Industrial and commercial information early warning method based on big data

Abstract

The invention relates to the technical field of data processing, in particular to an industrial and commercial information early warning method based on big data, and the method comprises the steps: capturing data; risk calculation; generating correction; evidence and freshness; calculating a correction index; simulating the influence; punishment and credibility; executing correction; performing high-risk judgment; the grabbing range is adjusted; and sending out early warning. According to the method, the enterprise judicial case newly-added number, the negative public opinion popularity, the industry and commerce change frequency, the registration capital change rate and the upstream and downstream enterprise abnormal number are dynamically collected, and standardization processing and weighted fusion are carried out on multi-source indexes to form a rolling risk index; and the risk index is continuously adjusted in combination with a correction instruction, an evidence speed value, freshness, an influence quantity and a penalty term, so that the problems of early warning lag and insufficient risk prevention and control caused by difficulty in timely and accurate identification of enterprise high-risk information due to data updating delay, single risk judgment and static capture strategy are effectively solved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method for early warning of industrial and commercial information based on big data. Background Technology

[0002] With the rapid development of information technology and big data, the business environment is becoming increasingly dynamic and complex. The large volume, frequent updates, and diverse sources of data present unprecedented challenges to the timely identification and assessment of business risks. Traditional static monitoring methods struggle to capture the evolving trends of risks and are ill-equipped to respond effectively in the early stages of risk accumulation. Therefore, how to quickly and accurately identify potentially high-risk entities from massive amounts of information has become a major challenge that urgently needs to be addressed in current regulatory and management work.

[0003] Chinese Patent Application Publication No. CN111179022A discloses a business data processing system and method. The method includes: S100, dividing and tracing the business data of an enterprise in a bidding process of an e-commerce platform system, and constructing a data quality problem domain and a business information tracking domain within a business credit big data system based on the information division results; S200, using a passive big data crawler system to perform structured crawling of the business data, forming temporary data and related data of the business database, and transforming the temporary data and the business information tracking domain using a heterogeneous data conversion tool; S300, placing the transformed temporary data and related data of the business database into the data quality problem domain and the business information tracking domain, respectively; S400, using expert rules and data anomaly analysis algorithms in the business credit big data system to perform data analysis on the temporary data of the business database, and feeding back the data analysis results and the related data of the business database in the business information tracking domain to the e-commerce platform system.

[0004] Therefore, the aforementioned business data processing system and method have the following problems: the method relies on passive big data crawlers for structured crawling, and there is a delay in the crawling and transformation of enterprise business data, which cannot quickly reflect the dynamic changes in the enterprise's operating status; the method mainly relies on expert rules and data anomaly analysis algorithms to analyze the data, and lacks mechanisms for multi-parameter weighting, simulation and credibility correction, making it difficult to form a comprehensive risk index and the risk judgment is relatively simple. Summary of the Invention

[0005] To address this, the present invention provides a business information early warning method based on big data. This method overcomes the problems in existing technologies, such as the difficulty in timely and accurate identification of high-risk enterprise information due to data update delays, single risk judgments, and static capture strategies, which lead to delayed early warnings and insufficient risk control. This is achieved through multi-dimensional indicator collection, rolling risk index calculation, and dynamic adjustment of correction instructions.

[0006] To achieve the above objectives, this invention provides a business information early warning method based on big data, comprising: The system retrieves data from a large database, including the number of new judicial cases, the level of negative public opinion, the frequency of business registration changes, the rate of change in registered capital, and the number of abnormal upstream and downstream enterprises for target companies, based on a preset retrieval period and a preset retrieval scope. A rolling risk index is determined based on the number of new judicial cases, the intensity of negative public opinion, and the frequency of business registration changes. A correction instruction is generated based on the registered capital change rate and the number of abnormal upstream and downstream enterprises. Simultaneously, an evidence overview value is determined based on the registered capital change rate and the number of abnormal upstream and downstream enterprises. The data freshness is determined based on the time when the correction instruction is generated. The correction index is calculated based on the evidence overview value and the data freshness, and the correction instruction is simulated according to the correction index and the preset correction threshold to obtain the simulation value. The impact amount is determined based on the simulation value and the rolling risk index, and the penalty item is determined by combining the preset sensitivity factor and the impact amount; The correction credibility is determined based on the evidence overview value, the data freshness, the impact, and the penalty item, and the correction instruction is executed based on the correction credibility to redetermine the rolling risk index, and the main contributing factors in the redetermining process of the rolling risk index are recorded simultaneously. The high-risk level is determined based on the redefined rolling risk index and the preset risk threshold; The preset capture range for the next round is adjusted according to the type of the main contributing factor; A warning is issued for high-risk levels that are redefined after adjusting the preset capture range in consecutive preset rounds.

[0007] Furthermore, the process of determining the rolling risk index based on the number of new judicial cases, the intensity of negative public opinion, and the frequency of business registration changes includes: The number of new judicial cases, the level of negative public opinion, and the frequency of business registration changes are standardized to obtain the corresponding normalized index values. The normalized index values ​​are weighted and summed according to the preset index weighting to obtain the basic risk index; The rolling risk index is obtained by combining the basic risk index with the rolling risk index of the previous preset capture period according to a preset decay coefficient.

[0008] Furthermore, when determining the rolling risk index for the first time, the basic risk index of the current preset capture period is used as the rolling risk index.

[0009] Furthermore, the process of generating correction instructions based on the registered capital change rate and the number of abnormal upstream and downstream enterprises includes: Calculate the difference between the registered capital change rate in the current preset capture period and the difference between the number of abnormal upstream and downstream enterprises in the previous preset capture period; When the absolute value of the difference in the registered capital change rate is greater than or equal to the preset capital change rate threshold, the correction instruction to adjust the preset judicial new weight in the preset indicator weight restructuring is generated. When the absolute value of the difference between the number of abnormal enterprises in the upstream and downstream enterprises is greater than or equal to the preset abnormal difference threshold, the correction instruction for adjusting the preset industrial and commercial change weight in the preset indicator weight reorganization is generated.

[0010] Furthermore, the process of simultaneously determining the evidence overview value based on the registered capital change rate and the number of anomalies in upstream and downstream enterprises includes: The evidence summary value is determined based on the absolute value of the difference in the registered capital change rate, the preset first standard registration evidence value, and the preset second standard registration evidence value; The evidence overview value is determined based on the absolute value of the difference between the number of anomalies of the upstream and downstream enterprises, the preset first standard outlier evidence value, and the preset second standard outlier evidence value.

