Financial risk control system based on block chain and public opinion

Abstract

The invention relates to the technical field of financial risk control, and discloses a financial risk control system based on a block chain and public opinions, and the system comprises a public opinion data collection module which is used for collecting financial-related public opinion data from multiple channels; the public opinion analysis and processing module is used for performing sentiment analysis, keyword extraction and trend prediction on the collected public opinion data; the block chain storage module is used for storing public opinion analysis results and risk related data, and is responsible for the non-tampering and transparency of the data; the intelligent contract execution module is used for dynamically triggering risk control measures according to the public opinion analysis result; the risk assessment and credit scoring module is used for performing dynamic risk assessment on the enterprise and generating a credit score; and the risk early warning and control module is used for generating risk early warning information and automatically executing a risk control strategy. According to the invention, through the distributed account book and the consensus mechanism of the block chain, the system ensures that the whole-process records of the public opinion analysis result, the risk score and the risk control strategy cannot be tampered.

Description

Technical Field

[0001] The present invention relates to the technical field of financial risk control, and in particular to a financial risk control system based on blockchain and public opinion. Background Art

[0002] With the digitalization and globalization of the financial industry, the complexity and dynamism of risk factors continue to increase. Various events in the market (such as corporate bankruptcy, policy changes, sudden public opinion, etc.) can quickly affect the credit status of enterprises and the stability of financial markets. Therefore, financial institutions urgently need a risk control system that can dynamically respond to changes in the external environment to capture potential risks in real time and take effective control measures. At the same time, with the rise of blockchain technology, its advantages in data credibility, immutability and distributed collaboration have brought new technical opportunities to the field of financial risk control.

[0003] However, the existing financial risk control system mainly relies on static information such as the company's financial data, transaction records and manual credit ratings. This single data source has great limitations. First, the credibility of traditional data is low and it is susceptible to human tampering or opaque operations, which makes it difficult for financial institutions to fully trust the data source in the risk control process. Second, the lack of data traceability makes it difficult to track historical risk control decisions and review processes, which is particularly prominent in financial supervision and audit scenarios. Summary of the invention

[0004] In order to make up for the above shortcomings, the present invention provides a financial risk control system based on blockchain and public opinion, which aims to improve the problem that traditional data has low credibility and is susceptible to human tampering or opaque operations, making it difficult for financial institutions to fully trust the source of data during the risk control process.

[0005] In a first aspect, the present invention provides the following technical solution: a financial risk control system based on blockchain and public opinion, comprising:

[0006] The public opinion data collection module is used to collect financial-related public opinion data from multiple channels and pre-process and store the data;

[0007] Public opinion analysis and processing module, used to perform sentiment analysis, keyword extraction and trend prediction on the collected public opinion data;

[0008] The blockchain storage module is used to store public opinion analysis results and risk-related data, and is responsible for the immutability and transparency of the data;

[0009] Smart contract execution module, used to dynamically trigger risk control measures based on public opinion analysis results;

[0010] The risk assessment and credit scoring module is used to dynamically assess the risk of enterprises and generate credit scores;

[0011] The risk warning and control module is used to generate risk warning information and automatically execute risk control strategies.

[0012] Preferably, the public opinion data collection module includes:

[0013] Data source management unit: used to configure the scope and target of public opinion collection, intelligently crawl social media and news website data through the Scrapy framework, and obtain real-time public opinion in combination with the API interface; use BlockSci to parse blockchain transaction data;

[0014] Data preprocessing unit: used to clean, denoise and normalize raw data, using Pandas and Spark to perform cleaning, word segmentation, stop word removal and language standardization on public opinion data;

[0015] Data classification and storage unit: used to classify and store public opinion data based on topics and sentiments, classify data by topic through the Transformer model, and use Elasticsearch to store classification results to support full-text retrieval.

[0016] Preferably, the public opinion analysis and processing module includes:

[0017] Sentiment analysis unit: used to evaluate the positive, negative or neutral sentiment of public opinion texts, train sentiment classification models based on the deep learning framework PyTorch, and use the fine-tuning capabilities of the BERT model for sentiment classification;

[0018] Risk keyword extraction unit: used to extract high-frequency keywords related to risks from text, calculate keyword weights using the TF-IDF algorithm, and extract core phrases using the TextRank algorithm;

[0019] Public opinion trend prediction unit: used to predict the future trend of public opinion, use the LSTM model to process time series data, and combine the ARIMA model to optimize the prediction results.

