Content verification method and system based on LLM security protection

Abstract

The invention discloses a content verification method and system based on LLM security protection, and the method comprises the steps: determining to-be-examined data from all data corresponding to LLM security protection, and carrying out the preliminary examination operation of the to-be-examined data based on a predetermined examination mechanism, and obtaining a preliminary examination result; and according to a predetermined target analysis algorithm and a data review strategy, executing a data verification operation on the preliminary review result to obtain a data verification result corresponding to the to-be-reviewed data. It can be seen that the data of the LLM safety protection system can be intelligently verified, the accuracy and reliability of data verification can be improved, and the operation safety of the LLM safety protection system can be further improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data verification, and in particular, to a content verification method and system based on LLM security protection. Background Art

[0002] A large language model (LLM) is an artificial intelligence model based on deep learning technology, aiming to understand and generate human language. In recent years, the research and application of large language model (LLM) technology have been developing rapidly.

[0003] With the further development of LLM technology, the security products of LLM also need to be continuously upgraded and improved. Currently, most of the data verification methods for LLM technology are to verify LLM data through preset software testing methods. However, the current method mainly scans the program to be detected, and detects whether there is a security verification function according to a large number of rule sets and static analysis strategies to obtain the data verification result. There is a technical problem of low data verification accuracy. Therefore, it is particularly important to provide a new content verification method based on LLM security protection to improve the accuracy and reliability of data verification. Summary of the Invention

[0004] The present invention provides a content verification method and system based on LLM security protection, which can intelligently verify the data of the LLM protection system, is beneficial to improving the accuracy and reliability of data verification, and further beneficial to improving the operation security of the LLM security protection system.

[0005] To solve the above technical problems, in the first aspect of the present invention, a content verification method based on LLM security protection is disclosed, and the method includes: Determine the data to be reviewed from all the data corresponding to the LLM security protection, and perform a preliminary review operation on the data to be reviewed based on a pre-determined review mechanism to obtain a preliminary review result; According to a pre-determined target analysis algorithm and data review strategy, perform a data verification operation on the preliminary review result to obtain a data verification result corresponding to the data to be reviewed.

[0006] As an optional implementation manner, in the first aspect of the present invention, the method further includes: According to the data verification result, determine whether the data verification result meets a pre-determined protection verification condition; When it is determined that the data verification result does not meet the pre-determined protection verification condition, based on the data verification result and the protection verification condition, determine the target reason why the data verification result does not meet the protection verification condition; Optimize the data review strategy according to the target reason.

[0007] As an alternative implementation manner, in the first aspect of the present invention, before performing a data verification operation on the preliminary review result according to the pre-determined target analysis algorithm and data review strategy to obtain the data verification result corresponding to the data to be reviewed, the method further includes: Obtain comprehensive review information, where the comprehensive review information includes one or more of language review information, cultural review information, and background review information; Perform an information preprocessing operation on the comprehensive review information to obtain preprocessed review information; Determine whether the preprocessed review information meets a preset information review condition; When it is determined that the preprocessed review information meets the preset information review condition, generate a data review strategy based on the preprocessed review information and the information preprocessing operation.

[0008] As an alternative implementation manner, in the first aspect of the present invention, the method further includes: When it is determined that the preprocessed review information does not meet the preset information review condition, determine the operation parameters corresponding to the information preprocessing operation and the difference parameters between the preprocessed review information and the information review condition; Generate an operation adjustment parameter for the information preprocessing operation based on the operation parameters and the difference parameters, update the information preprocessing operation based on the operation adjustment parameter, and re-trigger the execution of the operation of performing an information preprocessing operation on the comprehensive review information to obtain preprocessed review information and the operation of triggering the determination of whether the preprocessed review information meets the preset information review condition.

[0009] As an alternative implementation manner, in the first aspect of the present invention, the optimizing the data review strategy according to the target reason includes: Determine the review effectiveness of the data review strategy according to the target reason, and determine whether the review effectiveness is greater than or equal to a preset review effectiveness threshold; When it is determined that the review effectiveness is greater than or equal to the preset review effectiveness threshold, determine at least one first optimization factor that matches the key verification factor in the pre-determined optimization parameter database based on the target reason, and perform an optimization operation on the data review strategy based on all the first optimization factors; When it is determined that the review effectiveness is less than the preset review effectiveness threshold, analyze the review performance indicators in the data review strategy based on the target reason to obtain an indicator analysis result, generate a reason to be optimized based on the indicator analysis result and the target reason, determine at least one second optimization factor that matches the key verification factor in the pre-determined optimization parameter database based on the reason to be optimized, and perform an optimization operation on the data review strategy based on all the second optimization factors.

[0010] As an optional implementation manner, in the first aspect of the present invention, the performing a preliminary review operation on the data to be reviewed based on a pre-determined review mechanism to obtain a preliminary review result includes: Generating a target review model based on a pre-determined review mechanism and all the data corresponding to the LLM security protection; Inputting the data to be reviewed into the target review model to obtain a model review result, where the model review result includes the review result after performing a review operation on the data to be reviewed through the review mechanism; Generating a preliminary review result corresponding to the data to be reviewed based on the model review result.

[0011] As an optional implementation manner, in the first aspect of the present invention, before performing an optimization operation on the data review strategy according to the target reason, the method further includes: Obtaining policy feedback information for the data review strategy and determining whether the policy feedback information meets a preset policy optimization condition; When it is determined that the policy feedback information meets the preset policy optimization condition, generating feedback optimization parameters for the data review strategy based on the policy feedback information; Wherein, the performing an optimization operation on the data review strategy according to the target reason includes: Performing an optimization operation on the data review strategy according to the target reason and the feedback optimization parameters.

[0012] The second aspect of the present invention discloses a content verification system based on LLM security protection, the system includes: A determination module, configured to determine data to be reviewed from all the data corresponding to the LLM security protection; A review module, configured to perform a preliminary review operation on the data to be reviewed based on a pre-determined review mechanism to obtain a preliminary review result; A verification module, configured to perform a data verification operation on the preliminary review result according to a pre-determined target analysis algorithm and a data review strategy to obtain a data verification result corresponding to the data to be reviewed.

[0013] As an alternative implementation, in the second aspect of the present invention, the system further includes: A judgment module, configured to judge whether the data verification result meets a pre-determined protection verification condition according to the data verification result; The determination module is further configured to, when the judgment module determines that the data verification result does not meet the pre-determined protection verification condition, determine a target reason why the data verification result does not meet the protection verification condition based on the data verification result and the protection verification condition; An optimization module, configured to perform an optimization operation on the data review strategy according to the target reason.