[0011] Furthermore, the process of determining data freshness based on the time of generation of the correction instruction includes: The data freshness is determined based on the time interval between the current correction instruction generation time and the previous preset capture period, and the preset decay coefficient.

[0012] Furthermore, the process of determining the impact amount based on the simulated value and the rolling risk index, and determining the penalty term in conjunction with the preset sensitivity factor and the impact amount, includes: The difference between the simulated value and the rolling risk index is calculated to obtain the influence amount; The penalty term is determined based on the preset sensitivity factor, the amount of influence, the preset small influence threshold, and the preset scale factor.

[0013] Furthermore, the process of determining the credibility of the correction based on the evidence overview value, the data freshness, the impact, and the penalty item includes: The corrected credibility is obtained by weighting and summing the evidence overview value, the data freshness, the impact, the penalty item, and the preset corrected credibility weight.

[0014] Furthermore, the process of determining the high-risk level based on the redefined rolling risk index and the preset risk threshold includes: When the rolling risk index is greater than the preset risk threshold, the target enterprise is determined to be in a high-risk state, thereby determining the high-risk level.

[0015] Furthermore, the process of adjusting the preset crawling range for the next round based on the type of the main contributing factor includes: When the type of the main contributing factor is newly added judicial cases, the preset capture range of the number of newly added judicial cases is increased according to the preset judicial capture coefficient; When the type of the main contributing factor is business registration change, the preset capture range of the business registration change frequency is increased according to the preset capture coefficient.

[0016] Compared with existing technologies, the beneficial effects of this invention lie in its ability to dynamically collect data on the number of new corporate judicial cases, the intensity of negative public opinion, the frequency of business registration changes, the rate of change in registered capital, and the number of abnormal upstream and downstream enterprises. This multi-source indicator is then standardized and weighted to form a rolling risk index. This index is continuously adjusted based on correction instructions, evidence overview values, freshness, impact, and penalties, achieving a real-time depiction of the enterprise's risk status. Through the dynamic mapping relationship between indicators, changes in registered capital and abnormal upstream and downstream enterprises can reasonably influence the weight adjustments of judicial and business registration information. The difference between the simulated value and the rolling risk index guides sensitivity adjustment, ultimately generating a correction credibility to drive weight correction. This method can adaptively adjust the capture scope, strengthen the monitoring of major risk-contributing factors, and enable high-risk enterprises to be identified and warned in a timely manner. This achieves continuity in risk assessment, sensitivity in response, and maximizes data utilization, improving the accuracy and reliability of business registration information warnings. It effectively solves the problems of delayed and inadequate risk control caused by data update delays, singular risk judgments, and static capture strategies, which make it difficult to identify high-risk enterprise information in a timely and accurate manner.

[0017] Furthermore, by standardizing the number of new judicial cases, the intensity of negative public opinion, and the frequency of business registration changes, and then weighting and summing them according to preset weights to obtain a basic risk index, and combining it with the rolling risk index from the previous capture period using a decay coefficient, continuous tracking of dynamic changes in enterprise risk can be achieved. The design of weights and decay coefficients among the parameters reflects the interrelationships of risk indicators: the weight of new judicial cases reflects the direct contribution of legal events to overall risk; the weight of public opinion intensity reflects the role of social opinion in the perception of enterprise risk; the weight of business registration changes reflects the impact of internal operational changes on risk; and the decay coefficient ensures the reasonable transmission of historical information in rolling risk calculations. Through this comprehensive processing method, the system can smooth short-term fluctuations, highlight important risk factors, and achieve continuous evolution of the risk index over time, enhancing the accuracy of early warnings and the timeliness of response.

[0018] Furthermore, when initially determining the rolling risk index, the basic risk index of the current preset capture period is directly used as the rolling risk index. This can make full use of the latest collected data to reflect the company's real-time risk status, while avoiding the uncertainty caused by the lack of historical data in the initial stage. This allows risk assessment to be effectively quantified and continuously tracked in the early stages of system startup, thereby providing a reliable reference basis for subsequent rolling calculations and ensuring the smoothness and accuracy of the risk index evolution over time.

[0019] Furthermore, by calculating the difference between the rate of change in registered capital and the number of anomalies in upstream and downstream enterprises in real time, a correction instruction is automatically generated to adjust the corresponding weights when the difference exceeds a preset threshold, enabling a rapid response to dynamic changes in enterprise risk. By combining the difference with the weight adjustment coefficient, the corrected judicial addition weights and industrial and commercial change weights can accurately reflect the actual contribution of various risk factors to the rolling risk index, allowing risk assessment to take into account both recent data fluctuations and overall trends, while maintaining the continuity and rationality of weight changes, thereby improving the sensitivity, stability, and timeliness of risk prediction and early warning.

[0020] Furthermore, by mapping the difference in registered capital change rate and the difference in the number of outliers among upstream and downstream enterprises to corresponding evidence overview values, this embodiment can quantify the degree of anomaly in key enterprise indicators and directly reflect it in the weight adjustment of correction instructions. The difference in registered capital change rate is normalized using preset first and second standard registration evidence values ​​to ensure that the impact on risk judgment is limited when the capital change is small, while its effect can be rapidly amplified when the change is significant. Similarly, the difference in the number of outliers among upstream and downstream enterprises, after being normalized using first and second standard outlier evidence values, can accurately reflect abnormal fluctuations in the network of related enterprises. In this way, a clear proportional relationship is established between different indicators for the evidence overview values, making the adjustment range of correction instructions proportional to the actual change range of risk factors, thereby achieving dynamic and precise adjustment of the rolling risk index, improving the sensitivity and response speed of the early warning system to abnormal enterprise behavior, and ensuring system stability.

[0021] Furthermore, by combining the time interval between the current correction instruction generation time and the previous preset capture period with a preset decay coefficient to calculate data freshness, the timeliness of information can be quantitatively assessed. Data freshness decays exponentially with increasing time interval, making recently acquired data have a greater impact on the correction index and rolling risk index, while the impact of older data gradually weakens, thus reasonably reflecting the timeliness and dynamic characteristics of information. This approach ensures that real-time and significantly changing enterprise information is prioritized during rolling risk assessment, improving the accuracy and sensitivity of risk judgment, while avoiding excessive interference from outdated data in risk assessment.

