Dynamic enterprise ESG idea evaluation system and method based on social media index

Abstract

The invention relates to the cross technical field of computer science and business management, and discloses a dynamic enterprise ESG idea evaluation system and method based on social media indexes. According to the method, through linkage of triple technologies of social media public opinion analysis, policy intelligent analysis and adaptive weight adjustment, the core pain points of high data cost and policy response delay in traditional evaluation are effectively solved; the system captures public emotional tendency and hot topics in real time by using a map information retrieval technology, and realizes multi-dimensional accurate diagnosis of ESG performance in combination with enterprise operation data; an evaluation result is ensured to be continuously optimized along with a market environment through a dynamic weight mechanism, and a customized report with an action guidance value is generated; according to the scheme, the ESG management threshold of small and medium-sized enterprises is remarkably reduced, the sustainable development capability of the small and medium-sized enterprises is systematically improved under the condition that resources are limited, the green financing competitiveness and the social liability fulfillment efficiency are synchronously enhanced, and innovative support is provided for dealing with commercial challenges in the carbon neutralization era.

Description

Technical Field

[0001] The present invention relates to the interdisciplinary field of computer science and business management, and specifically to a dynamic enterprise ESG concept evaluation system and method based on social media indexes. Background Art

[0002] With the rapid development of the global economy and rising environmental awareness, companies are increasingly prioritizing environmental, social, and governance (ESG) performance while pursuing economic returns. However, small and medium-sized enterprises (SMEs) face significant challenges when conducting ESG assessments, primarily due to high data collection costs, delayed policy responses, inefficient stakeholder communication, and insufficient information transparency. Existing ESG assessment methods often rely on traditional manual data collection and qualitative analysis, making it difficult to reflect a company's true performance and changes in the external environment in real time.

[0003] Currently, many small and medium-sized enterprises (SMEs) lack effective tools to quickly access and analyze relevant data when conducting ESG assessments. Especially in today's age of social media, a company's image and public opinion have a significant impact on its ESG performance. However, traditional assessment methods cannot fully integrate public opinion information from social media, resulting in companies being unable to adjust their strategies in a timely manner to respond to changes, which in turn affects their overall performance and competitiveness.

[0004] Furthermore, potential issues with existing assessment systems include the lack of a dynamic feedback mechanism, making real-time adjustments and optimization difficult. Small and medium-sized enterprises often face high costs and low efficiency when conducting ESG assessments due to a lack of expertise and resources. Therefore, developing a dynamic assessment system that integrates social media data, policy texts, and AI algorithms is of urgent practical significance.

[0005] To address these issues, this paper proposes a dynamic ESG assessment system based on social media indices. This system leverages deep learning and big data technologies to collect and analyze public opinion on social media in real time. This system, combined with internal corporate data and policies and regulations, enables a comprehensive assessment of ESG performance. By establishing a dynamic weighting adjustment mechanism, the system can flexibly adjust assessment indicators based on the latest environmental changes and public opinion feedback, resulting in more accurate and reliable results.

[0006] Especially in the context of the growing sophistication of information technology, this system is expected to provide businesses with a comprehensive, dynamic ESG assessment tool, helping small and medium-sized enterprises improve their management and enhance their capabilities in social responsibility and sustainable development. Through intelligent approaches, this system will promote innovative applications of ESG management within businesses, achieving both economic benefits and social value, thereby promoting sustainable development across society. Summary of the Invention

[0007] In response to the shortcomings of the existing technology, the present invention provides a dynamic enterprise ESG concept evaluation system and method based on social media index to solve the background technical problems.

