ESG data, targeted disclosure strategies are generated, which solves the problems of scattered information and insufficient data accuracy in existing ESG information disclosure, and realizes the provision of high-quality ESG information and enterprises. Improved transparency.

CN118917748AInactive Publication Date: 2024-11-08FUZHOU PLANNING DESIGN & RES INST

Patent Information

Application Number
CN202411408178.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-10
Publication Date
2024-11-08
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing ESG information disclosure methods have problems such as scattered information, different standards, and insufficient data accuracy, which is difficult to meet the needs of investors, regulators and all sectors of society for high-quality ESG information.

Method used

A preferred method for enterprise ESG information disclosure is proposed. By collecting, preprocessing, data fusion, sentiment analysis and machine learning model prediction of ESG data from different channels, targeted disclosure strategies are generated, and data collection and optimization through user feedback Analysis process.

Benefits of technology

It improves the accuracy and reliability of ESG information, enhances corporate transparency, enhances the trust of stakeholders, meets the needs of high-quality ESG information, and ensures the timeliness and targeted information disclosure.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118917748A_ABST
    Figure CN118917748A_ABST
Patent Text Reader

Abstract

The invention provides an enterprise ESG information disclosure optimization method and a rating maintenance system. Belongs to the technical field of enterprise information disclosure. The method comprises the following steps: collecting ESG related data from different channels, storing the collected ESG related data into a cloud space for preprocessing, and integrating the preprocessed related data through a data fusion algorithm to form an ESG information data set; and analyzing related data in the data set through an NLP technology, and extracting key information and emotional tendency related to the ESG. By automatically collecting and preprocessing ESG related data and storing the data in the cloud space, the efficiency and security of data processing are greatly improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention proposes an enterprise ESG information disclosure optimization method and a rating maintenance system, belonging to the field of enterprise information batch technology. Background Art

[0002] As the world pays more attention to corporate sustainable development and social responsibility, ESG information disclosure has become an important means for companies to demonstrate their non-financial performance, enhance their brand image and attract investment. However, existing ESG information disclosure methods often have problems such as fragmented information, inconsistent standards, and insufficient data accuracy, which makes it difficult to meet the needs of investors, regulators and all sectors of society for high-quality ESG information. Summary of the invention

[0003] The present invention provides an enterprise ESG information disclosure optimization method and rating maintenance system to solve the problems mentioned in the above background technology: The present invention proposes a method for optimizing corporate ESG information disclosure, the method comprising: S1. Collect ESG-related data from different channels, store the collected ESG-related data in the cloud space, pre-process them, and integrate the pre-processed data through data fusion algorithms to form an ESG information data set; S2. Analyze the relevant data in the dataset through NLP technology to extract key information and sentiment trends related to ESG; S3. Train machine learning models based on historical ESG data and external macroeconomic data to predict corporate ESG performance trends and identify potential risks and opportunities; S4. Generate an ESG information disclosure strategy based on the results of intelligent analysis and the needs survey of stakeholders; and generate an ESG report or summary based on the strategy; S5. Collect relevant information on ESG information disclosure from stakeholders through user feedback channels, and continuously optimize the data collection, processing, analysis and disclosure processes based on user feedback and data analysis results.

[0004] Furthermore, the S1 includes: S11. Determine the scope and channels of data collection, and automatically collect ESG-related data from the above channels through crawler technology and API interface calls; S12. Preprocess the collected ESG-related data and upload the preprocessed ESG-related data to the cloud storage platform; and encrypt the sensitive data in the cloud storage platform using a data encryption algorithm; S13. Use multi-source data fusion algorithm to integrate data from different channels but related; S14. Construct an ESG information data set, where the ESG information data set includes structured data, semi-structured data, and unstructured data.

[0005] Furthermore, the S13 includes: Further clean up the preprocessed data and unify the metrics and formats between different data sources; Extract key entities from text data through named entity recognition algorithms, and match and link these entities with entities in known databases or knowledge graphs through entity linking techniques; Based on the expertise in the ESG field, we build an ontology model to clarify the relationships and hierarchical structures between data elements. We also use the ontology alignment algorithm to align similar or related concepts in different data sources to eliminate semantic heterogeneity. According to the characteristics and requirements of the data, a fusion strategy is selected, and the data fusion operation is performed according to the selected fusion strategy or model; after the fusion is completed, the fusion results are quantitatively evaluated through evaluation indicators; The fused data is integrated and reconstructed according to the structural requirements of the ESG information data set to form a unified data representation.

[0006] Furthermore, the S2 includes: S21. Preprocess the text data again through the NLP tool library and extract key information from the text; S22, using pre-trained models to represent text and capture the semantic features of the text; S23. Design a sentiment dictionary and apply sentiment analysis algorithms to judge the sentiment tendency of the text.

[0007] S24. Analyze the distribution trend of sentiment tendencies and identify the public’s attitudes and opinions towards the company in combination with ESG themes.

[0008] Furthermore, the S3 includes: S31. Integrate historical ESG data and macroeconomic indicators as input features of machine learning models and construct labels; S32. Select machine learning algorithms and optimize them for ESG prediction tasks; evaluate and train machine learning models; S33. Based on the trained model, make trend forecasts on future ESG performance; identify potential risk points and evaluate potential opportunities.

[0009] Furthermore, the S33 includes: The forecasting task is subdivided into short-term, medium-term and long-term forecasts; and the model parameters are adjusted for different time scales; Construct different scenarios, input models for simulation and prediction, and analyze the changes in corporate ESG performance under different scenarios; Use anomaly detection algorithms to identify outliers in forecast results that significantly deviate from historical trends or industry standards as a preliminary screening of potential risk points; Build an ESG risk assessment framework, combine industry characteristics, corporate realities and expert knowledge to further evaluate identified potential risk points; Assign a quantitative score to each potential risk point, taking into account various dimensions of risk and forming a risk priority ranking; Analyze positive trends in forecast results as clues to potential opportunities; compare and analyze the company's ESG forecast results with the industry average and leading companies; Utilize SWOT analysis to conduct in-depth assessment of identified potential opportunities, determine priorities and development strategies.

[0010] Furthermore, the S4 includes: S41. Collect feedback from stakeholders through multiple means; S42. Evaluate the ESG performance of enterprises by combining the sentiment analysis results of NLP analysis and the trends predicted by the machine learning model; S43. Use data analysis tools to visualize data and intuitively display ESG key indicators, trend changes and risk assessment results; S44. Analyze competitors’ ESG performance, identify industry practices, and determine the key areas and priorities of ESG information disclosure based on stakeholder needs and feedback; S45. Design a disclosure strategy and generate an ESG report or summary based on the disclosure strategy.

[0011] Furthermore, the S5 includes: S51. Regularly collect and analyze feedback from stakeholders through user feedback channels to identify users’ satisfaction, concerns and improvement suggestions on ESG information disclosure; S52. Through text mining and sentiment analysis technology, deeply analyze the feedback content and extract valuable information and insights; S53. Adjust the data collection strategy and optimize the data preprocessing and fusion algorithms according to user feedback and data analysis results; S54. Revise the ESG disclosure strategy, update the report content and format, and establish a continuous improvement mechanism to regularly review and evaluate the effectiveness of ESG information disclosure.

[0012] Furthermore, the S53 includes: Classify the collected user feedback according to dimensions and extract key information from a large amount of feedback through text mining technology; Use sentiment analysis algorithms to judge the sentiment tendency of feedback content and quantify the intensity of user satisfaction and dissatisfaction; Prioritize data collection channels and content based on the importance and urgency of user feedback, and give priority to collecting ESG data that users are most concerned about or have the most feedback on; Explore new data collection channels based on user feedback and optimize the frequency and depth of data collection in existing channels; Establish a data collection quality monitoring mechanism to regularly evaluate the accuracy and completeness of data collection and promptly correct errors and omissions; Based on user feedback and data analysis results, the multi-source data fusion algorithm is iteratively optimized.

[0013] A rating maintenance system proposed in the present invention includes a memory, a processor, and a computer program stored in and executable on the memory, wherein the processor executes the program to implement any of the above-mentioned methods for optimizing enterprise ESG information disclosure.