[0020] Preferably, the blockchain storage module includes:

[0021] Data on-chain unit: used to store the public opinion analysis results on the blockchain. Using the Hyperledger Fabric permission chain, the analysis results are serialized into JSON format through smart contracts and then compressed and stored on the blockchain.

[0022] Data verification and consensus unit: used to verify the authenticity of public opinion data and risk scores, using the SHA-256 algorithm to generate data hash values, and using the PBFT consensus algorithm to ensure verification consistency among distributed nodes;

[0023] Data access management unit: used to control access rights for users of different roles, perform identity authentication based on decentralized identity authentication DID technology, and define permission policies in combination with RBAC.

[0024] Preferably, the smart contract execution module includes:

[0025] Contract trigger unit: used to dynamically trigger risk control contracts, monitor the real-time changes of public opinion analysis results through Web3.js, and trigger smart contracts according to set thresholds;

[0026] Risk control strategy execution unit: used to execute operations such as freezing funds and adjusting credit scores, calling the enterprise account interface based on smart contracts, and using oracle technology to receive external data sources to update risk scores;

[0027] Audit and log unit: used to record the execution process of smart contracts, record contract operations through the built-in log module of the blockchain, and support traceability query in combination with the blockchain browser.

[0028] Preferably, the risk assessment and credit scoring module includes:

[0029] Data fusion unit: used to integrate public opinion data and historical transaction records based on the Kafka stream processing engine for real-time data integration, and use MongoDB for distributed storage;

[0030] Dynamic scoring model unit: used to generate real-time credit scores, perform feature engineering on fused data through the XGBoost model, and optimize the scoring results in combination with public opinion analysis results;

[0031] Risk level classification unit: used to dynamically adjust the enterprise risk level according to the score, set the scoring thresholds for high, medium and low risks based on the quantile method, and dynamically update the weights of the scoring model.

[0032] Preferably, the risk warning and control module includes:

[0033] Real-time warning unit: used to generate warnings when risk scores fluctuate, detect abnormal fluctuations using the IsolationForest model, and push low-latency messages through RabbitMQ;

[0034] Automation control unit: used to automatically execute risk control strategies according to risk levels, and batch execute transaction restrictions and fund freezing tasks through Python scripts;

[0035] Visualization display unit: used to intuitively display risk status and control effects, build dynamic dashboards based on Grafana, and display risk scores, enterprise risk distribution, and the implementation of control measures in real time.

[0036] In a second aspect, the present invention provides the following technical solution, a financial risk control method based on blockchain and public opinion, comprising:

[0037] S1. Collect public opinion data in real time from social media, news websites and blockchain transaction record channels, obtain unstructured text data through intelligent crawlers and API interfaces, and clean, denoise and normalize them. Use natural language processing technology to classify the text content by topic and sentiment, and finally store the structured public opinion data in a distributed database;

[0038] S2. Conduct in-depth analysis of the collected public opinion data, use sentiment analysis models to evaluate the sentiment attributes of the text, identify positive, negative or neutral sentiment, extract high-frequency keywords related to financial risks through keyword extraction algorithms, and analyze future changes in public opinion trends in combination with time series prediction models;

[0039] S3. Write the public opinion analysis results and related data into the blockchain through smart contracts, use data compression and hash verification technology to ensure storage efficiency and security, verify the authenticity and integrity of the data through a distributed consensus mechanism, and generate tamper-proof evidence records in the blockchain ledger;

[0040] S4. Based on the dynamic changes of public opinion analysis, the smart contract with preset threshold conditions is triggered by real-time monitoring data. When it is detected that the enterprise risk score or public opinion index reaches the risk trigger condition, the smart contract automatically executes the corresponding risk control operation, including freezing funds, adjusting financing quotas or restricting trading permissions;

[0041] S5. Integrate the public opinion analysis results stored in the blockchain with the company's historical credit data and transaction records, generate a dynamic credit score for the company through the trained machine learning model, and classify the company into high risk, medium risk or low risk levels based on the score results;

[0042] S6. Generate real-time risk warnings for score changes and public opinion fluctuations, and execute corresponding risk control strategies through automated scripts, including adjusting the company's credit rating or suspending high-risk transactions. At the same time, use visualization tools to display risk assessment results and the execution of control strategies to support decision makers' further actions.

