Private domain knowledge base construction and authority control method for large model

Abstract

The invention discloses a large model-oriented private domain knowledge base construction and authority control method. The control method comprises the following steps of: uniformly authenticating and logging in SSO; collecting and analyzing data, and generating analysis data; generating a knowledge base according to the analysis data, and setting permission for the knowledge base; and performing machine training and learning on the content of the knowledge base to obtain a data conversion result, and displaying the data conversion result to a user. According to the method, fine-grained authority management based on a user private domain is carried out on a knowledge base on the basis of a unified authority authentication system in a preorder of establishment of a private domain knowledge base for large model application.

Description

Technical Field

[0001] The present invention relates to the field of constructing a private domain knowledge base for large model applications, and in particular to a method for constructing a private domain knowledge base for large models and controlling permissions. Background Art

[0002] With the deepening of digital transformation and the rapid development of AIGC technology, many large model products have emerged. However, these general large models have defects in the permission management of knowledge bases and fail to implement effective data governance strategies based on user private domains, resulting in insufficient data security. This situation not only increases the risk of data leakage and abuse, but may also cause legal and data security issues, especially in terms of personal privacy and intellectual property rights. Therefore, it is urgent to strengthen the protection of user private domain data for large model applications, ensure the security and privacy protection of user private domain data during collection, processing, storage and sharing, so as to build a more secure and reliable artificial intelligence ecosystem.

[0003] The existing private domain knowledge base construction for large model applications generally adopts the measures of aggregating, collecting and cleaning all data at the data processing level. In the construction process, the differences in the permissions of private domain users are not fully considered to effectively divide and isolate data, and only a single knowledge base without subdivision is formed.

[0004] Disadvantages of existing technologies: The current construction of private domain knowledge bases can only ensure that users within a certain range can obtain their private domain knowledge indiscriminately, and cannot guarantee the data security of the knowledge base based on the fine-grained private domain of users, which may indirectly lead to security issues such as internal data leakage, corporate trade secrets or customer information leakage. Summary of the invention

[0005] In view of the above problems, the present invention is proposed to provide a method for constructing a private domain knowledge base and controlling permissions for a large model, which overcomes the above problems or at least partially solves the above problems.

[0006] According to one aspect of the present invention, a method for constructing a private domain knowledge base and controlling permissions for a large model is provided, and the control method comprises:

[0007] SSO unified authentication login;

[0008] Data collection and analysis, generating analytical data;

[0009] Generate a knowledge base based on the analysis data, and set permissions for the knowledge base;

[0010] The content of the knowledge base is machine trained and learned to obtain data conversion results, which are then displayed to users.

[0011] Optionally, the SSO unified authentication login specifically includes: after the user logs in to the business system through the SSO unified authentication, the system obtains the user's current system permissions.

[0012] Optionally, the system permissions specifically include: data permissions, button permissions and menu permissions.

[0013] Optionally, the data collection and analysis to generate analytical data specifically includes:

[0014] Collect and analyze data of the current login system according to the user's system permissions;

[0015] The collected data results are different according to the user's permissions. Each user's information is within the user's permissions, realizing data permission control and ensuring the security of user's private data.

[0016] Optionally, the performing machine training and learning on the contents of the knowledge base to obtain data conversion results and presenting them to the user specifically includes: performing machine training and learning based on the contents of the knowledge base, converting the data into visual and easily obtainable results and presenting them to the user.

[0017] Optionally, the large model for machine training and learning includes:

[0018] Large language models for processing text data and understanding natural language;

[0019] Vision big model for image processing and analysis;

[0020] Multimodal large models are large models used to process multiple types of data.

[0021] According to a method for constructing a private domain knowledge base and controlling permissions for a large model as described in claim 6, it is characterized in that the multiple types of data specifically include: multimodal data of text, image, and audio.

[0022] The present invention provides a method for constructing a private domain knowledge base and controlling permissions for a large model, the control method comprising: SSO unified authentication login; data collection and analysis to generate analysis data; generating a knowledge base based on the analysis data and setting permissions for the knowledge base; machine training and learning the content of the knowledge base to obtain data conversion results and display them to users. In the pre-order of establishing a private domain knowledge base for large model applications, relying on a unified permission authentication system, the knowledge base is managed with fine-grained permissions based on the user's private domain.

[0023] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0025] Figure 1 A flowchart of a method for constructing a private domain knowledge base and controlling permissions for a large model provided by an embodiment of the present invention;

[0026] Figure 2 A composition diagram of a system corresponding to a method for constructing a private domain knowledge base and controlling permissions for a large model provided by an embodiment of the present invention;

[0027] Figure 3 A schematic diagram of the composition of a user private domain system provided by an embodiment of the present invention;

[0028] Figure 4 A schematic diagram of a process for generating a private domain knowledge base provided by an embodiment of the present invention;

[0029] Figure 5 A flow chart of model training provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0030] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0031] The terms "comprises" and "having" and any variations thereof in the description embodiments, claims and drawings of the present invention are intended to cover non-exclusive inclusions, for example, including a series of steps or units.

[0032] The technical solution of the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments.

[0033] like Figure 1 and Figure 2As shown, a method for constructing a private domain knowledge base and controlling permissions for a large model includes: an SSO unified authentication login module, a data collection and analysis module, a knowledge base, a large model, and various business systems;

[0034] SSO unified authentication login module: After the user logs in to the business system through the "SSO unified authentication login module", the system obtains the user's current system permissions, including but not limited to data permissions, button permissions, menu permissions, etc.

[0035] Each business system: saves various data and permissions of users in this system, provides accurate permission information and data information to the "data collection and analysis module", and connects to the "SSO unified authentication login module".