[0014] As an alternative implementation, in the second aspect of the present invention, the system further includes: An acquisition module, configured to acquire comprehensive review information before the verification module performs a data verification operation on the preliminary review result according to a pre-determined target analysis algorithm and a data review strategy to obtain a data verification result corresponding to the data to be reviewed, where the comprehensive review information includes one or more of language review information, cultural review information, and background review information; A processing module, configured to perform an information preprocessing operation on the comprehensive review information to obtain preprocessed review information; The judgment module is further configured to judge whether the preprocessed review information meets a preset information review condition; A generation module, configured to generate a data review strategy based on the preprocessed review information and the information preprocessing operation when the judgment module determines that the preprocessed review information meets the preset information review condition.

[0015] As an alternative implementation, in the second aspect of the present invention, the determination module is further configured to determine operation parameters corresponding to the information preprocessing operation and determine a difference parameter between the preprocessed review information and the information review condition when the judgment module determines that the preprocessed review information does not meet the preset information review condition; The generation module is further configured to generate an operation adjustment parameter for the information preprocessing operation based on the operation parameters and the difference parameter, update the information preprocessing operation based on the operation adjustment parameter, and re-trigger the processing module to perform the operation of performing an information preprocessing operation on the comprehensive review information to obtain preprocessed review information and trigger the judgment module to perform the operation of judging whether the preprocessed review information meets a preset information review condition.

[0016] As an alternative implementation, in the second aspect of the present invention, the specific manner in which the optimization module performs an optimization operation on the data review policy according to the target reason includes: Determine the review effectiveness of the data review policy according to the target reason, and determine whether the review effectiveness is greater than or equal to a preset review effectiveness threshold; When it is determined that the review effectiveness is greater than or equal to the preset review effectiveness threshold, determine at least one first optimization factor matching the key verification factor in the pre-determined optimization parameter database based on the target reason, and perform an optimization operation on the data review policy based on all the first optimization factors; When it is determined that the review effectiveness is less than the preset review effectiveness threshold, analyze the review performance indicators in the data review policy according to the target reason to obtain an index analysis result, generate a reason to be optimized based on the index analysis result and the target reason, determine at least one second optimization factor matching the key verification factor in the pre-determined optimization parameter database based on the reason to be optimized, and perform an optimization operation on the data review policy based on all the second optimization factors.

[0017] As an alternative implementation, in the second aspect of the present invention, the specific manner in which the review module performs a preliminary review operation on the data to be reviewed based on a pre-determined review mechanism to obtain a preliminary review result includes: Generate a target review model based on a pre-determined review mechanism and all the data corresponding to the LLM security protection; Input the data to be reviewed into the target review model to obtain a model review result, where the model review result includes the review result after performing a review operation on the data to be reviewed through the review mechanism; Generate a preliminary review result corresponding to the data to be reviewed based on the model review result.

[0018] As an alternative implementation, in the second aspect of the present invention, the acquisition module is further configured to obtain policy feedback information on the data review policy before the optimization module performs an optimization operation on the data review policy according to the target reason; The judgment module is further configured to judge whether the policy feedback information meets a preset policy optimization condition; The generation module is further configured to generate feedback optimization parameters for the data review policy based on the policy feedback information when the judgment module determines that the policy feedback information meets the preset policy optimization condition; Among them, the specific manner in which the optimization module performs an optimization operation on the data review policy according to the target reason includes: Performing an optimization operation on the data review policy according to the target reason and the feedback optimization parameters.

[0019] The third aspect of the present invention discloses another content verification system based on LLM security protection, and the system includes: A memory storing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory and executes the content verification method based on LLM security protection disclosed in the first aspect of the present invention.

[0020] The fourth aspect of the present invention discloses a computer storage medium, and the computer storage medium stores computer instructions, which are used to execute the content verification method based on LLM security protection disclosed in the first aspect of the present invention when called.

[0021] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: In the embodiments of the present invention, the data to be reviewed is determined from all the data corresponding to LLM security protection, and a preliminary review operation is performed on the data to be reviewed based on the pre-determined review mechanism to obtain a preliminary review result; according to the pre-determined target analysis algorithm and data review policy, a data verification operation is performed on the preliminary review result to obtain the data verification result corresponding to the data to be reviewed. It can be seen that implementing the present invention can intelligently verify the data of the LLM protection system, which is beneficial to improving the accuracy and reliability of data verification, and further beneficial to improving the operation security of the LLM security protection system. Description of the Drawings

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings without creative efforts based on these drawings.

[0023] Figure 1 It is a flowchart of a content verification method based on LLM security protection disclosed in the embodiments of the present invention; Figure 2 It is a flowchart of another content verification method based on LLM security protection disclosed in the embodiments of the present invention; Figure 3 It is a structural diagram of a content verification system based on LLM security protection disclosed in the embodiments of the present invention; Figure 4 It is a schematic structural diagram of another content verification system based on LLM security protection disclosed in an embodiment of the present invention; Figure 5 It is a schematic structural diagram of yet another content verification system based on LLM security protection disclosed in an embodiment of the present invention. Detailed implementation manners

[0024] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below based on the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without making creative efforts belong to the scope of protection of the present invention.

[0025] The terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or terminal that includes a series of steps or units is not limited to the listed steps or units, but optionally further includes steps or units not listed, or optionally further includes other steps or units inherent to these processes, methods, products or terminals.

[0026] Referring to "embodiment" herein means that a specific feature, structure or characteristic described based on the embodiment may be included in at least one embodiment of the present invention. The phrase appears in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art explicitly and implicitly understand that the embodiments described herein may be combined with other embodiments.

[0027] The present invention discloses a content verification method and system based on LLM security protection, which can intelligently verify the data of the LLM protection system, is beneficial to improving the accuracy and reliability of data verification, and further beneficial to improving the operation security of the LLM security protection system. The following will be described in detail respectively.

[0028] Embodiment 1 Please refer to Figure 1 , Figure 1 It is a schematic flowchart of a content verification method based on LLM security protection disclosed in an embodiment of the present invention. Among them, Figure 1The described content verification method based on LLM security protection can be applied to a content verification system based on LLM security protection. Among them, the content verification system based on LLM security protection can be integrated into a cloud server or a local server, which is not limited in the embodiments of the present invention. As Figure 1 shown, the content verification method based on LLM security protection may include the following operations: 101. Determine the data to be reviewed from all the data corresponding to LLM security protection, and perform a preliminary review operation on the data to be reviewed based on a pre-determined review mechanism to obtain a preliminary review result.