[0043] In the third aspect, the invention provides the following technical solution: a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor; when the processor executes the computer program, the financial risk control method based on blockchain and public opinion is implemented.

[0044] In a fourth aspect, the present invention provides the following technical solution: a readable storage medium having a computer program stored thereon, and the computer program, when executed by a processor, implements the above-mentioned financial risk control method based on blockchain and public opinion.

[0045] The present invention has the following beneficial effects:

[0046] 1. In the present invention, through the distributed ledger and consensus mechanism of blockchain, the system ensures that the whole process records of public opinion analysis results, risk scores and risk control strategies cannot be tampered with. This feature greatly improves the data credibility of financial institutions in risk assessment and decision-making, and provides a transparent and reliable risk management tool for regulators and auditing agencies.

[0047] 2. In the present invention, by combining public opinion analysis with smart contracts, the system can monitor changes in market public opinion in real time and automatically trigger corresponding risk control strategies. For example, when the public opinion fluctuation of a certain enterprise reaches a preset threshold, the smart contract will immediately execute transaction restrictions or credit score adjustments. This real-time and automated response not only improves risk control efficiency, but also reduces delays and errors caused by manual intervention.

[0048] 3. In the present invention, the system establishes a multi-dimensional risk assessment framework by integrating public opinion data, historical credit records and blockchain transaction data. In particular, in public opinion analysis, the system uses sentiment analysis, keyword extraction and trend prediction technology to provide a more comprehensive data basis for risk assessment. This multi-dimensional assessment method significantly improves the accuracy and applicability of risk scoring.

[0049] 4. In the present invention, the dynamic credit scoring model of the system can adjust the scoring results in real time according to changes in public opinion and corporate behavior. This real-time optimization capability enables financial institutions to promptly detect potential credit risks and take preventive measures. In addition, the interpretable design of the scoring model (such as through SHAP value analysis) also helps decision makers to deeply understand the risk factors behind the scoring and optimize risk control decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 This is a system architecture diagram of a financial risk control system based on blockchain and public opinion proposed by the present invention;

[0051] Figure 2 This is a diagram of the architecture of the public opinion data collection module of a financial risk control system based on blockchain and public opinion proposed by the present invention;

[0052] Figure 3 This is a diagram of the architecture of the public opinion analysis and processing module of a financial risk control system based on blockchain and public opinion proposed by the present invention;

[0053] Figure 4This is a diagram of the blockchain storage module architecture of a financial risk control system based on blockchain and public opinion proposed by the present invention;

[0054] Figure 5 This is a diagram of the architecture of a smart contract execution module of a financial risk control system based on blockchain and public opinion proposed by the present invention;

[0055] Figure 6 This is a risk assessment and credit scoring module architecture diagram of a financial risk control system based on blockchain and public opinion proposed by the present invention;

[0056] Figure 7 This is a risk warning and control module architecture diagram of a financial risk control system based on blockchain and public opinion proposed by the present invention;

[0057] Figure 8 This is a method flow chart of a financial risk control method based on blockchain and public opinion proposed by the present invention. DETAILED DESCRIPTION

[0058] The technical solutions in the embodiments of the present invention will be described clearly and completely below 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 field without creative work are within the scope of protection of the present invention.

[0059] Embodiment 1

[0060] Reference Figure 1-Figure 7 In the first embodiment of the present invention, the present invention provides a financial risk control system based on blockchain and public opinion, including:

[0061] The public opinion data collection module is used to collect financial-related public opinion data from multiple channels and pre-process and store the data;

[0062] Public opinion analysis and processing module, used to perform sentiment analysis, keyword extraction and trend prediction on the collected public opinion data;

[0063] The blockchain storage module is used to store public opinion analysis results and risk-related data, and is responsible for the immutability and transparency of the data;

[0064] Smart contract execution module, used to dynamically trigger risk control measures based on public opinion analysis results;

[0065] The risk assessment and credit scoring module is used to dynamically assess the risk of enterprises and generate credit scores;

[0066] The risk warning and control module is used to generate risk warning information and automatically execute risk control strategies.

