Artificial intelligence-based enterprise informatization platform

Abstract

The application discloses an enterprise informatization platform based on artificial intelligence and relates to the field of enterprise information technology.The application comprises a labeling background, a database, a processing unit and an interactive module.The database comprises an enterprise information module and a personnel module.The processing unit comprises a verification module and a grading module.The ports of the labeling background are connected with the ports of the enterprise information module and the personnel module respectively.The ports of the enterprise information module are connected with the ports of the verification module and the grading module respectively.The ports of the personnel module are connected with the ports of the verification module and the grading module respectively.The ports of the verification module and the grading module are connected with the port of the interactive module.The labeling background labels the data in the enterprise information module and the personnel module.The application replaces the traditional account password verification with a decision tree, can quickly narrow the personnel range, can identify abnormal operations, limits the risk spread through a permission downgrade mechanism and reduces the data exposure risk.

Description

Technical Field

[0001] This invention relates to the field of enterprise information technology, specifically to an enterprise information platform based on artificial intelligence. Background Technology

[0002] Traditional enterprise information platforms primarily rely on account and password systems for identity verification and access control. This involves pre-setting a unique account and password for each user. When a user logs in using the interactive module, their identity is unknown. By pre-setting each user's account, password, and corresponding data permissions, the system requires the user to enter their account and password to log in. The system then determines whether the user can read or write specific enterprise data based on the pre-defined permissions. Access control is often based on organizational structure levels, such as setting fixed access scopes by department or position. Operational compliance verification typically relies on post-event auditing or static rule checks.

[0003] Patent CN115310063A discloses an artificial intelligence-based enterprise information platform. This platform divides existing information platforms into different modules, each requiring approval from a higher authority before allowing browsing, downloading, and deleting of information. However, to ensure data security, all user logins are processed using the same method, requiring approval for all operations. This approach cannot tailor security strategies to different users. For non-sensitive data, delayed approval reduces the efficiency of the enterprise information platform and increases the workload of pre-login preparation. Therefore, how to make enterprise information security strategies intelligent is a technical problem that those skilled in the art need to solve. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an artificial intelligence-based enterprise information platform, which solves the problems mentioned in the background section.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an enterprise information platform based on artificial intelligence, comprising a labeling backend, a database, a processing unit, and an interaction module. The database includes enterprise information and human resources modules. The processing unit includes a verification module and a grading module. The ports of the labeling backend are respectively connected to the ports of the enterprise information and human resources modules. The ports of the enterprise information are respectively connected to the ports of the verification module and the grading module. The ports of the human resources module are respectively connected to the ports of the verification module and the grading module. The ports of the verification module and the grading module are both connected to the port of the interaction module.

[0006] The annotation backend performs annotation actions on the data in the enterprise information and personnel modules. The annotation actions include administrators setting organizational structures and entering personnel data. The enterprise information is used to store enterprise data, and the personnel module is used to store internal personnel data. The verification module is used for personnel locking and verification. The hierarchical module classifies personnel data to obtain basic features, behavioral features, and relational features, and hierarchically classifies enterprise data to obtain a relational network. The hierarchical module is used to determine whether the data read or written by the user exceeds the permission range. The interaction module is specifically a touch screen display. Users interact with the verification module and the hierarchical module through the interaction module. The interaction actions include selecting elements and reading / writing data.

[0007] Furthermore, the labeling of data in the enterprise information and human resources modules by the hierarchical module specifically includes the following:

[0008] Personnel data is categorized and broken down into basic characteristics, behavioral characteristics, and relational characteristics. Basic characteristics include, but are not limited to, gender, age, and year of employment. Behavioral characteristics include, but are not limited to, business trip applications, leave applications, and expense reimbursement applications. Relational characteristics include, but are not limited to, business managers and project team members. Basic characteristics, behavioral characteristics, and relational characteristics serve as the data source for user-selected elements, used for user personnel locking and personnel verification. Basic characteristics, behavioral characteristics, and relational characteristics are saved to the HR module.

[0009] The enterprise data is classified and the organizational structure of personnel data is obtained. The organizational structure includes the department, position and hierarchical relationship of each person. A relationship network is formed according to the organizational structure topology. Each node in the relationship network corresponds to a part of the position. The line segments connecting each node in the relationship network correspond to the hierarchical relationship. The authorization scope of each file in the enterprise data is obtained. The authorization scope of each file is matched and bound to the relationship network as the basis for judging whether the user reads or writes enterprise data beyond the file permission scope. The relationship network is saved to the enterprise information.

