Data processing method and device, computer device, storage medium and program product

Abstract

The application relates to a data processing method and device, computer equipment, a storage medium and a program product. Based on a user identifier and a knowledge identifier, a candidate path between a user node corresponding to the user identifier and a knowledge node corresponding to the knowledge identifier is screened in associated data information, and privacy budget allocation is performed according to a candidate path risk. The method avoids the problems of excessive disturbance or insufficient protection caused by fixed budget, improves the utilization efficiency of privacy resources, and adapts to the privacy risk differences of different data queries.

Description

Technical Field

[0001] This application relates to the field of information security technology, and in particular to a data processing method, apparatus, computer equipment, storage medium, and program product. Background Technology

[0002] Differential privacy is a data privacy protection technology that involves adding controlled random noise to data query results, making it impossible for attackers to accurately deduce specific information about any user or query record from the data query results, while ensuring that the data query results still have usability.

[0003] Traditional technologies typically use a fixed privacy budget to introduce random noise, which can lead to either excessive perturbation affecting usability or insufficient protection causing data security risks. This results in a lack of adaptability to the different privacy risks associated with different data queries. Summary of the Invention

[0004] Therefore, it is necessary to provide a data processing method, apparatus, computer equipment, storage medium, and program product that can adapt to the differences in privacy risks associated with different data queries, in order to address the aforementioned technical problems.

[0005] In a first aspect, this application provides a data processing method, including:

[0006] Obtain a data query request; wherein the data query request includes a user identifier and a knowledge identifier;

[0007] Based on the user identifier and the knowledge identifier, candidate paths between the user node corresponding to the user identifier and the knowledge node corresponding to the knowledge identifier are filtered in the associated data information; wherein, the nodes in the associated data information include user nodes, knowledge nodes and user organizational structure nodes; the edges in the associated data information include query permission edges connecting user nodes and knowledge nodes, membership relationship edges connecting user nodes and user organizational structure nodes, and knowledge association edges between knowledge nodes.

[0008] Based on the candidate paths, determine the target privacy budget corresponding to the data query request;

[0009] Noisy data is determined based on the target privacy budget, and the noisy data is added to the data query result corresponding to the data query request.

[0010] In the above embodiments, based on user identifiers and knowledge identifiers, candidate paths between user nodes corresponding to user identifiers and knowledge nodes corresponding to knowledge identifiers are filtered in the associated data information, and privacy budgets are allocated according to the risks of candidate paths. This not only avoids the problem of excessive disturbance or insufficient protection caused by fixed budgets, but also improves the utilization efficiency of privacy resources and adapts to the differences in privacy risks of different data queries.

[0011] In one embodiment, determining the target privacy budget corresponding to the data query request based on the candidate path includes:

[0012] Based on the path attribute information of the candidate paths, the risk coefficient of the candidate paths is determined;

[0013] Based on the request attribute data of the data query request, determine the basic privacy budget;

[0014] The basic privacy budget is adjusted using the risk coefficient to obtain the target privacy budget corresponding to the data query request.

[0015] In the above embodiments, by adjusting the basic privacy budget using a risk coefficient, the query context and access path risks are organically combined, making the allocation of the privacy budget more reasonable. This avoids the problems of excessive disturbance or insufficient protection caused by a fixed budget, improves the utilization efficiency of privacy resources, and adapts to the differences in privacy risks of different data queries.

[0016] In one embodiment, the request attribute data includes query type, user role, and knowledge sensitivity level; determining the basic privacy budget based on the request attribute data of the data query request includes:

[0017] Determine the type score corresponding to the query type, the role score corresponding to the user role, and the first sensitivity score corresponding to the knowledge sensitivity level;

[0018] The type score, the role score, and the first sensitivity score are weighted to obtain the basic privacy budget.

[0019] In the above embodiments, on the one hand, by using a weighted model based on three dimensions—query type, user role, and knowledge sensitivity level—to calculate the basic privacy budget, the differences in privacy risks among different queries can be comprehensively reflected. On the other hand, through configurable weight coefficients, it can flexibly adapt to the security strategies and compliance requirements of different industries and enterprises. This multi-dimensional weighted model, while ensuring computational efficiency, achieves fine-grained privacy budget allocation, effectively solving the problems of excessive disturbance or insufficient protection that exist in traditional fixed budget methods.

[0020] In one embodiment, the path attribute information includes path length, knowledge sensitivity level of knowledge nodes, risk level of permission edges, and knowledge association strength; determining the risk coefficient of the candidate path based on the path attribute information of the candidate path includes:

[0021] Determine the length score corresponding to the path length, the second sensitivity score corresponding to the knowledge sensitivity level, the risk score corresponding to the risk level, and the association score corresponding to the knowledge association strength;

[0022] The length score, the second sensitivity score, the risk score, and the association score are weighted to obtain the risk coefficient of the candidate path.

[0023] In the above embodiments, by considering four dimensions of factors—path length, knowledge sensitivity, permission edge risk level, and knowledge association strength—the risk coefficient is calculated, which can comprehensively and accurately quantify the privacy risks of different access paths. Furthermore, through configurable weight coefficients, the importance of each risk factor can be adjusted according to different business scenarios and security requirements. This multi-dimensional risk assessment model can effectively identify indirect access risks and associated access risks that traditional methods cannot detect, thereby improving the overall security of data.

[0024] In one embodiment, determining the noisy data based on the target privacy budget includes:

[0025] Obtain the current remaining budget; wherein, the current remaining budget is the total remaining budget of the knowledge base or the remaining budget of the user who initiated the data query request;

[0026] If the current remaining budget is greater than or equal to the target privacy budget, noisy data is determined based on the target privacy budget.

[0027] In the above embodiments, on the one hand, by introducing a budget availability judgment mechanism, the excessive consumption of privacy budget can be effectively prevented, ensuring the long-term privacy security of the knowledge base; on the other hand, by dynamically adjusting the noise intensity according to the target privacy budget, the correspondence between privacy protection intensity and budget consumption is realized, and the perturbation operation is only performed when the budget is sufficient, avoiding unnecessary waste of privacy resources.