[0014] The beneficial effects of the present invention are as follows: by automatically collecting and preprocessing ESG-related data and storing the data in the cloud, the efficiency and security of data processing are greatly improved; by adopting data fusion algorithms and multi-source data fusion technology, data metrics and formats are unified, the inconsistency and ambiguity of data are reduced, and the accuracy and reliability of data are improved; by using NLP technology to parse ESG-related data, key information and sentiment tendencies are extracted, which helps to more comprehensively understand the ESG performance of enterprises; by training historical data and macroeconomic data through machine learning models, the ESG performance trends of enterprises can be predicted, and potential risks and opportunities can be identified in advance; by combining intelligent analysis results and stakeholder needs, targeted ESG information disclosure strategies are generated, which helps to improve the quality and effectiveness of information disclosure; by formulating and implementing effective ESG information disclosure strategies, enterprises can improve their transparency, enhance the trust of stakeholders, and enhance their corporate image; by collecting user feedback and analyzing data, the data collection, processing, analysis and disclosure processes are continuously optimized to ensure that information disclosure methods keep pace with the times and meet the needs of stakeholders. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a step diagram of the method described in the present invention. DETAILED DESCRIPTION

[0016] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0017] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. The embodiments described are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0019] One embodiment of the present invention, as Figure 1 As shown, a preferred method for corporate ESG information disclosure includes: S1. Collect ESG-related data from different channels, store the collected ESG-related data in the cloud space, pre-process them, and integrate the pre-processed data through data fusion algorithms to form an ESG information data set; S2. Analyze the relevant data in the dataset through NLP technology to extract key information and sentiment trends related to ESG; S3. Train machine learning models based on historical ESG data and external macroeconomic data to predict corporate ESG performance trends and identify potential risks and opportunities; S4. Generate an ESG information disclosure strategy based on the results of intelligent analysis and the survey of stakeholders’ needs. The strategy covers multiple aspects such as disclosure content, frequency, and channels to ensure the pertinence and effectiveness of information disclosure. Generate an ESG report or summary based on the strategy. S5. Collect relevant information from stakeholders on ESG information disclosure through user feedback channels, including satisfaction and suggestions; and continuously optimize the data collection, processing, analysis and disclosure processes based on user feedback and data analysis results.

[0020] The working principle of the above technical solution is as follows: collect ESG-related data from multiple channels (such as corporate annual reports, social responsibility reports, news media, etc.). The relevant data covers environmental performance (such as carbon emissions, resource utilization efficiency), social performance (such as employee benefits, community relations) and governance performance (such as corporate governance structure, transparency) and other aspects; store the collected data in the cloud space; clean the collected raw data (remove duplicate, erroneous or irrelevant data), format (unify data format) and standardize (such as converting non-numeric data into numerical data); use data fusion algorithms to integrate pre-processed data from different sources to form a unified and comprehensive ESG information data set; use natural language processing (NLP) technology to deeply analyze the text data in the ESG information data set, identify and extract key information related to ESG, such as specific indicators, event descriptions, etc.; further use the sentiment analysis function of NLP to evaluate the sentiment tendency (positive, negative or neutral) in the relevant text data to understand the attitudes and views of stakeholders on the company's ESG performance; train machine learning models based on historical ESG data and external macroeconomic data (such as industry trends, etc.). Among them, the machine learning model can learn and identify the key factors that affect ESG performance and predict future trends; use the trained model to predict the company's future ESG performance trends and identify potential risks and opportunities; combine the results of intelligent analysis (such as performance forecasts, risk assessments) and stakeholder demand surveys (through questionnaires, interviews, etc.) to generate targeted ESG information disclosure strategies. The content, frequency, channels and other elements of the disclosure are determined in the strategy; according to the disclosure strategy, an ESG report or summary is prepared. Through user feedback channels (such as online surveys, social media interactions, etc.), stakeholders' satisfaction and suggestions on ESG information disclosure are collected; based on user feedback and data analysis results, data collection, processing, analysis and disclosure processes are continuously adjusted and optimized. Specifically, it includes improving data collection methods, optimizing analysis models, adjusting disclosure strategies, etc.

[0021] The effects of the above technical solutions are as follows: by collecting ESG-related data from multiple channels, the comprehensiveness and diversity of the data are ensured; using cloud space for data storage helps to centrally manage and efficiently access data; pre-processing improves the accuracy and availability of data, laying a solid foundation for subsequent analysis; integrating data from different sources through data fusion algorithms eliminates data silos and forms a unified and comprehensive ESG information data set, which helps to more comprehensively evaluate the ESG performance of enterprises; the introduction of NLP technology makes the analysis of text data more in-depth, and can automatically extract key information and sentiment tendencies, improving the efficiency and accuracy of information processing; machine learning models trained based on historical data and external macroeconomic data can predict corporate ESG performance trends, identify potential risks and opportunities, and provide strong support for corporate decision-making; understanding stakeholders through demand surveys The focus and expectations of the enterprise are reflected in the information disclosure strategy, which is closer to the actual needs and improves the pertinence and effectiveness of information disclosure. The information disclosure strategy covers multiple aspects such as content, frequency, and channels, which can be flexibly adjusted according to the actual situation of the enterprise and the needs of stakeholders to achieve personalized disclosure. ESG reports or summaries are automatically generated according to the disclosure strategy, which reduces the workload of manual report preparation and improves the efficiency of report preparation. Since the reports are prepared based on comprehensive and accurate data and intelligent analysis results, the high quality of the report content is ensured, which helps to enhance the transparency and credibility of the enterprise. The satisfaction and suggestions of stakeholders are collected through user feedback channels, which helps to understand the actual effect of information disclosure and market feedback. The data collection, processing, analysis and disclosure processes are continuously optimized according to user feedback and data analysis results, forming a closed-loop mechanism for continuous improvement, ensuring the continuous optimization and improvement of information disclosure.

[0022] In one embodiment of the present invention, the S1 includes: S11. Determine the scope and channels of data collection, including but not limited to government announcements, industry association reports, corporate social responsibility reports, media reports, social media platforms, and non-governmental organization evaluations; and automatically collect ESG-related data from the above channels through crawler technology and API interface calls; S12. Preprocess the collected ESG-related data and upload the preprocessed ESG-related data to the cloud storage platform; and encrypt the sensitive data in the cloud storage platform using a data encryption algorithm; S13. Use multi-source data fusion algorithms (such as ontology alignment, machine learning fusion models, etc.) to integrate data from different channels but related; S14. Construct an ESG information data set, wherein the ESG information data set includes structured data, semi-structured data (such as tables and lists) and unstructured data (such as text and images).

[0023] The working principle of the above technical solution is: clarify the type and source of ESG data to be collected, and the data sources include government announcements, industry association reports, corporate social responsibility reports, media reports, social media platforms and non-governmental organization evaluations. The channels provide rich ESG information, covering all aspects of environment, society and governance; use crawler technology and API interface calls to automatically capture ESG-related data from the above public channels. The API interface call directly interacts with the data provider through the program to obtain real-time or regularly updated data; and performs pre-processing operations such as cleaning, deduplication, and formatting on the collected ESG-related data. Including correcting erroneous data, filling missing values, converting data types and formats, etc.; uploading the pre-processed ESG-related data to the cloud storage platform, and using the advantages of cloud computing to achieve centralized management and efficient access to data. At the same time, the data encryption algorithm is used to encrypt the sensitive data in the cloud storage platform; and multi-source data fusion algorithms such as ontology alignment and machine learning fusion models are used to integrate data from different channels but related. Ontology alignment matches entities and attributes in different data sources by defining common concepts and relationships; the machine learning fusion model uses machine learning algorithms to explore potential connections and patterns between data to achieve deep data fusion; through multi-source data fusion algorithms, data from different channels are integrated into a unified and comprehensive ESG information view to eliminate data silos; the constructed ESG information data set includes structured data (such as numerical indicators), semi-structured data (such as tables and lists) and unstructured data (such as text and images). These data store the company's ESG performance information in different forms, providing a rich data source for subsequent analysis and disclosure; manage and maintain the ESG information data set, including data update, verification and expansion. As new ESG data is continuously generated and old data is updated, the data set needs to be maintained and updated regularly to ensure the timeliness and accuracy of the data.