[0022] Furthermore, by quantifying the difference between the simulated value and the rolling risk index, the impact is obtained. This impact is then combined with a preset sensitivity factor, a minor impact threshold, and a scaling factor to generate a penalty term. This allows the system to dynamically reflect the actual effect of changes in each parameter on the risk index. When the impact is large, the penalty term is appropriately amplified to indicate that the system remains sensitive to abnormal fluctuations. When the impact is small, the penalty term is suppressed by combining the minor impact threshold and the scaling factor to avoid overreacting to minor fluctuations. This achieves reasonable adjustment of the rolling risk index by factors such as changes in registered capital, anomalies in upstream and downstream enterprises, and data freshness, making the risk assessment results more robust and responsive.

[0023] Furthermore, by weighting and summing the evidence overview value, data freshness, impact, and penalty items, and introducing the Sigmoid function to map the results, a revised credibility is obtained. This helps to achieve a dynamic balance among different indicators, ensuring that the contribution of each factor to the final risk judgment reflects its relative importance while avoiding the excessive influence of a single abnormal indicator on the judgment. At the same time, the preset weights and bias terms can be flexibly adjusted according to actual monitoring needs. Thus, considering changes in registered capital, upstream and downstream anomalies, and data timeliness, the reliability of the correction instructions can be reasonably quantified, making the adjustment of the rolling risk index more robust, continuous, and in line with the actual trend of risk changes.

[0024] Furthermore, by quantitatively comparing the continuously calculated rolling risk index with a preset risk threshold, enterprises in a high-risk state can be accurately identified. When the rolling risk index exceeds the threshold, the system promptly labels the target enterprise as high-risk, achieving dynamic risk capture and early warning. This method not only reflects the comprehensive impact of multi-dimensional parameters such as the number of new judicial cases, the intensity of negative public opinion, the frequency of business registration changes, and anomalies in upstream and downstream industries, but also ensures the stability and consistency of risk assessment, providing an operable and traceable decision-making basis for enterprise risk management, while improving the timeliness and accuracy of early warning response.

[0025] Furthermore, by identifying the main contributing factors from the rolling risk index calculation, the scope of data collection for the next round is dynamically adjusted in a targeted manner. When the main contributing factor is the addition of judicial cases, the scope of judicial case collection is expanded to enable the system to more comprehensively capture recent changes in the company's judicial risks; when the main contributing factor is business registration changes, the scope of business registration changes collection is expanded to enable the system to more fully grasp changes in the company's structure and registration information. This adaptive adjustment mechanism ensures that various key parameters receive reasonable attention in different collection cycles, closely matching the breadth and depth of collected data with the fluctuations in the company's risks, thereby improving the sensitivity and accuracy of risk warnings and enhancing the system's responsiveness to changes in multidimensional risk information. Attached Figure Description

[0026] Figure 1 This is a flowchart of the business information early warning method based on big data in this embodiment; Figure 2 As shown, this is a flowchart for determining the rolling risk index in this embodiment; Figure 3 This embodiment defines the logic diagram for determining the high-risk level. Figure 4 This is a logic diagram for adjusting the preset capture range in this embodiment. Detailed Implementation

[0027] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0028] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0029] Please see Figure 1 The diagram shown is a flowchart of the business information early warning method based on big data in this embodiment. This embodiment provides a business information early warning method based on big data, including: The system retrieves data from a large database, including the number of new judicial cases, the level of negative public opinion, the frequency of business registration changes, the rate of change in registered capital, and the number of abnormal upstream and downstream enterprises for target companies, based on a preset retrieval period and a preset retrieval scope. A rolling risk index is determined based on the number of new judicial cases, the intensity of negative public opinion, and the frequency of business registration changes. A correction instruction is generated based on the registered capital change rate and the number of abnormal upstream and downstream enterprises. Simultaneously, an evidence overview value is determined based on the registered capital change rate and the number of abnormal upstream and downstream enterprises. The data freshness is determined based on the time when the correction instruction is generated. The correction index is calculated based on the evidence overview value and the data freshness, and the correction instruction is simulated according to the correction index and the preset correction threshold to obtain the simulation value. The impact amount is determined based on the simulation value and the rolling risk index, and the penalty item is determined by combining the preset sensitivity factor and the impact amount; The correction credibility is determined based on the evidence overview value, the data freshness, the impact, and the penalty item, and the correction instruction is executed based on the correction credibility to redetermine the rolling risk index, and the main contributing factors in the redetermining process of the rolling risk index are recorded simultaneously. The high-risk level is determined based on the redefined rolling risk index and the preset risk threshold; The preset capture range for the next round is adjusted according to the type of the main contributing factor; A warning is issued for high-risk levels that are redefined after adjusting the preset capture range in consecutive preset rounds.

[0030] In this embodiment, the big data database includes a judicial information database, a business information database, a public opinion monitoring database, and an upstream and downstream related database. The judicial information database records the filing time, case type, and judgment documents of enterprises, and is used to count the number of new judicial cases. The business information database contains enterprise registration information, change records, and registered capital, and is used to obtain the frequency of business changes, the rate of change of registered capital, and the number of abnormal upstream and downstream enterprises. The public opinion monitoring database integrates news, social media, and professional commentary data, and calculates the heat of negative public opinion through sentiment analysis. The upstream and downstream related database stores supply chain and customer and supplier relationships, and is used to count the number of abnormal enterprises.

[0031] In this embodiment, the system collects key indicator data of target enterprises from a large database according to a preset crawling period and a preset crawling range. Specific operations include: the system determines the crawling time window (i.e., the preset crawling range) based on the current time and the last crawling time; and automatically obtains the target enterprise's number of new judicial cases, negative public opinion intensity, business registration change frequency, registered capital change rate, and the number of abnormal upstream and downstream enterprises by calling database interfaces or APIs. During the collection process, the system sets corresponding parameters for each indicator to ensure the integrity and accuracy of the data. For example, the number of new judicial cases is accumulated daily, negative public opinion intensity is normalized according to public opinion scoring standards, business registration change frequency is counted by enterprise change events, registered capital change rate is calculated as a percentage, and the number of abnormal upstream and downstream enterprises is statistically calculated based on the total number of abnormal events. After collection, the system performs preliminary data cleaning and normalization processing to generate standardized input values ​​that can be used for calculating rolling risk indices and correction instructions.

