Artificial intelligence-based enterprise risk multi-dimensional analysis method and system

Abstract

The application discloses a kind of based on artificial intelligence enterprise risk multidimensional analysis method and system, design artificial intelligence technical field, specifically based on artificial intelligence enterprise risk multidimensional analysis method, the method aims at solving the technical problems of data fragmentation, single analysis dimension, module isolation, lack of closed-loop optimization in enterprise analysis technology.This method includes enterprise portrait construction, enterprise knowledge base construction, public opinion analysis, multi-dimensional risk identification and AI assistant real-time interaction, hierarchical cooperation, two-way linkage, form the complete closed loop of "basic support-analysis processing-core application-interaction output-feedback optimization".The application based on artificial intelligence enterprise multidimensional analysis method and system realizes the comprehensive integration and accurate analysis of enterprise multidimensional data, improves the efficiency and accuracy of enterprise analysis, reduces the operation complexity, and is widely applicable, can provide integrated enterprise analysis service.

Description

Technical Field

[0001] This invention belongs to the field of artificial intelligence technology, specifically relating to a method and system for multi-dimensional analysis of enterprise risk based on artificial intelligence. Background Technology

[0002] With the deepening of digital transformation, the business environment is becoming increasingly complex, with risks from multiple dimensions such as operations, finance, law, and market intertwined and overlapping. This places higher demands on the real-time, accurate, and intelligent level of enterprise risk control. In recent years, AI technology, with its powerful data processing and intelligent analysis capabilities, has gradually penetrated into the field of enterprise risk control, providing new technological paths for scenarios such as risk identification and public opinion monitoring.

[0003] While existing technologies have enabled some intelligent functions in enterprise risk control, several shortcomings remain in practical applications: First, enterprise profile building is largely limited to static data integration, lacking AI-based dynamic updates and precise optimization capabilities, making it difficult to match changes in enterprise operating status and resulting in a disconnect between profiles and actual risk scenarios. Second, there is data fragmentation in the public opinion monitoring and risk identification stages; public opinion information cannot be effectively synchronized to the risk identification model, and risk identification often focuses on a single dimension, lacking multi-dimensional risk linkage analysis, leading to insufficient accuracy in early warnings. Third, existing risk control solutions lack user-friendly human-computer interaction and the support of structured knowledge bases, resulting in low data reuse rates, poor technical synergy across stages, and an inability to form a closed-loop risk control system. Fourth, some solutions rely on traditional algorithms, which have limited feature extraction capabilities for complex data, making it difficult to adapt to risk control scenarios with massive amounts of multi-source heterogeneous data and failing to fully leverage the technological advantages of AI.

[0004] Therefore, based on the shortcomings of the existing technologies, there is an urgent need for an AI-based multi-dimensional enterprise analysis method that can dynamically construct enterprise profiles, identify multi-dimensional risks, and coordinate technologies throughout the entire process, in order to make up for the deficiencies of existing technologies and meet the needs of enterprises for refined risk control. Summary of the Invention

[0005] This invention aims to address the following problems in existing enterprise risk control: First, enterprise profiles and risk identification data are fragmented and lack coordination. Profiles are mostly static data integrations, lacking dynamic optimization capabilities based on AI, and public opinion data cannot be effectively linked to the risk identification process, resulting in poor accuracy and timeliness of risk identification. Second, existing risk control solutions have not formed a closed-loop collaborative system, lacking structured knowledge base support and efficient human-computer interaction mechanisms, resulting in low data reuse rates, poor coordination among various technical links, and an inability to fully leverage the advantages of AI in processing massive amounts of multi-source heterogeneous data.

[0006] To address the aforementioned technical problems, this invention provides a method and system for multi-dimensional enterprise risk analysis based on artificial intelligence.

[0007] A multi-dimensional enterprise risk analysis method based on artificial intelligence includes the following steps:

[0008] Collect multi-dimensional raw data from enterprises, generate structured enterprise profiles through preprocessing and feature extraction, and establish a dynamic update mechanism to update the enterprise profiles. Collect knowledge information from multiple sources and build a structured enterprise knowledge base through integration and structuring. Collect relevant public opinion information about enterprises from multiple channels, and generate public opinion analysis results after text preprocessing, sentiment recognition and hot topic extraction; Based on the enterprise profile, public opinion analysis results, and enterprise knowledge base, risk characteristics are extracted and input into the risk assessment model. Risk scores are obtained through reasoning, and risk levels are determined. Early warnings are triggered based on risk levels, and corresponding response strategies are matched to form risk results. The risk results are fed back to the users and user feedback is obtained. Based on the user feedback, the results are transmitted back and dynamically updated and optimized.