[0008] To achieve the above objectives, the present invention provides the following technical solutions: First, a dynamic enterprise ESG concept assessment method based on social media index, characterized by comprising the following steps:

[0009] Step 1: Establish a multivariate data input module to obtain data;

[0010] Step 2: Based on the data obtained by the multivariate data input module, a social media index dynamic evaluation model is constructed, wherein the model identifies and analyzes social media public opinion data and infers the public image and reputation of the enterprise;

[0011] Step 3: Obtain the user's ESG assessment goals, including specific goals in environmental protection, social responsibility, and corporate governance dimensions;

[0012] Step 4: Obtain and analyze the user's enterprise characteristic data, including enterprise size, industry category, and current ESG performance indicators;

[0013] Step 5: Fusion analysis and calculation of the ESG assessment targets obtained in Step 3 and the enterprise characteristic data obtained in Step 4 with the data generated by the social media index dynamic assessment model in Step 2 to obtain a multi-dimensional ESG performance quantitative assessment report;

[0014] Step 6: Provide a dynamic weight adjustment mechanism that dynamically adjusts the weights of different ESG indicators based on real-time data feedback and changes in public opinion;

[0015] Step 7: Implement policy compliance verification by parsing relevant policy texts to ensure that corporate operations comply with existing laws and regulations, environmental standards, and social responsibility requirements, and generate a compliance report.

[0016] Step 8: Generate a visual report to intuitively display the ESG assessment and analysis results;

[0017] Step 9: After the user implements the recommended ESG management measures, the implementation effect data is obtained to evaluate the effectiveness of the recommended strategy and optimize the social media index dynamic evaluation model.

[0018] Preferably, the visual report includes a chart, a summary, and a detailed description.

[0019] Preferably, the multivariate data input module includes a data cleaning and preprocessing unit for improving the quality of data and the accuracy of analysis.

[0020] Preferably, the data acquired by the multivariate data input module includes three categories:

[0021] The first category is policy documents, including laws and regulations related to the environment, society and corporate governance;

[0022] The second category is social media platform data, specifically real-time hot topics and issues related to the three dimensions of ESG;

[0023] The third category is enterprise-related data, which includes passive collection and active disclosure. Passive collection data includes relevant news and customer comments on social media platforms; active disclosure data includes financial reports, environmental monitoring data, and social responsibility activity records disclosed by enterprises.

[0024] Preferably, the social media index dynamic evaluation model adopts GraphRAG technology to enhance the information retrieval and analysis process through a graph structure; GraphRAG technology organizes and presents the information distributed on social media in a visual form by constructing a graph structure of nodes and edges, thereby forming a relationship network and integrating multi-source data.

[0025] Preferably, the visual report supports user-defined display dimensions and indicators; the visual report supports exporting to PDF, Excel and interactive web page formats.

[0026] Preferably, the method also includes an intelligent monitoring function for real-time tracking and analysis of policy changes, market dynamics and social opinions, and triggering the update of ESG assessment targets in the process of obtaining the user's ESG assessment targets.

[0027] The second aspect is a dynamic enterprise ESG philosophy assessment system based on social media indices. This system is used to implement the dynamic enterprise ESG philosophy assessment method based on social media indices described in the first aspect, and specifically includes:

[0028] A first database unit, for storing data collected by the multivariate data input module;

[0029] A second information acquisition unit is used to obtain the user's ESG assessment goals and enterprise characteristic data;

[0030] The third intelligent analysis unit is used to analyze the user's evaluation goals and feature data with the social media index dynamic evaluation model to generate a quantitative evaluation report;

[0031] The fourth visualization output unit is used to generate and display a visualization report of the ESG assessment results.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] This invention has constructed an ESG dynamic assessment system based on multi-source data fusion for small and medium-sized enterprises. Through the triple technical linkage of social media public opinion analysis, policy intelligent analysis and adaptive weight adjustment, it effectively solves the core pain points of high data cost and slow policy response in traditional assessments; the system uses graphical information retrieval technology to capture public sentiment tendencies and hot topics in real time, and combines corporate operating data to achieve multi-dimensional and accurate diagnosis of ESG performance; through a dynamic weight mechanism, it ensures that the assessment results are continuously optimized with the market environment, and generates customized reports with action-oriented value; this solution significantly lowers the ESG management threshold for small and medium-sized enterprises, enabling them to systematically improve their sustainable development capabilities under resource-constrained conditions, and simultaneously enhance green financing competitiveness and social responsibility fulfillment effectiveness, providing innovative support for responding to business challenges in the carbon neutrality era.