[0029] In the embodiments of the present invention, optionally, all the data corresponding to LLM security protection may include all the data included in the LLM security protection system. Further, all the data corresponding to LLM security protection may include one or more of input data, training data, output data, personal privacy data, plugin data, and code data.

[0030] In the embodiments of the present invention, optionally, the number of the data to be reviewed may be one, or multiple, or all the data corresponding to LLM security protection, which is not specifically limited in the embodiments of the present invention.

[0031] 102. According to a pre-determined target analysis algorithm and a data review strategy, perform a data verification operation on the preliminary review result to obtain a data verification result corresponding to the data to be reviewed.

[0032] In the embodiments of the present invention, optionally, the pre-determined target analysis algorithm may include one or more of a multi-dimensional data analysis algorithm and a machine learning algorithm.

[0033] In the embodiments of the present invention, optionally, the data verification result corresponding to the data to be reviewed may include data review results for indicating the compliance and rationality of the data to be reviewed.

[0034] It can be seen that the implementation Figure 1The described content verification method for LLM security protection can determine the data to be reviewed from all the data corresponding to LLM security protection and perform a preliminary review operation on the data to be reviewed based on a pre-determined review mechanism to obtain a preliminary review result. According to the target analysis algorithm and the data review strategy, a data verification operation is performed on the preliminary review result to obtain the data verification result corresponding to the data to be reviewed. It can effectively identify and filter out potential harmful content through the pre-determined review mechanism and data review strategy, thereby enhancing the security of the LLM. Through the preliminary review and data verification operations on the data to be reviewed, this method helps to improve the resistance of the LLM to malicious inputs, enhance the robustness of the model, while enhancing security, it can ensure a high-quality response to benign instructions, maintain the usability of the LLM, and according to different application scenarios and security requirements, flexibly adjust the review mechanism and data review strategy to adapt to different security protection needs. And by ensuring that the output of the LLM complies with laws, regulations and industry standards, this method helps to promote the compliance of the LLM, can intelligently verify the data of the LLM protection system, is beneficial to improving the accuracy and reliability of data verification, and further beneficial to improving the operating security of the LLM security protection system.

[0035] Embodiment 2 Please refer to Figure 2 , Figure 2 which is a schematic flowchart of a content verification method for LLM security protection disclosed in an embodiment of the present invention. Among them, Figure 2 The described content verification method for LLM security protection can be applied to a content verification system for LLM security protection. Among them, the content verification system for LLM security protection can be integrated in a cloud server or a local server, and the embodiments of the present invention do not make limitations. As Figure 2 shown, the content verification method for LLM security protection can include the following operations: 201. Determine the data to be reviewed from all the data corresponding to LLM security protection, and perform a preliminary review operation on the data to be reviewed based on a pre-determined review mechanism to obtain a preliminary review result.

[0036] 202. According to a pre-determined target analysis algorithm and data review strategy, perform a data verification operation on the preliminary review result to obtain the data verification result corresponding to the data to be reviewed.

[0037] In the embodiments of the present invention, for the detailed description of steps 201 - 202, please refer to the other descriptions of steps 101 - 102 in Embodiment 1, and the embodiments of the present invention will not be repeated.

[0038] 203. According to the data verification result, determine whether the data verification result meets a pre-determined protection verification condition.

[0039] In an embodiment of the present invention, optionally, determining whether the data verification result meets the pre-determined protection verification condition may include: Determine the data security level corresponding to the data verification result, and determine whether the data security level is greater than or equal to the security level threshold corresponding to the pre-determined protection verification condition; When it is determined that the data security level is greater than or equal to the security level threshold corresponding to the pre-determined protection verification condition, it is determined that the data verification result meets the pre-determined protection verification condition; when it is determined that the data security level is less than the security level threshold corresponding to the pre-determined protection verification condition, it is determined that the data verification result does not meet the pre-determined protection verification condition.

[0040] 204. When it is determined that the data verification result does not meet the pre-determined protection verification condition, based on the data verification result and the protection verification condition, determine the target reason why the data verification result does not meet the protection verification condition.

[0041] In an embodiment of the present invention, further optionally, when it is determined that the data verification result meets the pre-determined protection verification condition, this process may be ended.

[0042] In an embodiment of the present invention, optionally, the above-mentioned determining the target reason why the data verification result does not meet the protection verification condition based on the data verification result and the protection verification condition may include: Based on the data verification result and the protection verification condition, determine the difference information between the data verification result and the protection verification condition, and generate the target reason why the data verification result does not meet the protection verification condition according to the difference information.

[0043] 205. According to the target reason, perform an optimization operation on the data review strategy.

[0044] It can be seen that the implementation Figure 2The described content verification method for LLM security protection can determine whether the data verification result meets the pre-determined protection verification conditions. If not, it determines the target reason why the data verification result does not meet the protection verification conditions based on the data verification result and the protection verification conditions, and performs an optimization operation on the data review strategy according to the target reason. By judging whether the data verification result meets the pre-determined protection verification conditions, it can accurately identify which data fails to pass the security review, thus improving the accuracy of security protection. After determining that the data verification result does not meet the protection verification conditions, it can determine the target reason for non-compliance based on the verification result and the protection conditions, and optimize the data review strategy according to the target reason, which can achieve dynamic adjustment to ensure that the review strategy can adapt to changing security threats and data environments, improve the overall protection ability. By optimizing the data review strategy, unnecessary reviews and processing can be reduced, the overall efficiency of the system can be improved, and the reasonable utilization of resources can be ensured. It can also enhance users' trust in the system and improve user satisfaction through effective security protection measures and continuous optimization processes. It can intelligently verify the data of the LLM protection system, which is beneficial to improving the accuracy and reliability of data verification, and further beneficial to improving the operating security of the LLM security protection system.

[0045] In an optional embodiment, before performing a data verification operation on the preliminary review result according to the pre-determined target analysis algorithm and data review strategy to obtain the data verification result corresponding to the data to be reviewed, the method further includes: Obtain comprehensive review information, where the comprehensive review information includes one or more of language review information, cultural review information, and background review information; Perform an information preprocessing operation on the comprehensive review information to obtain preprocessed review information; Judge whether the preprocessed review information meets the preset information review conditions; When it is determined that the preprocessed review information meets the preset information review conditions, generate a data review strategy based on the preprocessed review information and the information preprocessing operation.

[0046] In this optional embodiment, optionally, the comprehensive review information can be obtained in real time or at regular intervals according to a preset time period, and the embodiments of the present invention do not make specific limitations.