[0010] Furthermore, when users log in to the platform using the interactive module, since the specific identity information of the logging user is unknown, existing technology involves pre-setting each user's account and password, as well as their corresponding data permissions, and then logging in by entering the account and password. However, the above method carries the risk of password leakage. A solution based on the decision tree principle utilizes fragmented and non-sensitive personnel data to filter users. A decision tree is a bifurcated reasoning model that narrows the user pool from all possibilities to a single unique user through a series of questions, locking in the user's specific identity information. Subsequent personnel verification confirms the user's identity, preventing identity misuse and avoiding the account and password verification process, thus improving the platform's login security. Personnel locking specifically includes the following:

[0011] The verification module establishes a personnel set D, retrieves basic characteristics, behavioral characteristics, and relationship characteristics from the personnel module, and inputs them into personnel set D, according to the formula. Calculate the first set purity IG for each feature. The higher the value of the first set purity IG, the stronger the ability of the questions raised based on that feature to distinguish them. f is one of the features in the personnel set D. H(D) is the entropy state of the personnel set D, which specifically refers to the disorder of all feature data in the personnel set D. The larger the value of H(D), the more difficult it is for the feature data to distinguish the identity information of the personnel. Dv is the value of one subset of feature f. The subset of each feature is split into elements.

[0012] The first set of purity IG is selected in descending order of numerical value to generate the problem. Features that have been used will not be used to generate problems again to avoid using exhaustive methods to repeatedly select and brute-force the personnel locking process. Six elements corresponding to the feature are generated for the user to choose from based on the problem. After the user selects one element, the personnel corresponding to the elements not selected by the user are filtered out. After filtering, it is determined whether the remaining personnel are a single person. If the number of remaining personnel is 1, it means that the specific personnel can be identified in the personnel locking process, and the process jumps to the personnel verification. If the number of remaining personnel is greater than 1, it means that the personnel locking process has not yet identified a specific personnel, and the process proceeds to the next step.

[0013] The verification module sets a first judgment threshold n. When a user selects an element corresponding to a feature, it records the number of times the user narrows down the range of personnel data. Each time the user selects an element, the recorded number is incremented by 1. The verification module judges whether the recorded number of times the range of personnel data has been narrowed down is greater than the first judgment threshold n. If the number of times the range of personnel data has been narrowed down is less than the first judgment threshold n, it means that the number of times the user has selected an element is insufficient to determine the specific person. The module needs to jump to the user's element selection and generate a new question based on the feature. If the number of times the range of personnel data has been narrowed down is greater than the first judgment threshold n, it means that multiple screenings cannot determine the specific person. The user may have made an error when selecting an element, or the user may be an external person using an exhaustive method to try to break the user locking process. The module jumps to the creation of a personnel set D and re-splitting the elements.

[0014] Furthermore, personnel verification specifically includes the following:

[0015] The verification module establishes a set of enterprises E, retrieves the relationship network from the enterprise information and inputs it into the set of enterprises E, and then applies the formula... Calculate the second set purity JG for each file. The higher the value of the second set purity JG, the stronger the ability to distinguish the questions raised by this file. r is one of the files in the enterprise set E. Q(E) is the entropy state of the enterprise set E, which specifically refers to the disorder of all files in the enterprise set E in the relational network. The larger the value of Q(E), the more difficult it is for the file to distinguish the identity information of the personnel. Ev is the value of a subset of file r. The formula calculation process is the same as the calculation process of the first set purity IG. The subset of each file is split into elements.

[0016] The second set of purity JG is selected in descending order of numerical value to generate the problem. Each file is only allowed to generate the problem once to avoid using brute force to repeatedly select and crack the personnel locking process. Six elements corresponding to the file are generated based on the problem for the user to select. After the user selects one element, it is determined whether the selected element is consistent with the subset in the file. If the content of the selected element is inconsistent with the content of the subset in the file, it means that the person is not within the scope of the file relationship network permissions, and the verification fails. The process jumps to the step of selecting the feature element in the personnel locking process. If the content of the selected element is consistent with the content of the subset in the file, it means that the person is within the scope of the file relationship network permissions, and the verification passes. When verifying the problem again, the enterprise data range of the selected file is reduced and the next step is executed.

[0017] The verification module sets a second judgment threshold m1. When a user selects an element corresponding to a file, it records the number of times the enterprise data range is narrowed down. Each time the user successfully verifies, the recorded count is incremented by 1. The verification module determines whether the recorded number of successful verifications is greater than the second judgment threshold m1. If the number of successful verifications is less than the second judgment threshold m1, it means that the number of times the user selects an element is insufficient to determine that it is within the file's authorized range. The module needs to jump to the user's element selection and generate a new question based on the file. If the number of successful verifications is greater than the second judgment threshold m1, it means that the user has passed the verification multiple times. It can be determined that the user is within the file's authorized range. The user's verification is successful, and reading and writing files within the authorized range of the verified user is allowed.

[0018] Furthermore, before a user applies to read or write files in the enterprise data, the hierarchical module obtains the relationship network from the enterprise information and counts the number of nodes from the end of the relationship network upwards. The end of the relationship network is the lowest-level position in the enterprise's organizational structure. Since the organizational structure is generally a tree diagram structure, counting the number of nodes upwards can quickly identify the hierarchical levels of positions in the enterprise, and the number of nodes is defined as the enterprise's relationship level.

[0019] When a user requests to read or write files in the enterprise data, the hierarchical module determines whether the requested file exceeds the user's permission scope. If the requested file exceeds the permission scope of the verified user, the user is prohibited from reading or writing the file. At the same time, a temporary permission is created and assigned to the user. The temporary permission is one level lower than the user's verified user permission. The process then jumps to the user verification process and repeats the user element selection step. If the verification is successful, the user is allowed to read and write files in the enterprise data with temporary permission. If the verification fails, the process jumps to the user locking process and repeats the user element selection step.