[0024] The effects of the above technical solutions are: by determining a wide range of data collection and a variety of collection channels, such as government announcements, industry association reports, corporate social responsibility reports, media reports, social media platforms and non-governmental organization evaluations, comprehensive coverage of ESG-related data is ensured. It helps to understand the ESG performance of enterprises from multiple dimensions and angles, and improve the accuracy and objectivity of the evaluation; using crawler technology and API interface calling methods, the automatic collection of data is realized, the efficiency of data collection is improved, and manual intervention and errors are reduced, ensuring the timeliness and accuracy of data; pre-processing of the collected ESG-related data, such as cleaning, deduplication, formatting, etc., improves the quality and availability of data. It is helpful for subsequent data analysis and processing, reducing errors and deviations caused by data quality problems; uploading the pre-processed data to the cloud storage platform, and encrypting sensitive data through data encryption algorithms to ensure data security and privacy protection. It helps to prevent data leakage and illegal access and ensure the information security of enterprises; using multi-source data fusion algorithms, such as ontology alignment, machine learning fusion models, etc., to integrate data from different channels but related. This helps eliminate data silos, realize data interconnection and sharing, and improve the comprehensive utilization value of data; through data fusion, a unified and comprehensive ESG information view is constructed, so that companies can have a clearer understanding of their own ESG performance and compare and analyze with other companies; ESG information data sets include structured data, semi-structured data (such as tables, lists) and unstructured data (such as text, images). This diverse data structure makes the data set more flexible and rich, and can meet different analysis and disclosure needs; the richness of the data set content helps to more comprehensively reflect the ESG performance of enterprises, including quantitative indicators (such as carbon emissions, resource utilization efficiency) and qualitative information (such as corporate culture, social contribution, etc.), providing stakeholders with a more comprehensive and in-depth understanding.

[0025] In one embodiment of the present invention, the S12 includes: Identify and remove outliers in the data set through machine learning algorithms (such as Isolation Forest and DBSCAN clustering), and unify data formats (such as dates, currency units, etc.) from different channels into a standard format; Use hash tables combined with similarity algorithms (such as Jaccard similarity and cosine similarity) to identify and merge duplicate data; For numerical data, the minimum-maximum standardization or Z-score standardization method is used to eliminate the impact of different dimensions on data analysis; for semi-structured or unstructured text data, word segmentation, stop word removal, stem extraction and other processing are performed to improve the analyzability of text data; for texts containing emotional tendencies, sentiment dictionaries or deep learning models (such as BERT) are used to perform preliminary sentiment annotation to lay the foundation for subsequent sentiment analysis.

[0026] Use machine learning classification algorithms (such as support vector machines and random forests) to automatically classify ESG data into three categories, such as environment, society, and governance, and their subcategories. Build an ESG data labeling system based on business needs and industry standards, and label each piece of data accordingly to facilitate subsequent retrieval and analysis.

[0027] Based on regular expression matching and machine learning models, sensitive information in the data (such as personal information, trade secrets, etc.) is automatically identified; encryption algorithms of different strengths (such as AES-256, RSA) are used for encryption according to the sensitivity level of the data. For extremely sensitive data, consider using homomorphic encryption technology to allow calculations on encrypted data without exposing the original data.

[0028] The working principle of the above technical solution is: using machine learning algorithms (such as isolation forests, DBSCAN clustering) to identify outliers in the data set. The algorithm can regard points that deviate from the normal data distribution range as outliers and remove them based on the distribution characteristics or density characteristics of the data; convert data from different channels (such as dates, currency units, etc.) into a unified standard format. It helps to eliminate analysis barriers caused by inconsistent formats and ensure the consistency and comparability of data; use hash tables combined with similarity algorithms (such as Jaccard similarity and cosine similarity) to identify and merge duplicate data. The hash table can quickly locate whether the data already exists, and the similarity algorithm can further determine whether the content is highly similar, thereby effectively reducing data redundancy; for numerical data, the minimum-maximum standardization or Z-score standardization method is used to convert data of different dimensions to the same scale for subsequent comparison and analysis; for semi-structured or unstructured text data, word segmentation, stop word removal, stem extraction and other processing are performed. These steps can reduce the dimension of text data, improve the analyzability of text data, and lay the foundation for subsequent tasks such as sentiment analysis and topic extraction; for texts containing sentiment tendencies, use sentiment dictionaries or deep learning models (such as BERT) for preliminary sentiment annotation. Sentiment dictionaries can quickly identify sentiment words in texts and assign corresponding sentiment polarities, while deep learning models can more accurately capture complex sentiment information in texts; use machine learning classification algorithms (such as support vector machines and random forests) to automatically classify ESG data. These algorithms can learn classification rules based on data features, automatically divide ESG data into major categories such as environment, society, governance, and their subcategories, and improve the organization and retrieval efficiency of data organization; build an ESG data labeling system based on business needs and industry specifications, and label each piece of data accordingly. The labeling system can achieve refined management and rapid retrieval of data, facilitating subsequent data analysis and application; based on regular expression matching and machine learning models, automatically identify sensitive information in data (such as personal information, trade secrets, etc.). According to the sensitivity level of the data, encryption algorithms of different strengths (such as AES-256 and RSA) are used for encryption processing. For extremely sensitive data, consider using homomorphic encryption technology, which allows calculations to be performed on encrypted data without exposing the original data. Encryption processing can ensure the security of data during storage and transmission, preventing data leakage and illegal access.

[0029] The effects of the above technical solutions are as follows: by identifying and eliminating outliers in the data set through machine learning algorithms, the analysis bias caused by erroneous data can be effectively reduced, and the accuracy and reliability of the data set can be improved; unifying data formats and merging duplicate data can reduce data redundancy and noise, making the data set purer and providing a high-quality data source for subsequent analysis; using machine learning algorithms and data structures such as hash tables, the automated processing of tasks such as data cleaning, deduplication, and classification is realized, greatly improving the speed and efficiency of data processing; standardizing numerical data and pre-processing text data such as word segmentation and stop word removal lays the foundation for subsequent data analysis, modeling and other tasks, and reduces the need for manual intervention; preliminary sentiment annotation of texts containing emotional tendencies provides an important basis for subsequent sentiment analysis, helps to deeply explore the emotional information in ESG data, and understand the attitudes and views of the public or stakeholders; using machine learning classification algorithms to automatically classify ESG data and build an ESG data labeling system to make the data more organized, easy to retrieve and analyze. This helps to discover hidden patterns and correlations in ESG data, providing strong support for decision-making; it automatically identifies sensitive information in the data based on regular expression matching and machine learning models, and uses encryption algorithms of different strengths for encryption processing. This helps to protect personal privacy and business secrets from being leaked, and ensure the security of data during storage and transmission; for extremely sensitive data, consider using homomorphic encryption technology to allow calculations on encrypted data without exposing the original data. It further improves the security of data processing and meets the needs of data analysis while protecting privacy; by unifying data formats and building an ESG data label system, it enhances the standardization and interoperability of ESG data, allowing data from different sources and formats to be seamlessly integrated and shared. This helps promote the widespread application of ESG data and cross-sector cooperation, and promotes the realization of sustainable development goals.

[0030] In one embodiment of the present invention, the S13 includes: Further clean up the preprocessed data, including removing duplicate records, correcting format errors, filling or deleting missing values, identifying and processing outliers, etc., and unifying the metrics and formats between different data sources; such as date format, currency unit, measurement unit, etc.; Through the named entity recognition (NER) algorithm, key entities such as company names, project names, and geographic locations are extracted from text data, and through entity linking technology, these entities are matched and linked with entities in known databases or knowledge graphs, realizing data association across data sources.