[0034] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 This is a framework diagram of the dynamic enterprise ESG evaluation system based on social media indexes in the present invention;

[0036] Figure 2 This is a visualization diagram of the data flow of the ESG expert system of the present invention;

[0037] Figure 3 This is a flow chart of the dynamic enterprise ESG assessment method based on social media indexes of the present invention;

[0038] Figure 4 This is a diagram of the data collection module architecture of the present invention;

[0039] Figure 5 This is a diagram of the data processing module architecture of the present invention;

[0040] Figure 6 This is the GraphRAG technology stack architecture diagram of the present invention;

[0041] Figure 7 This is a flowchart of the division of tasks between LLM and MOE of the present invention;

[0042] Figure 8 This is a flow chart of the weight adjustment mechanism of LLM and MOE of the present invention. DETAILED DESCRIPTION

[0043] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this technical field without making any creative efforts shall fall within the scope of protection of the present invention.

[0044] See also Figures 1-8 In the specific implementation process of the present invention, the dynamic enterprise ESG evaluation system and method based on social media index realizes in-depth, real-time, diversified and refined evaluation of corporate environment, social responsibility and governance performance through full-process intelligent integration and high automation.

[0045] The present invention includes the following steps: Step 1: Establishing a multivariate data input module

[0046] like Figure 4 As shown, the system first establishes a data-driven infrastructure through a multi-data input module. This module utilizes a layered, integrated architecture to comprehensively collect, accurately, and diversely integrate ESG-related information from companies, further laying a solid data foundation for subsequent intelligent modeling and quantitative assessment. To ensure the comprehensiveness and timeliness of input data, the multi-data input module categorizes information sources into three main categories: policy documents, social media platform data, and company-related data. It also optimizes and standardizes the collection, processing, and integration processes for each type of data.

[0047] First, in terms of policy document collection, the system connects to authoritative databases and public channels of national and local governments, professional regulatory agencies, and industry associations to build a regulatory collection network covering environmental protection laws, social responsibility norms, and corporate governance standards. Policy documents not only include currently effective laws and regulations, but also promptly retrieve and include policy changes, drafts for comments, expert interpretations, judicial interpretations, and regulatory dynamics to ensure the timeliness and foresight of policy information. The system has built-in regulatory parsing and entity recognition tools that can automatically catalog, index, and semantically decompose policy texts, and establish a precise mapping between original regulatory clauses, core concepts, and compliance requirements and their corresponding ESG assessment indicators. For example, when new environmental protection regulations are issued, the system can automatically associate relevant clauses with indicators such as pollutant emissions, energy consumption limits, or ecological compensation under the environmental dimension, dynamically guiding the real-time adaptation of subsequent evaluation logic.

[0048] Secondly, in terms of acquiring and managing data from social media platforms, the system connects to mainstream social media and vertical industry forums in real time, and uses big data crawlers, NLP keyword filtering, and sentiment analysis algorithms to hierarchically collect hot topics, public opinion events, public evaluations, and dynamics of mainstream opinion leaders related to corporate ESG. The system regularly captures high-frequency topics and sudden hot searches related to environmental protection, corporate social responsibility fulfillment, governance behavior, ethical risks, etc. on social platforms, and uses semantic recognition and entity linking technology to effectively classify discussions scattered in different contexts. At the same time, it identifies positive and negative comments, forwarding volume, and interaction heat, and promptly captures potential crises, social concerns, and audience sentiment trends. The construction of this social media data system not only provides companies with ESG feedback from the public's perspective, but also provides data support for dynamic public opinion warnings, reputation risk identification, and social value mining.

