A big data risk management and analysis system based on big model
By designing a big data risk management and analysis system based on big models in a big data environment, using file data collection, risk identification, evaluation and early warning decision-making modules, dynamically adjusting the inspection cycle and number of times, solving the risk inspection in traditional risk management methods. Timely issues have achieved improvements in data security and efficiency.
Patent Information
- Application Number
- CN202410455066.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-04-16
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2044-04-16
AI Technical Summary
Traditional risk management methods are difficult to identify and respond to data risks in a timely manner in a big data environment, resulting in untimely risk investigation and inability to adapt to rapidly changing data security threats.
A big data risk management and analysis system based on big models was designed, including file data acquisition module, risk identification module, risk assessment module and early warning decision-making module. By dynamically adjusting the inspection cycle and number of inspections, the inspection is based on the access data and sample content. Deviation determines the risk reference value to achieve timely identification and response to data risks.
It improves data processing efficiency and security, can promptly detect and respond to data security threats, optimize resource allocation, improve system operation efficiency, dynamically adjust risk management strategies, and reduce losses caused by potential risks.
Smart Images

Figure CN118350627B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of data management, and in particular to a big data risk management and analysis system based on a big model. Background Art
[0002] With the rapid development of information technology, data has become an important asset for enterprises and organizations. Its security and effective management are crucial to ensuring business continuity and maintaining competitive advantage. However, data risk management in the big data environment faces unprecedented challenges, including the surge in data volume, the diversification of data types, and the increasing complexity of data security threats. Traditional risk management methods are often difficult to adapt to this rapidly changing environment, and an intelligent and automated solution is urgently needed to improve the efficiency and accuracy of risk identification, assessment and response.
[0003] Big data risk management and analysis systems based on big models have emerged, aiming to conduct in-depth analysis and processing of large-scale data sets through advanced artificial intelligence technology to identify potential risk factors and provide decision support.
[0004] Chinese Patent Publication No. CN116155627B discloses a display screen access data management system and method based on the Internet, which belongs to the field of display screen data management technology. The system includes an Internet access call module, a real-time monitoring module, a data test module and a security warning module; the output end of the Internet access call module is connected to the input end of the data test module; the output end of the real-time monitoring module is connected to the input end of the data test module; the output end of the data test module is connected to the input end of the security warning module. The system can judge the risk of the display screen being attacked based on security analysis, and provide a more accurate troubleshooting sequence in the troubleshooting process to improve troubleshooting efficiency. At the same time, it can also realize security warning analysis under different display screen access data. It can be seen that the troubleshooting process of this invention is passively triggered, and troubleshooting is not carried out in time. Summary of the invention
[0005] To this end, the present invention provides a big data risk management and analysis system based on a big model to overcome the problem of untimely risk screening caused by a fixed risk screening cycle in the prior art.
[0006] To achieve the above objectives, the present invention provides a big data risk management and analysis system based on a big model, comprising:
[0007] File data collection module, used to obtain access data of each file link;
[0008] A risk identification module connected to the file data acquisition module, used to determine the data risk identification check cycle, the expected number of accesses and the expected value of a single access duration for each file link according to the access data, and adjust the check cycle based on the preset number of accesses and the preset single access duration. The risk identification module determines the number of checks within a single check cycle according to the actual access frequency of a single file link, so as to check the file links within the check cycle;
[0009] A risk assessment module, which is connected to the risk identification module and the file data acquisition module respectively, and is used to determine a risk reference value according to the deviation between the large model output result of the sample input content in a single file link and the corresponding sample output content, determine a preset risk reference value according to the number of learned samples of the current large model and its actual accuracy, and determine whether to use the current sample content based on the preset risk reference value and the risk reference value;
[0010] An early warning decision module, which is connected to the risk assessment module and is used to determine its decision based on the assessment result, the decision including whether to adjust the investigation cycle and whether to adjust the number of investigations;
[0011] The access data includes the number of access terminals, the number of accesses of a single access terminal, the access duration of a single access terminal and access data definition, and the data definition includes sample input and acquisition results.
[0012] Furthermore, the risk identification module determines the data risk identification check cycle for each file link according to the access data definition, wherein:
[0013] The access data is defined as the investigation period corresponding to the sample input is less than the investigation period corresponding to the acquisition result defined as the access data.