[0009] As one possible implementation method, the enterprise profile is obtained through the following steps: The method combines web crawling and API calls to collect multi-dimensional raw data from enterprises. This multi-dimensional raw data includes basic enterprise information, operating data, industry position data, and credit-related data. The original data is cleaned, standardized, and structured to obtain preprocessed data. The structured processing includes filling in missing values, deleting duplicate values, and removing invalid data. Feature extraction is performed on the preprocessed data to obtain feature data, which includes basic features, business features, industry features and credit features. The feature data is weighted and assigned, and a comprehensive score for each enterprise is calculated and integrated to form a structured enterprise profile. It adopts a combination of triggered updates and periodic updates to retain historical enterprise profile versions and support version rollback.

[0010] As one possible implementation method, the construction of an enterprise knowledge base includes the following steps: Collect knowledge information from multiple sources, including internal historical data, industry reports, policies and regulations, risk response cases, and industry trends. Data collection methods include data import, web scraping, API calls, and manual entry. Perform deduplication, verification, and removal of redundant, invalid, and conflicting data from multi-source knowledge information; The integrated knowledge information is classified and tagged to build a hierarchical knowledge system tree. We regularly update the knowledge base content by receiving external feedback data, and optimize the knowledge base retrieval and accurate recommendation logic.

[0011] As one possible implementation method, the process of collecting multi-channel corporate public opinion information, performing text preprocessing, sentiment recognition, and hotspot extraction, and then generating public opinion analysis results includes the following steps: Collect relevant public opinion information about enterprises from multiple channels. The information collection methods include web scraping technology and API interface calls. The multiple channels include news media platforms, social platforms and industry forums and communities. The public opinion information is segmented, stop words are removed, and normalization is performed to remove irrelevant characters and meaningless words, and synonyms and near-synonyms are standardized to obtain the public opinion text. A pre-trained natural language processing model is used to perform sentiment recognition on public opinion texts, outputting sentiment tendency and confidence level, and low confidence level results are manually reviewed; Using keyword extraction and hot topic filtering algorithms, the core keywords of public opinion texts are extracted and integrated to form hot topics related to enterprises and industries, and the trend of hot topic dissemination is tracked in real time. Based on sentiment trends and relevant hot topics related to the company and industry, public opinion analysis results are generated. These results include sentiment distribution, hot topic reports, and public opinion dissemination trends. The results are then used to update the company profile and knowledge base.

[0012] As one possible implementation method, the risk outcome is obtained through the following steps: Based on enterprise profiles, public opinion analysis results, and enterprise knowledge base, feature extraction is performed to obtain risk feature vectors, which include operational risk features, financial risk features, legal risk features, and market risk features. Collect historical risk data and corresponding historical risk feature vectors of enterprises, and divide them into training set, validation set and test set after manual annotation. Use machine learning model combination for training and optimization. Input the risk feature vector into the trained risk assessment model to calculate the enterprise risk score and determine the risk level according to preset rules; A corresponding early warning level is set according to the risk level. An early warning mechanism is triggered based on the early warning level, and targeted response strategies are matched with risk response cases in the knowledge base, thereby forming a risk result. The risk result includes risk score, risk level, early warning information, and response strategy.

[0013] As one possible implementation method, it also includes real-time interaction based on an AI assistant, specifically: The system receives user query requests through an AI assistant, which include enterprise profiling, public opinion analysis, risk identification, and enterprise knowledge base. The query requests are formatted, segmented, and stop words are removed. Keywords are extracted and combined with algorithms to clarify the user's query intent. Based on the query intent, the corresponding module is invoked to output data, which is then organized into structured response content; The system provides structured response content to users, supporting viewing, previewing, and report downloading, and offers multi-round interactive functionality. Collect user query requests, operation feedback, and response evaluations, and transmit them synchronously to other modules to drive optimization.