[0047] In this optional embodiment, optionally, the language review information can include the language information corresponding to the review strategy. For example, it can include languages such as Chinese and English; the cultural review information includes the normative information of the cultural group corresponding to the review strategy; the background review information can include the normative information corresponding to the social background.

[0048] In this optional embodiment, optionally, the information preprocessing operation may include one or more of sensitive information detection operation, sensitive information marking operation, sensitive information deletion operation, and sensitive information replacement operation; for example, through the information preprocessing operation, content that is biased, discriminatory, or otherwise unethical can be automatically detected and marked, ensuring the ethical compliance of the output.

[0049] In this optional embodiment, optionally, the above determination of whether the preprocessing review information meets the preset information review conditions may include: Determine whether there is sensitive information related to ethical norms corresponding to the preset information review conditions in the preprocessing review information; When it is determined that there is no sensitive information related to ethical norms corresponding to the preset information review conditions in the preprocessing review information, it is determined that the preprocessing review information meets the preset information review conditions; when it is determined that there is sensitive information related to ethical norms corresponding to the preset information review conditions in the preprocessing review information, it is determined that the preprocessing review information does not meet the preset information review conditions.

[0050] In this optional embodiment, optionally, the above generation of a data review strategy based on the preprocessing review information and the information preprocessing operation may include: Determine a first preprocessing parameter according to the information preprocessing operation, and generate a data review strategy based on the preprocessing review information and the first preprocessing parameter; wherein, the data review strategy at least includes the preprocessing review information and the first preprocessing parameter.

[0051] It can be seen that implementing this optional embodiment can obtain comprehensive review information and perform information preprocessing operations on the comprehensive review information to obtain preprocessed review information, and determine whether the preprocessed review information meets the preset information review conditions. If it meets the conditions, a data review strategy is generated based on the preprocessed review information and the information preprocessing operations. By obtaining comprehensive review information including language, culture, and background, various aspects of the review can be more comprehensively covered, improving the meticulousness and accuracy of the review. And through the operation of preprocessing review information, the interference of redundant and irrelevant information can be reduced, making the review process more efficient and accelerating the review speed. Through the preset information review conditions, it can be ensured that only the preprocessed review information that meets specific standards is used to generate the data review strategy, thereby improving the accuracy of the review results. The data review strategy generated based on the preprocessed review information that meets the preset conditions can better adapt to the review requirements of different languages, cultures, and backgrounds, improving the applicability and flexibility of the strategy. By comprehensively considering various aspects of review information and performing preprocessing and condition judgment, a more accurate, efficient, and adaptable data review strategy can be generated, thereby improving the overall review effect. It can intelligently verify the data of the LLM protection system, which is beneficial to improving the accuracy and reliability of data verification, and further beneficial to improving the operation security of the LLM security protection system.

[0052] In another optional embodiment, the method further includes: When it is determined that the preprocessed review information does not meet the preset information review conditions, determine the operation parameters corresponding to the information preprocessing operations, and determine the difference parameters between the preprocessed review information and the information review conditions; Based on the operation parameters and the difference parameters, generate operation adjustment parameters for the information preprocessing operations, and update the information preprocessing operations based on the operation adjustment parameters, and re-trigger the operation of performing information preprocessing operations on the comprehensive review information to obtain the preprocessed review information, and trigger the operation of determining whether the preprocessed review information meets the preset information review conditions.

[0053] In this optional embodiment, optionally, the difference parameters between the preprocessed review information and the information review conditions may include the difference parameters between the preprocessed review information and the reference review information corresponding to the information review conditions.

[0054] In this optional embodiment, optionally, the above-mentioned generating operation adjustment parameters for the information preprocessing operations based on the operation parameters and the difference parameters may include: Based on the operation parameters and the difference parameters, determine the operation difference parameters between the operation parameters and the difference parameters, and based on the operation difference parameters, determine the operation adjustment parameters for the information preprocessing operations that match the operation difference parameters in the pre-determined adjustment parameter library.

[0055] It can be seen that implementing this optional embodiment can determine the operation parameters corresponding to the information preprocessing operation and the difference parameters between the preprocessing review information and the information review conditions when it is determined that the preprocessing review information does not meet the information review conditions, and generate subordinate operation adjustment parameters for the information preprocessing operation based on the operation parameters and the difference parameters, update the information processing operation, and re-determine whether the preprocessing review information meets the preset information review conditions. By determining the operation parameters corresponding to the information preprocessing operation and the difference parameters between the preprocessing review information and the information review conditions, the preprocessing operation can be more accurately identified and adjusted, thereby improving the accuracy and precision of the review. The operation adjustment parameters generated based on the operation parameters and the difference parameters enhance the adaptability and flexibility of the system. By re-triggering the information preprocessing operation, invalid or inefficient data processing can be reduced, resource utilization can be optimized, and the review efficiency can be improved. It is also possible to self-optimize the preprocessing operation based on the difference parameters, achieve continuous performance improvement and adapt to new review requirements, and can intelligently verify the data of the LLM protection system, which is beneficial to improving the accuracy and reliability of data verification, and further beneficial to improving the operation security of the LLM security protection system.

[0056] In another optional embodiment, an optimization operation is performed on the data review policy according to the target reason, including: According to the target reason, determine the review effectiveness of the data review policy, and determine whether the review effectiveness is greater than or equal to a preset review effectiveness threshold; When it is determined that the review effectiveness is greater than or equal to the preset review effectiveness threshold, determine at least one first optimization factor that matches the key verification factor in the pre-determined optimization parameter database based on the target reason, and perform an optimization operation on the data review policy based on all the first optimization factors; When it is determined that the review effectiveness is less than the preset review effectiveness threshold, analyze the review performance indicators in the data review policy according to the target reason to obtain an index analysis result, generate a reason to be optimized based on the index analysis result and the target reason, determine at least one second optimization factor that matches the key verification factor in the pre-determined optimization parameter database based on the reason to be optimized, and perform an optimization operation on the data review policy based on all the second optimization factors.

[0057] In this optional embodiment, optionally, the review effectiveness of the data review policy may include the effectiveness of reviewing the data to be reviewed through the data review policy.

[0058] In this optional embodiment, optionally, the key verification factor includes core evaluation indicators or key elements in the data review policy for measuring and verifying the review effect, accuracy, or reliability.

[0059] In this alternative embodiment, optionally, determining at least one first optimization factor that matches the key verification factor in the pre-determined optimization parameter database based on the target reason may include: Extract the first reason keyword of the target reason, and determine the optimization factor that matches the first reason keyword in the pre-determined optimization parameter database based on the first reason keyword, and determine all the optimization factors that match the first reason keyword as the first optimization factor that matches the key verification factor.