[0049] Thirdly, the input of enterprise-related data covers both proactively disclosed data and passively collected data, forming a dual-track parallel structure of enterprise-owned information and external verification. In the proactive disclosure dimension, the system connects to the enterprise's own ERP, ESG information system, financial and environmental annual reports, social responsibility / sustainable development reports, etc., and directly imports the latest publicly disclosed operating performance, environmental indicators, employee rights, innovation practices, governance documents and other data. In order to support the differentiated management needs of enterprises in multiple industries and sizes, the system also has a high-freedom data interface that can be linked with the enterprise's internal real-time monitoring equipment, production management platforms, third-party statistical interfaces, etc. to ensure the authenticity, traceability and timeliness of information. The passive collection dimension focuses mainly on public opinion and consumer feedback, automatically capturing news media exposure, partner evaluations, complaints and reports, consumer ratings and reports from third-party authoritative organizations, and continuously supplementing data perspectives and authenticity verification in combination with various industry credit ratings and social responsibility lists.

[0050] To unify the structure of multi-channel data and eliminate redundancy and errors, the system features a data cleansing and preprocessing unit. All raw data, once received, undergoes a series of processes, including deduplication, noise filtering, field standardization, timestamp correction, and error correction. At the semantic level, natural language processing and entity standardization are used to uniformly map data from diverse sources, formats, and granularities into a standardized ESG indicator system. The data cleansing unit also intelligently identifies information gaps and anomalies, proactively prompting businesses and administrators to complete key information or initiating manual review, significantly improving the data quality, stability, and objectivity of subsequent analysis.

[0051] Step 2: Unstructured data processing and multi-model fusion

[0052] like Figure 4As shown in the figure, after the system completes the input and initial stratification of multivariate data, it enters the core data processing and intelligent integration phase. This step not only cleans and standardizes the data, but more importantly, it systematically improves the usability, insight, and intelligent reasoning depth of unstructured, multi-source, heterogeneous data through the synergy of graph processing technologies such as GraphRAG and the MOE (Mixed Model of Experts) architecture, ultimately providing strong support for subsequent ESG indicator algorithm quantification and advanced decision-making analysis.

[0053] First of all, the foundation of any data processing process lies in strict cleaning and preprocessing. This system combines a number of NLP and big data cleaning algorithms to perform comprehensive deduplication, noise filtering and structured mapping of the original input policy texts, social media public opinion data and corporate multi-source collected information. For example, for policy texts, the system will automatically extract clauses, policy keywords, core regulatory logic and compliance requirements, and perform segmented indexing and semantic classification; for social media content, it not only captures the main body of the text, but also deeply analyzes complex dimensions such as user comments, interaction levels, and emotional polarity to identify opinion tendencies and risk signals; and corporate business data needs to be managed hierarchically according to source attributes to ensure that actively disclosed data and passively collected information can be seamlessly connected in terms of granularity, timeliness, semantics, etc. Once all data items are entered into the database, standardized labels, timestamp synchronization, anomaly detection, and automatic completion of missing or abnormal fields based on business scenarios are used to continuously optimize the overall data structure.

[0054] This high-quality data foundation lays a solid foundation for subsequent intelligent graph construction and deep semantic network analysis. The system then uses GraphRAG graph technology to transform the cleaned structured and semi-structured data into a multi-layered, dynamic ESG knowledge network centered on the enterprise, following a node-edge-relationship logic. The graph nodes encompass entities such as corporate behavior, policies and laws, social issues, public figures, and key public opinion events, while the edges between nodes reveal influence chains, attribution paths, information flows, and interaction patterns. With GraphRAG, the system can dynamically model complex driving forces such as "corporate environmental protection actions—public opinion response—policy pressure—governance adjustments," enabling multi-level information backtracking and automatic analysis of causal chains, effectively presenting an interactive and panoramic view of ESG practices. At the social media topic level, this dynamic network accurately captures discussion heat, the spread of positive and negative emotions, and the paths of key influencers, helping enterprises instantly understand external brand reputation, social image changes, and potential risks.