[0014] An AI-based multi-dimensional enterprise risk analysis system includes an enterprise profile building module, an enterprise knowledge base building module, a public opinion analysis module, a multi-dimensional risk identification module, and a feedback and update module. The enterprise profile building module collects multi-dimensional raw data of enterprises, generates structured enterprise profiles through preprocessing and feature extraction, and establishes a dynamic update mechanism to update the enterprise profiles. The enterprise knowledge base construction module collects knowledge information from multiple sources and constructs a structured enterprise knowledge base through integration and structuring. The public opinion analysis module collects relevant public opinion information about enterprises from multiple channels, and generates public opinion analysis results after text preprocessing, sentiment recognition and hot topic extraction. The multi-dimensional risk identification module extracts risk features based on the enterprise profile, public opinion analysis results, and enterprise knowledge base, inputs them into the risk assessment model, obtains a risk score through reasoning and determines the risk level, triggers an early warning based on the risk level and matches a response strategy to form a risk result. The feedback update module sends the risk results back to the user and obtains user feedback. Based on the user feedback, it performs reverse transmission and dynamic updates and optimizations.

[0015] A computer-readable storage medium storing a computer program, characterized in that the computer program, when executed by a processor, implements the following method: Collect multi-dimensional raw data from enterprises, generate structured enterprise profiles through preprocessing and feature extraction, and establish a dynamic update mechanism to update the enterprise profiles. Collect knowledge information from multiple sources and build a structured enterprise knowledge base through integration and structuring. Collect relevant public opinion information about enterprises from multiple channels, and generate public opinion analysis results after text preprocessing, sentiment recognition and hot topic extraction; Based on the enterprise profile, public opinion analysis results, and enterprise knowledge base, risk characteristics are extracted and input into the risk assessment model. Risk scores are obtained through reasoning, and risk levels are determined. Early warnings are triggered based on risk levels, and corresponding response strategies are matched to form risk results. The risk results are fed back to the users and user feedback is obtained. Based on the user feedback, the results are transmitted back and dynamically updated and optimized.

[0016] An artificial intelligence-based multi-dimensional enterprise risk analysis device includes a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that the processor, when executing the computer program, implements the following method: Collect multi-dimensional raw data from enterprises, generate structured enterprise profiles through preprocessing and feature extraction, and establish a dynamic update mechanism to update the enterprise profiles. Collect knowledge information from multiple sources and build a structured enterprise knowledge base through integration and structuring. Collect relevant public opinion information about enterprises from multiple channels, and generate public opinion analysis results after text preprocessing, sentiment recognition and hot topic extraction; Based on the enterprise profile, public opinion analysis results, and enterprise knowledge base, risk characteristics are extracted and input into the risk assessment model. Risk scores are obtained through reasoning, and risk levels are determined. Early warnings are triggered based on risk levels, and corresponding response strategies are matched to form risk results. The risk results are fed back to the users and user feedback is obtained. Based on the user feedback, the results are transmitted back and dynamically updated and optimized.

[0017] This invention, by adopting the above technical solutions, has significant technical effects: (1) By integrating data and knowledge from three major modules—enterprise profiling, public opinion analysis, and enterprise knowledge base—AI technology can be used to achieve intelligent identification, early warning, and precise matching of multi-dimensional enterprise risks and response strategies, thereby significantly improving the comprehensiveness and accuracy of enterprise analysis and solving the problems of single analysis dimensions and insufficient accuracy of existing technologies. (2) By constructing a two-way closed-loop collaboration mechanism of five modules, namely “basic support → analysis and processing → core application → interactive output → feedback optimization”, and combining dynamic update technology, the data of each module can be interconnected, feedback can be linked and the whole system can be adaptively optimized, which improves the timeliness and convenience of enterprise analysis and breaks the limitations of isolated operation of each analysis module and lack of iterative optimization in the existing technology.

[0018] Other advantages, objectives and features of the present invention will be apparent in part from the following description, and in part from the understanding of those skilled in the art through study and practice of the invention. Attached Figure Description

[0019] Figure 1 This is a schematic flowchart of the method of the present invention; Figure 2 This is a schematic diagram of the modules of the system of the present invention. Detailed Implementation

[0020] To clearly illustrate the present invention and make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings, so that those skilled in the art can implement them based on the description.