[0067] Specifically, this system ensures that the whole process of public opinion data, risk analysis results and risk control strategies are transparent and cannot be tampered with through the distributed ledger and consensus mechanism of blockchain, providing technical guarantee for the credibility of data and the reliability of risk control decisions. At the same time, the system combines public opinion analysis with dynamic credit scoring, and introduces a multi-dimensional data foundation for risk assessment through technical means such as sentiment analysis, keyword extraction and trend prediction. With the real-time response capability of smart contracts, the system can automatically trigger risk control strategies such as transaction restrictions and account freezes to quickly respond to changes in market public opinion. Finally, by dynamically adjusting the accuracy of credit scores and multi-dimensional risk analysis, the system not only optimizes the risk control capabilities of financial institutions, but also provides decision makers with comprehensive and real-time risk information support.

[0068] The public opinion data collection module includes:

[0069] Data source management unit: used to configure the scope and target of public opinion collection, intelligently crawl social media and news website data through the Scrapy framework, and obtain real-time public opinion in combination with the API interface; use BlockSci to parse blockchain transaction data;

[0070] Data preprocessing unit: used to clean, denoise and normalize raw data, using Pandas and Spark to perform cleaning, word segmentation, stop word removal and language standardization on public opinion data;

[0071] Data classification and storage unit: used to classify and store public opinion data based on topics and sentiments, classify data by topic through the Transformer model, and use Elasticsearch to store classification results to support full-text retrieval.

[0072] Specifically, the main goal of the public opinion data collection module is to efficiently and in real time obtain public opinion data related to financial risks from multiple sources, and to clean and preprocess the data to provide high-quality input data for subsequent analysis. Considering the diversity and complexity of data sources, the module has designed a flexible multi-channel collection mechanism and combined it with an efficient preprocessing process.

[0073] Multi-channel data collection strategy

[0074] The system supports obtaining data from social media (such as Twitter, Weibo), news aggregation websites (such as Google News, Yahoo Finance), forums (such as Reddit), and blockchain public transaction records. For social media and news websites, the system uses customized intelligent crawlers to achieve precise collection, and combines API interfaces to obtain real-time updated data. For example, when collecting public opinion related to a financial event, the system will give priority to crawling posts containing specific keywords (such as "bankruptcy" and "layoffs"). In addition, in order to obtain transaction data on the blockchain, the system integrates the BlockSci parsing tool, which can read the transaction behavior of specific companies or accounts in real time.

[0075] Combining real-time and batch

[0076] In order to balance real-time collection and batch processing, the system builds a task scheduling architecture based on Kafka message queues. Real-time data is quickly pushed to the queue and processed, while batch data is stored in a temporary cache and processed regularly. This design ensures that the system can efficiently respond to sudden public opinion events while reducing the consumption of system resources.

[0077] Data cleaning and structuring

[0078] In the data preprocessing stage, the system first performs language standardization operations on the original text, including word segmentation, stop word removal, and spelling correction. For multilingual public opinion data, the system integrates SpaCy and Jieba tools to support mixed processing of Chinese and English. In addition, in order to improve the accuracy of financial industry data, the system is configured with a customized financial field dictionary to accurately annotate terms (such as "IPO" and "credit rating"). At the same time, non-text data (such as images and PDF files) will be extracted through Tesseract-OCR technology to ensure the comprehensiveness of public opinion information.

[0079] The public opinion analysis and processing module includes:

[0080] Sentiment analysis unit: used to evaluate the positive, negative or neutral sentiment of public opinion texts, train sentiment classification models based on the deep learning framework PyTorch, and use the fine-tuning capabilities of the BERT model for sentiment classification;

[0081] Risk keyword extraction unit: used to extract high-frequency keywords related to risks from text, calculate keyword weights using the TF-IDF algorithm, and extract core phrases using the TextRank algorithm;

[0082] Public opinion trend prediction unit: used to predict the future trend of public opinion, use the LSTM model to process time series data, and combine the ARIMA model to optimize the prediction results.

[0083] Specifically, the task of the public opinion analysis and processing module is to transform the pre-processed public opinion data into analysis results with decision-making value. This module uses a variety of deep learning and statistical analysis techniques to comprehensively mine the sentiment, semantics and trends of public opinion.

