A business process-oriented adaptive data security and permission management method

Abstract

The application discloses a kind of adaptive data security and right management methods for business process, specifically related to data security and right management technical field, first to target business process modeling, collection node, subject, environment and data attribute, form access event, and build dynamic trust chain network with trust state.Secondly, in the sliding time window, generate collaborative behavior path around sensitive data, calculate risk score combining path frequency, topological structure and cross-organizational cross-role features, identify high-risk collaborative behavior path.Then, for high-risk path, generate multi-granularity candidate right intervention measures, comprehensively evaluate risk benefits and business impact, automatically generate subject-level, session-level, resource-level and path-level right adjustment strategies and issue for execution.Finally, based on policy execution result and audit feedback, dynamically update trust state and feature weight, realize continuous and accurate interception of abnormal collaborative behavior and minimization of business impact.

Description

Technical Field

[0001] This invention relates to the field of data security and access control technology, and more specifically, to an adaptive data security and access control method oriented towards business processes. Background Technology

[0002] With the continuous development of government systems, financial systems, enterprise information systems, and industry platforms, more and more critical business processes are being migrated to distributed, multi-system collaborative online environments. Business processing models have evolved from traditional single-system, single-department approvals to collaborative business processes involving multiple departments, organizations, and even multiple units. In such collaborative business environments, on the one hand, the types of participating entities are diverse. Within the same business process, there are personnel accounts, application system accounts, third-party interface accounts, and robotic process automation accounts, etc. These entities frequently exchange tasks and data between different process nodes, forming a complex network of collaborative relationships. On the other hand, the business process structure is complex and dynamic. Processes typically include conditional branches, parallel nodes, exception rollbacks, and temporary signatures and countersignatures. The actual operational path depends on factors such as data content, approval opinions, and risk control results, exhibiting significant dynamism. Simultaneously, business processes widely involve important or sensitive data such as customer information, transaction records, approval opinions, internal reports, and regulatory data. Unauthorized access, abnormal export, or collaborative data leakage will pose serious compliance and security risks.

[0003] Existing technologies mostly employ role-based access control (RBAC) or attribute-based access control (ABAC) models, supplemented by certain risk control strategies. While these models offer relatively mature permission constraints at the single-system and single-entity granularity, they exhibit significant shortcomings in complex collaborative scenarios: permission policies are often pre-configured static rules with weak correlation to the business process context, making it difficult to reflect current nodes, data sensitivity, and recent behavioral changes in a timely manner; risk identification typically uses single access requests as the unit of analysis, rarely explicitly modeling the collaborative relationships between accessing entities; for collaborative malicious behaviors implemented in stages by multiple entities, such as combined behaviors like "low-privilege data entry—intermediate batch modification—high-privilege export—external forwarding," traditional auditing and simple rules struggle to identify and accurately block them in a timely manner; when risks are detected, common practices such as tightening role permissions or temporarily banning accounts can significantly impact normal business operations, lacking a refined strategy selection mechanism to minimize business impact while ensuring risk control; furthermore, permission policies and risk identification rules are often configured only once, lacking closed-loop optimization capabilities based on actual policy execution effects.

[0004] Therefore, there is a need in this field for a data security and access control method that can identify abnormal collaborative behavior among multiple entities and adaptively adjust permissions. Summary of the Invention

[0005] To overcome the aforementioned deficiencies in the prior art, this invention provides an adaptive data security and access control method oriented towards business processes. This method involves modeling the target business process, constructing a dynamic trust chain network, identifying high-risk collaborative behavior paths around sensitive data, and adaptively generating and adjusting access control strategies based on risk-reward and business impact, thereby solving the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: An adaptive data security and access control method oriented towards business processes includes the following steps: Step 1: Model the target business process, identify process nodes, access subjects, and data resources, and collect node, subject, environment, and data attributes when an access occurs to generate access event records; Step 2: Using the accessing entity as the node and the delegation, data interaction or collaborative operation between entities as the edge, set the trust state and update it with access events to build a dynamic trust chain network. Step 3: Construct collaborative behavior paths around sensitive data in the dynamic trust chain network, and use preset collaborative malicious behavior identification rules to analyze path frequency, path topology and cross-organizational and cross-role combinations to identify high-risk collaborative behavior paths. Step 4: Generate candidate permission intervention measures for the high-risk collaborative behavior path. The candidate permission intervention measures include at least tightening permissions, adding identity verification or approval steps, and restricting access frequency or access scope. Determine the permission adjustment strategy through a permission adjustment algorithm based on a comprehensive assessment of risk-reward and business impact, and issue it to the permission control point for execution. Step 5: Collect the execution results of the permission adjustment policy and subsequent access behavior. Based on the collected results, update the trust status of the collaborative edges related to high-risk collaborative behavior paths in the dynamic trust chain network, and update the feature weights used to identify high-risk collaborative behavior paths in the collaborative malicious behavior identification rules, so that subsequent collaborative behavior identification and permission decisions on the dynamic trust chain network are executed based on the updated trust status and feature weights.