[0021] In the context of this invention, a trusted execution environment (TEE) provides an isolated runtime environment from the perspective of underlying hardware and operating system, protecting the code and data running within it from external attacks, including attacks from the operating system, hardware, and other applications. This technology has been used in some fields to achieve the objectives described above, and some of its basic principles are known to those skilled in the art. However, those skilled in the art will understand how to apply this technology in this context after reading this application, and will clearly recognize that the technology, combined with other features in a specific context, possesses novelty.

[0022] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0023] Example 1: A multi-dimensional enterprise risk analysis method based on artificial intelligence, such as Figure 1 As shown, the specific steps are as follows: S100. Collect multi-dimensional raw data of enterprises, generate structured enterprise profiles through preprocessing and feature extraction, and establish a dynamic update mechanism to update the enterprise profiles. S200: Collect knowledge information from multiple sources and build a structured enterprise knowledge base through integration and structuring. S300: Collects relevant public opinion information about enterprises from multiple channels, and generates public opinion analysis results after text preprocessing, sentiment recognition and hotspot extraction; S400. Based on the enterprise profile, public opinion analysis results and enterprise knowledge base, extract risk characteristics and input them into the risk assessment model. Obtain a risk score and determine the risk level through reasoning. Trigger an early warning and match response strategies according to the risk level to form a risk result. S500: Feedback the risk results to the user and obtain user feedback, and then transmit the feedback in reverse and dynamically update and optimize the system.

[0024] This invention proposes an AI-based multi-dimensional enterprise analysis method and system. The method includes enterprise profile construction, enterprise knowledge base construction, public opinion analysis, multi-dimensional risk identification, and real-time interaction with an AI assistant. It features layered collaboration and two-way linkage, forming a complete closed loop of "basic support - analysis and processing - core application - interactive output - feedback optimization". This achieves comprehensive integration and accurate analysis of multi-dimensional enterprise data, improves the efficiency and accuracy of enterprise analysis, reduces operational complexity, has a wide range of applications, and can provide integrated enterprise analysis services.

[0025] In step S100, multi-dimensional raw data of the enterprise is collected, preprocessed, and feature extracted to generate a structured enterprise profile, and a dynamic update mechanism is established. Specifically, the steps include: We use a combination of web crawling and API calls to collect multi-dimensional raw data from enterprises, including basic enterprise information, operational data, industry position data, and credit-related data, ensuring that the data sources are legal and comprehensive. The collected raw data is cleaned, standardized, and structured. Missing values ​​are filled, duplicate values ​​are deleted, and invalid data is removed. Unstructured data is transformed into structured data, eliminating the influence of units and ensuring data accuracy and usability. From the preprocessed structured data, the core characteristics of enterprises are extracted, including basic characteristics, operational characteristics, industry characteristics and credit characteristics, comprehensively covering the basic situation, operational level, industry position and credit status of enterprises; The extracted core features are weighted and assigned, and the enterprise's comprehensive score is calculated using a weighted summation method. The comprehensive score is combined with various feature details to form a structured enterprise profile, which is presented in an intuitive and easy-to-understand form. Establish a mechanism that combines triggered updates with periodic updates, receive external feedback data (including public opinion analysis results, risk identification results, and user feedback), update the enterprise profile in real time or periodically, retain historical versions and support version rollback, and ensure that the enterprise profile can reflect the latest status of the enterprise.

[0026] In step S200, the collection of multi-source knowledge information, after integration and structuring, constructs a structured enterprise knowledge base and continuously optimizes it, specifically including the following steps: We use a combination of data import, web scraping, API calls, and manual input to collect multi-source knowledge information, including internal historical data, industry reports, policies and regulations, risk response cases, and industry trends. The collected knowledge information is deduplicated and verified to remove redundant, invalid, and conflicting data. Conflicting information is verified and accurate content is retained to ensure the accuracy and completeness of the knowledge information. The integrated knowledge information is classified and tagged to build a hierarchical knowledge system tree, transforming the knowledge information into a searchable and associative structured form; Establish a knowledge update mechanism to receive external feedback data (including public opinion analysis results, risk identification results, and user feedback), regularly update the knowledge base content, optimize knowledge retrieval and precise recommendation logic, and improve the practicality and relevance of the knowledge base.