[0032] In this embodiment, the preset crawling period refers to the time interval at which the system crawls various indicator data of the target enterprise from the large database. This depends on the data update frequency and business monitoring needs, and is typically set between 1 hour and 7 days. In this embodiment, it is set to 24 hours to balance data real-time performance and system load, ensuring continuous updates to the rolling risk index. The preset crawling range refers to the amount of data or data time window included in each round of crawling. This depends on the amount of information of the target enterprise and the historical data coverage requirements, and is typically set between the most recent 1 month and 12 months. In this embodiment, it is set to the most recent 3 months to balance the sufficiency of risk assessment and computational efficiency. The preset correction threshold refers to the criterion for triggering a correction instruction when the correction index reaches this threshold. This depends on the fluctuation range of the enterprise's indicators and the expected risk sensitivity, and is typically set between 0.1 and 0.5. In this embodiment, it is set to 0.3 to ensure that the correction instruction is triggered only when the index is displayed. The system makes corrections when changes occur to avoid over-adjustment. The preset sensitivity factor refers to the weighting coefficient of the impact factor in the penalty calculation, depending on the risk assessment's response requirements to abnormal fluctuations. It is typically set between 0.5 and 2; in this embodiment, it is set to 1, which can moderately amplify the impact of abnormal indicators on the reliability of corrections and enhance risk identification capabilities. The preset risk threshold refers to the boundary at which the rolling risk index exceeds this value to determine that the enterprise is in a high-risk state. It depends on the enterprise's industry risk level and historical risk data distribution, typically set between 0.6 and 0.9; in this embodiment, it is set to 0.8, which can accurately identify high-risk enterprises and issue timely warnings. The preset rounds refer to the number of times the system continuously adjusts the preset capture range, depending on the stability of the monitoring strategy and the risk response frequency. It is typically set between 3 and 10 rounds; in this embodiment, it is set to 5 rounds, which can gradually optimize the capture range and improve the accuracy of warnings.

[0033] By dynamically collecting data on new corporate legal cases, negative public opinion intensity, frequency of business registration changes, registered capital change rate, and the number of abnormal upstream and downstream enterprises, this method standardizes and weights multi-source indicators to form a rolling risk index. This index is then continuously adjusted based on correction instructions, evidence overview values, freshness, impact, and penalties, providing a real-time picture of corporate risk status. Through the dynamic mapping relationship between indicators, changes in registered capital and upstream / downstream anomalies can reasonably influence the weighting of judicial and business registration information. The difference between simulated values ​​and the rolling risk index guides sensitivity adjustment, ultimately generating a correction credibility to drive weight correction. This method adaptively adjusts the scope of data collection, strengthens monitoring of key risk-contributing factors, and enables timely identification and early warning of high-risk enterprises. This achieves continuity in risk assessment, sensitivity in response, and maximizes data utilization, improving the accuracy and reliability of business registration information early warning. It effectively solves the problems of delayed and inaccurate identification of high-risk corporate information due to data update delays, singular risk judgments, and static data collection strategies, leading to delayed early warnings and insufficient risk control.

[0034] Please see Figure 2 The flowchart shown illustrates the process of determining the rolling risk index in this embodiment. In this embodiment, the process of determining the rolling risk index based on the number of new judicial cases, the intensity of negative public opinion, and the frequency of business registration changes includes: standardizing the number of new judicial cases, the intensity of negative public opinion, and the frequency of business registration changes to obtain corresponding normalized index values; and weighting and summing the normalized index values ​​according to a preset index weighting to obtain the basic risk index, where R0 = a × S + b × F + c × G, R0 is the basic risk index, a is the preset judicial case weighting in the preset index weighting, b is the preset public opinion weighting in the preset index weighting, and c is the preset public opinion weighting in the preset index weighting. The weight of public opinion heat is defined as follows: c is the preset industrial and commercial change weight in the preset indicator weight reorganization; S is the normalized value of judicial addition in the normalized indicator value; F is the normalized value of public opinion heat in the normalized indicator value; and G is the normalized value of industrial and commercial change in the normalized indicator value. The basic risk index is combined with the rolling risk index of the previous preset capture period according to a preset decay coefficient to obtain the rolling risk index, where R(t) = R0 + λ × R(t − Δt), R(t) is the rolling risk index of the current preset capture period, λ is the preset decay coefficient, R(t - Δt) is the rolling risk index of the previous preset capture period, and Δt is the preset capture period.

[0035] In this embodiment, a preset weight for newly added judicial cases is used to quantify the impact of the number of newly added judicial cases on the rolling risk index. The value depends on the sensitivity of judicial events to the overall risk of the enterprise and historical data fluctuations, and is typically set between 0.3 and 0.5; in this embodiment, it is set to 0.4. A preset weight for public opinion intensity is used to quantify the effect of negative public opinion on the rolling risk index. The value depends on the reliability of public opinion data and its impact on the enterprise's public image and market reaction; it is typically set between 0.1 and 0.3; in this embodiment, it is set to 0.3. A preset weight for business registration changes is used to quantify the contribution of the frequency of business registration changes to the rolling risk index. The value depends on the frequency of changes in the enterprise's operations and its potential impact on overall risk; it is typically set between 0.2 and 0.4; in this embodiment, it is set to 0.3. The sum of these three weights is 1, which ensures a reasonable allocation of the proportions of each indicator in the basic risk index, while comprehensively reflecting the impact of judicial events, public opinion, and business registration changes on the overall risk of the enterprise. The preset attenuation coefficient is used to adjust the transmission weight of historical risk information in the rolling risk index. The value depends on the reference value of historical data for current risk prediction and the timeliness of information. It is usually set between 0.3 and 0.8. In this embodiment, it is set to 0.5, which can smooth short-term fluctuations and allow the rolling risk index to retain an appropriate amount of historical trend information while reflecting the latest data, thereby improving the stability and reliability of risk warning.