[0084] High-precision optimization for sentiment analysis

[0085] The system performs sentiment analysis on public opinion texts based on the BERT pre-trained model and fine-tunes it in combination with a specific corpus in the financial field to ensure that the model can identify subtle sentiment changes in industry-specific contexts. For example, a "profit warning" may not be obvious in a normal context, but it may mean a significant risk in a financial scenario. In addition, in order to enhance the result dimension of sentiment analysis, the system has designed indicators such as "trust" and "risk index" to comprehensively evaluate the potential impact of public opinion on enterprises.

[0086] Keyword and risk factor extraction

[0087] The system uses TF-IDF to calculate high-frequency words in the text and combines it with the TextRank algorithm to identify core keywords. These keywords are further mapped to predefined risk categories, for example, "default" and "layoffs" are marked as trigger words for high-risk events. In order to handle complex semantics, the system also uses a graph network method based on a co-occurrence matrix to analyze the path of public opinion propagation between companies and the chain reactions it may cause.

[0088] Trend prediction and anomaly capture

[0089] The system uses the LSTM model to predict the future trend of public opinion changes, and uses the ARIMA model to capture the linear and seasonal characteristics in the time series. In addition, for emergencies, the system has designed a dynamic monitoring algorithm based on a sliding window, which automatically triggers an abnormal warning when the rate of change of public opinion exceeds a certain threshold of the historical mean.

[0090] The blockchain storage module includes:

[0091] Data on-chain unit: used to store the public opinion analysis results on the blockchain. Using the Hyperledger Fabric permission chain, the analysis results are serialized into JSON format through smart contracts and then compressed and stored on the blockchain.

[0092] Data verification and consensus unit: used to verify the authenticity of public opinion data and risk scores, using the SHA-256 algorithm to generate data hash values, and using the PBFT consensus algorithm to ensure verification consistency among distributed nodes;

[0093] Data access management unit: used to control access rights for users of different roles, perform identity authentication based on decentralized identity authentication DID technology, and define permission policies in combination with RBAC.

[0094] Specifically, the blockchain storage module aims to achieve transparent storage, traceability and non-tamperability of public opinion data and analysis results through distributed ledger technology. The module adopts a storage strategy that combines on-chain and off-chain to balance storage efficiency and security.

[0095] Distributed storage design

[0096] The system stores the analysis results (such as risk scores and public opinion sentiment data) in the blockchain in hash form, while the original public opinion data is stored in the IPFS off-chain. The on-chain data and the off-chain data correspond one-to-one through the hash value, ensuring that the authenticity of the off-chain data can be quickly verified when needed.

[0097] Efficient consensus mechanism

[0098] To improve storage and verification efficiency, the system selects the PBFT consensus algorithm and combines it with a weighted voting strategy to prioritize consensus among highly trusted nodes. For example, if a node has a higher transaction record credibility, its voting weight will be higher than other nodes, reducing the consumption of computing resources during the consensus process.

[0099] Dynamic permissions and privacy protection

[0100] The system manages data access permissions through decentralized identity (DID) technology to ensure that users with different roles can only access data that matches their permissions. For example, financial regulators can access the complete public opinion storage records, while ordinary users can only view summary information of risk scores.

[0101] The smart contract execution module includes:

[0102] Contract trigger unit: used to dynamically trigger risk control contracts, monitor the real-time changes of public opinion analysis results through Web3.js, and trigger smart contracts according to set thresholds;

[0103] Risk control strategy execution unit: used to execute operations such as freezing funds and adjusting credit scores, calling the enterprise account interface based on smart contracts, and using oracle technology to receive external data sources to update risk scores;

[0104] Audit and log unit: used to record the execution process of smart contracts, record contract operations through the built-in log module of the blockchain, and support traceability query in combination with the blockchain browser.

[0105] Specifically, the smart contract execution module achieves automated response to risk control through flexible rule configuration and trigger mechanism.

[0106] Flexibility in rule configuration

[0107] The triggering rules of the contract support multiple condition combinations, such as the public opinion sentiment value is lower than a certain threshold, the keyword hit rate exceeds a certain percentage, or the corporate credit score drops significantly, etc. When a rule is met, the contract will be triggered immediately and the response operation will be executed according to the predefined strategy.