[0020] If a user requests to read or write files that do not exceed the authorized personnel's permission scope, the user is allowed to read and write some encrypted corporate data. Initially, even if the user passes the personnel verification, there is still a risk of corporate data leakage. Therefore, it is necessary to encrypt part of the file content and judge the compliance of the user's operation by monitoring whether the user's file reading and writing exceeds the scope in real time. Only if the user's multiple read and write operations do not exceed the permission scope, it means that the user's operation process is compliant and risk-free, and then full permissions are granted.

[0021] The hierarchical module sets a third judgment threshold m2. When a user reads or writes files in the enterprise data, each read / write operation is recorded and accumulated. The hierarchical module determines whether the number of read / write operations by the user is greater than the second judgment threshold m2. If the number of read / write operations is less than or equal to m2, it means that the number of times the user has read / written enterprise data is not enough to meet the security limit, and the process jumps to the user element selection step in the personnel verification process. If the number of read / write operations is greater than m2, it means that the number of times the user has read / written enterprise data is sufficient to meet the security standard, and the user is granted full permissions to verified personnel, allowing them to read and write enterprise data files within the scope of the verified personnel's permissions.

[0022] Furthermore, the first judgment threshold n is determined based on the entropy state H(D) of the personnel set D. The formula for calculating the entropy state H(D) is as follows:

[0023]

[0024] Where k1 is the number of subsets of the feature, and pi is the proportion of personnel in the i-th subset;

[0025] If H(D) is less than or equal to 1, it means that the set of people D has high discrimination and is easy to filter. The first judgment threshold n is 4. If H(D) is greater than 1 and less than or equal to 2, it means that the set of people D has average discrimination and is moderately difficult to filter. The first judgment threshold n is 7. If H(D) is greater than 2, it means that the set of people D has low discrimination and is not easy to filter. The first judgment threshold n is 11.

[0026] Furthermore, the second judgment threshold m1 is determined based on the entropy state Q(E) of the enterprise set E, and the formula for calculating the entropy state Q(E) is as follows:

[0027]

[0028] Where k2 is the number of subsets of files, and qi is the percentage of people in the i-th subset;

[0029] If Q(E) is less than or equal to 1, it means that the set of enterprises E has high discrimination and is easy to screen. The second judgment threshold m1 is 3. If Q(E) is greater than 1 and less than or equal to 2, it means that the set of enterprises E has average discrimination and is moderately difficult to screen. The second judgment threshold m1 is 5. If Q(E) is greater than 2, it means that the set of enterprises E has low discrimination and is not easy to screen. The second judgment threshold m1 is 8.

[0030] Furthermore, the third judgment threshold m2 is determined based on the enterprise's relationship level. The hierarchical module classifies the enterprise relationship level into high-level, middle-level, and low-level. The first third of the relationship levels are defined as high-level, the last third of the relationship levels are positioned as low-level, and the remaining relationship levels are positioned as middle-level.

[0031] When the verified user is at a high-level position, the confidentiality level of the files in the enterprise data is high, and the data is sensitive. Data security is the primary concern, and a conservative policy should be adopted for the management of enterprise data. The third judgment threshold m2 is set to 15. When the verified user is at a mid-level position, the confidentiality level of the files in the enterprise data is moderate, and a general policy should be adopted for the management of enterprise data. The third judgment threshold m2 is set to 10. When the verified user is at a low-level position, the confidentiality level of the files in the enterprise data is low, and the data is not sensitive. An open policy should be adopted for the management of enterprise data. The third judgment threshold m2 is set to 5 to facilitate verification and avoid excessive verification that may affect the efficiency of enterprise operations.

[0032] Furthermore, to prevent external personnel from using brute-force methods to crack personnel data, the interaction module is locked and interaction is prohibited when a user fails verification more than 3 times during the personnel verification process. After being locked, the administrator can unlock it through the annotation backend.

[0033] Furthermore, the encryption of partially encrypted enterprise data is specifically defined as follows: when a user reads a file in the enterprise data, the text content displayed in the file is encrypted. Files read with temporary permissions are also encrypted. The text content encryption is based on RSA public key encryption technology. When a user writes a file in the enterprise data, the hierarchical module creates a temporary copy to save the file written by the user. After the user obtains full permissions later, the temporary copy replaces the file that the user originally wanted to write. The same applies to files written to the enterprise data with temporary permissions.

[0034] The present invention has the following beneficial effects:

[0035] 1. Based on the decision tree principle, it replaces traditional account password verification by filtering fragmented and non-sensitive personnel features layer by layer, thereby avoiding risks such as password leakage and brute-force attacks from the root and preventing identity theft. By calculating the purity of the first set of features (IG), it prioritizes high-discrimination features to generate problems, and dynamically controls the number of filtering times in combination with the first judgment threshold n. This can quickly narrow down the personnel to a single target and identify abnormal operations, balancing efficiency and security.