[0060] In this alternative embodiment, optionally, analyzing the review performance indicators in the data review policy based on the target reason to obtain an indicator analysis result, and generating a reason to be optimized based on the indicator analysis result and the target reason may include: Extract the second reason keyword of the target reason, and extract the target review performance indicator in the data review policy. Generate an indicator analysis model based on the target review performance indicator, input the second reason keyword into the indicator analysis model to obtain an indicator analysis result, and generate a reason to be optimized based on the indicator analysis result and the second reason keyword.

[0061] It can be seen that implementing this alternative embodiment can determine the review effectiveness of the data review policy according to the target reason and judge whether it is greater than or equal to the preset review effectiveness threshold. If so, determine the first optimization factor in the optimization parameter database based on the target reason and then perform an optimization operation on the data review policy. If not, analyze the review performance indicators in the data review policy based on the target reason to obtain an indicator analysis result, and generate a reason to be optimized based on the indicator analysis result and the target reason. Determine the second optimization factor that matches the key verification factor in the optimization parameter database based on the reason to be optimized and then perform an optimization operation on the data review policy. By judging the review effectiveness and selecting the optimization factor accordingly, the effect of the review policy can be improved, ensuring the accuracy and reliability of the review result. And through precise optimization operations, the occurrence of false positives and false negatives can be reduced, improving the accuracy of the review. Through continuous optimization operations, the review policy can be continuously improved to adapt to new review challenges. Moreover, the optimized review policy can better comply with laws, regulations and industry standards, reducing compliance risks. The dynamic optimization based on the target reason not only improves the effectiveness and adaptability of the review policy, but also enhances the automation level and self-improving ability of the system, enabling intelligent verification of the data of the LLM protection system, which is beneficial to improving the accuracy and reliability of data verification, and further beneficial to improving the operational security of the LLM security protection system.

[0062] In yet another alternative embodiment, performing a preliminary review operation on the data to be reviewed based on the pre-determined review mechanism to obtain a preliminary review result includes: Generate a target review model based on a pre-determined review mechanism and all data corresponding to LLM security protection; Input the data to be reviewed into the target review model to obtain a model review result, where the model review result includes the review result after performing a review operation on the data to be reviewed through the review mechanism; Generate a preliminary review result corresponding to the data to be reviewed based on the model review result.

[0063] In this optional embodiment, optionally, the generating a target review model based on a pre-determined review mechanism and all data corresponding to LLM security protection may include: Input all data corresponding to a pre-determined review mechanism and LLM security protection into a pre-determined alternative review model to obtain a model output result; Determine whether the model output result meets a preset model operation condition; When it is determined that the model output result meets the preset model operation condition, determine the alternative review model as the target review model; When it is determined that the model output result does not meet the preset model operation condition, analyze the result difference value between the model output result and the preset model operation condition, and perform an update operation on the alternative review model based on the result difference value, and re-trigger the operation of determining whether the model output result meets the preset model operation condition.

[0064] In this optional embodiment, optionally, the preliminary review result corresponding to the data to be reviewed at least includes the model review result.

[0065] It can be seen that implementing this optional embodiment can generate a target review model based on a pre-determined review mechanism and all data corresponding to LLM security protection, input the data to be reviewed into the target review model to obtain a model review result, and generate a preliminary review result corresponding to the data to be reviewed based on the model review result. By generating a target review model and inputting the data to be reviewed, the model review result can be obtained automatically, which not only improves the review efficiency, but also improves the review accuracy by utilizing the capabilities of the LLM. And by combining all data corresponding to LLM security protection, potential security threats can be better identified and prevented, enhancing the overall security. By analyzing the model review result, the review mechanism and model can be continuously optimized to support the continuous improvement and update of the review strategy, and can intelligently verify the data of the LLM protection system, which is beneficial to improving the accuracy and reliability of data verification, and further beneficial to improving the operation security of the LLM security protection system.

[0066] In another optional embodiment, before performing an optimization operation on the data review strategy according to the target reason, the method further includes: Obtain policy feedback information for the data review policy, and determine whether the policy feedback information meets the preset policy optimization conditions; When it is determined that the policy feedback information meets the preset policy optimization conditions, generate feedback optimization parameters for the data review policy based on the policy feedback information; Among them, according to the target reason, perform an optimization operation on the data review policy, including: Perform an optimization operation on the data review policy according to the target reason and the feedback optimization parameters.

[0067] In this optional embodiment, optionally, the policy feedback information may include usage feedback information for the data review policy.

[0068] In this optional embodiment, optionally, the above determination of whether the policy feedback information meets the preset policy optimization conditions may include: Determine the feedback satisfaction corresponding to the policy feedback information, and determine whether the feedback satisfaction is greater than or equal to the feedback satisfaction threshold corresponding to the preset policy optimization conditions; When it is determined that the feedback satisfaction is greater than or equal to the feedback satisfaction threshold corresponding to the preset policy optimization conditions, determine that the policy feedback information meets the preset policy optimization conditions; when it is determined that the feedback satisfaction is less than the feedback satisfaction threshold corresponding to the preset policy optimization conditions, determine that the policy feedback information does not meet the preset policy optimization conditions.

[0069] In this optional embodiment, further optionally, when it is determined that the policy feedback information does not meet the preset policy optimization conditions, this process may be ended.

[0070] In this optional embodiment, optionally, the feedback optimization parameters for the data review policy match the policy feedback information.

[0071] In this optional embodiment, optionally, the above performing an optimization operation on the data review policy according to the target reason and the feedback optimization parameters may include: Extract the first optimization keyword in the target reason, and extract the second optimization keyword in the feedback optimization parameters, and generate a target optimization keyword based on the first optimization keyword and the second optimization keyword; Based on the target optimization keyword, determine the target optimization parameter, and perform an optimization operation on the data review policy based on the target optimization parameter.

[0072] It can be seen that implementing this optional embodiment can obtain policy feedback information for the data review policy and determine whether the preset policy optimization conditions are met. If so, feedback optimization parameters for the data review policy are generated based on the policy feedback information, and an optimization operation is performed on the data review policy according to the target reason and the feedback optimization parameters. By collecting policy feedback information and determining whether it meets the optimization conditions, it can ensure that only valid feedback is used in the optimization process, thereby improving the effectiveness of the data review policy. Based on real-time or periodic policy feedback information, the data review policy can be dynamically adjusted to adapt to changing data review requirements and environments. The optimized policy can allocate review resources more effectively, reduce resource waste in low-risk areas, and concentrate resources on handling high-risk areas. The optimized review policy can better comply with laws, regulations, and industry standards, reducing compliance risks. By combining policy feedback information and target reasons, not only the effectiveness and adaptability of the review policy are improved, but also the automation level and self-improving ability of the system are enhanced. It can intelligently verify the data of the LLM protection system, which is beneficial to improving the accuracy and reliability of data verification, and further beneficial to improving the operation security of the LLM security protection system.