[0108] External data source integration

[0109] In order to enhance the applicability of smart contracts, the system obtains off-chain data through Oracle, such as real-time market fluctuation information or international news events, and incorporates them into the trigger rules. In this way, the contract can quickly respond to changes in the external environment.

[0110] Full life cycle records

[0111] Each time a contract is triggered and executed, it will be recorded in the blockchain, including the triggering conditions, execution results, and external data sources called. This transparent design not only supports audit requirements, but also enhances the trust of contract execution.

[0112] The risk assessment and credit scoring modules include:

[0113] Data fusion unit: used to integrate public opinion data and historical transaction records based on the Kafka stream processing engine for real-time data integration, and use MongoDB for distributed storage;

[0114] Dynamic scoring model unit: used to generate real-time credit scores, perform feature engineering on fused data through the XGBoost model, and optimize the scoring results in combination with public opinion analysis results;

[0115] Risk level classification unit: used to dynamically adjust the enterprise risk level according to the score, set the scoring thresholds for high, medium and low risks based on the quantile method, and dynamically update the weights of the scoring model.

[0116] Specifically, this module provides enterprises with dynamic risk assessment and credit rating by integrating multi-source data and advanced machine learning algorithms.

[0117] Dynamic weight adjustment of risk factors

[0118] The system introduces a dynamic weighting mechanism into the credit scoring model to adjust the scoring weight according to the real-time changes in public opinion data. For example, when the negative public opinion of a company increases significantly, the weight of the public opinion index will be automatically adjusted upward to reflect its direct impact on credit risk.

[0119] Adaptive Optimization of Models

[0120] The credit scoring model is implemented based on the XGBoost algorithm, which continuously optimizes the model parameters through incremental learning technology so that it can adapt to the risk feature distribution in different time periods. At the same time, in order to avoid the model "overfitting" public opinion data, the system has designed a feature screening process to eliminate redundant variables that have a low contribution to the score.

[0121] Risk grading strategy

[0122] The system divides the credit score results of enterprises into high risk, medium risk and low risk levels, and judges the potential development direction of enterprises based on the historical trend of risk scores. For example, high-risk enterprises may trigger multi-level warnings under continuous negative public opinion.

[0123] The risk warning and control module includes:

[0124] Real-time warning unit: used to generate warnings when risk scores fluctuate, detect abnormal fluctuations using the IsolationForest model, and push low-latency messages through RabbitMQ;

[0125] Automation control unit: used to automatically execute risk control strategies according to risk levels, and batch execute transaction restrictions and fund freezing tasks through Python scripts;

[0126] Visualization display unit: used to intuitively display risk status and control effects, build dynamic dashboards based on Grafana, and display risk scores, enterprise risk distribution, and the implementation of control measures in real time.

[0127] Specifically, the risk warning and control module is responsible for converting analysis results into actual operations and improving risk control efficiency through real-time warning and automated control.

[0128] Multi-level response of early warning mechanism

[0129] The system has designed a three-level early warning system: low-level warnings indicate potential risks, intermediate warnings require manual review, and high-level warnings directly trigger automated control measures. For example, when the public opinion sentiment value of a company drops for more than two consecutive days and the keyword hit rate increases abnormally, the system will trigger a high-level warning.

[0130] Automated policy enforcement

[0131] The system automatically executes risk control measures through the Python script engine, including suspending high-risk account transactions, freezing specific funds, or adjusting credit limits. These measures can be dynamically adjusted according to different warning levels.

[0132] Visualization of risk results

[0133] The system uses Grafana to build an interactive dashboard to display the company's risk score change trends, public opinion propagation paths, and the implementation of risk control measures, supporting decision makers to quickly formulate response strategies based on specific data.