[0031] Based on the expertise in the ESG field, we build an ontology model to clarify the relationships and hierarchical structures between data elements. We also use the ontology alignment algorithm to align similar or related concepts in different data sources to eliminate semantic heterogeneity. Select a fusion strategy based on the characteristics and requirements of the data, including but not limited to rule-based fusion, statistics-based fusion, machine learning-based fusion, etc.

[0032] For complex ESG data fusion tasks, machine learning technology can be used to build a fusion model. First, select a suitable machine learning algorithm (such as deep learning, ensemble learning, etc.) and design a reasonable feature representation and model structure. Then, use historical data and labeled data to train and optimize the model to improve the fusion effect. Perform data fusion operations based on the selected fusion strategy or model; during the fusion process, attention should be paid to issues such as maintaining data integrity, conflict resolution, and consistency checking. After the fusion is completed, the fusion results are quantitatively evaluated through evaluation indicators (such as accuracy, recall, F1 score, etc.); The fused data is integrated and reconstructed according to the structural requirements of the ESG information data set to form a unified data representation.

[0033] The working principle of the above technical solution is: deep cleaning of pre-processed data, removal of duplicate records, correction of format errors, filling or deletion of missing values, and identification and processing of outliers. These steps ensure the quality and consistency of the data; then, the metrics and formats between different data sources are unified, such as date format, currency unit, measurement unit, etc. This helps to eliminate analysis barriers caused by inconsistent data formats; through the NER algorithm, key entities such as company name, project name, and geographic location are extracted from text data. These entities are the basis for subsequent data association and analysis; using entity linking technology, the extracted entities are matched and linked with entities in known databases or knowledge graphs. This process realizes data association across data sources and enhances data interoperability; based on professional knowledge in the ESG field, an ontology model is constructed. The ontology model clarifies the relationship and hierarchy between data elements and provides a semantic basis for data fusion; through the ontology alignment algorithm, similar or related concepts in different data sources are aligned. This eliminates semantic heterogeneity, allowing data from different channels to be compared and analyzed under the same semantic framework; according to the characteristics and needs of the data, a suitable fusion strategy is selected. These strategies include rule-based fusion, statistics-based fusion, machine learning-based fusion, etc. For complex ESG data fusion tasks, machine learning technology can be used to build a fusion model; for machine learning-based fusion strategies, it is necessary to select a suitable machine learning algorithm and design a reasonable feature representation and model structure. Then, the model is trained and optimized using historical data and annotated data to improve the fusion effect; data fusion operations are performed according to the selected fusion strategy or model. During the fusion process, attention should be paid to issues such as data integrity maintenance, conflict resolution, and consistency checking. After the fusion is completed, the fusion results are quantitatively evaluated through evaluation indicators (such as accuracy, recall, F1 score, etc.) to ensure that the fusion effect meets expectations; the fused data is integrated according to the structural requirements of the ESG information data set. This includes organizing structured data, semi-structured data, and unstructured data in a unified format and rules; on the basis of integration, the data is reconstructed to form a unified data representation. This helps subsequent data analysis and information disclosure work and improves the availability and comprehensibility of the data.

[0034] The effects of the above technical solutions are: by further cleaning and standardizing the pre-processed data, such as removing duplicate records, correcting format errors, filling or deleting missing values, identifying and processing outliers, etc., the accuracy and consistency of the data are ensured. At the same time, the metrics and formats between different data sources, such as date format, currency unit, measurement unit, etc., are unified, eliminating the problems caused by inconsistent data formats and improving the comparability and analyzability of data; the named entity recognition (NER) algorithm is used to extract key entities from text data, and these entities are matched and linked with entities in known databases or knowledge graphs through entity linking technology, realizing data association across data sources. This not only enhances the interoperability of data, but also enables data to be understood and utilized in a wider context; an ontology model is built based on professional knowledge in the ESG field, clarifying the relationship and hierarchy between data elements. Through the ontology alignment algorithm, similar or related concepts in different data sources are aligned to eliminate semantic heterogeneity. This makes the data consistent at the semantic level, providing a solid foundation for subsequent data analysis and mining; select appropriate fusion strategies according to the characteristics and needs of the data, such as rule-based fusion, statistics-based fusion, machine learning-based fusion, etc. In particular, for complex ESG data fusion tasks, machine learning technology is used to build a fusion model, which improves the intelligence level and efficiency of data fusion; the fusion results are quantitatively evaluated through evaluation indicators (such as accuracy, recall rate, F1 score, etc.) to ensure the reliability and effectiveness of the fusion effect; the fused data is integrated and reconstructed according to the structural requirements of the ESG information data set to form a unified data representation. This not only improves the readability and ease of use of the data, but also enables the data to be more efficiently used in ESG information disclosure, risk assessment, performance evaluation and other aspects; high-quality, unified ESG information data sets provide more comprehensive, accurate and timely ESG information support for the decision-making level of enterprises, which helps enterprises better fulfill their social responsibilities, enhance their brand image and enhance their market competitiveness; through the open and transparent disclosure of ESG information, enterprises can enhance communication and trust with stakeholders such as investors, consumers, and governments, and lay a solid foundation for the sustainable development of enterprises.

[0035] In one embodiment of the present invention, the S2 includes: S21. Preprocess the text data again through NLP tool libraries such as NLTK, SpaCy, Transformers, etc., apply technologies such as word segmentation, part-of-speech tagging, named entity recognition, and extract key information from the text; S22. Use pre-trained models (such as BERT and RoBERTa) to represent text and capture the semantic features of text. S23. Design a sentiment dictionary that includes positive, negative, and neutral ESG-related words; apply sentiment analysis algorithms, such as rule-based methods, machine learning classifiers, etc., to judge the sentiment tendency of the text.

[0036] S24. Analyze the distribution trend of sentiment tendencies and identify the public’s attitudes and opinions towards the company in combination with ESG themes.

[0037] The working principle of the above technical solution is as follows: Use NLP tool libraries (such as NLTK, SpaCy, Transformers, etc.) to pre-process text data again, in order to further improve the quality of text data and make it suitable for subsequent analysis; Divide continuous text into several meaningful vocabulary units, which is a basic step in NLP and helps with subsequent part-of-speech tagging and named entity recognition; Label each vocabulary unit with its part of speech (such as noun, verb, adjective, etc.), which helps to understand the grammatical role and semantic function of vocabulary in sentences; Identify and extract entities with specific meanings from text (such as company names, project names, geographic locations, etc.), which are often key information in text; Represent text through pre-trained models (such as BERT, RoBERTa), which are trained with a large amount of text data and can capture the deep semantic features of text; Convert text data into vector form so that computers can process and understand it. These vectors not only contain the semantic information of the vocabulary, but also take into account the contextual relationship, so as to more accurately represent the meaning of the text; Design a sentiment dictionary containing positive, negative and neutral vocabulary according to the specific context of the ESG field. These words are closely related to ESG topics, which helps to accurately judge the sentiment tendency of the text; apply sentiment analysis algorithms (such as rule-based methods, machine learning classifiers, etc.) to judge the sentiment tendency of the text. Rule-based methods rely on predefined sentiment dictionaries and rule sets, while machine learning classifiers learn sentiment classification models through training data; statistically analyze the sentiment tendencies of text data to understand the distribution trend of sentiment tendencies. This helps to discover the general attitudes and opinions of the public towards corporate ESG practices; combine ESG topics (such as environmental protection, social responsibility, corporate governance, etc.) to conduct a deeper interpretation of sentiment tendencies. By analyzing the sentiment tendencies under different ESG topics, the public's specific attitudes and opinions on companies in different aspects can be identified.