[0028] In one embodiment, the method further includes:

[0029] If the current remaining budget is less than the target privacy budget, the data query results corresponding to the data query request shall be prohibited from being obtained.

[0030] In the above embodiments, prohibiting the acquisition of data query results when the privacy budget is insufficient can strictly control the consumption of the privacy budget and ensure that the knowledge base is always maintained at a secure level of privacy protection. Furthermore, this budget control mechanism can effectively prevent attackers from exhausting the privacy budget through high-frequency queries, thereby breaking through the privacy protection defenses and ensuring the long-term secure and stable operation of the knowledge base.

[0031] In one embodiment, the method further includes:

[0032] If the current remaining budget is less than the target privacy budget, request a new privacy budget from the privacy budget manager;

[0033] If the sum of the requested privacy budget and the current remaining budget is greater than or equal to the target privacy budget, noisy data is determined based on the target privacy budget.

[0034] In the above embodiments, by introducing a budget application and approval mechanism, while strictly controlling the consumption of privacy budget, users are provided with a channel to obtain additional budget, which meets the reasonable needs of users and improves the usability of the knowledge base.

[0035] In one embodiment, the method further includes:

[0036] If the current remaining budget is less than or equal to the budget threshold, a budget warning message is sent to the privacy budget manager.

[0037] In the above embodiments, by setting budget thresholds and early warning mechanisms, situations where the budget is consumed too quickly can be detected in a timely manner, and the privacy budget manager can be reminded in advance to take measures to avoid the situation where the budget is suddenly exhausted, causing the knowledge base to be unable to provide data query services, thereby improving the availability of the knowledge base.

[0038] In one embodiment, the step of filtering candidate paths between user nodes corresponding to the user identifier and knowledge nodes corresponding to the knowledge identifier from associated data information based on the user identifier and the knowledge identifier includes:

[0039] Retrieve the data query result corresponding to the data query request from the cache medium;

[0040] If no data query results are found, candidate paths between the user node corresponding to the user identifier and the knowledge node corresponding to the knowledge identifier are filtered from the associated data information based on the user identifier and the knowledge identifier.

[0041] In the above embodiments, on the one hand, by introducing a caching mechanism, the repeated execution of privacy protection operations for the same query request can be avoided, reducing the consumption of privacy budget; on the other hand, by directly returning the data query results from the cache, the response speed of data query is improved, and the user experience is enhanced.

[0042] Secondly, this application also provides a data processing apparatus, comprising:

[0043] The acquisition module is used to acquire data query requests; wherein, the data query request includes a user identifier and a knowledge identifier;

[0044] The filtering module is used to filter candidate paths between user nodes corresponding to the user identifier and knowledge nodes corresponding to the knowledge identifier in the associated data information based on the user identifier and the knowledge identifier; wherein, the nodes in the associated data information include user nodes, knowledge nodes and user organizational structure nodes; the edges in the associated data information include query permission edges connecting user nodes and knowledge nodes, membership relationship edges connecting user nodes and user organizational structure nodes, and knowledge association edges between knowledge nodes;

[0045] The determination module is used to determine the target privacy budget corresponding to the data query request based on the candidate path;

[0046] An add module is used to determine noisy data based on the target privacy budget and add the noisy data to the data query result corresponding to the data query request.

[0047] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0048] Obtain a data query request; wherein the data query request includes a user identifier and a knowledge identifier;

[0049] Based on the user identifier and the knowledge identifier, candidate paths between the user node corresponding to the user identifier and the knowledge node corresponding to the knowledge identifier are filtered in the associated data information; wherein, the nodes in the associated data information include user nodes, knowledge nodes and user organizational structure nodes; the edges in the associated data information include query permission edges connecting user nodes and knowledge nodes, membership relationship edges connecting user nodes and user organizational structure nodes, and knowledge association edges between knowledge nodes.

[0050] Based on the candidate paths, determine the target privacy budget corresponding to the data query request;

[0051] Noisy data is determined based on the target privacy budget, and the noisy data is added to the data query result corresponding to the data query request.

[0052] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0053] Obtain a data query request; wherein the data query request includes a user identifier and a knowledge identifier;

[0054] Based on the user identifier and the knowledge identifier, candidate paths between the user node corresponding to the user identifier and the knowledge node corresponding to the knowledge identifier are filtered in the associated data information; wherein, the nodes in the associated data information include user nodes, knowledge nodes and user organizational structure nodes; the edges in the associated data information include query permission edges connecting user nodes and knowledge nodes, membership relationship edges connecting user nodes and user organizational structure nodes, and knowledge association edges between knowledge nodes.

[0055] Based on the candidate paths, determine the target privacy budget corresponding to the data query request;

[0056] Noisy data is determined based on the target privacy budget, and the noisy data is added to the data query result corresponding to the data query request.

[0057] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0058] Obtain a data query request; wherein the data query request includes a user identifier and a knowledge identifier;

[0059] Based on the user identifier and the knowledge identifier, candidate paths between the user node corresponding to the user identifier and the knowledge node corresponding to the knowledge identifier are filtered in the associated data information; wherein, the nodes in the associated data information include user nodes, knowledge nodes and user organizational structure nodes; the edges in the associated data information include query permission edges connecting user nodes and knowledge nodes, membership relationship edges connecting user nodes and user organizational structure nodes, and knowledge association edges between knowledge nodes.

[0060] Based on the candidate paths, determine the target privacy budget corresponding to the data query request;

[0061] Noisy data is determined based on the target privacy budget, and the noisy data is added to the data query result corresponding to the data query request.

[0062] The aforementioned data processing methods, apparatus, computer equipment, storage media, and program products, based on user identifiers and knowledge identifiers, filter candidate paths between user nodes corresponding to user identifiers and knowledge nodes corresponding to knowledge identifiers in associated data information, and allocate privacy budgets according to the risks of candidate paths. This not only avoids the problems of excessive disturbance or insufficient protection caused by fixed budgets, but also improves the utilization efficiency of privacy resources and adapts to the differences in privacy risks of different data queries. Attached Figure Description

[0063] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0064] Figure 1 This is a flowchart illustrating a data processing method in one embodiment;

[0065] Figure 2 This is a flowchart illustrating the process of determining the target privacy budget corresponding to a data query request in one embodiment.