[0007] As a further aspect of the present invention, in step two, the trust state includes a high-trust state, a normal state, a cautious state, and a high-risk state; when constructing the dynamic trust chain network, a trust state is initialized for each collaborative edge; when frequent abnormal access or suspicious behavior is detected in the collaborative operation corresponding to the collaborative edge, the trust state of the corresponding edge is adjusted from a high-trust state or a normal state to a cautious state or a high-risk state; when an audit confirms that no malicious behavior exists, the trust state of the corresponding edge is upgraded from a cautious state or a high-risk state to a normal state or a high-trust state; when an audit confirms that malicious behavior exists, the trust state of the corresponding edge is downgraded to a high-risk state, and the collaborative operation corresponding to that collaborative edge is prohibited.

[0008] As a further aspect of the present invention, step two further includes: gradually reducing the trust status of collaborative edges that have not engaged in collaborative behavior within a preset time period; when the trust status is reduced to below a preset threshold, removing the collaborative edge from the dynamic trust chain network to reflect the timeliness of the collaborative relationship and control the network size.

[0009] As a further aspect of the present invention, step three employs a sliding time window to process access events. Within each time window: access events related to the target sensitive data are filtered from the access event records; a collaborative behavior path is constructed based on the order of business process nodes and the collaborative relationship between access subjects; and the update results of the dynamic trust chain network are shared between adjacent time windows to ensure that the path construction process is consistent with the update of the trust state.

[0010] As a further aspect of the present invention, the analysis of the path topology in step three includes identifying at least one of the following structures: a multi-hop long chain structure formed within a short time window; a closed-loop structure containing multiple access subjects; and a star-shaped cluster structure centered on the same access subject.

[0011] As a further aspect of the present invention, the process of comprehensively evaluating the business impact of each candidate permission intervention measure in step four includes: evaluating the impact of the candidate permission intervention measures on the overall processing time, failure rate and service level of the target business process; setting business impact weights for each process node according to the importance of the process node in the business process; and combining the impact of each candidate permission intervention measure on different process nodes with the corresponding business impact weights to determine the business impact level of each candidate permission intervention measure.

[0012] As a further aspect of the present invention, the permission adjustment strategy in step four includes at least one level of permission adjustment strategy: subject-level, session-level, resource-level, and collaborative behavior path-level. The subject-level strategy is used to adjust the overall permission configuration of a specific access subject; the session-level strategy is used to constrain the access frequency, request source, or authentication strength within the validity period of the current access session; the resource-level strategy is used to set differentiated access control requirements for data resources of specific types or specific sensitivity levels; and the collaborative behavior path-level strategy is used to partially or completely block collaborative behavior paths that are determined to be high-risk collaborative behavior paths.

[0013] As a further aspect of the present invention, the candidate permission intervention measures in step four include at least one or more of the following: setting stricter identity verification methods or adding manual approval steps for access subjects in a cautious or high-risk state; setting an access frequency cap for access subjects whose access frequency exceeds a threshold within a preset time window; setting a single access data volume cap or export frequency cap for access requests involving highly sensitive data resources; and temporarily prohibiting or forcing manual review of collaborative operations corresponding to collaborative edges in a high-risk state.

[0014] As a further aspect of the present invention, step five, the process of updating the trust status and feature weights based on the execution result of the permission adjustment policy, includes: for high-risk collaborative behavior paths that are confirmed by manual verification to be false alarms, reducing the weight of abnormal features in the collaborative malicious behavior identification rules, increasing the weight of normal behavior features, and improving the trust status of the relevant collaborative edges; for actual security events that are successfully blocked during policy execution, increasing the weight of the corresponding abnormal features in the collaborative malicious behavior identification rules, and keeping the trust status of the relevant collaborative edges in a cautious or high-risk state to enhance the interception capability of similar collaborative behavior paths.