[0134] Embodiment 2:

[0135] Reference Figure 8 In a second embodiment of the present invention, the present invention provides a financial risk control method based on blockchain and public opinion, comprising:

[0136] S1. Collect public opinion data in real time from social media, news websites and blockchain transaction record channels, obtain unstructured text data through intelligent crawlers and API interfaces, and clean, denoise and normalize them. Use natural language processing technology to classify the text content by topic and sentiment, and finally store the structured public opinion data in a distributed database;

[0137] S2. Conduct in-depth analysis of the collected public opinion data, use sentiment analysis models to evaluate the sentiment attributes of the text, identify positive, negative or neutral sentiment, extract high-frequency keywords related to financial risks through keyword extraction algorithms, and analyze future changes in public opinion trends in combination with time series prediction models;

[0138] S3. Write the public opinion analysis results and related data into the blockchain through smart contracts, use data compression and hash verification technology to ensure storage efficiency and security, verify the authenticity and integrity of the data through a distributed consensus mechanism, and generate tamper-proof evidence records in the blockchain ledger;

[0139] S4. Based on the dynamic changes of public opinion analysis, the smart contract with preset threshold conditions is triggered by real-time monitoring data. When it is detected that the enterprise risk score or public opinion index reaches the risk trigger condition, the smart contract automatically executes the corresponding risk control operation, including freezing funds, adjusting financing quotas or restricting trading permissions;

[0140] S5. Integrate the public opinion analysis results stored in the blockchain with the company's historical credit data and transaction records, generate a dynamic credit score for the company through the trained machine learning model, and classify the company into high risk, medium risk or low risk levels based on the score results;

[0141] S6. Generate real-time risk warnings for score changes and public opinion fluctuations, and execute corresponding risk control strategies through automated scripts, including adjusting the company's credit rating or suspending high-risk transactions. At the same time, use visualization tools to display risk assessment results and the execution of control strategies to support decision makers' further actions.

[0142] Embodiment 3

[0143] The third embodiment of the present invention is based on the same inventive concept. The present invention proposes a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps of the financial risk control method based on blockchain and public opinion of the above embodiment.

[0144] Embodiment 4

[0145] The fourth embodiment of the present invention is based on the same inventive concept. A computer device proposed by the present invention comprises: a processor and a memory; the processor and the memory communicate with each other; the memory is used to store instructions; the processor is used to execute the instructions in the memory, and execute the financial risk control method based on blockchain and public opinion of the above embodiment.

[0146] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0147] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A financial risk control system based on blockchain and public opinion, characterized in that: include: The public opinion data collection module is used to collect financial-related public opinion data from multiple channels and pre-process and store the data; Public opinion analysis and processing module, used to perform sentiment analysis, keyword extraction and trend prediction on the collected public opinion data; The blockchain storage module is used to store public opinion analysis results and risk-related data, and is responsible for the immutability and transparency of the data; Smart contract execution module, used to dynamically trigger risk control measures based on public opinion analysis results; The risk assessment and credit scoring module is used to dynamically assess the risk of enterprises and generate credit scores; The risk warning and control module is used to generate risk warning information and automatically execute risk control strategies.

2. The financial risk control system based on blockchain and public opinion according to claim 1 is characterized in that: The public opinion data collection module includes: Data source management unit: used to configure the scope and target of public opinion collection, intelligently crawl social media and news website data through the Scrapy framework, and obtain real-time public opinion in combination with the API interface; use BlockSci to parse blockchain transaction data; Data preprocessing unit: used to clean, denoise and normalize raw data, using Pandas and Spark to perform cleaning, word segmentation, stop word removal and language standardization on public opinion data; Data classification and storage unit: used to classify and store public opinion data based on topics and sentiments, classify data by topic through the Transformer model, and use Elasticsearch to store classification results to support full-text retrieval.

3. The financial risk control system based on blockchain and public opinion according to claim 1 is characterized in that: The public opinion analysis and processing module includes: Sentiment analysis unit: used to evaluate the positive, negative or neutral sentiment of public opinion texts, train sentiment classification models based on the deep learning framework PyTorch, and use the fine-tuning capabilities of the BERT model for sentiment classification; Risk keyword extraction unit: used to extract high-frequency keywords related to risks from text, calculate keyword weights using the TF-IDF algorithm, and extract core phrases using the TextRank algorithm; Public opinion trend prediction unit: used to predict the future trend of public opinion, use the LSTM model to process time series data, and combine the ARIMA model to optimize the prediction results.

4. The financial risk control system based on blockchain and public opinion according to claim 1 is characterized in that: The blockchain storage module includes: Data on-chain unit: used to store the public opinion analysis results on the blockchain. Using the Hyperledger Fabric permission chain, the analysis results are serialized into JSON format through smart contracts and then compressed and stored on the blockchain. Data verification and consensus unit: used to verify the authenticity of public opinion data and risk scores, using the SHA-256 algorithm to generate data hash values, and using the PBFT consensus algorithm to ensure verification consistency among distributed nodes; Data access management unit: used to control access rights for users of different roles, perform identity authentication based on decentralized identity authentication DID technology, and define permission policies in combination with RBAC.