[0027] In step S300, the collection of multi-channel enterprise-related public opinion information, after text preprocessing, sentiment recognition, and hot topic extraction, generates public opinion analysis results, specifically including the following steps: By combining web crawling technology with API calls, we collect relevant public opinion information about enterprises from multiple channels, covering public opinion channels such as news media platforms, social media platforms, and industry forums and communities, to ensure that the public opinion information is comprehensive and real-time. The collected public opinion text is processed by word segmentation, stop word removal and normalization, irrelevant characters and meaningless words are removed, and synonyms and near-synonyms are standardized to improve the accuracy of text analysis; A pre-trained natural language processing model is used to perform sentiment recognition on the pre-processed public opinion text, outputting the sentiment tendency and confidence level. Low-confidence results are manually reviewed to ensure the accuracy of sentiment recognition. By combining keyword extraction and hot topic filtering algorithms, core keywords are extracted from public opinion texts, integrated to form hot topics related to enterprises and industries, and the spread of hot topics is tracked in real time. Fading hot topics are archived. Generate public opinion analysis results, including public opinion sentiment distribution, hot topic reports, and public opinion dissemination trends, and push them simultaneously to the multi-dimensional risk identification module and the AI ​​assistant real-time interaction module, and provide feedback to the enterprise profile building module and the enterprise knowledge base building module for updates and optimization.

[0028] In step S400, the enterprise profile, public opinion analysis results, and knowledge base knowledge are integrated to extract risk features and input them into the trained model. A risk score is calculated, the risk level is determined, an early warning is triggered, and a response strategy is matched to generate a risk result. Specifically, this includes the following steps: By integrating the core characteristics of enterprises output by the enterprise profiling module, the public opinion results output by the public opinion analysis module, and the knowledge information output by the enterprise knowledge base construction module, multi-dimensional risk-related characteristics are extracted, including operational risk characteristics, financial risk characteristics, legal risk characteristics, and market risk characteristics, forming a complete risk characteristic vector. Collect historical risk data and corresponding risk feature vectors of enterprises, manually label the data, divide it into training set, validation set and test set, and use machine learning model combination for training and optimization to ensure the model recognition accuracy and calculation accuracy. The extracted risk feature vectors are input into the trained model to calculate the enterprise risk score, and the enterprise risk level is determined according to the preset rules to clearly distinguish different risk levels. Set corresponding early warning levels based on risk levels. When the enterprise's risk level reaches the early warning threshold, the early warning mechanism is triggered. Combine risk response cases in the enterprise knowledge base construction module to match targeted risk response strategies. Generate complete risk results, including risk score, risk level, early warning information and response strategies, and push them simultaneously to the AI ​​assistant real-time interaction module, and provide feedback to the enterprise profile building module and enterprise knowledge base building module for updates and optimization.

[0029] In step S500, the AI ​​assistant receives user query requests in real time, parses them, calls the data corresponding to the above steps, generates a structured response feedback to the user, and collects user feedback, specifically including the following steps: It provides multiple interactive entry points to receive query requests from users in natural language, covering various query needs related to enterprise profiling, public opinion analysis, risk identification, and enterprise knowledge base; The received user query requests are formatted, segmented, and stop word removed. Core keywords are extracted, and keyword matching and intent recognition algorithms are combined to clarify the user's query intent and core needs. Based on the user's query intent, the corresponding module's output data (enterprise profile, public opinion results, risk results, and knowledge base content) is retrieved, and the data is organized into structured response content. Visual charts are added when necessary to improve the readability of the response. The structured response content is fed back to the user, allowing the user to view, preview, and download the report. It provides multi-round interactive functions and accurately responds to the user's follow-up questions. Collect user query requests, operation feedback, and response evaluations, and synchronously transmit the feedback information to the other four major modules of the system to drive each module to perform precise optimization and achieve closed-loop iteration of the system.

[0030] In summary, the AI-based multi-dimensional enterprise analysis method and system provided by this invention aims to solve the technical problems of data fragmentation, single analysis dimensions, isolated modules, and lack of closed-loop optimization in enterprise analysis technology. This method includes enterprise profile construction, enterprise knowledge base construction, public opinion analysis, multi-dimensional risk identification, and real-time interaction with an AI assistant. It features layered collaboration and two-way linkage, forming a complete closed loop of "basic support - analysis and processing - core application - interactive output - feedback optimization." This AI-based multi-dimensional enterprise analysis method and system achieves comprehensive integration and accurate analysis of multi-dimensional enterprise data, improving the efficiency and accuracy of enterprise analysis, reducing operational complexity, and has a wide range of applications, providing integrated enterprise analysis services.