[0036] By standardizing the number of new judicial cases, the intensity of negative public opinion, and the frequency of business registration changes, and then weighting and summing them according to preset weights to obtain a basic risk index, and combining it with the rolling risk index from the previous capture period using a decay coefficient, continuous tracking of dynamic changes in enterprise risk can be achieved. The design of weights and decay coefficients among the parameters reflects the interrelationships of risk indicators: the weight of new judicial cases reflects the direct contribution of legal events to overall risk; the weight of public opinion intensity reflects the role of social opinion in the perception of enterprise risk; the weight of business registration changes reflects the impact of internal operational changes on risk; and the decay coefficient ensures the reasonable transmission of historical information in rolling risk calculations. Through this comprehensive processing method, the system can smooth short-term fluctuations, highlight important risk factors, and achieve continuous evolution of the risk index over time, enhancing the accuracy of early warnings and the timeliness of response.

[0037] Specifically, when determining the rolling risk index for the first time, the basic risk index of the current preset capture period is used as the rolling risk index.

[0038] When initially determining the rolling risk index, the basic risk index of the current preset capture period is directly used as the rolling risk index. This can make full use of the latest collected data to reflect the company's real-time risk status, while avoiding the uncertainty caused by the lack of historical data in the initial stage. This allows risk assessment to be effectively quantified and continuously tracked in the early stages of system startup, thus providing a reliable reference basis for subsequent rolling calculations and ensuring the smoothness and accuracy of the risk index evolution over time.

[0039] Specifically, the process of generating a correction instruction based on the registered capital change rate and the number of abnormal upstream and downstream enterprises includes: calculating the difference between the registered capital change rate in the current preset capture period and the difference between the number of abnormal upstream and downstream enterprises in the previous preset capture period; when the absolute value of the difference in the registered capital change rate is greater than or equal to a preset capital change rate threshold, generating a correction instruction to adjust the preset judicial new weight in the preset index weight restructuring, where a'=a×[1+k×(Z-Z0) / Z0], a' is the adjusted preset judicial new weight, k is the preset weight adjustment coefficient, Z is the absolute value of the difference in the registered capital change rate, and Z0 is the preset capital change rate threshold; when the absolute value of the difference in the number of abnormal upstream and downstream enterprises is greater than or equal to a preset abnormality difference threshold, generating a correction instruction to adjust the preset industrial and commercial change weight in the preset index weight restructuring, where c'=c×[1+k×(V-V0) / V0], c' is the adjusted preset industrial and commercial change weight, V is the absolute value of the difference in the number of abnormal upstream and downstream enterprises, and V0 is the preset abnormality difference threshold.

[0040] The preset capital change rate threshold is used to determine whether the change in registered capital is significant. Its magnitude depends on the industry characteristics, capital size, and historical fluctuation range of the enterprise, and is usually set between 1% and 10%. In this embodiment, it is set to 5%, which can effectively distinguish between normal fluctuations and abnormal capital changes and trigger corresponding weight adjustments. The preset weight adjustment coefficient is used to map the difference in registered capital change rate or the difference in the number of abnormalities between upstream and downstream enterprises to the weight adjustment range. It depends on the control strategy of the required sensitivity and correction range, and is usually set between 0.1 and 1. In this embodiment, it is set to 0.5, which can balance the sensitivity of risk response and system stability and achieve reasonable risk weight correction. The preset abnormality difference threshold is used to determine whether the change in the number of abnormalities of upstream and downstream enterprises is significant. Its magnitude depends on the fluctuation range of historical abnormal data and the sensitivity of the enterprise's related network, and is usually set between 1 and 20. In this embodiment, it is set to 5, which can effectively distinguish between occasional fluctuations and real abnormal events, thereby guiding the adjustment of industrial and commercial change weights.

[0041] By calculating the difference between the rate of change in registered capital and the number of anomalies in upstream and downstream enterprises in real time, and automatically generating correction instructions to adjust the corresponding weights when the difference exceeds a preset threshold, a rapid response to dynamic changes in enterprise risk is achieved. By combining the difference with the weight adjustment coefficient, the corrected judicial addition weights and industrial and commercial change weights can accurately reflect the actual contribution of various risk factors to the rolling risk index, enabling risk assessment to take into account both recent data fluctuations and overall trends, while maintaining the continuity and rationality of weight changes, thereby improving the sensitivity, stability, and timeliness of risk prediction and early warning.

[0042] Specifically, the process of simultaneously determining the evidence overview value based on the registered capital change rate and the number of abnormal upstream and downstream enterprises includes: determining the evidence overview value based on the absolute value of the difference in the registered capital change rate, the preset first standard registration evidence value, and the preset second standard registration evidence value, where Mz=min[1, (Z-Z1) / (Z2-Z1)], Mz is the evidence overview value corresponding to the correction instruction of the preset judicial new weight adjustment, Z1 is the preset first standard registration evidence value, and Z2 is the preset second standard registration evidence value; determining the evidence overview value based on the absolute value of the difference in the number of abnormal upstream and downstream enterprises, the preset first standard outlier evidence value, and the preset second standard outlier evidence value, where My=min[1, (V-V1) / (V2-V1)], My is the evidence overview value corresponding to the correction instruction of the preset industrial and commercial change weight adjustment in the preset indicator weight restructuring, V1 is the preset first standard outlier evidence value, and V2 is the preset second standard outlier evidence value.

[0043] In this embodiment, a preset first standard registration evidence value is used to define a threshold where the rate of change in registered capital is small but meaningful. This threshold depends on the company's historical capital fluctuations and the industry average, and is typically set between 0.5% and 3%. In this embodiment, it is set to 1%, which provides a reference for minor capital changes without causing excessive correction. A preset second standard registration evidence value is used to define a threshold where the rate of change in registered capital is significant and has a large impact. This threshold depends on the company's capital adjustment inertia and industry fluctuation characteristics, and is typically set between 3% and 10%. In this embodiment, it is set to 5%, which enhances the correction weight when capital changes significantly. A preset first standard outlier evidence value is used to define a threshold where the number of anomalies in upstream and downstream companies is small but requires attention. This threshold depends on the historical distribution and average level of anomalies in the supply chain, and is typically set between 1 and 5. In this embodiment, it is set to 2, which provides a reference for initial anomaly fluctuations. A preset second standard outlier evidence value is used to define a threshold where the number of anomalies in upstream and downstream companies changes significantly. This threshold depends on the size of the upstream and downstream network and the historical magnitude of anomalies, and is typically set between 5 and 20. In this embodiment, it is set to 10, which allows for significant adjustment of correction instructions when the number of anomalies increases significantly, improving the accuracy of supply chain risk response and early warning.