[0014] Furthermore, the risk identification module determines the expected number of accesses and the expected duration of a single access of each file link according to the number of access terminals, the number of accesses of a single access terminal and the duration of access of a single access terminal.
[0015] Further, the risk identification module compares the expected value of the number of accesses with the preset value of the number of accesses of the corresponding file link, and compares the expected value of the single access duration with the preset value of the single access duration of the corresponding file link, and adjusts the screening period of the corresponding file link based on the comparison result;
[0016] If the expected number of visits is less than or equal to the preset number of visits, and the expected single visit duration is less than or equal to the preset single visit duration, the risk identification module determines not to adjust the investigation period, and sets the currently determined investigation period as the initially determined investigation period;
[0017] If the expected value of the number of accesses is greater than the preset value of the number of accesses, and / or the expected value of the duration of a single access is greater than the preset value of the duration of a single access, the risk identification module adjusts the initially determined screening period according to the difference between the expected value of the number of accesses and the preset value of the number of accesses of the corresponding file link and / or the difference between the expected value of the duration of a single access and the preset value of the duration of a single access of the corresponding file link.
[0018] Furthermore, the risk identification module determines the corresponding preset value of the number of accesses, the preset value of the duration of a single access, and the number of inspections within a single inspection cycle according to the actual access frequency of a single file link.
[0019] Further, the risk assessment module determines a risk reference value of a single file link using at least one set of samples, including:
[0020] In a single file link, the risk assessment module determines the corresponding large model output result according to the single sample input content, and determines the risk reference value according to the deviation between the large model output result and the sample output content;
[0021] The sample includes sample input content and corresponding sample output content.
[0022] Further, the risk assessment module determines whether to use the current sample content to train the large model according to the risk reference value and the number of learned samples of the current large model and transmits the assessment result to the early warning decision module;
[0023] If the risk reference value is less than or equal to the preset risk reference value corresponding to the number of samples, the risk assessment module determines to use the content of the file link corresponding to the current sample content to train the large model;
[0024] If the risk reference value is greater than the preset risk reference value corresponding to the number of samples, the risk assessment module determines not to use the content of the file link corresponding to the current sample content to train the large model.
[0025] Furthermore, the risk assessment module determines the preset risk reference value according to the number of learned samples of the current large model and the actual accuracy of the large model.
[0026] Furthermore, the early warning decision module determines its decision according to the evaluation result, including:
[0027] When the evaluation result is to use the current sample content to train the large model, the early warning decision module adjusts the inspection cycle according to the actual accuracy change of the large model after using the sample and adjusts the number of inspections within a single cycle according to the actual accuracy change value of the large model after using the sample.
[0028] Furthermore, the early warning decision module determines its decision according to the evaluation result, including:
[0029] When the cumulative evaluation results of the adjacent preset number of times are that the current sample content is not used to train the large model, the early warning decision module determines that there is a risk of data destruction and issues an alarm.
[0030] Compared with the prior art, the invention has the beneficial effect that the big data risk management and analysis system provided by the invention uses the file data acquisition module to obtain and define the access data of the file link, the risk identification module determines the investigation cycle and importance of data risk identification according to the access data, the risk assessment module assesses the sample risk reference value, and finally the early warning decision module makes corresponding decisions according to the assessment results. This system can not only improve the efficiency of data processing, but also timely discover and respond to various data security threats, thereby supporting data-driven decision-making for enterprises and organizations while ensuring data security.
[0031] Furthermore, the big data risk management and analysis system provided by the present invention realizes effective management and analysis of risks in a big data environment by comprehensively using multiple modules such as file data acquisition, risk identification, risk assessment and early warning decision-making, so that it can timely discover and deal with potential data risks, improve data security and reliability, and optimize resource allocation and improve system operation efficiency.
[0032] Furthermore, by determining the investigation cycle, expected number of accesses and expected duration of a single access for data risk identification of file links, the present invention can dynamically adjust the investigation cycle, so that the system can flexibly respond to different risk conditions and adjust risk management strategies in a timely manner.