[0036] 2. Based on enterprise data generation verification issues, high-discrimination files are filtered through the purity of the second set JG to ensure that the verification content is closely related to the user's actual permissions and business, and only authorized personnel can pass through, further confirming that it is the person's operation. A second judgment threshold m1 is set according to the enterprise data entropy state Q(E). The higher the entropy value, the more verifications are required. In complex business scenarios, multiple verifications reduce the risk of impersonation, while avoiding unnecessary redundant verifications, balancing security and user experience.

[0037] 3. By combining the hierarchical relationship network to determine read and write permissions in real time, temporary permissions are dynamically granted and secondary verification is performed on operations that exceed the permissions. This satisfies the needs of cross-level collaboration and limits the spread of risks through the permission downgrade mechanism. A third judgment threshold m2 is set to control the risk of data exposure through intermediate states such as encrypted files and temporary copies. Full permissions are only granted after a user has performed multiple compliant operations. This achieves dynamic adaptation from the accumulation of operational compliance to the improvement of security level, reducing the risk of data leakage. Sensitive content is protected by RSA encryption technology. By combining read and write count statistics with threshold judgment, abnormal operations are identified in real time and secondary verification is triggered to form a security closed loop throughout the entire process.

[0038] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0039] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 This is a system block diagram of the enterprise information platform based on artificial intelligence of the present invention;

[0041] Figure 2 This is a schematic diagram of the interface of the interactive module of the present invention;

[0042] Figure 3 This is a schematic diagram of the personnel locking and personnel verification process of this invention;

[0043] Figure 4This is a schematic diagram illustrating the process of a user reading and writing enterprise data according to the present invention. Detailed Implementation

[0044] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0045] Please see Figure 1-4 This invention provides a technical solution: an enterprise information platform based on artificial intelligence, such as... Figure 1 As shown, it includes a labeling backend, a database, a processing unit, and an interaction module. The database includes enterprise information and personnel modules. The processing unit includes a verification module and a classification module. The ports of the labeling backend are connected to the ports of the enterprise information and personnel modules, respectively. The ports of the enterprise information modules are connected to the ports of the verification and classification modules, respectively. The ports of the personnel modules are connected to the ports of the verification and classification modules, respectively. The ports of the verification and classification modules are both connected to the ports of the interaction module.

[0046] The annotation backend performs annotation actions on the data in the enterprise information and human resources modules. Annotation actions include administrators setting organizational structures and entering personnel data. The enterprise information module is used to store enterprise data, the human resources module is used to store internal personnel data, the verification module is used for personnel locking and verification, the hierarchical module classifies personnel data to obtain basic features, behavioral features, and relational features, and hierarchically classifies enterprise data to obtain a relational network. The hierarchical module is used to determine whether the data read or written by the user exceeds the permission scope. The interaction module is specifically a touch screen. Users interact with the verification module and the hierarchical module through the interaction module. Interaction actions include selecting elements and reading / writing data.

[0047] The labeling of data in the enterprise information and human resources modules by the hierarchical module specifically includes the following:

[0048] Personnel data is categorized and broken down into basic characteristics, behavioral characteristics, and relational characteristics. Basic characteristics include, but are not limited to, gender, age, and year of employment. Behavioral characteristics include, but are not limited to, business trip applications, leave applications, and expense reimbursement applications. Relational characteristics include, but are not limited to, business managers and project team members. For example, employee A's basic characteristics are male, 26 years old, and year of employment in 2021. His behavioral characteristic is that he submitted a business trip application on May 1, 2022. His relational characteristics are business manager B and project team member C. Basic characteristics, behavioral characteristics, and relational characteristics serve as the data source for user-selected elements, used for user personnel locking and personnel verification. Basic characteristics, behavioral characteristics, and relational characteristics are saved in the HR module.

[0049] The enterprise data is classified and the organizational structure of personnel data is obtained. The organizational structure includes the department, position and hierarchical relationship of each person. A relationship network is formed according to the organizational structure topology. Each node in the relationship network corresponds to a part of the position. The line segments connecting each node in the relationship network correspond to the hierarchical relationship. The authorization scope of each file in the enterprise data is obtained. The authorization scope of each file is matched and bound to the relationship network as the basis for judging whether the user reads or writes enterprise data beyond the file permission scope. The relationship network is saved to the enterprise information.