[0066] Figure 3 This is a flowchart illustrating a method for dynamically allocating a privacy budget in one embodiment;

[0067] Figure 4 This is a schematic diagram of the structure of a privacy budget dynamic allocation system in one embodiment;

[0068] Figure 5 This is a structural block diagram of a data processing device in one embodiment;

[0069] Figure 6 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0070] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0071] The data processing method provided in this application can be applied to application scenarios that protect data security when querying data. This method can be executed by a server or a terminal.

[0072] The terminals can be, but are not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart in-vehicle systems, and projection devices. Portable wearable devices can include smartwatches, smart bracelets, and head-mounted displays. Head-mounted displays can be virtual reality (VR) devices, augmented reality (AR) devices, and smart glasses. Servers can be independent physical servers, server clusters or distributed systems composed of multiple physical servers, or cloud servers providing cloud computing services.

[0073] In one exemplary embodiment, such as Figure 1 As shown, a data processing method is provided. Taking the application of this method to a server as an example, it includes the following steps:

[0074] S101, Get data query request.

[0075] Data query requests are requests initiated by users to the knowledge base to retrieve specific knowledge content, including but not limited to keyword retrieval, full-text extraction, statistical aggregation, and model fine-tuning. Data query requests include user identifiers and knowledge identifiers. User identifiers are used to uniquely identify system users, such as employee identifiers (IDs), usernames, or email addresses. Knowledge identifiers are used to uniquely identify specific knowledge entries in the knowledge base, such as document IDs, knowledge entry numbers, and Uniform Resource Locators (URLs).

[0076] For example, data query requests sent by user terminals can be received via Hypertext Transfer Protocol (HTTP) or Hypertext Transfer Protocol Secure (HTTPS). The request body can be encapsulated in JSON format and includes information such as user identifier, knowledge identifier, query type, and query parameters.

[0077] After receiving a data query request, the system can first authenticate the user's identity, verifying the validity of the user's identifier and login status. If the authentication is successful, the system will perform permission verification to check whether the user has the basic permissions to access the target knowledge item. If the permission verification is successful, the request will be parsed into a format that the system can process and stored in the request queue for processing.

[0078] S102, based on user identifier and knowledge identifier, filter candidate paths between user nodes corresponding to user identifiers and knowledge nodes corresponding to knowledge identifiers in the associated data information.

[0079] The associated data information is a graph data structure used to represent the relationships between users, knowledge, and organizational structures. It consists of nodes and edges. The graph data structure can be a knowledge graph, hypergraph, knowledge graph network, or a data table in a relational database, such as a Structured Query Language (SQL) table. The nodes in the associated data information include user nodes, knowledge nodes, and user organizational structure nodes; the edges in the associated data information include query permission edges connecting user nodes and knowledge nodes, membership relationship edges connecting user nodes and user organizational structure nodes, and knowledge association edges between knowledge nodes.

[0080] User nodes are nodes that represent system users in the associated data information, and each user node corresponds to a unique user identifier.

[0081] A knowledge node is a node in the associated data information that represents a knowledge entry in the knowledge base. Each knowledge node corresponds to a unique knowledge identifier.

[0082] User organizational structure nodes are nodes that represent organizational structure units within the associated data information, such as departments, teams, and project groups.

[0083] The query permission edge is an edge connecting a user node and a knowledge node, indicating that the user has permission to access the knowledge entry.

[0084] A membership edge is an edge that connects a user node to a user organizational structure node, indicating that the user belongs to that organizational structure unit.

[0085] Knowledge association edges are edges that connect knowledge nodes, indicating that there is a content association or reference relationship between two knowledge items.

[0086] Candidate paths are the set of all paths from user nodes to knowledge nodes in the associated data information.

[0087] For example, the corresponding user node can be located in the associated data information based on the user identifier, and the corresponding knowledge node can be located based on the knowledge identifier. Then, a breadth-first search algorithm can be used to traverse the associated data information and filter out all paths from the user node to the knowledge node.

[0088] Optionally, during the traversal, paths can be filtered based on a preset path length threshold, such as a maximum path length of 3, retaining paths whose length does not exceed the threshold as candidate paths. Simultaneously, the validity of edges on each path can be checked, such as whether query permission edges are valid or whether membership edges have changed. Invalid edges can be removed from the graph network, and the path can be recalculated.

[0089] S103, Based on the candidate paths, determine the target privacy budget corresponding to the data query request.

[0090] The target privacy budget is the differential privacy budget value allocated to this data query request, which can be represented by ε and determines the intensity of noise added to the query results.

[0091] A risk assessment can be performed on each candidate path, calculating a risk coefficient for each path. The calculation of the risk coefficient needs to consider factors such as path length, the sensitivity level of knowledge nodes, the risk level of permission edges, and the strength of knowledge associations. Then, based on the risk coefficients of all candidate paths, a comprehensive risk value is calculated, for example, using a weighted average method or a maximum value method.

[0092] Furthermore, the target privacy budget for a data query request can be determined based on the overall risk value. For example, the higher the overall risk value, the smaller the target privacy budget; the lower the overall risk value, the larger the target privacy budget.

[0093] S104, determine the noisy data based on the target privacy budget, and add the noisy data to the data query result corresponding to the data query request.

[0094] The noise data is random noise generated according to the target privacy budget and is used to add to the data query results to achieve differential privacy protection. Laplace noise or Gaussian noise can be used.

[0095] Noise data can be determined based on the noise generation algorithm and the target privacy budget. For example, for numerical data, Laplace noise can be generated using the Laplace mechanism; for non-numerical data or scenarios requiring higher practicality, Gaussian noise can be generated using the Gaussian mechanism.