[0044] By mapping the difference in registered capital change rate and the difference in the number of outliers among upstream and downstream enterprises to corresponding evidence overview values, this embodiment can quantify the degree of anomaly in key enterprise indicators and directly reflect it in the weight adjustment of correction instructions. The difference in registered capital change rate is normalized using preset first and second standard registration evidence values ​​to ensure that the impact on risk judgment is limited when the capital change is small, but can be rapidly amplified when the change is significant. Similarly, the difference in the number of outliers among upstream and downstream enterprises, after being normalized using first and second standard outlier evidence values, can accurately reflect abnormal fluctuations in the network of related enterprises. In this way, a clear proportional relationship is established between different indicators for the evidence overview values, making the adjustment range of correction instructions proportional to the actual change range of risk factors, thereby achieving dynamic and precise adjustment of the rolling risk index, improving the sensitivity and response speed of the early warning system to abnormal enterprise behavior, and ensuring system stability.

[0045] Specifically, the process of determining data freshness based on the time of the correction instruction's generation includes: determining the data freshness based on the time interval between the current time of the correction instruction's generation and the previous preset fetching period, and a preset decay coefficient, where E=exp(−Δt1 / τ), E is the data freshness, Δt1 is the time interval between the current time of the correction instruction's generation and the previous preset fetching period, and τ is the preset decay coefficient.

[0046] The preset decay coefficient is a parameter used to control the rate at which data freshness decays. It depends on the frequency of enterprise information updates and the sensitivity of the indicators being monitored to risk assessment. It is usually set between 0 and 1. In this embodiment, it is set to 0.5, which can balance the influence of recent data and historical data, so that the rolling risk index reflects recent changes while taking into account historical trends, thereby improving the accuracy and stability of risk warnings.

[0047] By calculating data freshness using the time interval between the current correction instruction generation time and the previous preset capture period, combined with a preset decay coefficient, the timeliness of information can be quantitatively assessed. Data freshness decays exponentially with increasing time interval, making recently acquired data have a greater impact on the correction index and rolling risk index, while the impact of older data gradually weakens, thus reasonably reflecting the timeliness and dynamic characteristics of information. This approach ensures that real-time and significantly changing enterprise information is prioritized during rolling risk assessment, improving the accuracy and sensitivity of risk judgment, while avoiding excessive interference from outdated data in risk assessment.

[0048] Specifically, the process of calculating the correction index based on the evidence overview value and the data freshness includes: X = xz × Mz + xy × My + xe × E, where xz is the preset registration correction weight, xy is the preset outlier correction weight, and xe is the preset freshness correction weight.

[0049] Specifically, the process of simulating the correction instruction according to the correction index and the preset correction threshold to obtain the simulation value includes: when the correction index is greater than the preset correction threshold, it is determined that simulation is required, and the number of new judicial cases, negative public opinion heat, frequency of industrial and commercial changes, rate of change of registered capital, number of abnormal upstream and downstream enterprises, and preset indicator weight reorganization of the current preset capture period are input into the historical data-driven regression model to calculate the possible change value of the current rolling risk index, and the change value is used as the simulation value output.

[0050] The preset registration correction weight, preset outlier correction weight, and preset freshness correction weight are used to measure the relative contributions of the registered capital change rate, the number of outliers in upstream and downstream enterprises, and data freshness to the correction index, respectively. Their specific values ​​depend on the importance of different parameters to the enterprise's risk and are typically set between 0 and 1, with the sum of the three being 1. In this embodiment, the preset registration correction weight is set to 0.4, the preset outlier correction weight is set to 0.3, and the preset freshness correction weight is set to 0.3. This ensures that the influence of each factor is reasonably allocated when calculating the correction index, so that the correction index reflects both the enterprise's own changes and the timeliness of data and its correlation with upstream and downstream enterprises.

[0051] By calculating the correction index by weighting the evidence overview value and data freshness, and combining it with a preset correction threshold to conduct historical data-driven regression simulation of the correction instruction, it is possible to quantify the potential impact of changes in registered capital, abnormalities in upstream and downstream enterprises, and judicial and public opinion information on the rolling risk index. This allows the simulation value to truly reflect the possible trend of enterprise risk changes, thereby providing a reliable risk reference before executing the correction instruction, enabling scientific prediction and dynamic optimization of risk index adjustments, and improving the timeliness and accuracy of early warning.

[0052] Specifically, the process of determining the impact amount based on the simulation value and the rolling risk index, and determining the penalty term in combination with the preset sensitivity factor and the impact amount, includes: calculating the difference between the simulation value and the rolling risk index to obtain the impact amount, where ΔR = |Rf - R(t)|, ΔR is the impact amount, and Rf is the simulation value; determining the penalty term based on the preset sensitivity factor, the impact amount, the preset small impact threshold, and the preset scale factor, where J = D × max[0, (ΔR - δ) / γ], J is the penalty term, δ is the preset small impact threshold, γ is the preset scale factor, and D is the preset sensitivity factor.

[0053] The preset minor impact threshold is used to distinguish between negligible fluctuations and actual impacts. Its magnitude depends on the fluctuation range of the rolling risk index in the historical capture period, and is usually set between 0.005 and 0.02. In this embodiment, it is set to 0.01, which can effectively filter out the interference caused by minor fluctuations. The preset scaling factor is used to adjust the amplification of the penalty term when the impact exceeds the minor impact threshold. Its value depends on the sensitivity requirements of the risk index change, and is usually set between 0.3 and 0.7. In this embodiment, it is set to 0.5, which can reasonably amplify the response to significant abnormal fluctuations and make the penalty term match the actual risk change.

[0054] By quantifying the difference between simulated values ​​and the rolling risk index, the impact is obtained. This impact is then combined with preset sensitivity factors, a small impact threshold, and a scaling factor to generate a penalty term. This allows the system to dynamically reflect the actual effect of changes in various parameters on the risk index. When the impact is large, the penalty term is appropriately amplified to indicate that the system remains sensitive to abnormal fluctuations. When the impact is small, the penalty term is suppressed by combining the small impact threshold and the scaling factor to avoid overreacting to minor fluctuations. This achieves reasonable adjustment of the rolling risk index by factors such as changes in registered capital, anomalies in upstream and downstream enterprises, and data freshness, making the risk assessment results more robust and responsive.