[0073] Embodiment III Please refer to Figure 3 , Figure 3 which is a schematic structural diagram of a content verification system based on LLM security protection disclosed in an embodiment of the present invention. As Figure 3 shown, the content verification system based on LLM security protection may include: A determination module 301, configured to determine the data to be reviewed from all the data corresponding to the LLM security protection; A review module 302, configured to perform a preliminary review operation on the data to be reviewed based on a pre-determined review mechanism to obtain a preliminary review result; A verification module 303, configured to perform a data verification operation on the preliminary review result according to a pre-determined target analysis algorithm and a data review policy to obtain a data verification result corresponding to the data to be reviewed.

[0074] It can be seen that implementing Figure 3The described system can determine the data to be reviewed from all the data corresponding to the LLM security protection and perform a preliminary review operation on the data to be reviewed based on a pre-determined review mechanism to obtain a preliminary review result. It performs a data verification operation on the preliminary review result according to the target analysis algorithm and the data review strategy to obtain the data verification result corresponding to the data to be reviewed. It can effectively identify and filter out potential harmful content through the pre-determined review mechanism and data review strategy, thereby enhancing the security of the LLM. Through the preliminary review and data verification operations on the data to be reviewed, this method helps to improve the resistance of the LLM to malicious inputs and enhance the robustness of the model. While enhancing security, it can ensure a high-quality response to benign instructions, maintain the usability of the LLM, and flexibly adjust the review mechanism and data review strategy according to different application scenarios and security requirements to adapt to different security protection needs. And by ensuring that the output of the LLM complies with laws, regulations, and industry standards, this method helps to promote the compliance of the LLM. It can intelligently verify the data of the LLM protection system, which is beneficial to improving the accuracy and reliability of data verification, and further beneficial to improving the operational security of the LLM security protection system.

[0075] In an alternative embodiment, as Figure 4 shown, the system further includes: A judgment module 304, configured to judge whether the data verification result meets the pre-determined protection verification conditions according to the data verification result; A determination module 301 is further configured to, when the judgment module 304 determines that the data verification result does not meet the pre-determined protection verification conditions, determine the target reason why the data verification result does not meet the protection verification conditions based on the data verification result and the protection verification conditions; An optimization module 305, configured to perform an optimization operation on the data review strategy according to the target reason.

[0076] It can be seen that implementing Figure 4The described system can determine whether the data verification result meets the pre-determined protection verification conditions. If it does not meet, it determines the target reason why the data verification result does not meet the protection verification conditions based on the data verification result and the protection verification conditions, and performs an optimization operation on the data review strategy according to the target reason. By judging whether the data verification result meets the pre-determined protection verification conditions, it can accurately identify which data fails to pass the security review, thereby improving the accuracy of security protection. After determining that the data verification result does not meet the protection verification conditions, it can determine the target reason for non-compliance based on the verification result and the protection conditions, and perform an optimization operation on the data review strategy, which can achieve dynamic adjustment to ensure that the review strategy can adapt to the changing security threats and data environment, improve the overall protection ability. By optimizing the data review strategy, unnecessary reviews and processing can be reduced, the overall efficiency of the system can be improved, and the reasonable utilization of resources can be ensured. It can also enhance users' trust in the system and improve user satisfaction through effective security protection measures and continuous optimization processes. It can intelligently verify the data of the LLM protection system, which is beneficial to improving the accuracy and reliability of data verification, and further beneficial to improving the operating security of the LLM security protection system.

[0077] In another alternative embodiment, as Figure 4 shown, the system further includes: An acquisition module 306, configured to acquire comprehensive review information before the verification module 303 performs a data verification operation on the preliminary review result according to the pre-determined target analysis algorithm and the data review strategy to obtain the data verification result corresponding to the data to be reviewed, where the comprehensive review information includes one or more of language review information, cultural review information, and background review information; A processing module 307, configured to perform an information preprocessing operation on the comprehensive review information to obtain preprocessed review information; A judgment module 304, further configured to judge whether the preprocessed review information meets the preset information review conditions; A generation module 308, configured to generate a data review strategy based on the preprocessed review information and the information preprocessing operation when the judgment module 304 determines that the preprocessed review information meets the preset information review conditions.

[0078] It can be seen that implementing Figure 4The described system can obtain comprehensive review information and perform information preprocessing operations on the comprehensive review information to obtain preprocessed review information, and determine whether the preprocessed review information meets the preset information review conditions. If it meets the conditions, a data review strategy is generated based on the preprocessed review information and the information preprocessing operations. By obtaining comprehensive review information including language, culture, and background, it can cover all aspects of the review more comprehensively, improve the meticulousness and accuracy of the review, and reduce the interference of redundant and irrelevant information through the operation of preprocessing review information, making the review process more efficient and accelerating the review speed. Through the preset information review conditions, it can ensure that only the preprocessed review information that meets specific standards is used to generate the data review strategy, thereby improving the accuracy of the review results. The data review strategy generated based on the preprocessed review information that meets the preset conditions can better adapt to the review requirements of different languages, cultures, and backgrounds, improving the applicability and flexibility of the strategy. By comprehensively considering various aspects of review information and performing preprocessing and condition judgment, a more accurate, efficient, and adaptable data review strategy can be generated, thereby improving the overall review effect. It can intelligently verify the data of the LLM protection system, which is beneficial to improving the accuracy and reliability of data verification, and further beneficial to improving the operation security of the LLM security protection system.

[0079] In yet another alternative embodiment, as Figure 4 shown, the determination module 301 is further configured to, when the judgment module 304 determines that the preprocessed review information does not meet the preset information review conditions, determine the operation parameters corresponding to the information preprocessing operation, and determine the difference parameters between the preprocessed review information and the information review conditions; The generation module 308 is further configured to generate an operation adjustment parameter for the information preprocessing operation based on the operation parameters and the difference parameters, update the information preprocessing operation based on the operation adjustment parameter, and re-trigger the operation of the processing module 307 to perform the information preprocessing operation on the comprehensive review information to obtain the preprocessed review information and trigger the operation of the judgment module 304 to determine whether the preprocessed review information meets the preset information review conditions.