[0036] Data collection and analysis module: Collect and analyze data of the currently logged-in system according to the user's system permissions. The collected data results will be different according to the user's permissions. Each user's information is within the user's permissions, realizing data permission control and ensuring the security of user private data.

[0037] Knowledge base: Based on the data generated after collection and analysis by the "Data Collection and Analysis Module", a knowledge base is generated and permissions are set for the knowledge base. Only the current user has access rights to ensure the security of knowledge base data storage.

[0038] Big model: Perform machine training and learning on the content of the knowledge base, and convert the data into visual and easily accessible results to display to users.

[0039] Large model knowledge base permissions generation includes:

[0040] like Figure 3 As shown, the user logs in through the "SSO unified authentication login module". After successful identity authentication, the user can obtain the user rights of the current system, including the user's data rights, button rights, and menu rights.

[0041] The data collection and analysis module will collect data and generate data for the user's private domain system based on the various permissions of the user after logging into the system.

[0042] like Figure 4 As shown, after the data is cleaned, the user's private domain knowledge base is generated, and the knowledge base is uploaded to the large language model (LLM) for model training.

[0043] like Figure 5As shown in the figure, after the user completes the identity authentication through the "SSO unified authentication login module", the user calls the "model capability application service" capability through the question and answer interface (Q / A), capability service interface (API), intelligent agent (Agent), etc. Based on the user's permissions, LLM recalls the results from its exclusive private domain knowledge base to answer, and then feeds back the results to LLM for re-training based on the user's satisfaction.

[0044] The built-in LLM will assist in data cleaning. In order to maximize the utilization of computing power, deduplication based on the granularity of paragraphs and texts is preferred.

[0045] According to the different input data types, large models can be divided into the following three categories:

[0046] Language big models: refers to a type of big models in the field of natural language processing (NLP), which are usually used to process text data and understand natural language. The main feature of this type of big models is that they are trained on large-scale corpora to learn various grammatical, semantic and contextual rules of natural language.

[0047] Visual large models (CV): Large models used in the field of Computer Vision (CV), usually for image processing and analysis. Such models are trained on large-scale image data to achieve various visual tasks, such as image classification, object detection, image segmentation, pose estimation, face recognition, etc.

[0048] Multimodal large models: large models that can handle multiple different types of data, such as text, images, audio, and other multimodal data. This type of model combines the capabilities of NLP and CV to achieve comprehensive understanding and analysis of multimodal information, and can more comprehensively understand and process complex data.

[0049] According to different application fields, large models can be divided into three levels: L0, L1, and L2:

[0050] Universal large model L0: A large model that is universal in multiple fields and tasks. It uses large computing power, massive open data, and deep learning algorithms with huge parameters to train on large-scale unlabeled data to find features and discover patterns, thereby forming a powerful generalization ability that can "draw inferences from one example to another". It can complete multi-scenario tasks without or with little fine-tuning, which is equivalent to AI completing "general education".

[0051] Industry large model L1: A large model for a specific industry or field. It is usually pre-trained or fine-tuned using industry-related data to improve performance and accuracy in that field, which is equivalent to AI becoming an "industry expert."

[0052] Vertical large model L2: A large model for a specific task or scenario. It is usually pre-trained or fine-tuned using task-related data to improve performance and results on the task.

[0053] Beneficial effect: The user private domain data permission of the present invention is applicable to the permission design of all the above large models, achieving the effect of protecting data security.

[0054] Before establishing a private domain knowledge base for large model applications, we rely on a unified permission authentication system to manage the knowledge base in a fine-grained manner based on the user's private domain. That is, we first determine the knowledge permissions and then build the knowledge base. The establishment of each knowledge base must be completed under the control of the user's personal permissions to ensure that the knowledge base has private permission coverage, realize the user's private knowledge base and user private domain permission control, ensure that data is not abused or leaked, and improve data security.

[0055] The above specific implementation methods further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above are only specific implementation methods of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for constructing a private domain knowledge base and controlling permissions for a large model, characterized in that: The control method comprises: SSO unified authentication login; Data collection and analysis, generating analytical data; Generate a knowledge base based on the analysis data, and set permissions for the knowledge base; The content of the knowledge base is machine trained and learned to obtain data conversion results, which are then displayed to users.

2. According to the method for constructing a private domain knowledge base and controlling permissions for a large model according to claim 1, it is characterized in that: The SSO unified authentication login specifically includes: after the user logs in to the business system through the SSO unified authentication, the system obtains the user's current system permissions.

3. According to claim 2, a method for constructing a private domain knowledge base and controlling permissions for a large model is characterized in that: The system permissions specifically include: data permissions, button permissions and menu permissions.

4. According to the method for constructing a private domain knowledge base and controlling permissions for a large model according to claim 1, it is characterized in that: The data collection and analysis to generate analytical data specifically includes: Collect and analyze data of the current login system according to the user's system permissions; The collected data results are different according to the user's permissions. Each user's information is within the user's permissions, realizing data permission control and ensuring the security of user's private data.

5. According to the method for constructing a private domain knowledge base and controlling permissions for a large model according to claim 1, it is characterized in that: The performing machine training and learning on the content of the knowledge base to obtain data conversion results and presenting them to the user specifically includes: performing machine training and learning according to the content of the knowledge base to convert the data into visual and easily obtainable results and presenting them to the user.

6. According to the method for constructing a private domain knowledge base and controlling permissions for a large model according to claim 1, it is characterized in that: The large model for machine training and learning includes: Large language models for processing text data and understanding natural language; Vision big model for image processing and analysis; Multimodal large models are large models used to process multiple types of data.

7. A method for constructing a private domain knowledge base and controlling permissions for a large model according to claim 6, characterized in that: The multiple types of data specifically include: multimodal data of text, image, and audio.

Images