[0038] The effects of the above technical solutions are: by using NLP tool libraries (such as NLTK, SpaCy, Transformers, etc.), it is possible to efficiently perform tasks such as word segmentation, part-of-speech tagging, and named entity recognition, which are essential for understanding and analyzing text data. These tool libraries are optimized to provide accurate and efficient processing results; automated processing of text data reduces the need for manual intervention and improves processing speed and accuracy. This advantage is particularly evident when processing large amounts of text data; text representation through pre-trained models (such as BERT and RoBERTa) can capture the deep semantic features of text. These models are trained with a large amount of text data and can understand the complex relationships and contextual information between words, thereby providing more accurate text representation; capturing the semantic features of text helps to understand the text content more deeply, providing strong support for subsequent sentiment analysis and topic identification; designing a sentiment dictionary containing positive, negative, and neutral words related to ESG, which can make accurate sentiment tendency judgments for the ESG field. This targeted design improves the accuracy and reliability of sentiment analysis; applying a variety of sentiment analysis algorithms such as rule-based methods and machine learning classifiers, it is possible to select appropriate algorithms for sentiment tendency judgments according to different scenarios and needs. This flexibility improves the applicability and accuracy of sentiment analysis; by analyzing the distribution trend of sentiment tendencies, we can understand the public's overall attitude and views on corporate ESG practices. This analysis helps companies grasp public sentiment and adjust strategies in a timely manner; sentiment tendency analysis combined with ESG themes can identify the public's specific attitudes and views on companies in different ESG aspects. This in-depth analysis helps companies discover their own strengths and weaknesses in ESG practices, so as to improve and enhance them in a targeted manner; by identifying the public's attitudes and views on corporate ESG practices, companies can discover their own shortcomings in ESG and take corresponding measures to improve and optimize. This continuous improvement process helps to enhance the company's ESG performance and social responsibility image.

[0039] In one embodiment of the present invention, S3 includes: S31. Integrate historical ESG data and macroeconomic indicators (such as GDP growth rate, policy changes, etc.) as input features of the machine learning model; and construct labels; that is, the historical performance of corporate ESG performance or expert evaluation results.

[0040] S32. Select machine learning algorithms (such as random forest, gradient boosting tree, LSTM, etc.) and optimize them for ESG prediction tasks; evaluate and train machine learning models; measure model performance through indicators such as AUC, accuracy, and recall.

[0041] S33. Based on the trained model, make trend forecasts on future ESG performance; identify potential risk points and evaluate potential opportunities.

[0042] The working principle of the above technical solution is: collect and integrate historical ESG data and related macroeconomic indicators (such as GDP growth rate, policy changes, etc.). These data will be used as input features of the machine learning model to capture various factors that affect ESG performance; then, build the model's labels, that is, the historical performance of the company's ESG performance or the results of expert evaluation. These labels represent the target variables that the model needs to predict and are an important basis for evaluating the performance of the model; select appropriate machine learning algorithms based on the characteristics of the ESG prediction task. These algorithms may include algorithms such as random forests, gradient boosting trees (such as XGBoost, LightGBM) for processing traditional structured data, and algorithms such as LSTM (Long Short-Term Memory Network) for processing time series data; optimize the algorithm for ESG prediction tasks to improve the prediction accuracy of the model. This may include adjusting the parameters of the algorithm, using feature engineering techniques to improve the quality of input features, etc.; evaluate the model through methods such as cross-validation to ensure that it has good generalization ability. Then, use the training data to train the model so that it can learn the relationship between input features and ESG performance; measure the performance of the model through indicators such as AUC (area under the curve), accuracy, and recall. These indicators can intuitively reflect the accuracy and reliability of the model in predicting ESG performance; based on the trained machine learning model, current and future macroeconomic indicators and possible ESG-related data are input to make trend forecasts for future ESG performance. These forecast results can help companies understand the changing trends of future ESG performance; by analyzing the forecast results, potential risk points that may affect ESG performance can be identified. These risk points may include policy changes, changes in the market environment, supply chain issues, etc.; in addition to identifying risk points, ESG-related opportunities that may arise in the future can also be evaluated. For example, the government may introduce new environmental protection policies or incentives, which will bring opportunities for companies to improve ESG performance and brand image.

[0043] The effects of the above technical solutions are: integrating historical ESG data and macroeconomic indicators as input features, so that the model can comprehensively consider multiple factors of the internal and external environment of the enterprise. This helps to more comprehensively evaluate the complexity and dynamics of the company's ESG performance; select appropriate machine learning algorithms according to the characteristics of the ESG prediction task and optimize them. This targeted selection and optimization can significantly improve the prediction accuracy of the model and provide companies with more reliable ESG performance prediction results; measure model performance through indicators such as AUC, accuracy, and recall, and conduct continuous evaluation and feedback. This helps to promptly discover and correct deficiencies in the model and further improve prediction accuracy; predict future ESG performance trends based on trained models, so that companies can understand the changing trends of ESG performance in advance. This helps companies to be prepared to respond and adjust strategies and plans in a timely manner; identify potential risk points and opportunities that may affect ESG performance based on the prediction results. This helps companies take measures in advance to avoid risks or seize opportunities and achieve sustainable development; by predicting ESG performance trends and identifying potential risks and opportunities, companies can allocate resources more reasonably, optimize ESG management inputs, and improve resource utilization efficiency; by continuously improving ESG performance and responding to potential risks, companies can enhance their ESG performance and social responsibility image. This helps enhance the company's brand value and market competitiveness; good ESG performance can attract more investors and stakeholders who pay attention to sustainable development. This brings more financial support and market opportunities to companies, promoting their long-term development.

[0044] In one embodiment of the present invention, the S33 includes: Divide the forecasting task into short-term (e.g. quarterly), medium-term (e.g. annual) and long-term (e.g. multi-year) forecasts; and adjust the model parameters for different time scales; Construct different scenarios of macroeconomics, policy changes, and industry development trends, input them into the model for simulation and forecasting, and analyze the possible changes in corporate ESG performance under different scenarios; Use anomaly detection algorithms to identify outliers in forecast results that significantly deviate from historical trends or industry standards as a preliminary screening of potential risk points; Build an ESG risk assessment framework, combine industry characteristics, corporate reality and expert knowledge to further evaluate the identified potential risk points; including the possibility of risk occurrence, degree of impact and duration.

[0045] Assign a quantitative score to each potential risk point, comprehensively consider various dimensions of risk, and form a risk priority ranking so that enterprises can formulate targeted response measures.

[0046] Analyze positive trends in the forecast results, such as significant improvements in ESG performance and outstanding performance in certain aspects, as clues to potential opportunities; compare the company's ESG forecast results with the industry average and leading companies to identify areas in which the company has the potential to surpass competitors or lead industry development.

[0047] Use SWOT analysis (strengths, weaknesses, opportunities, threats) to conduct in-depth assessments of identified potential opportunities, consider factors such as market potential, competition, and corporate resource matching, and determine priorities and development strategies.

[0048] The working principle of the above technical solution is: subdividing the ESG performance prediction task into short-term (such as quarterly), medium-term (such as annual) and long-term (such as multi-year) predictions. This subdivision helps companies formulate corresponding strategies and plans according to the needs of different time scales; adjusting the parameters of the machine learning model for prediction tasks of different time scales. Since the data characteristics and change patterns of different time scales may be different, it is necessary to optimize the prediction effect by adjusting the model parameters; constructing different scenarios such as macroeconomics, policy changes and industry development trends to reflect possible environmental changes in the future; inputting the constructed scenarios into the model for simulation prediction, and analyzing the possible changes in the company's ESG performance under different scenarios. This helps companies understand the potential fluctuations and risks of ESG performance in different environments; using anomaly detection algorithms to identify outliers in the prediction results that significantly deviate from historical trends or industry standards. These outliers may represent potential risk points or uncertainties; constructing an ESG risk assessment framework, and further evaluating the identified potential risk points in combination with industry characteristics, actual corporate conditions and expert knowledge. The assessment content includes the possibility of risk occurrence, impact and duration; assigning quantitative scores to each potential risk point, and comprehensively considering various dimensions of risk to form a risk priority ranking. This helps companies develop targeted response measures based on the importance and urgency of risks; analyze positive trends in forecast results, such as significant improvements in ESG performance and outstanding performance in certain aspects, as clues to potential opportunities; compare and analyze the company's ESG forecast results with the industry average and leading companies to identify in which areas the company has the potential to surpass competitors or lead industry development; use SWOT analysis (strengths, weaknesses, opportunities, threats) to conduct in-depth evaluation of identified potential opportunities. Consider factors such as market potential, competitive conditions, and corporate resource matching to determine priorities and development strategies; based on the above analysis results, companies can comprehensively evaluate ESG performance forecasts, risk assessments, and opportunity identification results at different time scales, and develop targeted strategies and plans. These strategies and plans are designed to optimize ESG performance, reduce potential risks, and seize development opportunities.