[0015] An adaptive data security and access control system oriented towards business processes includes a processor and a memory, wherein the memory stores a computer program, and the processor is configured to implement the steps of an adaptive data security and access control method oriented towards business processes when executing the computer program.

[0016] The technical effects and advantages of this invention, an adaptive data security and access control method oriented towards business processes, are as follows: This invention collects multi-dimensional attributes such as process nodes, access subjects, environment, and data at the business process level, constructing a dynamic trust chain network with trust states. This makes sensitive data flow paths and subject collaboration relationships explicit, facilitating the identification of abnormal collaborative behavior in the business context. By generating collaborative behavior paths within a sliding time window, and combining preset malicious collaboration identification rules with features such as path frequency, topology, and cross-organizational / cross-role characteristics to calculate risk scores, it can discover high-risk collaborative paths that are difficult to capture using traditional single-point auditing. For high-risk paths, multi-granularity access control adjustment strategies are generated, and the trust state and identification rules are updated in a closed loop based on the strategy execution results and subsequent access behavior. This achieves continuous improvement in the adaptability and refinement of access control while controlling sensitive data risks. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the overall process of an adaptive data security and access control method for business processes according to the present invention. Figure 2 This is a schematic diagram illustrating an application scenario of the present invention in a typical cross-departmental approval process; Figure 3 This is a schematic diagram of the dynamic trust chain network of the present invention; Figure 4 This is a schematic diagram illustrating the construction of cooperative behavior paths based on sliding time windows according to the present invention; Figure 5 This is a schematic diagram illustrating the adaptive permission policy generation, execution, and closed-loop feedback process of the present invention. Detailed Implementation

[0018] 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.

[0019] Example 1

[0020] This embodiment uses a government approval business platform as an application scenario to describe in detail the specific implementation process of the method of the present invention. The method is executed by an adaptive data security and access control system deployed in an independent security domain. The system interfaces with the government approval business platform, unified identity authentication service, and log auditing platform, achieving continuous monitoring of access behavior in the business process, identification of collaborative behavior risks, and adaptive optimization of access control policies without changing the existing business process configuration.

[0021] like Figure 1 As shown, the overall process of this method can be divided into five consecutive stages: business process modeling and context collection; construction and updating of dynamic trust chain network; collaborative malicious behavior identification based on dynamic trust chain network; adaptive permission decision based on dynamic trust chain network; policy execution and closed-loop feedback update.

[0022] like Figure 2 As shown, a typical government approval process in the target business platform usually includes sequentially linked process nodes such as handling, preliminary review by the department, secondary review by the department, risk control assessment, and approval by the supervising leader. This embodiment first obtains process definition information from the business platform, abstracts the above nodes as process nodes in a directed graph, and models the sequence, branching, and merging relationships between nodes as directed edges. Simultaneously, it abstracts handling personnel, reviewers, approving leaders, and risk control personnel as access subjects, and associates each access subject with its account identifier, role information, organizational affiliation, and business responsibilities.

[0023] In actual access, this embodiment uses "access event" as the basic data collection unit and records contextual information around four categories of attributes: "node, subject, environment, and data." Specifically: node attributes include the process to which it belongs, node type, node position in the process, and node processing time limit; subject attributes include the accessing subject's identity, role category, organization, and permission level; environment attributes include access time, access source IP, terminal type, and session identifier; and data attributes include the accessed data resource identifier, data sensitivity level, operation type (query, modification, export, etc.), and operation result.

[0024] The data sensitivity level can be configured based on existing data classification and grading specifications. For example, data can be divided into different levels such as public data, internal data, sensitive data, and highly sensitive data, or a grading standard can be formulated according to whether it involves personal privacy information, trade secrets, state secrets, etc., and stored in the system's configuration library in the form of a configuration table for reference by the method of this invention.

[0025] The above information is uniformly packaged into an access event record and associated with the corresponding process node and approval record. Through continuous collection and storage of access events, this embodiment obtains the complete business context of data interaction relationships and collaborative operation behaviors between various roles in cross-departmental approval business without interfering with the original business processing flow. This provides basic data support for subsequent construction of a dynamic trust chain network, identification of high-risk collaborative behavior paths, and automatic generation of permission adjustment strategies.