[0055] like Figure 6As shown, based on the intelligent retrieval and reasoning of graphs, the system further integrates the MOE (hybrid expert model) + LLM (large language model) architecture to achieve the complementary advantages of multi-dimensional expert networks. MOE introduces sub-experts such as "policy-enterprise matching experts", "governance compliance assessment experts" and "ESG value calculation experts" to conduct in-depth interpretation and analysis of data in a division of labor and cooperation in different fields such as policy clauses, social hot spot evaluations, and corporate governance. For example, policy matching experts can accurately match regulatory provisions with specific business events of enterprises and automatically determine the degree of compliance and rectification needs; governance assessment experts proactively identify internal governance weaknesses based on governance data and industry benchmarks, and provide scientific public opinion risk assessments and performance improvement suggestions; ESG value calculation experts apply rating algorithms to dynamically weight, calculate scores and predict trends for the three major sectors of environment, society and governance and social media dimensions.

[0056] MOE's "gating mechanism" adaptively adjusts the weighting of multiple expert analyses based on real-time feedback from social media trends, regulatory changes, and actual business operations. If the system detects a company sparking a public opinion storm due to a negative social media topic, it automatically increases the weighting of the public opinion analysts and, in conjunction with the large-scale LLM model, generates precise crisis response recommendations and corrective action plans. When new policies are introduced across the industry, the weighting of policy experts is automatically increased, and key indicators are dynamically adjusted simultaneously. This ensures that each assessment closely reflects the external environment and data flow, avoiding the risk of misjudgment caused by static, templated results.

[0057] At the same time, with GraphRAG's visualization and multi-path search capabilities, corporate managers, analysts, or third-party users can single-clickly trace the entire chain of data from source to decision-making analysis, achieving highly transparent, traceable, and verified ESG assessment results. The structured results and indicators generated by all deeply integrated analyses directly provide a standardized data source for subsequent corporate diagnostic reports, social media ESG index reports, and personalized strategic recommendations. This allows companies to continuously optimize their ESG governance performance and brand reputation in a complex and dynamic environment using a scientific, data-driven approach, achieving comprehensive improvements in compliance, transparency, and competitiveness.

[0058] After the system completes unstructured data processing and deep integration of multiple models, it enters the demand-driven intelligent analysis phase centered on the enterprise. This step fully relies on the second information acquisition unit to support enterprise users in independently setting the target system, weight distribution, and focus priorities of ESG assessments based on actual management goals and development directions. The system interface provides convenient and flexible parameter configuration functions for enterprise managers, ESG leaders, or decision-making teams. It can set weights, thresholds, and personalized focus items for various sub-indicators in the three major sectors of environmental protection, social responsibility, and corporate governance, such as carbon emissions, employee rights, and governance transparency, and accurately connect with the company's strategic goals or challenges in a specific cycle, such as upgrading industry regulations, maintaining market reputation, and responding to risk events.

[0059] After companies enter key feature data (such as company size, industry category, organizational structure, regional distribution, historical and current ESG performance scores, etc.), the system automatically connects structured company attribute data with the dynamic public opinion semantic network previously generated by GraphRAG processing. At this point, it can not only sense the external issue environment and policy changes in real time, but also form a unique ESG data profile of the current company based on the similarities and differences in public opinion popularity, social impact, and policy thresholds. By extracting target-related features and integrating them with the semantic relationships of social media indexes and policy and regulatory nodes, the model effectively bridges the "data gap" between a company's actual needs and the external environment, making quantitative assessments more accurate and relevant to real-world situations.