[0050] In existing technologies, when users log in to the platform using the interactive module, their specific identity information is unknown. Existing technology involves pre-setting each user's account, password, and corresponding data permissions, then logging in by entering the account and password. However, this method carries the risk of password leakage. A solution based on decision tree principles utilizes fragmented and non-sensitive personnel data to filter users. A decision tree is a bifurcated reasoning model that narrows the user pool from all possibilities to a single unique user through a series of questions, locking in the user's specific identity information. Subsequent personnel verification confirms the user's identity, preventing identity misuse and avoiding the need for account and password verification, thus improving the platform's login security. Figure 3 As shown, personnel locking specifically includes the following:

[0051] The verification module establishes a personnel set D, retrieves basic characteristics, behavioral characteristics, and relationship characteristics from the personnel module, and inputs them into personnel set D, according to the formula. Calculate the first set purity IG for each feature. The higher the value of the first set purity IG, the stronger the ability to distinguish the questions raised by that feature. f is one of the features in the set of people D, such as the basic feature age. H(D) is the entropy state of the set of people D, specifically the disorder of all feature data in the set of people D. The larger the value of H(D), the more difficult it is to distinguish the identity information of people by feature data. Dv is the value of a subset of feature f. For example, the value of the basic feature age subset 25 years old is 10. Here, the value 10 represents that there are 10 people in the set of people D with the basic feature age of 25 years old. The value of the subset can also represent the number of times the subset of the feature appears. When the formula is calculated, only the numerical value is extracted for calculation. It is dimensionless and the subset of each feature is split into elements.

[0052] The first set of features corresponding to purity IG is selected in descending order of numerical value. Features already used are not used again to generate problems, thus avoiding the use of brute-force methods to repeatedly select and lock the process for cracking. Figure 2 As shown, the generated question is displayed in the interactive module Label2 for the user to read. Text2 is the replacement object for the question. Label1 on the interactive module is used to display the feature corresponding to the generated question. Text1 is the replacement object for the feature. For example, Label1 displays "age" to replace Text1, Label2 displays "How old are you?" to replace Text2, and Label3 displays 6 Buttons. The content displayed in the Buttons is "21 years old, 22 years old, 23 years old, 24 years old, 27 years old, and 30 years old" for the user to select. The user can select one element to exclude people corresponding to the other elements, thereby achieving the purpose of filtering, narrowing the range of personnel data, and realizing the user's personnel locking. Based on the question, 6 elements corresponding to the feature are generated for the user to select. After the user selects one element, the filter removes people corresponding to the elements not selected by the user. After filtering, it is determined whether the remaining people are a single person. If the number of remaining people is 1, it means that a specific person can be identified in the personnel locking process, and the process jumps to execute personnel verification. If the number of remaining people is greater than 1, it means that the personnel locking process has not yet identified a specific person, and the process proceeds to the next step.

[0053] The verification module sets a first judgment threshold n. When a user selects an element corresponding to a feature, it records the number of times the user narrows down the range of personnel data. Each time the user selects an element, the recorded number is incremented by 1. The verification module judges whether the recorded number of times the range of personnel data has been narrowed down is greater than the first judgment threshold n. If the number of times the range of personnel data has been narrowed down is less than the first judgment threshold n, it means that the number of times the user has selected an element is insufficient to determine the specific person. The module needs to jump to the user's element selection and generate a new question based on the feature. If the number of times the range of personnel data has been narrowed down is greater than the first judgment threshold n, it means that multiple screenings cannot determine the specific person. The user may have made an error when selecting an element, or the user may be an external person using an exhaustive method to try to break the user locking process. The module jumps to the creation of a personnel set D and re-splitting the elements.

[0054] Among them, such as Figure 3 As shown, personnel verification specifically includes the following:

[0055] The verification module establishes a set of enterprises E, retrieves the relationship network from the enterprise information and inputs it into the set of enterprises E, and then applies the formula... Calculate the second set purity JG for each file. The higher the value of the second set purity JG, the stronger the ability to distinguish the questions raised by this file. r is one of the files in the enterprise set E, such as the expense reimbursement file submitted by employee A. Q(E) is the entropy state of the enterprise set E, specifically referring to the disorder of all files in the enterprise set E in the relational network. The larger the value of Q(E), the more difficult it is for the file to distinguish the identity information of the personnel. Ev is the value of a subset of file r, such as the value of the expense reimbursement personnel in the subset of the expense reimbursement file submitted by employee A, which is 3. Here, the value 3 represents that there are 3 expense reimbursement personnel in the expense reimbursement file in the enterprise set E. The value of the subset can also represent the number of times the subset of the file appears, such as the number of modifications of the C project group file submitted by employee A. When the formula is calculated, only the numerical value is extracted for calculation. It is dimensionless. The calculation process of the formula is the same as the calculation process of the first set purity IG. The subset of each file is split into elements.

[0056] The files corresponding to the purity JG of the second set are selected in descending order of numerical value to generate questions. Each file is only allowed to generate a question once to avoid using brute-force methods to repeatedly select brute-force crackers to lock the process. The generated questions are displayed in the interactive module Label2 for users to read. Text2 is the replacement object for the question. Label1 on the interactive module is used to display the file name corresponding to the generated question. Text1 is the replacement object for the file name. For example, Label1 displays "Business B File" instead of Text1, and Label2 displays "What is the monthly performance of Business B?" instead of Text2. Label3 displays 6 buttons, and the content displayed in the buttons is "100,000, 120,000, 160,000, 190,000, 2 ... Users are given two options: "10,000" and "260,000". Only one option is correct. Selecting one element determines whether the user is a relevant person for business document B. Based on the question, six elements corresponding to the document are generated for the user to select. After the user selects an element, it is checked whether the selected element is consistent with the subset in the document. If the content of the selected element is inconsistent with the content of the subset in the document, it means that the person is not within the scope of the document relationship network permissions, and the verification fails. The process then jumps to the step of selecting the characteristic element in the personnel locking process. If the content of the selected element is consistent with the content of the subset in the document, it means that the person is within the scope of the document relationship network permissions, and the verification passes. When verifying the question again, the enterprise data range of the selected document is reduced, and the next step is executed.