[0096] The system can retrieve the original query results corresponding to the data query request from the knowledge base, add the generated noise data to the original query results to obtain the perturbed query results, and return the perturbed query results to the user terminal.

[0097] In the above embodiments, based on user identifiers and knowledge identifiers, candidate paths between user nodes corresponding to user identifiers and knowledge nodes corresponding to knowledge identifiers are filtered in the associated data information, and privacy budgets are allocated according to the risks of candidate paths. This not only avoids the problem of excessive disturbance or insufficient protection caused by fixed budgets, but also improves the utilization efficiency of privacy resources and adapts to the differences in privacy risks of different data queries.

[0098] In some alternative implementations, see [link to relevant documentation]. Figure 2 , Figure 2 A flowchart illustrating the process of determining the target privacy budget corresponding to a data query request is provided, which includes the following steps:

[0099] S201, Determine the risk coefficient of the candidate path based on the path attribute information of the candidate path.

[0100] Path attribute information describes the characteristics of candidate paths and is used to determine the risk coefficient of candidate paths. The risk coefficient is a numerical value used to quantify the degree of privacy risk of candidate paths, with a value range of [0,1]. The larger the value, the higher the risk.

[0101] For example, the path attribute information of each candidate path can be extracted first, and the risk coefficient of the candidate path can be determined based on the mapping relationship between the path attribute information and the path's risk coefficient. The mapping relationship between the path attribute information and the path's risk coefficient can be determined based on historical data.

[0102] S202, Determine the basic privacy budget based on the request attribute data of the data query request.

[0103] The request attribute data describes the characteristics of the data query request, including query type, user role, knowledge sensitivity level, query time, and Internet Protocol (IP) address. The basic privacy budget is a privacy budget value calculated based on the request attribute data, without considering the impact of access path risks.

[0104] For example, the basic privacy budget corresponding to a data query request can also be determined based on the mapping relationship between request attribute data and privacy budget. The mapping relationship between request attribute data and privacy budget can be determined based on historical data.

[0105] S203 uses a risk factor to adjust the basic privacy budget to obtain the target privacy budget corresponding to the data query request.

[0106] For example, the comprehensive risk coefficient of all candidate paths can be calculated. The comprehensive risk coefficient can be calculated using various methods, such as the maximum value method (i.e., taking the maximum value of the risk coefficients of all candidate paths), the weighted average method (i.e., weighting the risk coefficients of each path according to the path length or frequency of occurrence), and the product method (i.e., multiplying the risk coefficients of all candidate paths), etc.

[0107] For example, the maximum value method can be used to calculate the comprehensive risk coefficient, and then the basic privacy budget can be multiplied by the comprehensive risk coefficient to obtain the target privacy budget.

[0108] In the above embodiments, by adjusting the basic privacy budget using a risk coefficient, the query context and access path risks are organically combined, making the allocation of the privacy budget more reasonable. This avoids the problems of excessive disturbance or insufficient protection caused by a fixed budget, improves the utilization efficiency of privacy resources, and adapts to the differences in privacy risks of different data queries.

[0109] In some alternative implementations, during the process of determining the privacy budget, the type score corresponding to the query type, the role score corresponding to the user role, and the first sensitivity score corresponding to the knowledge sensitivity level can be determined; then, the type score, role score, and first sensitivity score are weighted to obtain the basic privacy budget.

[0110] The type score is a normalized score determined based on the query type, reflecting the degree of privacy risk of different query types, and its value range is [0,1].

[0111] The role score is a normalized score determined based on the user role, reflecting the degree of operational risk for different user roles, and the value range is [0,1].

[0112] The first sensitivity score is a normalized score determined based on the knowledge sensitivity level, reflecting the sensitivity of different knowledge items, and its value ranges from [0,1].

[0113] For example, mapping tables can be pre-established for query types and type scores, user roles and role scores, and knowledge sensitivity levels and first sensitivity scores. These mapping tables can be stored in the server's configuration database and can be dynamically modified by administrators through the configuration interface.

[0114] When a data query request is received, the query type, user role, and knowledge sensitivity level information can be extracted from the request. Then, the corresponding mapping table is queried to obtain the type score, role score, and primary sensitivity score.

[0115] Optionally, if the query type is not explicitly specified in the request, the server can analyze the query content using natural language processing technology to automatically identify the query type. If user role information is not included in the request, the user's role information can be retrieved from the user management system based on the user identifier. Knowledge sensitivity level information is stored in the knowledge base, and each knowledge entry has a corresponding sensitivity level label.

[0116] Furthermore, the weight coefficients w corresponding to query type, user role, and knowledge sensitivity level can be obtained from the configuration database. q w r w k Satisfying w q +w r +w k =1. The initial value of the weighting coefficient can be set based on actual empirical data; for example, it can be set to w. q =0.4, w r =0.3, w k =0.3.

[0117] Administrators can adjust the weighting coefficients based on specific data security policies and business needs. For example, for industries with high data security requirements, such as the financial industry, the weight of the knowledge sensitivity level can be appropriately increased; for organizations with strict user access control, such as government departments, the weight of user roles can be appropriately increased.

[0118] It can be done according to the formula ε base =w q ×f(q)+w r ×g(r)+w k ×h(k) calculates the basic privacy budget ε base Where f(q) is the type score, g(r) is the role score, and h(k) is the first sensitivity score. The calculated basic privacy budget ranges from [0,1].

[0119] In the above embodiments, on the one hand, by using a weighted model based on three dimensions—query type, user role, and knowledge sensitivity level—to calculate the basic privacy budget, the differences in privacy risks among different queries can be comprehensively reflected. On the other hand, through configurable weight coefficients, it can flexibly adapt to the security strategies and compliance requirements of different industries and enterprises. This multi-dimensional weighted model, while ensuring computational efficiency, achieves fine-grained privacy budget allocation, effectively solving the problems of excessive disturbance or insufficient protection that exist in traditional fixed budget methods.