5. The financial risk control system based on blockchain and public opinion according to claim 1 is characterized in that: The smart contract execution module includes: Contract trigger unit: used to dynamically trigger risk control contracts, monitor the real-time changes of public opinion analysis results through Web3.js, and trigger smart contracts according to set thresholds; Risk control strategy execution unit: used to execute operations such as freezing funds and adjusting credit scores, calling the enterprise account interface based on smart contracts, and using oracle technology to receive external data sources to update risk scores; Audit and log unit: used to record the execution process of smart contracts, record contract operations through the built-in log module of the blockchain, and support traceability query in combination with the blockchain browser.

6. The financial risk control system based on blockchain and public opinion according to claim 1 is characterized in that: The risk assessment and credit scoring module includes: Data fusion unit: used to integrate public opinion data and historical transaction records based on the Kafka stream processing engine for real-time data integration, and use MongoDB for distributed storage; Dynamic scoring model unit: used to generate real-time credit scores, perform feature engineering on fused data through the XGBoost model, and optimize the scoring results in combination with public opinion analysis results; Risk level classification unit: used to dynamically adjust the enterprise risk level according to the score, set the scoring thresholds for high, medium and low risks based on the quantile method, and dynamically update the weights of the scoring model.

7. The financial risk control system based on blockchain and public opinion according to claim 1 is characterized in that: The risk warning and control module includes: Real-time warning unit: used to generate warnings when risk scores fluctuate, detect abnormal fluctuations using the IsolationForest model, and push low-latency messages through RabbitMQ; Automation control unit: used to automatically execute risk control strategies according to risk levels, and batch execute transaction restrictions and fund freezing tasks through Python scripts; Visualization display unit: used to intuitively display risk status and control effects, build dynamic dashboards based on Grafana, and display risk scores, enterprise risk distribution, and the implementation of control measures in real time.

8. A financial risk control method based on blockchain and public opinion, characterized in that: The financial risk control system based on blockchain and public opinion as described in any one of claims 1 to 7 comprises: S1. Collect public opinion data in real time from social media, news websites and blockchain transaction record channels, obtain unstructured text data through intelligent crawlers and API interfaces, and clean, denoise and normalize them. Use natural language processing technology to classify the text content by topic and sentiment, and finally store the structured public opinion data in a distributed database; S2. Conduct in-depth analysis of the collected public opinion data, use sentiment analysis models to evaluate the sentiment attributes of the text, identify positive, negative or neutral sentiment, extract high-frequency keywords related to financial risks through keyword extraction algorithms, and analyze future changes in public opinion trends in combination with time series prediction models; S3. Write the public opinion analysis results and related data into the blockchain through smart contracts, use data compression and hash verification technology to ensure storage efficiency and security, verify the authenticity and integrity of the data through a distributed consensus mechanism, and generate tamper-proof evidence records in the blockchain ledger; S4. Based on the dynamic changes of public opinion analysis, the smart contract with preset threshold conditions is triggered by real-time monitoring data. When it is detected that the enterprise risk score or public opinion index reaches the risk trigger condition, the smart contract automatically executes the corresponding risk control operation, including freezing funds, adjusting financing quotas or restricting trading permissions; S5. Integrate the public opinion analysis results stored in the blockchain with the company's historical credit data and transaction records, generate a dynamic credit score for the company through the trained machine learning model, and classify the company into high risk, medium risk or low risk levels based on the score results; S6. Generate real-time risk warnings for score changes and public opinion fluctuations, and execute corresponding risk control strategies through automated scripts, including adjusting the company's credit rating or suspending high-risk transactions. At the same time, use visualization tools to display risk assessment results and the execution of control strategies to support decision makers' further actions.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the financial risk control method based on blockchain and public opinion as described in claim 8 is implemented.

10. A readable storage medium, characterized in that: The readable storage medium stores a computer program, and when the computer program is executed by the processor, the financial risk control method based on blockchain and public opinion as claimed in claim 8 is implemented.

Images