[0057] The verification module sets a second judgment threshold m1. When a user selects an element corresponding to a file, it records the number of times the enterprise data range is narrowed down. Each time the user successfully verifies, the recorded count is incremented by 1. The verification module determines whether the recorded number of successful verifications is greater than the second judgment threshold m1. If the number of successful verifications is less than the second judgment threshold m1, it means that the number of times the user selects an element is insufficient to determine that it is within the file's authorized range. The module needs to jump to the user's element selection and generate a new question based on the file. If the number of successful verifications is greater than the second judgment threshold m1, it means that the user has passed the verification multiple times. It can be determined that the user is within the file's authorized range. The user's verification is successful, and reading and writing files within the authorized range of the verified user is allowed.

[0058] Among them, such as Figure 4 As shown, before a user applies to read or write files in the enterprise data, the hierarchical module obtains the relationship network from the enterprise information and counts the number of nodes from the end of the relationship network upwards. The end of the relationship network is the lowest-level position in the enterprise's organizational structure. Since the organizational structure is generally a tree diagram structure, counting the number of nodes upwards can quickly identify the hierarchical level of the enterprise's positions. The number of nodes is defined as the enterprise's relationship level.

[0059] When a user requests to read or write files in the enterprise data, the hierarchical module determines whether the requested file exceeds the user's permission scope. If the requested file exceeds the permission scope of the verified user, the user is prohibited from reading or writing the file. At the same time, a temporary permission is created and assigned to the user. The temporary permission is one level lower than the user's verified permission level. The process then jumps to the user verification process and repeats the user element selection step. If the verification is successful, the user is allowed to read and write files in the enterprise data with temporary permission. If the verification fails, the process jumps to the user locking process and repeats the user element selection step. For example, if employee A is in a position at level M2 in the enterprise, and the user exceeds their own permission scope when using the interaction module to read or write files, employee A is prohibited from accessing files that exceed their permission scope, and their level is downgraded to temporary permission M1. The user element selection step in the user verification process is then repeated.

[0060] If a user requests to read or write files that do not exceed the authorized personnel's permission scope, the user is allowed to read and write some encrypted corporate data. Initially, even if the user passes the personnel verification, there is still a risk of corporate data leakage. Therefore, it is necessary to encrypt part of the file content and judge the compliance of the user's operation by monitoring whether the user's file reading and writing exceeds the scope in real time. Only if the user's multiple read and write operations do not exceed the permission scope, it means that the user's operation process is compliant and risk-free, and then full permissions are granted.

[0061] The hierarchical module sets a third judgment threshold m2. When a user reads or writes files in the enterprise data, each read / write operation is recorded and accumulated. The hierarchical module determines whether the number of read / write operations by the user is greater than the second judgment threshold m2. If the number of read / write operations is less than or equal to m2, it means that the number of times the user has read / written enterprise data is not enough to meet the security limit, and the process jumps to the user element selection step in the personnel verification process. If the number of read / write operations is greater than m2, it means that the number of times the user has read / written enterprise data is sufficient to meet the security standard, and the user is granted full permissions to verified personnel, allowing them to read and write enterprise data files within the scope of the verified personnel's permissions.

[0062] The first judgment threshold n is determined based on the entropy state H(D) of the personnel set D. The formula for calculating the entropy state H(D) is as follows:

[0063]

[0064] Where k1 is the number of subsets of the feature. For example, if the feature "department" has 3 subsets: "R&D", "Marketing", and "Administration", then k1 = 3. pi is the percentage of personnel in the i-th subset. For example, if R&D accounts for 60%, then p1 = 0.6. If the subsets of the department feature are R&D 60%, Marketing 30%, and Administration 10%, then... ;

[0065] If H(D) is less than or equal to 1, it means that the set of people D has high discrimination and is easy to filter. The first judgment threshold n is 4. If H(D) is greater than 1 and less than or equal to 2, it means that the set of people D has average discrimination and is moderately difficult to filter. The first judgment threshold n is 7. If H(D) is greater than 2, it means that the set of people D has low discrimination and is not easy to filter. The first judgment threshold n is 11.

[0066] The second judgment threshold m1 is determined based on the entropy state Q(E) of the enterprise set E. The formula for calculating the entropy state Q(E) is as follows:

[0067]

[0068] Where k2 is the number of subsets of files, and qi is the percentage of people in the i-th subset;

[0069] If Q(E) is less than or equal to 1, it means that the set of enterprises E has high discrimination and is easy to screen. The second judgment threshold m1 is 3. If Q(E) is greater than 1 and less than or equal to 2, it means that the set of enterprises E has average discrimination and is moderately difficult to screen. The second judgment threshold m1 is 5. If Q(E) is greater than 2, it means that the set of enterprises E has low discrimination and is not easy to screen. The second judgment threshold m1 is 8.