[0060] like Figure 7 and Figure 8 As shown, the core of this process lies in the dynamic weighting mechanism and deep learning intelligent analysis of the MOE (Mixed Expert Model). Each sub-expert network within the MOE scientifically allocates attention to different ESG sectors based on the company's set target weights, real-time feedback, and comprehensive environmental signals. For example, when new governance regulations are frequently released, the system automatically identifies external governance pressures and increases the weighting of governance analysis experts in the assessment process, deepening its analysis of corporate governance, compliance operations, and decision-making transparency. If public opinion at a certain stage focuses on environmental issues, the environmental sub-experts automatically increase their scoring weight, and the comprehensive model output will significantly amplify the impact of environmental responsibility fulfillment or carbon emissions management. This adaptive, dynamic weighting mechanism ensures that ESG assessment results dynamically reflect policy trends, social risks, and corporate realities. The MOE also integrates with the Large Language Model (LLM) to output context-aware text analysis, strategic recommendations, and trend interpretation when risk events or compliance shortcomings emerge, making the results more practical and valuable.

[0061] The quantitative ESG performance reports obtained by companies cover the four key areas of "diagnosis - trends - risks - countermeasures." First, the comprehensive score provides a clear reference for companies' industry positioning, while the historical trends of sub-indicators help managers understand the dynamic changes. Automatically generated risk alerts, based on the joint scoring results of policy matching experts, social media opinion experts, and governance and compliance experts, provide companies with efficient early warnings for various risks, including "potential non-compliance, the spread of high-risk topics, and internal governance shortcomings," and offer recommendations for rectification and optimization. All report content is traceable to the source of the graph network, facilitating full understanding, verification, and reuse of the process by companies, third-party organizations, and stakeholders.

[0062] Step 3: Integrate intelligent decision-making based on enterprise needs and dynamically adjust weights

[0063] According to the actual needs of the enterprise, the system supports users to flexibly set the target dimensions and weights of ESG assessment through the second information acquisition unit. Enterprise managers can customize the target thresholds in dimensions such as environmental protection, social responsibility, and corporate governance based on the current industry background, enterprise scale, development strategy, and internal and external challenges in a specific period, and clarify preference items and core focus indicators. This process not only enhances the initiative of enterprises to participate in data governance independently, but also ensures that the output results of the analysis model are fully consistent with the actual management needs of the enterprise itself. After the enterprise has entered the target and enterprise characteristic data (such as scale, industry category, current ESG performance, etc.), the system automatically calls the map and model to timely connect the structured features with dynamic social media public opinion, and complete data fusion and standard feature extraction.

[0064] The system's third intelligent analysis unit undertakes the core tasks of multi-source data fusion, dynamic weight adjustment, indicator quantitative modeling, trend perception and strategy feedback. Its uniqueness lies in the fact that the algorithm realizes an automatic weight adjustment mechanism based on real-time monitoring, which can dynamically adjust the influence ratio of each ESG indicator on the comprehensive evaluation according to the latest market opinions, policy trends, sudden events and changes in the own data cycle. For example, when the society pays close attention to the topic of carbon neutrality at a certain stage, the weight of the environmental dimension in the overall score will be appropriately increased; if new governance regulations are issued at the legal and regulatory level, the system will increase the weight of governance indicator analysis. Based on multi-dimensional quantitative analysis, the system generates a detailed ESG performance score report, which is divided into four major sections: comprehensive score, historical trend of sub-indicators, risk warnings and strategic recommendations, so that companies can quickly identify their own strengths and weaknesses.

[0065] Regarding policy compliance verification, the system deeply analyzes policy texts and, combined with existing enterprise data, automatically determines the compliance of current operational management measures. Whether it's a newly enacted environmental law, a social equity initiative, or corporate governance regulations, the system matches a company's actual actions with policy requirements, automatically providing compliance diagnoses, risk warnings, and corrective action recommendations, significantly reducing legal risks and potential violations. Furthermore, the system allows companies to independently set and dynamically update assessment targets based on industry trends and policy changes, enabling truly "on-demand" and "policy-driven" assessment and management.