[0049] The effect of the above technical solution is: by subdividing the prediction task into short-term, medium-term and long-term, and adjusting the model parameters for different time scales, the accuracy of the prediction can be significantly improved. This subdivision method can more accurately capture the trend of ESG performance changes in different time ranges; the adjustment for different time scales makes the model more flexible and adaptable, and can better respond to market changes and policy adjustments in different time periods; constructing different scenarios such as macroeconomics, policy changes and industry development trends for simulation and prediction, which helps companies to fully understand the possible changes in ESG performance under different environments. This provides an important reference for companies to formulate response strategies; using anomaly detection algorithms to identify potential risk points, and further evaluating them by constructing an ESG risk assessment framework, including the possibility of risk occurrence, impact and duration. This comprehensive risk assessment helps companies to identify and respond to potential risks in a timely manner and reduce losses; assigning quantitative scores to each potential risk point and forming a risk priority ranking helps companies to formulate targeted response measures based on the importance and urgency of the risk and improve risk management efficiency; by analyzing the positive trends in the prediction results and comparing the company with the industry average and leading companies, potential opportunities for companies in the ESG field can be identified. This provides strong support for enterprises to formulate development strategies; using SWOT analysis to conduct in-depth evaluation of identified potential opportunities, considering factors such as market potential, competitive conditions, and corporate resource matching, can help companies determine priorities and development strategies and better seize ESG-related opportunities; by predicting ESG performance trends and assessing potential risks and opportunities, companies can formulate strategies and plans that are more in line with actual conditions. This helps companies achieve sustainable development and enhance their competitive advantages; by continuously improving ESG performance and responding to potential risks, companies can enhance their ESG performance and social responsibility image. This helps enhance the brand value and market competitiveness of companies; companies that perform well in the ESG field have the opportunity to become industry benchmarks and leaders, attract more investors and stakeholders who focus on sustainable development, and further promote the long-term development of companies.

[0050] In one embodiment of the present invention, the S4 includes: S41. Collect feedback from stakeholders through multiple methods, including online surveys, interviews, and seminars; S42. Evaluate the ESG performance of enterprises by combining the sentiment analysis results of NLP analysis and the trends predicted by the machine learning model; S43. Use data analysis tools (such as Tableau and PowerBI) to visualize data and present ESG key indicators, trend changes and risk assessment results; S44. Analyze competitors’ ESG performance, identify industry best practices, and determine the key areas and priorities of ESG information disclosure based on stakeholder needs and feedback; S45. Design a disclosure strategy, including the depth and breadth of the disclosure content, disclosure frequency, disclosure channels (such as annual reports, official websites, social media, etc.) and interactive mechanisms; generate an ESG report or summary based on the disclosure strategy.

[0051] The working principle of the above technical solution is: through online surveys, interviews, seminars and other methods, feedback information from different stakeholders (such as investors, customers, employees, suppliers, governments and non-governmental organizations, etc.) is widely collected. These methods can ensure the diversity and comprehensiveness of information, covering different perspectives and concerns; the collected feedback information is sorted, classified and summarized to form a systematic feedback database. This is helpful for subsequent analysis and evaluation; natural language processing (NLP) technology is used to analyze the sentiment tendency of the collected text information. This helps to identify the emotional attitudes of stakeholders towards the company's ESG performance (such as positive, negative or neutral), as well as specific concerns and opinions; combined with the company's ESG performance trends predicted by the machine learning model, the sentiment tendency analysis results of NLP analysis are combined with the predicted trends for comprehensive evaluation. This combination can more comprehensively reflect the current status and future trends of the company's ESG performance; data analysis tools such as Tableau and PowerBI are used to visualize ESG key indicators, trend changes and risk assessment results. Through charts, graphs and other forms, the key information of the company's ESG performance is intuitively displayed; data visualization helps stakeholders quickly understand the key points of the company's ESG performance, as well as the company's strengths and weaknesses in different aspects. At the same time, it also provides strong data support for subsequent decision-making; conduct in-depth analysis of competitors' ESG performance to understand the best practices and practices of benchmark companies in the industry. This helps companies clarify their position in the industry and the direction that needs to be improved and enhanced; according to the needs and feedback of stakeholders and the results of competitor analysis, determine the key areas and priorities of the company's ESG information disclosure. This helps companies concentrate resources and prioritize the issues that stakeholders are most concerned about; design specific disclosure strategies based on the determined key areas and priorities of ESG information disclosure. This includes the depth and breadth of the disclosure content, the frequency of disclosure, the disclosure channels (such as annual reports, official websites, social media, etc.) and the interactive mechanism. The design of the disclosure strategy should fully consider the needs and preferences of stakeholders, as well as the actual situation and resource constraints of the company; generate ESG reports or summaries based on the disclosure strategy. ESG reports should comprehensively and accurately reflect the company's performance and efforts in ESG, while highlighting key points and highlights. The summary is a refinement and summary of the core content of the ESG report, making it easier for stakeholders to quickly understand the key information on a company’s ESG performance.

[0052] The effect of the above technical solution is: through online surveys, interviews and seminars, etc., feedback from stakeholders can be collected more comprehensively. This multi-channel collection method not only improves the coverage and accuracy of feedback, but also enhances the sense of participation and importance of stakeholders; data visualization and ESG report disclosure enable companies to clearly and intuitively display their ESG performance, and enhance the transparency of communication with stakeholders. This helps to build trust and reduce misunderstandings and suspicions; using NLP to analyze sentiment tendencies and machine learning models to predict trends can more scientifically evaluate the ESG performance of companies. This combined technology approach can capture more subtle emotional changes and potential trends, improve the accuracy and foresight of the assessment; combining stakeholder feedback, NLP analysis results and machine learning prediction trends for comprehensive evaluation can more comprehensively reflect the ESG performance of companies. This comprehensive evaluation method helps companies discover their own strengths and weaknesses in ESG and provide strong support for subsequent improvements; by analyzing the ESG performance of competitors, identifying industry best practices, and determining the key areas and priorities of ESG information disclosure based on the needs and feedback of stakeholders. This targeted disclosure strategy can ensure that companies concentrate resources, give priority to disclosing information that stakeholders are most concerned about, and improve the pertinence and effectiveness of disclosure; designing a variety of disclosure channels (such as annual reports, official websites, social media, etc.) and interactive mechanisms can meet the information acquisition needs of different stakeholders. This diversified disclosure method helps to expand the scope of information dissemination and improve the accessibility and influence of information; through ESG assessment and information disclosure, companies can clearly understand their own strengths and weaknesses in ESG, thereby identifying opportunities and areas for improvement. This helps companies to formulate targeted improvement plans and promote continuous improvement and development in the field of ESG; by identifying industry best practices and striving to meet or exceed these standards, companies have the opportunity to become benchmarks and leaders in the industry. This can not only enhance the brand image and competitiveness of the company, but also promote the progress and development of the entire industry in ESG; through continuous ESG assessment, information disclosure and improvement, companies can continuously improve their ESG performance and enhance their social responsibility and sustainable development capabilities. This helps companies win the trust and support of more stakeholders and lay a solid foundation for the long-term development of the company; the ESG performance of the company is closely linked to the global sustainable development goals. By strengthening ESG management, companies can contribute to achieving these goals and promote the coordinated development of society, economy and environment.

[0053] In one embodiment of the present invention, S5 includes: S51. Regularly collect and analyze feedback from stakeholders through user feedback channels to identify users’ satisfaction, concerns and improvement suggestions on ESG information disclosure; S52. Through text mining and sentiment analysis technology, deeply analyze the feedback content and extract valuable information and insights; S53. Adjust the data collection strategy and optimize the data preprocessing and fusion algorithms according to user feedback and data analysis results; S54. Revise the ESG disclosure strategy, update the report content and format, and establish a continuous improvement mechanism to regularly review and evaluate the effectiveness of ESG information disclosure.