[0031] Example 2: An AI-based multi-dimensional enterprise risk analysis system, such as Figure 2 As shown, it includes an enterprise profile building module 100, an enterprise knowledge base building module 200, a public opinion analysis module 300, a multi-dimensional risk identification module 400, and a feedback update module 500; The enterprise profile building module 100 collects multi-dimensional raw data of enterprises, generates a structured enterprise profile through preprocessing and feature extraction, and establishes a dynamic update mechanism to update the enterprise profile. The enterprise knowledge base construction module 200 collects knowledge information from multiple sources and constructs a structured enterprise knowledge base through integration and structuring. The public opinion analysis module 300 collects relevant public opinion information about enterprises from multiple channels, and generates public opinion analysis results after text preprocessing, sentiment recognition and hotspot extraction. The multidimensional risk identification module 400 extracts risk features and inputs them into the risk assessment model based on the enterprise profile, public opinion analysis results and enterprise knowledge base. It obtains a risk score and determines the risk level through reasoning, triggers an early warning and matches a response strategy according to the risk level, and forms a risk result. The feedback update module 500 feeds back the risk results to the user and obtains user feedback. Based on the user feedback, it performs reverse transmission and dynamic updates and optimizations.

[0032] Various changes and modifications made without departing from the spirit and scope of this invention, and all equivalent technical solutions, also fall within the scope of this invention.

[0033] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0034] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0035] This invention is described with reference to flowchart illustrations and / or block diagrams of the method, terminal device (system), and computer program product according to the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0036] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing terminal device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0037] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal equipment, causing a series of operational steps to be performed on the computer or other programmable terminal equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable terminal equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0038] It should be noted that: The phrase "an embodiment" or "an embodiment" used in this specification means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the invention. Therefore, the phrase "an embodiment" or "an embodiment" appearing in various places throughout the specification does not necessarily refer to the same embodiment.

[0039] The above description of the embodiments is provided to enable those skilled in the art to understand and apply the present invention. It will be apparent to those skilled in the art that various modifications can be made to the above embodiments, and the general principles described herein can be applied to other embodiments without inventive effort. Therefore, the present invention is not limited to the above embodiments, and any improvements and modifications made to the present invention by those skilled in the art based on the disclosure thereof should be within the scope of protection of the present invention.

Claims

1. A multi-dimensional enterprise risk analysis method based on artificial intelligence, characterized in that, Includes the following steps: Collect multi-dimensional raw data from enterprises, generate structured enterprise profiles through preprocessing and feature extraction, and establish a dynamic update mechanism to update the enterprise profiles. Collect knowledge information from multiple sources and build a structured enterprise knowledge base through integration and structuring. Collect relevant public opinion information about enterprises from multiple channels, and generate public opinion analysis results after text preprocessing, sentiment recognition and hot topic extraction; Based on the enterprise profile, public opinion analysis results, and enterprise knowledge base, risk characteristics are extracted and input into the risk assessment model. Risk scores are obtained through reasoning, and risk levels are determined. Early warnings are triggered based on risk levels, and corresponding response strategies are matched to form risk results. The risk results are fed back to the users and user feedback is obtained. Based on the user feedback, the results are transmitted back and dynamically updated and optimized.

2. The multi-dimensional enterprise risk analysis method based on artificial intelligence according to claim 1, characterized in that, The enterprise profile is obtained through the following steps: The method combines web crawling and API calls to collect multi-dimensional raw data from enterprises. This multi-dimensional raw data includes basic enterprise information, operating data, industry position data, and credit-related data. The original data is cleaned, standardized, and structured to obtain preprocessed data. The structured processing includes filling in missing values, deleting duplicate values, and removing invalid data. Feature extraction is performed on the preprocessed data to obtain feature data, which includes basic features, business features, industry features and credit features. The feature data is weighted and assigned, and a comprehensive score for each enterprise is calculated and integrated to form a structured enterprise profile. It adopts a combination of triggered updates and periodic updates to retain historical enterprise profile versions and support version rollback.