[0070] Among them, the third judgment threshold m2 is determined according to the relationship level of the enterprise. The hierarchical module classifies the relationship level of the enterprise into high-level, middle-level and low-level. The first third of the relationship level is defined as high-level, the last third of the relationship level is positioned as low-level, and the remaining relationship level is positioned as middle-level.

[0071] When the verified user is at a high-level position, the confidentiality level of the files in the enterprise data is high, and the data is sensitive. Data security is the primary concern, and a conservative policy should be adopted for the management of enterprise data. The third judgment threshold m2 is set to 15. When the verified user is at a mid-level position, the confidentiality level of the files in the enterprise data is moderate, and a general policy should be adopted for the management of enterprise data. The third judgment threshold m2 is set to 10. When the verified user is at a low-level position, the confidentiality level of the files in the enterprise data is low, and the data is not sensitive. An open policy should be adopted for the management of enterprise data. The third judgment threshold m2 is set to 5 to facilitate verification and avoid excessive verification that may affect the efficiency of enterprise operations.

[0072] To prevent external personnel from using brute-force methods to crack personnel data, the interaction module is locked and interaction is prohibited when a user fails verification more than 3 times during the personnel verification process. After being locked, the administrator can unlock it through the backend.

[0073] Specifically, the encryption of enterprise data for users reading and writing means that when a user reads a file in the enterprise data, the text content displayed in the file is encrypted. Files read with temporary permissions are also encrypted. The text content encryption is based on RSA public key encryption technology. When a user writes a file in the enterprise data, the hierarchical module creates a temporary copy to save the file written by the user. After the user obtains full permissions later, the temporary copy replaces the file that the user originally wanted to write. The same applies to files written to the enterprise data with temporary permissions.

[0074] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. An enterprise information platform based on artificial intelligence, characterized by: The system includes a labeling backend, a database, a processing unit, and an interaction module. The database includes enterprise information and human resources modules. The processing unit includes a verification module and a classification module. The ports of the labeling backend are connected to the ports of the enterprise information and human resources modules, respectively. The ports of the enterprise information modules are connected to the ports of the verification and classification modules, respectively. The ports of the human resources modules are connected to the ports of the verification and classification modules, respectively. The ports of the verification and classification modules are both connected to the ports of the interaction module. The annotation backend performs annotation actions on the data in the enterprise information and personnel modules. The annotation actions include setting the organizational structure and entering personnel data. The enterprise information is used to store enterprise data, and the personnel module is used to store internal personnel data. The verification module is used for personnel locking and verification. The hierarchical module classifies personnel data to obtain basic features, behavioral features, and relational features, and hierarchically classifies enterprise data to obtain a relational network. The hierarchical module is used to determine whether the data read or written by the user exceeds the permission range. The interaction module is a touch screen display. The user interacts with the verification module and the hierarchical module through the interaction module. The interaction actions include selecting elements and reading and writing data. Personnel locking specifically includes the following: The verification module establishes a personnel set D, retrieves basic characteristics, behavioral characteristics, and relationship characteristics from the personnel module, and inputs them into personnel set D, according to the formula. Calculate the first set purity IG for each feature, where f is one of the features in the set of people D, H(D) is the entropy state of the set of people D, and Dv is the value of one subset of feature f. Split the subset of each feature into elements. The first set of purity IG is selected in descending order of numerical value to generate questions. Used features will not be used to generate questions again. Six feature elements are generated based on the questions for the user to select. After the user selects one element, the personnel corresponding to the elements not selected by the user are filtered out. After filtering, it is determined whether the remaining personnel are single. If the number of remaining personnel is 1, the process jumps to personnel verification. If the number of remaining personnel is greater than 1, the process proceeds to the next step. The verification module sets a first judgment threshold n. When the user selects an element, it records the number of times the range of personnel data is narrowed. Each time the user selects an element, the number of times the recorded range of personnel data is increased by 1. The verification module judges whether the number of times the recorded range of personnel data is narrowed is greater than the first judgment threshold n. If the number is less than n, it jumps to the user's element selection and generates a new question based on the features. If the number is greater than n, it jumps to the establishment of personnel set D and re-splitting the elements.

2. The enterprise information platform based on artificial intelligence according to claim 1, characterized in that, The labeling of data in the enterprise information and human resources modules by the hierarchical module specifically includes the following: The personnel data is categorized, and all personnel data is broken down into basic characteristics, behavioral characteristics, and relational characteristics. The basic characteristics, behavioral characteristics, and relational characteristics are then saved to the personnel module. The enterprise data is classified and the organizational structure of personnel data is obtained. The organizational structure includes the department, position and hierarchical relationship of each person. A relationship network is formed according to the organizational structure topology. Each node in the relationship network corresponds to a part of the position. The line segments connecting each node in the relationship network correspond to the hierarchical relationship. The authorization scope of each file in the enterprise data is obtained. The authorization scope of each file is matched and bound to the relationship network. The relationship network is saved to the enterprise information.