[0033] Furthermore, the present invention can determine whether to use the current sample content based on the comparison between the preset risk reference value and the actual risk reference value, thereby ensuring the quality and security of the data.
[0034] Furthermore, the present invention determines whether to adjust the inspection cycle and the number of inspections based on the risk assessment results, thereby realizing dynamic optimization of the risk management strategy. The existence of the early warning decision module enables the system to respond to risk changes in a timely manner, quickly take preventive or response measures, and reduce losses caused by potential risks. BRIEF DESCRIPTION OF THE DRAWINGS
[0035] Figure 1 It is a connection diagram of a big data risk management and analysis system based on a big model according to an embodiment of the present invention;
[0036] Figure 2 This is a workflow diagram of the risk identification module according to an embodiment of the present invention;
[0037] Figure 3A flow chart of determining the expected value of the number of visits by the risk identification module according to an embodiment of the present invention;
[0038] Figure 4 The following is a workflow diagram of the risk assessment module according to an embodiment of the present invention. DETAILED DESCRIPTION
[0039] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0040] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.
[0041] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the drawings. This is merely for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.
[0042] In addition, it should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0043] See also Figure 1 As shown, it is a connection diagram of a big data risk management and analysis system based on a big model according to an embodiment of the present invention. An embodiment of the present invention provides a big data risk management and analysis system based on a big model, including:
[0044] File data collection module, used to obtain access data of each file link;
[0045] A risk identification module, which is connected to the file data acquisition module, is used to determine the data risk identification investigation cycle, the expected number of accesses and the expected value of a single access duration for each file link according to the access data, and adjust the investigation cycle based on the preset number of accesses and the preset value of a single access duration. The risk identification module determines the number of investigations within a single investigation cycle according to the actual access frequency of a single file link, so as to investigate the file links within the investigation cycle; by determining the data risk identification investigation cycle, the expected number of accesses and the expected value of a single access duration for the file link, the present invention can dynamically adjust the investigation cycle, so that the system can flexibly respond to different risk conditions and adjust the risk management strategy in a timely manner;
[0046] A risk assessment module, which is connected to the risk identification module and the file data acquisition module respectively, and is used to determine a risk reference value according to the deviation between the large model output result of the sample input content in a single file link and the corresponding sample output content, determine a preset risk reference value according to the number of learned samples of the current large model and its actual accuracy, and determine whether to use the current sample content based on the preset risk reference value and the risk reference value; the present invention can determine whether to use the current sample content based on the comparison between the preset risk reference value and the actual risk reference value, thereby ensuring the quality and security of the data;
[0047] An early warning decision module is connected to the risk assessment module to determine its decision according to the assessment results, and the decision includes whether to adjust the investigation cycle and the number of investigations. The present invention determines whether to adjust the investigation cycle and the number of investigations according to the risk assessment results, thereby realizing dynamic optimization of the risk management strategy. The existence of the early warning decision module enables the system to respond to risk changes in a timely manner, quickly take preventive or countermeasures, and reduce losses caused by potential risks.
[0048] The access data includes the number of access terminals, the number of accesses of a single access terminal, the access duration of a single access terminal and access data definition, and the data definition includes sample input and acquisition results.
[0049] It can be understood that the big data risk management and analysis system provided by the present invention realizes the effective management and analysis of risks in the big data environment by comprehensively using multiple modules such as file data acquisition, risk identification, risk assessment and early warning decision-making, so that it can timely discover and deal with potential data risks, improve data security and reliability, and optimize resource allocation and improve system operation efficiency.
[0050] See also Figure 2 As shown, it is a workflow diagram of the risk identification module of the embodiment of the present invention. The risk identification module of the present invention determines the troubleshooting cycle for data risk identification of each file link according to the access data definition, wherein:
[0051] The access data is defined as the investigation period corresponding to the sample input is less than the investigation period corresponding to the acquisition result defined as the access data.
[0052] It is understandable that when the access data of a single file link is defined as sample input, it means that the current file link needs to be modified or added, which will increase the risk of data errors in the file link. Therefore, a relatively small troubleshooting cycle can be set for the file link. On the contrary, when the access data of a single file link is defined as obtaining results, it means that the current file link does not need to be modified or added, and only the original data needs to be relied upon to obtain the corresponding results. Therefore, the risk of data errors in the file link will not be increased. Therefore, a relatively large troubleshooting cycle can be set for the file link.