[0054] The working principle of the above technical solution is as follows: Establish a variety of user feedback channels, such as online surveys, telephone interviews, social media interactions, etc., in order to widely collect feedback from stakeholders; Regularly organize and analyze the collected feedback to identify stakeholders' satisfaction level, main concerns and specific improvement suggestions for ESG information disclosure. This information is the basis for subsequent adjustments and optimization work; Use text mining technology to automatically process feedback content and extract key information, themes and trends. This helps to quickly understand the overall attitude and main views of stakeholders; Use sentiment analysis technology to judge the sentiment tendency of feedback content and identify positive, negative or neutral sentiment attitudes. This helps to more accurately grasp the emotional reactions and potential needs of stakeholders; Adjust the data collection strategy based on user feedback and data analysis results. For example, increase the frequency and depth of data collection for key areas of concern to stakeholders; For areas with less feedback, appropriately reduce the resource investment in data collection; When preprocessing and integrating the collected data, optimize the relevant algorithms based on user feedback and data analysis results. For example, improve data cleaning and denoising algorithms to improve data quality; optimize data fusion algorithms so that data from different sources can more accurately reflect the company's ESG performance; revise the ESG disclosure strategy based on user feedback and data analysis results. This includes updating the content and format of reports to better meet the needs and expectations of stakeholders; adjusting the frequency and channels of disclosure to ensure the timeliness and breadth of information; and establishing a mechanism for regular review and evaluation of the effectiveness of ESG information disclosure. By setting clear evaluation indicators and cycles, conduct a comprehensive evaluation of information disclosure; and adjust and optimize work strategies in a timely manner based on the evaluation results to ensure continuous improvement and development of information disclosure.

[0055] The effects of the above technical solutions are as follows: by regularly collecting and analyzing feedback from stakeholders, companies can timely understand users' satisfaction, concerns and improvement suggestions for ESG information disclosure. This practice of actively listening to users' voices can enhance users' sense of participation and being valued; improvements based on user feedback can more directly meet users' needs and expectations, thereby improving user satisfaction; through in-depth analysis of feedback content using text mining and sentiment analysis technology, companies can extract valuable information and insights. This information helps companies to more accurately grasp users' concerns and needs, thereby optimizing the content and format of information disclosure; adjusting data collection strategies based on user feedback and data analysis results, and optimizing data preprocessing and fusion algorithms can ensure that the collected data is more accurate, comprehensive and useful. This helps to improve the quality and relevance of information disclosure; revising ESG disclosure strategies, updating report content and formats to make them more in line with user expectations and needs. This dynamic revision strategy can ensure the flexibility and adaptability of information disclosure; establishing a continuous improvement mechanism, regularly reviewing and evaluating the effectiveness of ESG information disclosure, can help companies to promptly identify problems and deficiencies, and take effective measures to improve them. This cycle of continuous improvement can drive companies to continuously improve the level and effectiveness of ESG information disclosure; through timely, accurate and comprehensive disclosure of ESG information, companies can enhance their own transparency. This helps reduce information asymmetry and reduce the doubts and concerns of stakeholders; actively responding to user feedback and continuously optimizing information disclosure can establish a good image of the company as responsible and trustworthy. This helps to enhance the company's brand value and market competitiveness; using user feedback as an important driving force for improving ESG management and practices can help companies continuously identify and solve problems and promote continuous improvement of ESG management and practices; by continuously optimizing ESG information disclosure, companies have the opportunity to become benchmarks and leaders in the industry and promote the progress and development of the entire industry in ESG.

[0056] In one embodiment of the present invention, the S53 includes: The collected user feedback is classified according to dimensions such as satisfaction, concerns, and improvement suggestions. Through text mining technology, key information such as high-frequency words, phrases or themes are extracted from a large amount of feedback to quickly locate the ESG issues and potential problems that users are most concerned about.

[0057] Use sentiment analysis algorithms to judge the sentiment tendency of feedback content (positive, negative, neutral) and quantify the intensity of user satisfaction and dissatisfaction; Prioritize data collection channels and content based on the importance and urgency of user feedback, and give priority to collecting ESG data that users are most concerned about or have the most feedback on; Based on user feedback, we will explore new data collection channels, such as emerging social media platforms, professional ESG databases, etc., and optimize the frequency and depth of data collection in existing channels. Establish a data collection quality monitoring mechanism to regularly evaluate the accuracy and completeness of data collection and promptly correct errors and omissions; Based on user feedback and data analysis results, the multi-source data fusion algorithm is iteratively optimized; for example, according to the ESG issues that users are concerned about, the granularity of ontology alignment is adjusted or the feature selection of the machine learning fusion model is improved to improve the accuracy and relevance of the fusion results.

[0058] The working principle of the above technical solution is as follows: the collected user feedback is classified according to the dimensions of satisfaction, focus, improvement suggestions, etc. This classification helps enterprises to understand the different needs and feedback types of users more clearly; text mining technology is used to extract key information from a large amount of feedback, such as high-frequency words, phrases or themes. These key information can quickly locate the ESG issues and potential problems that users are most concerned about, and provide direction for subsequent optimization work; sentiment analysis algorithms are used to judge the sentiment tendency (positive, negative, neutral) of the feedback content. This step aims to quantify the intensity of user satisfaction and dissatisfaction, and provide a reference for the sentiment dimension for subsequent optimization work; through sentiment analysis, enterprises can understand the overall attitude of users towards ESG information disclosure, as well as the sentiment tendency on specific issues. This helps enterprises to grasp the emotional response of users more accurately, so as to formulate more targeted optimization measures; data collection channels and content are prioritized according to the importance and urgency of user feedback. Prioritize the collection of ESG data that users are most concerned about or have the most feedback to ensure that enterprises can respond quickly to user needs and improve the pertinence and effectiveness of information disclosure; based on user feedback, explore new data collection channels, such as emerging social media platforms, professional ESG databases, etc. These new channels can provide enterprises with richer and more diverse data sources, which will help improve the comprehensiveness and timeliness of data; at the same time, optimize the frequency and depth of data collection in existing data collection channels. Adjust the collection strategy based on user feedback and data analysis results to ensure the accuracy and completeness of data; establish a data collection quality monitoring mechanism to regularly evaluate the accuracy and completeness of data collection. By regularly checking and verifying data, timely discover and correct errors and omissions to ensure the reliability and availability of data; iteratively optimize the multi-source data fusion algorithm based on user feedback and data analysis results. According to the ESG issues that users are concerned about, adjust the granularity of ontology alignment or improve the feature selection of machine learning fusion models to improve the accuracy and relevance of fusion results. This optimization can ensure that enterprises can more accurately reflect the real needs and concerns of users when processing multi-source data.

[0059] The effect of the above technical solution is: classifying user feedback according to dimensions such as satisfaction, focus, and improvement suggestions can help companies understand user needs more accurately and formulate targeted improvement measures. This precise feedback processing can significantly improve user satisfaction and enhance user participation; through sentiment analysis algorithms, companies can quantify the intensity of user satisfaction and dissatisfaction, and provide a reference for the sentiment dimension for subsequent optimization work. This helps companies grasp user emotions more carefully and formulate improvement strategies that are more in line with user expectations; prioritizing data collection channels and content according to the importance and urgency of user feedback can ensure that companies prioritize the collection and processing of ESG data that users are most concerned about or have the most feedback. This strategy can significantly improve the efficiency and pertinence of data collection; exploring new data collection channels, such as social media platforms, professional ESG databases, etc., can provide companies with more comprehensive and diverse data sources. At the same time, optimize the collection frequency and depth of existing channels, ensure the comprehensiveness and timeliness of data, and improve the overall quality of data collection; establish a data collection quality monitoring mechanism, regularly evaluate the accuracy and completeness of data collection, and correct errors and omissions in a timely manner. This mechanism can ensure the quality of the data collected by the enterprise, and provide a solid foundation for subsequent data analysis and decision-making; iterative optimization of the multi-source data fusion algorithm based on user feedback and data analysis results can improve the accuracy and relevance of the fusion results. This optimization helps enterprises better understand and use data, and improve the scientificity and credibility of decision-making; through the implementation of S53, enterprises can form a set of data collection and optimization mechanisms based on user feedback. This mechanism can be dynamically adjusted according to changes in market dynamics and user needs, and promote the continuous improvement of ESG information disclosure; by continuously optimizing data collection and fusion algorithms, and improving the quality and effectiveness of ESG information disclosure, enterprises have the opportunity to become benchmarks and leaders in the industry. This can not only enhance the brand image and market competitiveness of the enterprise, but also promote the progress and development of the entire industry in ESG.