[0026] like Figure 3As shown, after completing the collection of access events, this embodiment constructs a dynamic trust chain network with the access subject as the node and the delegation relationship, data interaction relationship or collaborative operation relationship between the access subjects as the edge.

[0027] In the initial phase, the system uses historical access logs and audit results to statistically analyze the frequency and type of collaboration between entities, as well as known security event records. It then initializes nodes for the accessing entities and establishes collaborative edges between nodes. Each collaborative edge is associated with a trust state representing the level of trust, including high trust, normal, cautious, and high-risk states. Collaborative edges with high historical collaboration frequency and no security issues can be initialized to a high trust state; those with limited historical records or certain suspicious aspects can be initialized to a normal or cautious state.

[0028] During system operation, whenever an access event involves collaborative operations between two or more access subjects (such as delegated approval, countersigning, transfer, joint query, etc.), the system updates the trust status of the corresponding collaborative edge based on the access event: when the access behavior is detected to conform to the established business rules and no risk alarm is triggered, the trust level of the collaborative edge is appropriately increased; when frequent abnormal access, abnormal time period operations, or behaviors significantly different from historical patterns are detected, the trust status of the corresponding collaborative edge is gradually reduced from high trust or normal to cautious or high risk; when the audit confirms the existence of actual security events such as malicious export or unauthorized approval, the corresponding collaborative edge is downgraded to a high risk status, and the collaborative operation corresponding to the collaborative edge is prohibited if necessary; when the audit confirms that the previous alarm was a false alarm, the trust status of the corresponding collaborative edge can be upgraded from a cautious or high risk status to a normal or high trust status.

[0029] To prevent the trust chain network from expanding indefinitely and to reflect the timeliness of collaborative relationships, this embodiment gradually reduces the trust status of collaborative edges that have not engaged in collaborative activity within a preset time period. When the trust status drops below a preset threshold, the collaborative edge is removed from the dynamic trust chain network. Through this mechanism, the dynamic trust chain network can always maintain a sensitive awareness of current business collaborative relationships, highlighting both frequent and stable business collaboration paths while downplaying historical collaborative relationships that no longer occur.

[0030] In one optional implementation, the preset time period can be set to 30 days or 90 days, and the specific value can be adjusted according to the average collaboration frequency of the business system. For collaboration edges that do not have any collaboration behavior during the entire period, the trust level is reduced by one level after each period. When the trust status decays from the high trust status to below the threshold corresponding to the cautious status, the collaboration edge is automatically deleted from the network.

[0031] like Figure 4As shown, in order to identify collaborative behaviors that may be related to sensitive data leakage or unauthorized operations from a continuous access event stream, this embodiment uses a sliding time window to construct collaborative behavior paths.

[0032] Within each time window, the system first filters access events related to the target sensitive data from the access event records and forms an access event sequence according to the order of their occurrence. For each access event within the sliding time window, the system maps it to the corresponding access subject node in the dynamic trust chain network based on the access subject identifier. Subsequently, the system constructs multiple collaborative behavior paths based on the dependencies between access events and the collaborative edge relationships in the dynamic trust chain network. Each path reflects the collaborative operation sequence of multiple subjects around the same sensitive data or the same business instance within that time window.

[0033] After obtaining a large number of collaborative behavior paths, this embodiment uses preset collaborative malicious behavior identification rules to comprehensively analyze path frequency, path topology, and cross-organizational and cross-role combinations. Simultaneously, it refers to the trust status of collaborative edges on the path to calculate a risk score for each collaborative behavior path. When the risk score exceeds a preset threshold, the collaborative behavior path is marked as a high-risk collaborative behavior path, providing input for subsequent permission decision-making stages.

[0034] In one optional implementation of this embodiment, the collaborative malicious behavior identification rule may include one or more of the following features: path length features, such as the number of access subjects and the number of operation steps included in the path; cross-organization features, such as the number of different organizations or departments involved in the path; cross-role features, such as the number of different role categories involved in the path; time distribution features, such as the proportion of access during non-working hours (night, holidays) and the degree of high-frequency access in a short period of time; operation type features, such as the proportion of export operations, the proportion of batch modification operations, and the frequency of access to sensitive fields. Historical security event association characteristics, such as whether the path contains collaborative edges or access subjects where security events have occurred; access result characteristics, such as access failure rate and the proportion of requests rejected by risk control.