[0053] Therefore, in implementation, the investigation period corresponding to the access data defined as sample input is set to 1 to 5 days, and the investigation period corresponding to the access data defined as acquisition results is set to 5 to 20 days; preferably, the investigation period corresponding to the sample input is 2 days, and the investigation period corresponding to the acquisition results is 12 days.
[0054] Specifically, the risk identification module determines the expected number of accesses and the expected duration of a single access of each file link according to the number of access terminals, the number of accesses of a single access terminal, and the duration of access of a single access terminal.
[0055] See also Figure 3 As shown, it is a flow chart of determining the expected value of the number of accesses by the risk identification module of an embodiment of the present invention. In implementation, for a single file link, the expected value of the number of accesses E1 is determined based on the number of access terminals and the number of accesses of a single access terminal. The determination method includes: (1) classifying the single access terminals with the same number of accesses into the same category; (2) determining that there are a total of j categories; (3) determining the probability of the number of accesses Pi for each category, Pi = the number of access terminals corresponding to the i-th category ÷ the number of access terminals, where (4) Calculate the expected number of visits E1 according to the formula, Among them, xi is the number of visits corresponding to the i-th category.
[0056] In implementation, for a single file link, the expected value E2 of a single access duration is determined based on the number of access terminals and the access duration of a single access terminal. The determination method includes: (1) classifying the single access terminals with the same access duration into the same category; (2) determining that there are a total of v categories; (3) determining the access probability Pu of each category, Pu = the number of access terminals corresponding to the uth category ÷ the number of access terminals, where (4) Calculate the expected number of visits E2 according to the formula, Among them, xu is the number of visits corresponding to the u-th category.
[0057] Specifically, the risk identification module compares the expected number of accesses with the preset number of accesses of the corresponding file link, and compares the expected value of a single access duration with the preset value of a single access duration of the corresponding file link, and adjusts the screening period of the corresponding file link based on the comparison result;
[0058] If the expected number of visits is less than or equal to the preset number of visits, and the expected single visit duration is less than or equal to the preset single visit duration, the risk identification module determines not to adjust the investigation period, and sets the currently determined investigation period as the initially determined investigation period;
[0059] If the expected value of the number of accesses is greater than the preset value of the number of accesses, and / or the expected value of the duration of a single access is greater than the preset value of the duration of a single access, the risk identification module adjusts the initially determined screening period based on the difference between the expected value of the number of accesses and the preset value of the number of accesses of the corresponding file link and / or the difference between the expected value of the duration of a single access and the preset value of the duration of a single access of the corresponding file link.
[0060] It can be understood that (1) if the expected value of the number of accesses is greater than the preset value of the number of accesses and the expected value of the duration of a single access is less than or equal to the preset value of the duration of a single access, the risk identification module initially determines the adjustment value of the investigation period only based on the difference between the expected value of the number of accesses and the preset value of the number of accesses of the corresponding file link, so as to reduce the initially determined investigation period; (2) if the expected value of the duration of a single access is greater than the preset value of the duration of a single access and the expected value of the number of accesses is less than or equal to the preset value of the number of accesses, the risk identification module initially determines the adjustment value of the investigation period only based on the difference between the expected value of the duration of a single access and the preset value of the duration of a single access. The difference between the single access time preset value of the corresponding file link is used to initially determine the adjustment value of the investigation period to reduce the initially determined investigation period; (3) if the expected single access time is greater than the single access time preset value and the expected number of accesses is greater than the access number preset value, then the risk identification module determines the adjustment value of the initially determined investigation period based on the difference between the expected number of accesses and the access number preset value of the corresponding file link, and the difference between the expected single access time and the single access time preset value of the corresponding file link, using the two differences to jointly determine the adjustment value of the initially determined investigation period.
[0061] In implementation, the adjustment value = (expected number of visits - preset number of visits) ÷ preset number of visits × variable number of days for investigation, where the variable number of days for investigation is determined according to the definition of the access data. If the access data is defined as sample input, the variable number of days for investigation is 5 days; if the access data is defined as obtaining results, the variable number of days for investigation is 10 days. The adjusted investigation period = the investigation period determined initially - the adjustment value.