[0120] In some optional implementations, in the process of determining the risk coefficient of a candidate path, the length score corresponding to the path length, the second sensitivity score corresponding to the knowledge sensitivity level, the risk score corresponding to the risk level, and the association score corresponding to the knowledge association strength can be determined; and the length score, second sensitivity score, risk score and association score are weighted to obtain the risk coefficient of the candidate path.

[0121] The length score is a normalized score determined based on the length of the candidate path, reflecting the complexity of the path, and its value ranges from [0,1]. The second sensitivity score is a normalized score determined based on the knowledge sensitivity level of knowledge nodes in the candidate path, reflecting the sensitivity of knowledge in the path, and its value ranges from [0,1]. The risk score is a normalized score determined based on the risk level of permission edges in the candidate path, reflecting the security risk level of permission edges, and its value ranges from [0,1]. The association score is a normalized score determined based on the knowledge association strength of knowledge association edges in the candidate path, reflecting the tightness of the association between knowledge, and its value ranges from [0,1]. The knowledge association strength can be determined based on factors such as the number of citations between knowledge items, content similarity, and topic relevance. For example, the content similarity and topic similarity between two knowledge items can be calculated, and the number of citations, content similarity, and topic similarity between knowledge items can be fused to obtain the knowledge association strength.

[0122] For example, a mapping function between path length and length score, a mapping table between knowledge sensitivity level and second sensitivity score, a mapping table between permission edge risk level and risk score, and a mapping table between knowledge association strength and association score can be pre-established.

[0123] The mapping function for path length can be an inverse proportional function, where the length score is the reciprocal of the path length. This means that the longer the path, the lower the length score, the smaller the risk coefficient, the smaller the allocated privacy budget, and the stronger the privacy protection.

[0124] The mapping table for knowledge sensitivity levels is the same as the mapping table for the first sensitivity score, ensuring that the sensitivity score for the same knowledge item remains consistent across different calculation stages.

[0125] The risk level of a permission edge can be determined based on factors such as the method of granting permissions, validity period, and access scope. For example, it can be divided into three levels: low, medium, and high, with corresponding risk scores of 0.3, 0.6, and 0.9, respectively.

[0126] Knowledge association strength can be calculated based on factors such as the number of times knowledge items are cited, content similarity, and topic relevance. The value range is [0,1]. The higher the association strength, the higher the association score.

[0127] Upon receiving a candidate path, the system extracts the path length, the highest knowledge sensitivity level of all knowledge nodes in the path, the highest risk level of all permission edges in the path, and the average knowledge association strength of all knowledge-related edges in the path. Then, it queries the corresponding mapping function or mapping table to obtain the length score, second sensitivity score, risk score, and association score.

[0128] Furthermore, the weight coefficients w corresponding to path length, knowledge sensitivity, permission edge risk level, and knowledge association strength can be obtained from the configuration database. l l, w s w p w a Satisfying w l +w s +w p +w a =1.

[0129] The initial value of the weighting coefficients can be set to w. l =0.3, w s =0.4, w p =0.2, w a =0.1. Administrators can adjust the weighting coefficients based on the data's security policy. For example, for data with complex knowledge associations, the weight of the knowledge association strength can be appropriately increased; for data with strict access control, the weight of the access risk level can be appropriately increased. Furthermore, the weighting can be adjusted according to the formula risk=w. l ×l score +w s ×s score +w p ×p score +w a ×a score Calculate the risk coefficient of the candidate path, where l score For length score, s score For the second sensitivity score, p score For risk score, a score This is the correlation score. The calculated risk coefficient ranges from [0,1].

[0130] In the above embodiments, by considering four dimensions of factors—path length, knowledge sensitivity, permission edge risk level, and knowledge association strength—the risk coefficient is calculated, which can comprehensively and accurately quantify the privacy risks of different access paths. Furthermore, through configurable weight coefficients, the importance of each risk factor can be adjusted according to different business scenarios and security requirements. This multi-dimensional risk assessment model can effectively identify indirect access risks and associated access risks that traditional methods cannot detect, thereby improving the overall security of data.

[0131] In some alternative implementations, to avoid the abuse of privacy budgets and to ensure data security after the budget is exhausted, it is necessary to first determine whether the remaining privacy budget is sufficient before allocating the corresponding privacy budget to the current data query request.

[0132] Based on this, the current remaining budget can be obtained; where the current remaining budget is the total remaining budget of the knowledge base or the remaining budget of the user who initiated the data query request; if the current remaining budget is greater than or equal to the target privacy budget, noisy data is determined according to the target privacy budget.

[0133] The current remaining budget is the currently available privacy budget value, which can be determined based on the difference between the total privacy budget and the cumulative privacy budget consumed. The total remaining budget can be the global remaining privacy budget for the entire knowledge base, applicable to all users and all knowledge entries, or it can be the remaining budget for a single user who initiated a data query request.

[0134] For example, if the current remaining budget is the total remaining budget of the knowledge base, the server can maintain a global budget pool, with a total budget ε in the global budget pool. total This can be configured by the administrator according to the company's security policies and compliance requirements, for example, set to version 1.0. The server can record the global cumulative consumption budget in real time.

[0135] If the current remaining budget is the remaining budget of the user who initiated the data query request, the server can maintain a user budget pool. The total budget in the user budget pool can be set differently according to the user's role; for example, the daily budget for a regular employee is 0.5, and the daily budget for an administrator is 1.0. The server can record the cumulative budget consumption of each user in real time.

[0136] Then, the current remaining budget can be compared with the target privacy budget. If the current remaining budget is greater than or equal to the target privacy budget, it means there is sufficient privacy budget to execute this query operation. Therefore, based on the target privacy budget and the data type of the query result, a corresponding noise generation algorithm can be selected, and noise data can be generated based on the target privacy budget using the corresponding noise generation algorithm. Simultaneously, the global cumulative consumption budget or the user's cumulative consumption budget is updated.

[0137] In the above embodiments, on the one hand, by introducing a budget availability judgment mechanism, the excessive consumption of privacy budget can be effectively prevented, ensuring the long-term privacy security of the knowledge base; on the other hand, by dynamically adjusting the noise intensity according to the target privacy budget, the correspondence between privacy protection intensity and budget consumption is realized, and the perturbation operation is only performed when the budget is sufficient, avoiding unnecessary waste of privacy resources.