[0060] One embodiment of the present invention is a rating maintenance system, comprising a memory, a processor, and a computer program stored in and executable on the memory, wherein the processor executes the program to implement any of the above-described preferred methods for enterprise ESG information disclosure.

[0061] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for optimizing corporate ESG information disclosure. It is characterized in that The method comprises: S1. Collect ESG-related data from different channels, store the collected ESG-related data in the cloud space, pre-process it, and integrate the pre-processed related data through data fusion algorithms to form an ESG information data set; S2. Analyze the relevant data in the dataset through NLP technology to extract key information and sentiment trends related to ESG; S3. Train machine learning models based on historical ESG data and external macroeconomic data to predict corporate ESG performance trends and identify potential risks and opportunities; S4. Generate an ESG information disclosure strategy based on the results of intelligent analysis and the needs survey of stakeholders; and generate an ESG report or summary based on the strategy; S5. Collect relevant information from stakeholders on ESG information disclosure through user feedback channels, and continuously optimize the data collection, processing, analysis and disclosure processes based on user feedback and data analysis results.

2. According to the preferred method for enterprise ESG information disclosure as claimed in claim 1, It is characterized in that Said S1 comprises: S11. Determine the scope and channels of data collection, and automatically collect ESG-related data from the above channels through crawler technology and API interface calls; S12. Preprocess the collected ESG-related data and upload the preprocessed ESG-related data to the cloud storage platform; and encrypt the sensitive data in the cloud storage platform using a data encryption algorithm; S13. Use multi-source data fusion algorithm to integrate data from different channels but related; S14. Construct an ESG information data set, where the ESG information data set includes structured data, semi-structured data, and unstructured data.

3. According to the preferred method for enterprise ESG information disclosure according to claim 2, It is characterized in that The S13 comprises: Further clean up the preprocessed data and unify the metrics and formats between different data sources; Extract key entities from text data through named entity recognition algorithms, and match and link these entities with entities in known databases or knowledge graphs through entity linking techniques; Based on the expertise in the ESG field, we build an ontology model to clarify the relationships and hierarchical structures between data elements. We also use the ontology alignment algorithm to align similar or related concepts in different data sources to eliminate semantic heterogeneity. According to the characteristics and requirements of the data, a fusion strategy is selected, and the data fusion operation is performed according to the selected fusion strategy or model; after the fusion is completed, the fusion results are quantitatively evaluated through evaluation indicators; The fused data is integrated and reconstructed according to the structural requirements of the ESG information data set to form a unified data representation.

4. According to the preferred method for enterprise ESG information disclosure according to claim 1, It is characterized in that The S2 comprises: S21. Preprocess the text data again through the NLP tool library and extract key information from the text; S22, using pre-trained models to represent text and capture the semantic features of the text; S23. Design sentiment dictionary and apply sentiment analysis algorithm to judge the sentiment tendency of text; S24. Analyze the distribution trend of sentiment tendencies and identify the public’s attitudes and opinions towards the company in combination with ESG themes.

5. According to the preferred method for enterprise ESG information disclosure according to claim 1, It is characterized in that The S3 includes: S31. Integrate historical ESG data and macroeconomic indicators as input features of machine learning models and construct labels; S32. Select machine learning algorithms and optimize them for ESG prediction tasks; evaluate and train machine learning models; S33. Based on the trained model, make trend forecasts on future ESG performance; identify potential risk points and evaluate potential opportunities.

6. According to the preferred method for enterprise ESG information disclosure according to claim 5, It is characterized in that The S33 comprises: The forecasting task is subdivided into short-term, medium-term and long-term forecasts; and the model parameters are adjusted for different time scales; Construct different scenarios, input models for simulation and prediction, and analyze the changes in corporate ESG performance under different scenarios; Use anomaly detection algorithms to identify outliers in forecast results that significantly deviate from historical trends or industry standards as a preliminary screening of potential risk points; Build an ESG risk assessment framework, combine industry characteristics, corporate realities and expert knowledge to further evaluate identified potential risk points; Assign a quantitative score to each potential risk point, taking into account various dimensions of risk and forming a risk priority ranking; Analyze positive trends in forecast results as clues to potential opportunities; compare and analyze the company's ESG forecast results with the industry average and leading companies; Utilize SWOT analysis to conduct in-depth assessment of identified potential opportunities, determine priorities and development strategies.

7. According to the preferred method for enterprise ESG information disclosure according to claim 1, It is characterized in that The S4 comprises: S41. Collect feedback information from relevant parties through multiple means; S42. Evaluate the ESG performance of enterprises by combining the sentiment analysis results of NLP analysis and the trends predicted by the machine learning model; S43. Use data analysis tools to visualize data and intuitively display ESG key indicators, trend changes and risk assessment results; S44. Analyze competitors’ ESG performance, identify industry practices, and determine the key areas and priorities of ESG information disclosure based on stakeholder needs and feedback; S45. Design a disclosure strategy and generate an ESG report or summary based on the disclosure strategy.

8. According to the preferred method for enterprise ESG information disclosure according to claim 1, It is characterized in that The S5 comprises: S51. Regularly collect and analyze feedback from stakeholders through user feedback channels to identify users’ satisfaction, concerns and improvement suggestions on ESG information disclosure; S52. Through text mining and sentiment analysis technology, deeply analyze the feedback content and extract valuable information and insights; S53. Adjust the data collection strategy and optimize the data preprocessing and fusion algorithms according to user feedback and data analysis results; S54. Revise the ESG disclosure strategy, update the report content and format, and establish a continuous improvement mechanism to regularly review and evaluate the effectiveness of ESG information disclosure.

9. According to the preferred method for enterprise ESG information disclosure according to claim 1, It is characterized in that The S53 comprises: Classify the collected user feedback according to dimensions and extract key information from a large amount of feedback through text mining technology; Use sentiment analysis algorithms to judge the sentiment tendency of feedback content and quantify the intensity of user satisfaction and dissatisfaction; Prioritize data collection channels and content based on the importance and urgency of user feedback, and give priority to collecting ESG data that users are most concerned about or have the most feedback on; Explore new data collection channels based on user feedback and optimize the frequency and depth of data collection in existing channels; Establish a data collection quality monitoring mechanism to regularly evaluate the accuracy and completeness of data collection and correct errors and omissions; Based on user feedback and data analysis results, the multi-source data fusion algorithm is iteratively optimized.

10. A rating maintenance system, It is characterized in that It includes a memory, a processor, and a computer program stored in the memory and executable on the memory, wherein the processor executes the program to implement the preferred method for enterprise ESG information disclosure as described in any one of claims 1-9.

Citation Information

Patent Citations

  • Enterprise ESG index determination method based on data completion and related product

    CN113313362A

  • Method for calculating and predicting ESG decision expectation value of power grid enterprise and related equipment

    CN118227982A

  • Enterprise comprehensive evaluation method and system based on large language model technology

    CN118313727A

  • Financial and economic customs management system

    CN118552020A

  • ESG Forecasting

    US20220327567A1

Cited By

  • Big data enterprise environment public service system

    CN119809897A

  • Enterprise ESG index determination method and system based on artificial intelligence

    CN120106636A

  • Enterprise demand generation method and system

    CN120746236A

  • Information processing systems, information processing methods, and programs

    JP7872628B1

  • Information processing systems, information processing methods, and programs

    JP7872629B1