[0062] Specifically, the risk identification module determines the corresponding preset access times, preset single access duration, and number of checks within a single check cycle according to the actual access frequency of a single file link.
[0063] It is understandable that the preset number of accesses is positively correlated with the actual access frequency of the file link; the preset access duration is positively correlated with the actual access frequency of the file link; and the number of inspections within a single inspection cycle is positively correlated with the actual access frequency of the file link.
[0064] In implementation, the actual access frequency is divided into at least three intervals according to the actual access frequency of each file link, wherein the actual access frequency of the first interval is less than the actual access frequency of the second interval and less than the actual access frequency of the third interval, and the actual access frequency ranges of each interval are equal, that is, the difference between the maximum actual access frequency and the minimum actual access frequency in any interval is equal. In implementation, each interval is set with a corresponding preset number of accesses, preset access duration, and number of inspections in a single inspection cycle. Preferably, for example, it can be implemented by using a query table with preset values of actual access frequency and number of accesses, preset value of single access duration, and number of single inspections.
[0065] See also Figure 4 As shown, it is a workflow diagram of the risk assessment module of an embodiment of the present invention. The risk assessment module of the present invention uses at least one set of samples to determine the risk reference value of a single file link, including:
[0066] In a single file link, the risk assessment module determines the corresponding large model output result according to the single sample input content, and determines the risk reference value according to the deviation between the large model output result and the sample output content;
[0067] The sample includes sample input content and corresponding sample output content.
[0068] In implementation, machine learning is used to correspond the sample output content to the standard output value, and the output result of the large model is also corresponded to the actual output value. The risk reference value = (standard output value - actual output value) ÷ standard output value × 100%.
[0069] If the output content is a paragraph of text, the standard output value is set to 100, and the machine learning method is used to assign points to the output result to determine the actual output value.
[0070] If the output content is a picture, the total area of all elements in the standard image corresponding to the sample output content is determined to be 1, and then the area value (less than or equal to 1) that is the same as the element contained in the standard image is determined on the actual image corresponding to the output result, and then the risk reference value is determined.
[0071] Specifically, the risk assessment module determines whether to use the current sample content to train the large model according to the risk reference value and the number of learned samples of the current large model, and transmits the assessment result to the early warning decision module;
[0072] If the risk reference value is less than or equal to the preset risk reference value corresponding to the number of samples, the risk assessment module determines to use the content of the file link corresponding to the current sample content to train the large model;
[0073] If the risk reference value is greater than the preset risk reference value corresponding to the number of samples, the risk assessment module determines not to use the content of the file link corresponding to the current sample content to train the large model.
[0074] Specifically, the risk assessment module determines the preset risk reference value K0 according to the number of learned samples of the current large model and the actual accuracy of the large model.
[0075] In practice, K0 = 1-[T×(1-e -a )], where T is the actual accuracy of the large model and a is the number of learned samples. The actual accuracy of the large model can be calculated according to any model accuracy determination method in the prior art, which will not be described in detail here.
[0076] Specifically, the early warning decision module determines its decision according to the evaluation result, including:
[0077] When the evaluation result is to use the current sample content to train the large model, the early warning decision module readjusts the inspection cycle according to the actual accuracy change of the large model after using the sample and adjusts the number of inspections within a single cycle according to the actual accuracy change value of the large model after using the sample.
[0078] During implementation, if the actual accuracy of the large model after using the sample increases, the early warning decision module determines to increase the investigation cycle; if the actual accuracy of the large model after using the sample decreases, the early warning decision module determines to reduce the investigation cycle; actual accuracy change value = actual accuracy after using the sample - actual accuracy before using the sample, number of investigations after adjustment = number of investigations before adjustment - actual accuracy change value × number of investigations before adjustment.
[0079] Specifically, the early warning decision module determines its decision according to the evaluation result, including:
[0080] When the cumulative evaluation results of the adjacent preset number of times are that the current sample content is not used to train the large model, the early warning decision module determines that there is a risk of data destruction and issues an alarm.
[0081] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.