[0138] In some alternative implementations, if the remaining budget is less than the target privacy budget, the data query results corresponding to the data query request are prohibited from being retrieved.

[0139] If the remaining budget is less than the target privacy budget, the data query results corresponding to the data query request can be refused. In other words, the user's data query request can be rejected, no query results can be returned, or a message "Insufficient budget, unable to execute query" can be returned.

[0140] In the above embodiments, prohibiting the acquisition of data query results when the privacy budget is insufficient can strictly control the consumption of the privacy budget and ensure that the knowledge base is always maintained at a secure level of privacy protection. Furthermore, this budget control mechanism can effectively prevent attackers from exhausting the privacy budget through high-frequency queries, thereby breaking through the privacy protection defenses and ensuring the long-term secure and stable operation of the knowledge base.

[0141] In some alternative implementations, if the current remaining budget is less than the target privacy budget, a new privacy budget is requested from the privacy budget manager; and if the sum of the requested privacy budget and the current remaining budget is greater than or equal to the target privacy budget, noisy data is determined based on the target privacy budget.

[0142] For example, if the remaining budget is less than the target privacy budget, the request can be rejected directly, or a new privacy budget can be automatically requested from the privacy budget manager. For instance, the server can generate a budget request form containing information such as user ID, user role, knowledge ID, knowledge sensitivity level, query type, target privacy budget, current remaining budget, and reason for the request. The reason for the request can be automatically generated or manually filled in by the user. The server sends the budget request form to the privacy budget manager, which can be via email, SMS, or in-system message. Simultaneously, the server returns a notification to the user's terminal, informing them that the budget request has been submitted and is awaiting approval. Optionally, the server can maintain a budget request queue to record all pending budget requests.

[0143] After receiving a budget request, the privacy budget manager reviews it. The review includes verifying the user's identity, the necessity of the query operation, whether the knowledge sensitivity level matches the user's permissions, and the reasonableness of the requested budget amount. The manager can approve, reject, or partially approve the budget. If the manager approves the request, the server adds the approved budget amount to the user's remaining budget or the global remaining budget.

[0144] If a new privacy budget is granted, the server recalculates the current remaining budget and compares it with the target privacy budget. If the sum of the granted privacy budget and the current remaining budget is greater than or equal to the target privacy budget, noisy data can be identified based on the target privacy budget.

[0145] If the management rejects the budget request, the server returns a rejection message to the user terminal, explaining the reason for the rejection.

[0146] In the above embodiments, by introducing a budget application and approval mechanism, while strictly controlling the consumption of privacy budget, users are provided with a channel to obtain additional budget, which meets the reasonable needs of users and improves the usability of the knowledge base.

[0147] In some alternative implementations, to avoid data query failures or privacy leaks due to the depletion of the privacy budget, a budget warning message can be sent to the privacy budget manager when the privacy budget is about to run out. That is, a budget warning message is sent to the privacy budget manager when the current remaining budget is less than or equal to the budget threshold.

[0148] The budget threshold is a preset value used to trigger budget alerts. The preset threshold can be fixed, such as a certain percentage of the total budget, like 80%. Alternatively, it can be dynamically set based on historical privacy budget consumption data to predict the average daily budget consumption. For example, the product of the predicted average daily budget consumption and a preset number of days can be used as the budget threshold. The preset number of days can be set according to actual business needs. The budget alert information is a notification sent to the privacy budget manager to inform them that the privacy budget consumption has reached or exceeded the preset threshold and that the privacy budget usage needs to be monitored.

[0149] For example, the server can monitor the current remaining budget in real time. If the remaining budget is less than or equal to a budget threshold, a global budget alert is sent to the privacy budget management system. The budget alert may include the current remaining budget, total budget, percentage consumed, and estimated time to exhaustion. It can be sent via email, SMS, or in-system message. Simultaneously, the server can log the alert event in the security log, recording the alert time, alert type, current remaining budget, and budget threshold.

[0150] In addition, the server can take some automatic measures according to preset policies, such as reducing the privacy protection strength of non-critical queries and limiting the query frequency of ordinary users, in order to extend the usage time of the budget.

[0151] In the above embodiments, by setting budget thresholds and early warning mechanisms, situations where the budget is consumed too quickly can be detected in a timely manner, and the privacy budget manager can be reminded in advance to take measures to avoid the situation where the budget is suddenly exhausted, causing the knowledge base to be unable to provide data query services, thereby improving the availability of the knowledge base.

[0152] In some alternative implementations, to conserve privacy budget, data query results can be cached. The next time a similar data query request is made, the cached results can be returned directly to the user's terminal without having to query the knowledge base again, thus avoiding the consumption of privacy budget.

[0153] For example, after obtaining a data query request, the data query result corresponding to the data query request can be retrieved from the cache medium.

[0154] If no data query results are found, candidate paths between user nodes corresponding to user identifiers and knowledge nodes corresponding to knowledge identifiers are filtered from the associated data information based on user identifiers and knowledge identifiers.

[0155] Caching media are high-speed storage devices used to store frequently accessed data, such as memory and cache.

[0156] For example, upon receiving a data query request, the request can first be hashed to generate a unique request hash value. Then, the cache can be checked to see if a cache entry exists with that hash value as the key. The cache entry can contain information such as the query result, generation time, validity period, and consumed privacy budget. If a corresponding entry exists in the cache and is still valid, the query result can be directly retrieved from the cache and returned to the user terminal without performing subsequent path filtering, privacy budget calculation, or noise addition operations, and without deducting privacy budget.

[0157] If the corresponding entry does not exist in the cache, or the entry has expired, the operation of filtering candidate paths in the associated data information based on user identifier and knowledge identifier will continue.