3. The multi-dimensional enterprise risk analysis method based on artificial intelligence according to claim 1, characterized in that, The construction of the enterprise knowledge base includes the following steps: Collect knowledge information from multiple sources, including internal historical data, industry reports, policies and regulations, risk response cases, and industry trends. Data collection methods include data import, web scraping, API calls, and manual entry. Perform deduplication, verification, and removal of redundant, invalid, and conflicting data from multi-source knowledge information; The integrated knowledge information is classified and tagged to build a hierarchical knowledge system tree. We regularly update the knowledge base content by receiving external feedback data, and optimize the knowledge base retrieval and accurate recommendation logic.

4. The multi-dimensional enterprise risk analysis method based on artificial intelligence according to claim 1, characterized in that, The process of collecting relevant public opinion information about enterprises from multiple channels, and generating public opinion analysis results after text preprocessing, sentiment recognition, and hot topic extraction includes the following steps: Collect relevant public opinion information about enterprises from multiple channels. The information collection methods include web scraping technology and API interface calls. The multiple channels include news media platforms, social platforms and industry forums and communities. The public opinion information is segmented, stop words are removed, and normalization is performed to remove irrelevant characters and meaningless words, and synonyms and near-synonyms are standardized to obtain the public opinion text. A pre-trained natural language processing model is used to perform sentiment recognition on public opinion texts, outputting sentiment tendency and confidence level, and low confidence level results are manually reviewed; Using keyword extraction and hot topic filtering algorithms, the core keywords of public opinion texts are extracted and integrated to form hot topics related to enterprises and industries, and the trend of hot topic dissemination is tracked in real time. Based on sentiment trends and relevant hot topics related to the company and industry, public opinion analysis results are generated. These results include sentiment distribution, hot topic reports, and public opinion dissemination trends. The results are then used to update the company profile and knowledge base.

5. The multi-dimensional enterprise risk analysis method based on artificial intelligence according to claim 1, characterized in that, The risk outcome is obtained through the following steps: Based on enterprise profiles, public opinion analysis results, and enterprise knowledge base, feature extraction is performed to obtain risk feature vectors, which include operational risk features, financial risk features, legal risk features, and market risk features. Collect historical risk data and corresponding historical risk feature vectors of enterprises, and divide them into training set, validation set and test set after manual annotation. Use machine learning model combination for training and optimization. Input the risk feature vector into the trained risk assessment model to calculate the enterprise risk score and determine the risk level according to preset rules; A corresponding early warning level is set according to the risk level. An early warning mechanism is triggered based on the early warning level, and targeted response strategies are matched with risk response cases in the knowledge base, thereby forming a risk result. The risk result includes risk score, risk level, early warning information and response strategy.

6. The multi-dimensional enterprise risk analysis method based on artificial intelligence according to claim 1, characterized in that, It also includes real-time interaction based on AI assistants, specifically: The system receives user query requests through an AI assistant, which include enterprise profiling, public opinion analysis, risk identification, and enterprise knowledge base. The query requests are formatted, segmented, and stop words are removed. Keywords are extracted and combined with algorithms to clarify the user's query intent. Based on the query intent, the corresponding module is invoked to output data, which is then organized into structured response content; The system provides structured response content to users, supporting viewing, previewing, and report downloading, and offers multi-round interactive functionality. Collect user query requests, operation feedback, and response evaluations, and transmit them synchronously to other modules to drive optimization.

7. A multi-dimensional enterprise risk analysis system based on artificial intelligence, characterized in that, It includes modules for building enterprise profiles, building enterprise knowledge bases, analyzing public opinion, identifying multi-dimensional risks, and providing feedback and updates. The enterprise profile building module collects multi-dimensional raw data of enterprises, generates a structured enterprise profile through preprocessing and feature extraction, and establishes a dynamic update mechanism to update the enterprise profile. The enterprise knowledge base construction module collects knowledge information from multiple sources and constructs a structured enterprise knowledge base through integration and structuring. The public opinion analysis module collects relevant public opinion information about enterprises from multiple channels, and generates public opinion analysis results after text preprocessing, sentiment recognition and hot topic extraction. The multi-dimensional risk identification module extracts risk features based on the enterprise profile, public opinion analysis results, and enterprise knowledge base, inputs them into the risk assessment model, obtains a risk score through reasoning and determines the risk level, triggers an early warning based on the risk level and matches a response strategy to form a risk result. The feedback update module sends the risk results back to the user and obtains user feedback. Based on the user feedback, it performs reverse transmission and dynamic updates and optimizations.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 6.

9. A multi-dimensional enterprise risk analysis device based on artificial intelligence, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1 to 6.

Images