[0082] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A big data risk management and analysis system based on big models, It is characterized in that include: File data collection module, used to obtain access data of each file link; A risk identification module connected to the file data acquisition module, used to determine the data risk identification check cycle, the expected number of accesses and the expected value of a single access duration for each file link according to the access data, and adjust the check cycle based on the preset number of accesses and the preset single access duration. The risk identification module determines the number of checks within a single check cycle according to the actual access frequency of a single file link, so as to check the file links within the check cycle; The risk identification module determines the data risk identification check cycle for each file link according to the access data definition; Wherein, the access data is defined as the investigation period corresponding to the sample input is less than the investigation period corresponding to the acquisition result when the access data is defined as; The risk identification module compares the expected number of accesses with a preset number of accesses of the corresponding file link, and compares the expected value of a single access duration with a preset value of a single access duration of the corresponding file link, and adjusts the screening period of the corresponding file link based on the comparison result; If the expected number of visits is less than or equal to the preset number of visits, and the expected single visit duration is less than or equal to the preset single visit duration, the risk identification module determines not to adjust the investigation period, and sets the currently determined investigation period as the initially determined investigation period; If the expected value of the number of accesses is greater than the preset value of the number of accesses, and / or the expected value of the duration of a single access is greater than the preset value of the duration of a single access, the risk identification module adjusts the initially determined investigation period according to the difference between the expected value of the number of accesses and the preset value of the number of accesses of the corresponding file link and / or the difference between the expected value of the duration of a single access and the preset value of the duration of a single access of the corresponding file link; A risk assessment module, which is connected to the risk identification module and the file data acquisition module respectively, and is used to determine a risk reference value according to the deviation between the large model output result of the sample input content in a single file link and the corresponding sample output content, determine a preset risk reference value according to the number of learned samples of the current large model and its actual accuracy, and determine whether to use the current sample content based on the preset risk reference value and the risk reference value; The risk assessment module determines a risk reference value of a single file link using at least one set of samples, including: In a single file link, the risk assessment module determines the corresponding large model output result according to the single sample input content, and determines the risk reference value according to the deviation between the large model output result and the sample output content; An early warning decision module, which is connected to the risk assessment module and is used to determine its decision based on the assessment result, the decision including whether to adjust the investigation cycle and whether to adjust the number of investigations; The access data includes the number of access terminals, the number of visits of a single access terminal, the access duration of a single access terminal and the access data definition. The access data definition includes sample input and acquisition results. A single sample includes the sample input content and the sample output content.
2. According to the big data risk management and analysis system based on big model according to claim 1, It is characterized in that The risk identification module determines the expected number of accesses and the expected duration of a single access of each file link according to the number of access terminals, the number of accesses of a single access terminal and the duration of access of a single access terminal.
3. According to the big data risk management and analysis system based on big model as claimed in claim 2, It is characterized in that The risk identification module determines the corresponding preset value of the number of accesses, the preset value of the duration of a single access, and the number of inspections within a single inspection cycle according to the actual access frequency of a single file link.
4. According to the big data risk management and analysis system based on big model as claimed in claim 1, It is characterized in that The risk assessment module determines whether to use the current sample content to train the large model according to the risk reference value and the number of samples learned by the current large model, and transmits the assessment result to the early warning decision module; If the risk reference value is less than or equal to the preset risk reference value corresponding to the number of samples, the risk assessment module determines to use the content of the file link corresponding to the current sample content to train the large model; If the risk reference value is greater than the preset risk reference value corresponding to the number of samples, the risk assessment module determines not to use the content of the file link corresponding to the current sample content to train the large model.
5. According to the big data risk management and analysis system based on big model according to claim 1, It is characterized in that The early warning decision module determines its decision according to the evaluation result, including: When the evaluation result is to use the current sample content to train the large model, the early warning decision module adjusts the inspection cycle according to the actual accuracy change of the large model after using the sample and adjusts the number of inspections within a single cycle according to the actual accuracy change value of the large model after using the sample.
6. According to the big data risk management and analysis system based on big model according to claim 5, It is characterized in that The early warning decision module determines its decision according to the evaluation result, including: When the cumulative evaluation result of the adjacent preset times is that the current sample content is not used to train the large model, the early warning decision module determines that there is a risk of data destruction and issues an alarm.