[0158] In the above embodiments, on the one hand, by introducing a caching mechanism, the repeated execution of privacy protection operations for the same query request can be avoided, reducing the consumption of privacy budget; on the other hand, by directly returning the data query results from the cache, the response speed of data query is improved, and the user experience is enhanced.

[0159] In some alternative implementations, see [link to relevant documentation]. Figure 3 , Figure 3 A flowchart illustrating a method for dynamically allocating a privacy budget is provided, which includes the following steps:

[0160] Step 1: Initialize the privacy budget pool. For example, initialize the total privacy budget ε. total And the cumulative consumption ∑ε.

[0161] Step 2: Obtain the data query request.

[0162] Step 3: Parse the request attribute data in the data query request.

[0163] Step 4: Determine the target privacy budget ε corresponding to the data query request. i .

[0164] Step 5: Determine whether the current remaining budget is greater than or equal to the target privacy budget, i.e., determine ε. total -∑ε≥ε i If yes, proceed to step six; otherwise, proceed to step eight.

[0165] Step 6: Determine the noisy data based on the target privacy budget.

[0166] Step 7: Update the cumulative budget consumption. That is, update ∑ε = ∑ε + ε i .

[0167] Step 8: Deny the request.

[0168] Step nine: Determine whether the early warning mechanism is triggered, i.e., determine whether ∑ε≥0.8×ε is satisfied. total If yes, proceed to step ten; otherwise, proceed to step eleven.

[0169] Step 10: Send a budget warning message to the privacy budget management provider.

[0170] Step 11, End.

[0171] In some alternative implementations, see [link to relevant documentation]. Figure 4 , Figure 4 A schematic diagram of a privacy budget dynamic allocation system is provided, specifically including:

[0172] The user terminal is used to initiate various data query requests and receive data query results.

[0173] The context parsing module performs structured parsing and validity verification on user query requests, extracts context parameters such as query type, user role, and knowledge sensitivity level, and sends these context parameters to the corresponding modules.

[0174] The anti-abuse monitoring module is used to calculate hash values ​​based on request characteristics through a caching mechanism, identify duplicate queries, and trigger a skip budget deduction instruction to prevent privacy budget abuse and cumulative leakage risks caused by high-frequency duplicate queries.

[0175] The configuration interface provides a visual parameter configuration entry point, allowing administrators to dynamically adjust the weighting coefficients and scoring rules for budget calculations online. Configuration changes take effect in real time without requiring a system restart.

[0176] The budget calculation engine combines context parameters, adjustable weights, and budget pool status to calculate the target privacy budget ε required for a single operation in real time based on a multi-dimensional weighted model. i .

[0177] The budget pool management module is used to maintain the system's global privacy budget pool, store the total budget and cumulative consumption value, respond to budget availability queries, update consumption records, and trigger warnings or operation restrictions when the budget reaches the budget threshold.

[0178] The access control decision module is used to execute the budget availability judgment logic. If the budget is sufficient, data access and disturbance operations are allowed; if the budget is insufficient, the request is rejected and an emergency approval process is prompted.

[0179] The differential privacy perturbation module adds random noise to the raw data returned by the knowledge base according to the allocated target privacy budget, outputs the results after privacy protection is completed, and synchronizes the budget consumption.

[0180] The knowledge base is used to centrally store all structured and unstructured knowledge data of an enterprise at different sensitivity levels, providing the original data source for all query operations.

[0181] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0182] Based on the same inventive concept, this application also provides a data processing apparatus for implementing the data processing method described above. The solution provided by this apparatus is similar to the implementation scheme described in the above method; therefore, the specific limitations in one or more data processing apparatus embodiments provided below can be found in the limitations of the data processing method described above, and will not be repeated here.

[0183] In one exemplary embodiment, such as Figure 5As shown, a data processing apparatus is provided, comprising:

[0184] The acquisition module 10 is used to acquire data query requests; wherein, the data query request includes user identifier and knowledge identifier;

[0185] The filtering module 20 is used to filter candidate paths between user nodes corresponding to user identifiers and knowledge nodes corresponding to knowledge identifiers in the associated data information based on user identifiers and knowledge identifiers; wherein, the nodes in the associated data information include user nodes, knowledge nodes and user organizational structure nodes; the edges in the associated data information include query permission edges connecting user nodes and knowledge nodes, membership relationship edges connecting user nodes and user organizational structure nodes, and knowledge association edges between knowledge nodes.

[0186] The determination module 30 is used to determine the target privacy budget corresponding to the data query request based on the candidate path;

[0187] Add module 40 to determine noisy data based on the target privacy budget and add the noisy data to the data query results corresponding to the data query request.

[0188] In the above embodiments, based on user identifiers and knowledge identifiers, candidate paths between user nodes corresponding to user identifiers and knowledge nodes corresponding to knowledge identifiers are filtered in the associated data information, and privacy budgets are allocated according to the risks of candidate paths. This not only avoids the problem of excessive disturbance or insufficient protection caused by fixed budgets, but also improves the utilization efficiency of privacy resources and adapts to the differences in privacy risks of different data queries.

[0189] In one embodiment, the determining module 30 is specifically used for:

[0190] Based on the path attribute information of the candidate paths, the risk coefficient of the candidate paths is determined; based on the request attribute data of the data query request, the basic privacy budget is determined; the basic privacy budget is adjusted using the risk coefficient to obtain the target privacy budget corresponding to the data query request.

[0191] In one embodiment, the determining module 30 is specifically used for:

[0192] Determine the type score corresponding to the query type, the role score corresponding to the user role, and the first sensitivity score corresponding to the knowledge sensitivity level; weight the type score, role score, and first sensitivity score to obtain the basic privacy budget.

[0193] In one embodiment, the path attribute information includes path length, knowledge sensitivity level of knowledge nodes, risk level of permission edges, and knowledge association strength; the determination module 30 is specifically used for:

[0194] Determine the length score corresponding to the path length, the second sensitivity score corresponding to the knowledge sensitivity level, the risk score corresponding to the risk level, and the association score corresponding to the knowledge association strength; weight the length score, second sensitivity score, risk score, and association score to obtain the risk coefficient of the candidate path.