3. The enterprise information platform based on artificial intelligence according to claim 1, characterized in that, Personnel verification specifically includes the following: The verification module establishes a set of enterprises E, retrieves the relationship network from the enterprise information and inputs it into the set of enterprises E, and then applies the formula... Calculate the second set purity JG for each file, where r is one of the files in the enterprise set E, Q(E) is the entropy state of the enterprise set E, and Ev is the value of a subset of file r. Split the subset of each file into elements. The second set of purity JG is selected in descending order of numerical value to generate the problem. Each problem can only be generated once. Six files are generated based on the problem for the user to select. After the user selects one element, it is determined whether the selected element is consistent with the subset in the file. If the element content is inconsistent with the subset, the verification fails and the process jumps to the personnel lock. If the element content is consistent with the subset, the verification passes and the next step is executed. The verification module sets a second judgment threshold m1. When the user selects an element, the number of records is incremented by 1 each time the user passes the verification. The verification module judges whether the number of records that have passed the verification is greater than the second judgment threshold m1. If the number is less than m1, it jumps to the user selecting an element and generates a new question based on the file. If the number is greater than m1, the user passes the verification and is allowed to read and write files within the authorized scope of the verified personnel.

4. The enterprise information platform based on artificial intelligence according to claim 1, characterized in that, Before a user applies to read or write files in the enterprise data, the hierarchical module obtains the relationship network from the enterprise information, counts the number of nodes from the end of the relationship network upwards, and defines the number of nodes as the enterprise's relationship level. When a user requests to read or write files in the enterprise data, the hierarchical module determines whether the requested file exceeds the user's permission scope. If the requested file exceeds the user's permission scope, the user is prohibited from reading or writing the file. At the same time, a temporary permission is created and assigned to the user. The temporary permission is one level lower than the user's verified personnel. The system then jumps to personnel verification. If the verification is successful, the user is allowed to read and write files in the enterprise data with temporary permission. If the verification fails, the system jumps to personnel locking. If the files a user requests to read or write do not exceed the user's authorized access level, the user is allowed to read and write some encrypted corporate data. The hierarchical module sets a third judgment threshold m2. When a user reads or writes files in the enterprise data, each read / write operation is recorded and accumulated. The hierarchical module determines whether the number of times the user reads or writes enterprise data is greater than the second judgment threshold m2. If the number of read / write operations is less than or equal to m2, the process jumps to personnel verification. If the number of read / write operations is greater than m2, it means that the number of times the user reads or writes enterprise data is sufficient to meet the security standards, and the user is granted full permissions, allowing them to read and write enterprise data files within the scope of their permissions.

5. The enterprise information platform based on artificial intelligence according to claim 1, characterized in that, The first judgment threshold n is determined based on the entropy state H(D) of the personnel set D. The formula for calculating the entropy state H(D) is as follows: ； Where k1 is the number of subsets of the feature, and pi is the proportion of personnel in the i-th subset; If H(D) is less than or equal to 1, the first judgment threshold n is 4; if H(D) is greater than 1 and less than or equal to 2, the first judgment threshold n is 7; if H(D) is greater than 2, the first judgment threshold n is 11.

6. The enterprise information platform based on artificial intelligence according to claim 3, characterized in that, The second threshold m1 is determined based on the entropy state Q(E) of the enterprise set E. The formula for calculating the entropy state Q(E) is as follows: ； Where k2 is the number of subsets of files, and qi is the percentage of people in the i-th subset; If Q(E) is less than or equal to 1, the second judgment threshold m1 is 3; if Q(E) is greater than 1 and less than or equal to 2, the second judgment threshold m1 is 5; if Q(E) is greater than 2, the second judgment threshold m1 is 8.

7. The enterprise information platform based on artificial intelligence according to claim 4, characterized in that, The third judgment threshold m2 is determined based on the enterprise's relationship level. The hierarchical module classifies the enterprise's relationship level into high-level, middle-level, and low-level. The first third of the relationship levels are defined as high-level, the last third of the relationship levels are positioned as low-level, and the remaining relationship levels are positioned as middle-level. When the user who has passed verification is located at a high level, the third judgment threshold m2 is set to 15; when the user who has passed verification is located at a middle level, the third judgment threshold m2 is set to 10; and when the user who has passed verification is located at a low level, the third judgment threshold m2 is set to 5.

8. The enterprise information platform based on artificial intelligence according to claim 1, characterized in that, If a user fails to verify their identity more than three times, the interaction module will be locked, prohibiting any interaction. Once locked, the administrator can unlock it through the backend.

9. The enterprise information platform based on artificial intelligence according to claim 4, characterized in that, When a user reads or writes partially encrypted enterprise data, the text content displayed in the file is encrypted. Files read with temporary permissions are also encrypted. The text content encryption is based on public key encryption technology. When a user writes to a file in the enterprise data, the hierarchical module creates a temporary copy to save the file written by the user. After the user obtains full permissions later, the temporary copy replaces the file that the user originally wanted to write. The same applies to files written to the enterprise data with temporary permissions.

Images