[0055] Specifically, the process of determining the revised credibility based on the evidence overview value, the data freshness, the impact, and the penalty item includes: performing a weighted summation of the evidence overview value, the data freshness, the impact, the penalty item, and a preset revised credibility weight to obtain the revised credibility, where P = σ(wm×Mz + wy×My + we×E−wr×ΔR−wj×J + q), P is the revised credibility, and σ(x) = 1 / (1+e −x ) is the Sigmoid function, wm is the preset registration overview value weight, wy is the preset anomaly count overview value weight, we is the preset freshness weight, wr is the preset influence weight, wj is the preset penalty weight, and q is the preset bias term.

[0056] The preset registration overview value weight, preset anomaly number overview value weight, preset freshness weight, preset impact weight, preset penalty weight, and preset bias term are parameters used by the system to measure the contribution of each indicator when calculating the corrected credibility. Their values ​​depend on the importance of each indicator to risk judgment and the system setting strategy, and are usually set between 0 and 1 (the bias term can be a positive or negative decimal). In this embodiment, they are set to specific values ​​(wm=0.15, wy=0.25, we=0.3, wr=0.2, wj=0.1, q=0.05), which can achieve an effective balance of various influencing factors, so that the corrected credibility can accurately reflect the comprehensive effect of the evidence overview value, data freshness, impact, and penalty term on the rolling risk index adjustment.

[0057] By weighting and summing the evidence overview value, data freshness, impact, and penalty items, and introducing the Sigmoid function to map the results, a corrected credibility is obtained. This helps to achieve a dynamic balance among different indicators, ensuring that the contribution of each factor to the final risk judgment reflects its relative importance while avoiding the excessive influence of a single abnormal indicator on the judgment. At the same time, the preset weights and biases can be flexibly adjusted according to actual monitoring needs. This allows for the reasonable quantification of the reliability of correction instructions, taking into account changes in registered capital, upstream and downstream anomalies, and data timeliness. This makes the adjustment of the rolling risk index more robust, continuous, and in line with the actual trend of risk changes.

[0058] Specifically, the process of executing the correction instruction based on the correction confidence level includes: executing the correction instruction when the correction confidence level is greater than a preset confidence level threshold.

[0059] The preset confidence threshold is a key parameter used to determine whether a correction instruction needs to be executed. Its value depends on the system's risk sensitivity requirements. It is usually set between 0 and 1. In this embodiment, it is set to 0.7, which ensures that the correction instruction is only executed when the correction confidence is high and the comprehensive signals of various indicators clearly indicate that there is a risk. This improves the reliability of risk assessment and avoids over-adjustment.

[0060] Specifically, the process of synchronously recording the main contributing factors in the re-determination of the rolling risk index includes: comparing the product of the preset judicial new weight after the execution of the correction instruction and the judicial new normalized value in the normalized index value, and the product of the preset industrial and commercial change weight after the execution of the correction instruction and the industrial and commercial change normalized value in the normalized index value. The larger value and its corresponding index type are recorded as the main contributing factors of the current rolling risk index. If the two are equal, the main contributing factor is determined to be the product of the preset judicial new weight after the execution of the correction instruction and the judicial new normalized value in the normalized index value, and its corresponding index type is the main contributing factor.

[0061] By identifying key indicators that play a dominant role in the re-determination of the rolling risk index, the main contributors can be quickly identified; even if two indicators are equal, the consistency and stability of the judgment can be ensured through priority rules. This enables quantitative comparison of changes in various parameters, making risk assessment more targeted and traceable. Simultaneously, it ensures that major risk factors are accurately recorded and fed back during continuous adjustments and corrections, providing a reliable basis for the next round of scope adjustments and high-risk level determinations, thus improving the system's ability to perceive and respond to potential enterprise risks.

[0062] Please see Figure 3 As shown, this is the logic diagram for determining the high-risk level in this embodiment. In this embodiment, the process of determining the high-risk level based on the re-determined rolling risk index and the preset risk threshold includes: when the rolling risk index is greater than the preset risk threshold, determining that the target enterprise is in a high-risk state, thereby determining the high-risk level.

[0063] By quantitatively comparing the continuously calculated rolling risk index with a preset risk threshold, enterprises in a high-risk state can be accurately identified. When the rolling risk index exceeds the threshold, the system promptly labels the target enterprise as high-risk, achieving dynamic risk capture and early warning. This method not only reflects the comprehensive impact of multiple dimensions of parameters such as the number of new judicial cases, the intensity of negative public opinion, the frequency of business registration changes, and anomalies in upstream and downstream industries, but also ensures the stability and consistency of risk assessment, providing an operable and traceable decision-making basis for enterprise risk management, while improving the timeliness and accuracy of early warning response.

[0064] Please see Figure 4As shown, this is the logic diagram for adjusting the preset capture range in this embodiment. In this embodiment, the process of adjusting the preset capture range in the next round according to the type of the main contribution factor includes: when the type of the main contribution factor is the addition of judicial cases, increasing the preset capture range of the number of new judicial cases according to the preset judicial capture coefficient, Hs'=Hs×[1+ks], where Hs' is the preset capture range of the number of new judicial cases after the increase, Hs is the preset capture range of the number of new judicial cases before the increase, and ks is the preset judicial capture coefficient; when the type of the main contribution factor is the change of business registration, increasing the preset capture range of the frequency of the change of business registration according to the preset frequency capture coefficient, Hg'=Hg×[1+kg], where Hg' is the preset capture range of the frequency of the change of business registration after the increase, Hg is the preset capture range of the frequency of the change of business registration before the increase, and kg is the preset frequency capture coefficient.