[0195] In one embodiment, the added module 40 is specifically used for:

[0196] Obtain the current remaining budget; where the current remaining budget is the total remaining budget of the knowledge base or the remaining budget of the user who initiated the data query request; if the current remaining budget is greater than or equal to the target privacy budget, determine the noisy data based on the target privacy budget.

[0197] In one embodiment, the adding module 40 is further configured to:

[0198] If the remaining budget is less than the target privacy budget, the data query results corresponding to the data query request shall be prohibited.

[0199] In one embodiment, the adding module 40 is further configured to:

[0200] If the current remaining budget is less than the target privacy budget, apply for a new privacy budget from the privacy budget manager; if the sum of the applied privacy budget and the current remaining budget is greater than or equal to the target privacy budget, determine the noisy data based on the target privacy budget.

[0201] In one embodiment, the device further includes a transmitting module for:

[0202] If the current remaining budget is less than or equal to the budget threshold, send a budget warning to the privacy budget manager.

[0203] In one embodiment, the filtering module 20 is specifically used for:

[0204] The system retrieves the data query results corresponding to the data query request from the cached medium. If no data query results are found, it filters the candidate paths between the user node corresponding to the user identifier and the knowledge node corresponding to the knowledge identifier from the associated data information based on the user identifier and the knowledge identifier.

[0205] Each module in the aforementioned data processing device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0206] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 6 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores knowledge data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements a data processing method.

[0207] Those skilled in the art will understand that Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0208] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the data processing method described in any of the above embodiments.

[0209] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the data processing method described in any embodiment.

[0210] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps of the data processing method described in any embodiment.

[0211] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0212] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0213] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0214] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A data processing method, characterized by, The method includes: Obtain a data query request; wherein the data query request includes a user identifier and a knowledge identifier; Based on the user identifier and the knowledge identifier, candidate paths between the user node corresponding to the user identifier and the knowledge node corresponding to the knowledge identifier are filtered in the associated data information; wherein, the nodes in the associated data information include user nodes, knowledge nodes and user organizational structure nodes; the edges in the associated data information include query permission edges connecting user nodes and knowledge nodes, membership relationship edges connecting user nodes and user organizational structure nodes, and knowledge association edges between knowledge nodes. Based on the candidate paths, determine the target privacy budget corresponding to the data query request; Noisy data is determined based on the target privacy budget, and the noisy data is added to the data query result corresponding to the data query request.

2. The method of claim 1, wherein, The step of determining the target privacy budget corresponding to the data query request based on the candidate path includes: Based on the path attribute information of the candidate paths, the risk coefficient of the candidate paths is determined; Based on the request attribute data of the data query request, determine the basic privacy budget; The basic privacy budget is adjusted using the risk coefficient to obtain the target privacy budget corresponding to the data query request.

3. The method of claim 2, wherein, The request attribute data includes query type, user role, and knowledge sensitivity level; determining the basic privacy budget based on the request attribute data of the query request includes: Determine the type score corresponding to the query type, the role score corresponding to the user role, and the first sensitivity score corresponding to the knowledge sensitivity level; The type score, the role score, and the first sensitivity score are weighted to obtain the basic privacy budget.

4. The method of claim 2, wherein, The path attribute information includes path length, knowledge sensitivity level of knowledge nodes, risk level of permission edges, and knowledge association strength; determining the risk coefficient of the candidate path based on the path attribute information includes: Determine the length score corresponding to the path length, the second sensitivity score corresponding to the knowledge sensitivity level, the risk score corresponding to the risk level, and the association score corresponding to the knowledge association strength; The length score, the second sensitivity score, the risk score, and the association score are weighted to obtain the risk coefficient of the candidate path.

5. The method of claim 1, wherein, The step of determining noisy data based on the target privacy budget includes: Obtain the current remaining budget; wherein, the current remaining budget is the total remaining budget of the knowledge base or the remaining budget of the user who initiated the data query request; If the current remaining budget is greater than or equal to the target privacy budget, noisy data is determined based on the target privacy budget.

6. The method of claim 5, wherein, The method further includes: If the current remaining budget is less than the target privacy budget, the data query results corresponding to the data query request shall be prohibited from being obtained.

7. The method of claim 5, wherein, The method further includes: If the current remaining budget is less than the target privacy budget, request a new privacy budget from the privacy budget manager; If the sum of the requested privacy budget and the current remaining budget is greater than or equal to the target privacy budget, noisy data is determined based on the target privacy budget.

8. The method of claim 5, wherein, The method further includes: If the current remaining budget is less than or equal to the budget threshold, a budget warning message is sent to the privacy budget manager.

9. The method according to any one of claims 1 to 8, characterized in that, The step of filtering candidate paths between user nodes corresponding to the user identifier and knowledge nodes corresponding to the knowledge identifier in the associated data information based on the user identifier and the knowledge identifier includes: Retrieve the data query result corresponding to the data query request from the cache medium; If no data query results are found, candidate paths between the user node corresponding to the user identifier and the knowledge node corresponding to the knowledge identifier are filtered from the associated data information based on the user identifier and the knowledge identifier.

10. A data processing apparatus, characterized by, The device includes: The acquisition module is used to acquire data query requests; wherein, the data query request includes a user identifier and a knowledge identifier; The filtering module is used to filter candidate paths between user nodes corresponding to the user identifier and knowledge nodes corresponding to the knowledge identifier in the associated data information based on the user identifier and the knowledge identifier; wherein, the nodes in the associated data information include user nodes, knowledge nodes and user organizational structure nodes; the edges in the associated data information include query permission edges connecting user nodes and knowledge nodes, membership relationship edges connecting user nodes and user organizational structure nodes, and knowledge association edges between knowledge nodes; The determination module is used to determine the target privacy budget corresponding to the data query request based on the candidate path; An add module is used to determine noisy data based on the target privacy budget and add the noisy data to the data query result corresponding to the data query request.

11. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 9.

12. A computer readable storage medium having stored thereon a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 9.

13. A computer program product comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 9.

Images