[0080] It can be seen that implementing Figure 4The described system can determine the operation parameters corresponding to the information preprocessing operation and the difference parameters between the preprocessing review information and the information review conditions when it is determined that the preprocessing review information does not meet the information review conditions. Based on the operation parameters and the difference parameters, it generates subordinate operation adjustment parameters for the information preprocessing operation, updates the information processing operation, and re-determines whether the preprocessing review information meets the preset information review conditions. By determining the operation parameters corresponding to the information preprocessing operation and the difference parameters between the preprocessing review information and the information review conditions, it can more accurately identify and adjust the preprocessing operation, thereby improving the accuracy and precision of the review. The operation adjustment parameters generated based on the operation parameters and the difference parameters enhance the adaptability and flexibility of the system. And by re-triggering the information preprocessing operation, it can reduce invalid or inefficient data processing, optimize the utilization of resources, and improve the review efficiency. It can also self-optimize the preprocessing operation based on the difference parameters to achieve continuous performance improvement and adaptation to new review requirements. It can intelligently verify the data of the LLM protection system, which is beneficial to improving the accuracy and reliability of data verification, and further beneficial to improving the operation security of the LLM security protection system.

[0081] In yet another alternative embodiment, as Figure 4 shown, the specific manner in which the optimization module 305 performs an optimization operation on the data review policy according to the target reason includes: According to the target reason, determine the review effectiveness of the data review policy, and determine whether the review effectiveness is greater than or equal to a preset review effectiveness threshold; When it is determined that the review effectiveness is greater than or equal to the preset review effectiveness threshold, based on the target reason, determine at least one first optimization factor that matches the key verification factor in the pre-determined optimization parameter database, and perform an optimization operation on the data review policy based on all the first optimization factors; When it is determined that the review effectiveness is less than the preset review effectiveness threshold, analyze the review performance indicators in the data review policy based on the target reason to obtain an index analysis result, and generate a reason to be optimized based on the index analysis result and the target reason. Based on the reason to be optimized, determine at least one second optimization factor that matches the key verification factor in the pre-determined optimization parameter database, and perform an optimization operation on the data review policy based on all the second optimization factors.

[0082] It can be seen that implementing Figure 4The described system can determine the review effectiveness of the data review strategy based on the target reason and judge whether it is greater than or equal to the preset review effectiveness threshold. If so, it determines the first optimization factor in the optimization parameter database based on the target reason and then performs an optimization operation on the data review strategy. If not, it analyzes the review performance indicators in the data review strategy based on the target reason to obtain an indicator analysis result, and generates a reason to be optimized based on the indicator analysis result and the target reason. Then, it determines the second optimization factor matching the key verification factor in the optimization parameter database based on the reason to be optimized and performs an optimization operation on the data review strategy. By judging the review effectiveness and selecting the optimization factor accordingly, the effect of the review strategy can be improved, ensuring the accuracy and reliability of the review results. And through precise optimization operations, the occurrence of false positives and false negatives can be reduced, improving the accuracy of the review. Through continuous optimization operations, the review strategy can be continuously improved to adapt to new review challenges. Moreover, based on the optimized review strategy, it can better comply with laws, regulations and industry standards, reducing compliance risks. The dynamic optimization based on the target reason not only improves the effectiveness and adaptability of the review strategy, but also enhances the automation level and self-improving ability of the system. It can intelligently verify the data of the LLM protection system, which is beneficial to improving the accuracy and reliability of data verification, and further beneficial to improving the operation safety of the LLM security protection system.

[0083] In another alternative embodiment, as Figure 4 shown, the specific way for the review module 302 to perform a preliminary review operation on the data to be reviewed based on the pre-determined review mechanism and obtain a preliminary review result includes: Generating a target review model based on the pre-determined review mechanism and all the data corresponding to the LLM security protection; Inputting the data to be reviewed into the target review model to obtain a model review result, where the model review result includes the review result after performing a review operation on the data to be reviewed through the review mechanism; Generating a preliminary review result corresponding to the data to be reviewed based on the model review result.

[0084] It can be seen that implementing Figure 4The described system can generate a target review model based on a pre-determined review mechanism and all the data corresponding to the LLM security protection, input the data to be reviewed into the target review model to obtain the model review result, and generate a preliminary review result corresponding to the data to be reviewed based on the model review result. By generating the target review model and inputting the data to be reviewed, the model review result can be obtained automatically, which not only improves the review efficiency but also enhances the review accuracy by leveraging the capabilities of the LLM. Moreover, by combining all the data corresponding to the LLM security protection, potential security threats can be better identified and prevented, enhancing the overall security. By analyzing the model review result, the review mechanism and model can be continuously optimized to support the continuous improvement and update of the review strategy. It can intelligently verify the data of the LLM protection system, which is beneficial to improving the accuracy and reliability of data verification and further beneficial to improving the operational security of the LLM security protection system.

[0085] In yet another alternative embodiment, as Figure 4 shown, the acquisition module 306 is further configured to acquire the policy feedback information for the data review policy before the optimization module 305 performs the optimization operation on the data review policy according to the target reason; The judgment module 304 is further configured to judge whether the policy feedback information meets the preset policy optimization conditions; The generation module 308 is further configured to generate the feedback optimization parameters for the data review policy based on the policy feedback information when the judgment module 304 determines that the policy feedback information meets the preset policy optimization conditions; Among them, the specific manner in which the optimization module 305 performs the optimization operation on the data review policy according to the target reason includes: Performing an optimization operation on the data review policy according to the target reason and the feedback optimization parameters.

[0086] It can be seen that implementing Figure 4The described system can obtain policy feedback information for the data review policy and determine whether it meets the preset policy optimization conditions. If it meets, it generates feedback optimization parameters for the data review policy based on the policy feedback information, and performs an optimization operation on the data review policy according to the target reason and the feedback optimization parameters. By collecting policy feedback information and judging whether it meets the optimization conditions, it can ensure that only effective feedback is used in the optimization process, thereby improving the effectiveness of the data review policy. Based on real-time or periodic policy feedback information, the data review policy can be dynamically adjusted to adapt to the changing data review requirements and environment. The optimized policy can allocate review resources more effectively, reduce resource waste in low-risk areas, and concentrate resources on handling high-risk areas. The optimized review policy can better comply with laws, regulations, and industry standards, reducing compliance risks. By combining policy feedback information and the target reason, it not only improves the effectiveness and adaptability of the review policy, but also enhances the automation level and self-improving ability of the system. It can intelligently verify the data of the LLM protection system, which is beneficial to improving the accuracy and reliability of data verification, and further beneficial to improving the operation security of the LLM security protection system.