[0065] The preset judicial case capture coefficient is a proportional parameter used to adjust the scope of the next round of judicial case capture when the main contributing factor is the addition of judicial cases. It depends on the dominance of judicial case information in the risk change trend and the sensitivity setting of the monitoring system to judicial information. It is usually set between 0.05 and 0.3. In this embodiment, it is set to 0.2, which can significantly increase the capture frequency when the impact of the addition of judicial cases is significant, thereby capturing risk change signals more quickly. The preset frequency capture coefficient is a proportional parameter used to adjust the scope of the next round of business registration change frequency capture when the main contributing factor is business registration change. It depends on the sensitivity of business registration change activities to enterprise risk fluctuations and the data capture resource allocation strategy. It is usually set between 0.05 and 0.3. In this embodiment, it is set to 0.15, which can promptly expand the data coverage when business registration changes occur frequently, improving the accuracy and real-time performance of risk monitoring.

[0066] By identifying the main contributing factors from the rolling risk index calculation, the scope of data collection for the next round is dynamically adjusted in a targeted manner. When the main contributing factor is the addition of judicial cases, the scope of judicial case collection is expanded to enable the system to more comprehensively capture recent changes in the company's judicial risks; when the main contributing factor is business registration changes, the scope of business registration changes collection is expanded to enable the system to more fully grasp changes in the company's structure and registration information. This adaptive adjustment mechanism ensures that various key parameters receive reasonable attention in different collection cycles, closely matching the breadth and depth of collected data with the fluctuations in the company's risks, thereby improving the sensitivity and accuracy of risk warnings and enhancing the system's responsiveness to changes in multidimensional risk information.

[0067] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A business information early warning method based on big data, characterized in that, include: The system retrieves data from a large database, including the number of new judicial cases, the level of negative public opinion, the frequency of business registration changes, the rate of change in registered capital, and the number of abnormal upstream and downstream enterprises for target companies, based on a preset retrieval period and a preset retrieval scope. A rolling risk index is determined based on the number of new judicial cases, the intensity of negative public opinion, and the frequency of business registration changes. A correction instruction is generated based on the registered capital change rate and the number of abnormal upstream and downstream enterprises. Simultaneously, an evidence overview value is determined based on the registered capital change rate and the number of abnormal upstream and downstream enterprises. The data freshness is determined based on the time when the correction instruction is generated. The correction index is calculated based on the evidence overview value and the data freshness, and the correction instruction is simulated according to the correction index and the preset correction threshold to obtain the simulation value. The impact amount is determined based on the simulation value and the rolling risk index, and the penalty item is determined by combining the preset sensitivity factor and the impact amount; The correction credibility is determined based on the evidence overview value, the data freshness, the impact, and the penalty item, and the correction instruction is executed based on the correction credibility to redetermine the rolling risk index, and the main contributing factors in the redetermining process of the rolling risk index are recorded simultaneously. The high-risk level is determined based on the redefined rolling risk index and the preset risk threshold; The preset capture range for the next round is adjusted according to the type of the main contributing factor; A warning is issued for high-risk levels that are redefined after adjusting the preset capture range in consecutive preset rounds.

2. The business information early warning method based on big data according to claim 1, characterized in that, The process of determining the rolling risk index based on the number of new judicial cases, the intensity of negative public opinion, and the frequency of business registration changes includes: The number of new judicial cases, the level of negative public opinion, and the frequency of business registration changes are standardized to obtain the corresponding normalized index values. The normalized index values ​​are weighted and summed according to the preset index weighting to obtain the basic risk index; The rolling risk index is obtained by combining the basic risk index with the rolling risk index of the previous preset capture period according to a preset decay coefficient.

3. The business information early warning method based on big data according to claim 2, characterized in that, When determining the rolling risk index for the first time, the basic risk index of the current preset capture period is used as the rolling risk index.

4. The business information early warning method based on big data according to claim 3, characterized in that, The process of generating a correction instruction based on the registered capital change rate and the number of abnormal upstream and downstream enterprises includes: Calculate the difference between the registered capital change rate in the current preset capture period and the difference between the number of abnormal upstream and downstream enterprises in the previous preset capture period; When the absolute value of the difference in the registered capital change rate is greater than or equal to the preset capital change rate threshold, the correction instruction to adjust the preset judicial new weight in the preset index weight restructuring is generated. When the absolute value of the difference between the number of abnormal enterprises in the upstream and downstream enterprises is greater than or equal to the preset abnormal difference threshold, the correction instruction for adjusting the preset industrial and commercial change weight in the preset indicator weight reorganization is generated.

5. The business information early warning method based on big data according to claim 4, characterized in that, The process of simultaneously determining the evidence overview value based on the registered capital change rate and the number of anomalies in upstream and downstream enterprises includes: The evidence summary value is determined based on the absolute value of the difference in the registered capital change rate, the preset first standard registration evidence value, and the preset second standard registration evidence value; The evidence overview value is determined based on the absolute value of the difference between the number of anomalies of the upstream and downstream enterprises, the preset first standard outlier evidence value, and the preset second standard outlier evidence value.

6. The business information early warning method based on big data according to claim 5, characterized in that, The process of determining data freshness based on the time of generation of the correction instruction includes: The data freshness is determined based on the time interval between the current correction instruction generation time and the previous preset capture period, and the preset decay coefficient.

7. The business information early warning method based on big data according to claim 6, characterized in that, The process of determining the impact amount based on the simulation value and the rolling risk index, and determining the penalty item by combining the preset sensitivity factor and the impact amount, includes: The difference between the simulated value and the rolling risk index is calculated to obtain the influence amount; The penalty term is determined based on the preset sensitivity factor, the amount of influence, the preset small influence threshold, and the preset scale factor.

8. The business information early warning method based on big data according to claim 7, characterized in that, The process of determining the credibility of the correction based on the evidence overview value, the data freshness, the impact, and the penalty item includes: The corrected credibility is obtained by weighting and summing the evidence overview value, the data freshness, the impact, the penalty item, and the preset corrected credibility weight.

9. The business information early warning method based on big data according to claim 8, characterized in that, The process of determining a high-risk level based on the redefined rolling risk index and the preset risk threshold includes: When the rolling risk index is greater than the preset risk threshold, the target enterprise is determined to be in a high-risk state, thereby determining the high-risk level.

10. The business information early warning method based on big data according to claim 9, characterized in that, The process of adjusting the preset capture range for the next round based on the type of the main contributing factor includes: When the type of the main contributing factor is newly added judicial cases, the preset capture range of the number of newly added judicial cases is increased according to the preset judicial capture coefficient; When the type of the main contributing factor is business registration change, the preset capture range of the business registration change frequency is increased according to the preset capture coefficient.

Images