[0066] Step 4: Visualization of results output and closed-loop self-evolution mechanism

[0067] When it comes to presenting the final assessment results, the system's visual output module implements three core features: high customization, intelligent interaction, and professional traceability, both in terms of technical architecture and user experience. Aimed at corporate managers, decision-makers, and external stakeholders, this module transcends the limitations of traditional static reporting by providing powerful interactive analysis tools and multi-level output capabilities, significantly enhancing the efficiency and impact of ESG data utilization. Users can flexibly filter and combine environmental, social, governance, and derivative indicators based on their company's management priorities. Through visual drag-and-drop, hierarchical drill-down, and template customization, they can independently generate integrated, customized ESG analysis reports. For example, mid- and senior-level management can generate comprehensive trend reports focused on strategic decision-making; for front-line business units or compliance departments, a detailed operational overview of a specific thematic indicator can be displayed. Reports support standard document formats such as PDF and Excel, as well as real-time, dynamic, interactive web pages that can be embedded in corporate intranets and official websites, ensuring that diverse audiences can intuitively and efficiently understand, compare, and track a company's true ESG performance.

[0068] The system fully considers the diverse needs of different corporate organizational structures and participating roles. It supports report generation in multiple languages ​​and across multiple regions, enabling companies to disclose information from a global perspective and achieve unified governance across multiple business segments within the group. Furthermore, a multi-level, role-based authorization mechanism establishes differentiated data access and editing permissions. Whether at headquarters, decision-makers, subsidiaries, business units, or external partners and regulators, all can efficiently and securely access the required ESG data and insights based on their specific responsibilities, achieving a seamless connection between accountability and transparent governance.

[0069] The report not only includes the ESG composite score, historical trends of each sub-dimension, compliance score, social media index popularity, and a review of key events, but also embeds a queryable traceability path for the MOE and GraphRAG model reasoning chain, allowing users to see the key data basis and evaluation process behind each conclusion with one click. This "analysis is explanation" mechanism greatly enhances the credibility of the report and its value for third-party adoption. The system also integrates a configurable reminder and early warning module, which automatically generates highlighted warnings and response suggestions for companies with major compliance risks, public opinion crises, or negative environmental and social events; for outstanding ESG projects, it can push positive incentives and external communication suggestions to help companies build a good brand image and social capital.

[0070] More importantly, the system has opened up an intelligent closed-loop link for data tracking, strategy implementation, and continuous learning. After the company adopts the recommendations in the report and implements relevant ESG management measures, the system automatically enters a new round of real-time data monitoring and performance results recovery, collecting multi-dimensional data including business changes, social public opinion trends, policy and regulatory updates, and internal governance adjustments. All feedback data is re-entered into the MOE expert model and GraphRAG knowledge graph to drive the self-adjustment of model parameters and analysis rules, and achieve adaptive evolution of the algorithm and continuous improvement of evaluation accuracy. This process not only allows the system's ESG assessment capabilities to maintain long-term coordinated updates with the external environment, but also provides corporate management with scientific decision-making tools under the "action-feedback-optimization" cycle, greatly improving the initiative and foresight of early warning and improvement.

[0071] Furthermore, the system supports in-depth permission grading and multi-perspective presentation modes. Companies can flexibly specify the operational boundaries of different user roles in data access, report editing, or output processes based on multiple levels of authorization, from headquarters to branches to departments. This protects core data security while ensuring controlled information disclosure and efficient collaboration. Customized presentation templates and summaries can be set for different external stakeholders, such as investors, the public, and third-party organizations, to meet the multiple public disclosure requirements of laws and regulations, market communication, and social responsibility, empowering corporate ESG evaluation systems with unprecedented transparency, explainability, and influence.

[0072] In summary, this invention forms an end-to-end, intelligent, and dynamically adaptive enterprise ESG quantitative assessment and management system through high-level data collection and preprocessing, dynamic social media index modeling, flexible indicator customization, real-time weighting optimization, automatic regulatory identification, intelligent visualization output, and continuous closed-loop optimization. Leveraging graph fusion technologies such as GraphRAG and an adaptive algorithm framework, this system significantly improves the scientific and timely nature of assessments, reduces manual effort and operational barriers, and serves as a strategic support platform for sustainable enterprise development and innovation in green governance within the industry.