[0087] Embodiment 4 Please refer to Figure 5 , Figure 5 which is a schematic structural diagram of another content verification system based on LLM security protection disclosed in the embodiments of the present invention. As Figure 5 shown, the content verification system based on LLM security protection may include: A memory 401 storing executable program code; A processor 402 coupled to the memory 401; The processor 402 calls the executable program code stored in the memory 401 and executes the steps in the content verification method based on LLM security protection described in Embodiment 1 or Embodiment 2 of the present invention.

[0088] Embodiment 5 The embodiments of the present invention disclose a computer storage medium storing computer instructions, which are used to execute the steps in the content verification method based on LLM security protection described in Embodiment 1 or Embodiment 2 of the present invention when the computer instructions are called.

[0089] Embodiment 6 The embodiments of the present invention disclose a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to execute the steps in the content verification method based on LLM security protection described in Embodiment 1 or Embodiment 2.

[0090] The system embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separated. The components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed to multiple network modules. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative work.

[0091] Through the specific descriptions of the above embodiments, those skilled in the art can clearly understand that each implementation can be achieved by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, and the storage medium includes Read-Only Memory (ROM), Random Access Memory (RAM), Programmable Read-only Memory (PROM), Erasable Programmable Read Only Memory (EPROM), One-time Programmable Read-Only Memory (OTPROM), Electrically-Erasable Programmable Read-Only Memory (EEPROM), Compact Disc Read-Only Memory (CD-ROM) or other optical disc memories, magnetic disk memories, tape memories, or any other computer-readable medium capable of carrying or storing data.

[0092] Finally, it should be noted that: The content verification method and system based on LLM security protection disclosed in the embodiments of the present invention only disclose the preferred embodiments of the present invention, and are only used to illustrate the technical solutions of the present invention, rather than limiting them; Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; And these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A content verification method based on LLM security protection, characterized in that, The method includes: Determine the data to be reviewed from all the data corresponding to LLM security protection, and perform a preliminary review operation on the data to be reviewed based on a pre-determined review mechanism to obtain a preliminary review result; According to a pre-determined target analysis algorithm and a data review strategy, perform a data verification operation on the preliminary review result to obtain a data verification result corresponding to the data to be reviewed.

2. The content verification method based on LLM security protection according to claim 1, wherein, The method further includes: According to the data verification result, determine whether the data verification result meets a pre-determined protection verification condition; When it is determined that the data verification result does not meet the pre-determined protection verification condition, based on the data verification result and the protection verification condition, determine the target reason why the data verification result does not meet the protection verification condition; According to the target reason, perform an optimization operation on the data review strategy.

3. The content verification method based on LLM security protection according to claim 2, wherein Before performing the data verification operation on the preliminary review result according to a pre-determined target analysis algorithm and a data review strategy to obtain a data verification result corresponding to the data to be reviewed, the method further includes: Obtain comprehensive review information, where the comprehensive review information includes one or more of language review information, cultural review information, and background review information; Perform an information preprocessing operation on the comprehensive review information to obtain preprocessed review information; Determine whether the preprocessed review information meets a preset information review condition; When it is determined that the preprocessed review information meets the preset information review condition, generate a data review strategy based on the preprocessed review information and the information preprocessing operation.

4. The content verification method based on LLM security protection according to claim 3, wherein The method further includes: When it is determined that the preprocessed review information does not meet the preset information review condition, determine the operation parameters corresponding to the information preprocessing operation and the difference parameters between the preprocessed review information and the information review condition; Based on the operation parameters and the difference parameters, generate an operation adjustment parameter for the information preprocessing operation, update the information preprocessing operation based on the operation adjustment parameter, and re-trigger the operation of performing the information preprocessing operation on the comprehensive review information to obtain preprocessed review information and the operation of triggering the determination of whether the preprocessed review information meets the preset information review condition.

5. The content verification method based on LLM security protection according to claim 2, wherein, The performing an optimization operation on the data review strategy according to the target reason includes: According to the target reason, determine the review effectiveness of the data review strategy, and determine whether the review effectiveness is greater than or equal to a preset review effectiveness threshold; When it is determined that the review effectiveness is greater than or equal to the preset review effectiveness threshold, based on the target reason, determine at least one first optimization factor that matches the key verification factor in a pre-determined optimization parameter database, and perform an optimization operation on the data review strategy based on all the first optimization factors; When it is determined that the review effectiveness is less than the preset review effectiveness threshold, analyze the review performance indicators in the data review strategy based on the target reason to obtain an indicator analysis result, generate a reason to be optimized based on the indicator analysis result and the target reason, determine at least one second optimization factor matching the key verification factor in the pre-determined optimization parameter database based on the reason to be optimized, and perform an optimization operation on the data review strategy based on all the second optimization factors.

6. The content verification method based on LLM security protection according to claim 1, characterized in that, Performing a preliminary review operation on the data to be reviewed based on the pre-determined review mechanism to obtain a preliminary review result, including: Generating a target review model based on the pre-determined review mechanism and all the data corresponding to the LLM security protection; Inputting the data to be reviewed into the target review model to obtain a model review result, where the model review result includes the review result after performing a review operation on the data to be reviewed through the review mechanism; Generating a preliminary review result corresponding to the data to be reviewed based on the model review result.

7. The content verification method based on LLM security protection according to claim 3, wherein, Before performing the optimization operation on the data review strategy according to the target reason, the method further includes: Obtaining policy feedback information for the data review strategy and determining whether the policy feedback information meets the preset policy optimization conditions; When it is determined that the policy feedback information meets the preset policy optimization conditions, generating feedback optimization parameters for the data review strategy based on the policy feedback information; Wherein, performing the optimization operation on the data review strategy according to the target reason includes: Performing an optimization operation on the data review strategy according to the target reason and the feedback optimization parameters.

8. A content verification system based on LLM security protection, characterized in that, The system includes: A determination module for determining the data to be reviewed from all the data corresponding to the LLM security protection; A review module for performing a preliminary review operation on the data to be reviewed based on the pre-determined review mechanism to obtain a preliminary review result; A verification module for performing a data verification operation on the preliminary review result according to the pre-determined target analysis algorithm and the data review strategy to obtain the data verification result corresponding to the data to be reviewed.

9. A content verification system based on LLM security protection, characterized in that, The system includes: A memory storing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the content verification method based on LLM security protection according to any one of claims 1-7.

10. A computer storage medium, characterized in that, The computer storage medium stores computer instructions, which are used to execute the content verification method based on LLM security protection according to any one of claims 1-7 when called.

Images