Claims

1. A dynamic corporate ESG concept assessment method based on social media index, characterized by: The following steps are involved: Step 1: Establish a multivariate data input module to obtain data; Step 2: Based on the data obtained by the multivariate data input module, a social media index dynamic evaluation model is constructed, wherein the model identifies and analyzes social media public opinion data and infers the public image and reputation of the enterprise; Step 3: Obtain the user's ESG assessment goals, including specific goals in environmental protection, social responsibility, and corporate governance dimensions; Step 4: Obtain and analyze the user's enterprise characteristic data, including enterprise size, industry category, and current ESG performance indicators; Step 5: Fusion analysis and calculation of the ESG assessment targets obtained in Step 3 and the enterprise characteristic data obtained in Step 4 with the data generated by the social media index dynamic assessment model in Step 2 to obtain a multi-dimensional ESG performance quantitative assessment report; Step 6: Provide a dynamic weight adjustment mechanism that dynamically adjusts the weights of different ESG indicators based on real-time data feedback and changes in public opinion; Step 7: Implement policy compliance verification by parsing relevant policy texts to ensure that corporate operations comply with existing laws and regulations, environmental standards, and social responsibility requirements, and generate a compliance report. Step 8: Generate a visual report to intuitively display the ESG assessment and analysis results; Step 9: After the user implements the recommended ESG management measures, the implementation effect data is obtained to evaluate the effectiveness of the recommended strategy and optimize the social media index dynamic evaluation model.

2. The dynamic enterprise ESG concept assessment method based on social media index according to claim 1 is characterized by: The visual report in step 8 includes charts, summaries and detailed descriptions.

3. The dynamic enterprise ESG concept assessment method based on social media index according to claim 1 is characterized by: The multivariate data input module includes a data cleaning and preprocessing unit for improving the quality of data and the accuracy of analysis.

4. The dynamic enterprise ESG concept assessment method based on social media index according to claim 1 is characterized by: The data obtained by the multivariate data input module includes three categories: The first category is policy documents, including laws and regulations related to the environment, society and corporate governance; The second category is social media platform data, specifically real-time hot topics and issues related to the three dimensions of ESG; The third category is enterprise-related data, which includes passive collection and active disclosure. Passive collection data includes relevant news and customer comments on social media platforms; active disclosure data includes financial reports, environmental monitoring data, and social responsibility activity records disclosed by enterprises.

5. The dynamic enterprise ESG concept assessment method based on social media index according to claim 1 is characterized by: The social media index dynamic evaluation model adopts GraphRAG technology to enhance the information retrieval and analysis process through graph structure; GraphRAG technology organizes and presents the information distributed on social media in a visual form by constructing a graph structure of nodes and edges, thereby forming a relationship network and integrating multi-source data.

6. The dynamic enterprise ESG concept assessment method based on social media index according to claim 1 is characterized by: The visual report supports user-defined display dimensions and indicators; the visual report supports exporting to PDF, Excel and interactive web page formats.

7. The dynamic enterprise ESG concept assessment method based on social media index according to claim 1 is characterized by: The method also includes an intelligent monitoring function for real-time tracking and analysis of policy changes, market dynamics and social opinions, and triggering the update of ESG assessment targets in the process of obtaining users' ESG assessment targets.

8. A dynamic enterprise ESG concept assessment system based on social media index, characterized by: The system is used to implement the dynamic enterprise ESG concept assessment method based on social media index as described in any one of claims 1 to 7, specifically comprising: A first database unit, for storing data collected by the multivariate data input module; A second information acquisition unit is used to obtain the user's ESG assessment goals and enterprise characteristic data; The third intelligent analysis unit is used to analyze the user's evaluation goals and feature data with the social media index dynamic evaluation model to generate a quantitative evaluation report; The fourth visualization output unit is used to generate and display a visualization report of the ESG assessment results.

Images