[0035] Of the aforementioned features, some can be considered "abnormal features," such as "exporting a large amount of highly sensitive data in a short period of time" or "frequently accessing the same approval item across multiple non-adjacent departments"; others can be considered "normal features," such as "processing in sequence during working hours" or "the combination of roles involved is consistent with historical normal approval patterns." The collaborative malicious behavior identification rules assign corresponding weights to each feature, and the system can calculate the risk score of the collaborative behavior path through weighted summation or other scoring functions.

[0036] For example, in a simple implementation, the following weighted scoring model can be used: First, normalize each feature to the interval [0,1], where the larger the value, the more it is biased towards abnormality; then assign a weight coefficient to each feature, with positive weights for abnormal features and negative weights for normal features, or reflect "biased towards normality" by inverting the weights; finally, the risk score can be expressed as a linear or non-linear combination of each feature value and its corresponding weight.

[0037] In a typical collaborative attack scenario, business entity A initiates multiple low-value transactions within a short period to trigger an approval process. Subsequently, entity B modifies these transactions in batches at an intermediate approval node, changing the amounts or sensitive fields to abnormal values. Finally, entity C, at a high-privilege node, centrally exports and forwards the same batch of transactions to external systems. If, within a 30-minute time window, it is detected that business instances sequentially pass through three entities with significantly different roles and belonging to different departments around the same batch of sensitive data, and the path contains a high proportion of batch modification and export operations, and at least one collaborative edge's trust status has been downgraded to a cautious or high-risk state, then the collaborative malicious behavior identification rule can raise the risk score of this collaborative behavior path above a preset threshold, thereby identifying the path as a high-risk collaborative behavior path.

[0038] In this embodiment, the size of the sliding time window and the risk scoring threshold can be configured according to business characteristics. For example, the time window can be set to 30 minutes and the risk scoring threshold can be set to 0.8; collaborative behavior paths with a risk score greater than or equal to 0.8 are marked as high-risk paths, and paths with scores between 0.5 and 0.8 are marked as medium-risk paths; the above parameters can be configured based on experience in the early stages of system launch, and then iteratively adjusted based on the strategy execution effect and audit results.

[0039] By using a path construction method based on a sliding time window and topology analysis based on a dynamic trust chain network, this embodiment can focus on multi-entity collaborative operations that occur around the same sensitive data within a short period of time, while ensuring time continuity, thus significantly improving the ability to detect sudden, highly covert collaborative attacks or violations.

[0040] like Figure 5 As shown, after identifying high-risk collaborative behavior paths that may lead to the leakage of sensitive data or unauthorized operations, this embodiment generates a corresponding adaptive permission adjustment strategy based on the path.

[0041] First, the system generates multiple candidate permission intervention measures for high-risk collaborative behavior paths. The candidate measures include at least: increasing the authentication strength or adding a manual approval process for relevant access subjects; setting an access frequency cap or temporarily limiting the flow for subjects whose access frequency exceeds the threshold; setting a single access data volume cap or export frequency cap for access requests involving highly sensitive data resources; and temporarily prohibiting or forcing manual review of collaborative operations corresponding to collaborative edges in a high-risk state.

[0042] To avoid unnecessary impacts on business operations due to excessive tightening of permissions, this embodiment conducts a comprehensive assessment of the business impact of each candidate permission intervention measure before generating the final permission adjustment strategy. Specifically, the system evaluates the impact of each candidate measure on the overall processing time, failure rate, and service level of the target business process, and sets business impact weights based on the importance of the process nodes in the business process.

[0043] In one optional implementation, the performance of each candidate permission intervention measure in terms of "risk reduction effect" and "business impact" can be normalized to the range of [0,1], where a larger value for risk reduction effect indicates that it is more conducive to reducing the probability of high-risk collaborative behavior paths, and a larger value for business impact indicates that it has a greater negative impact on the business. Then, a weighted scoring method is used to calculate the comprehensive score of each candidate measure.

[0044] Specifically, let R be the risk reduction effect of a candidate measure, and B be the business impact, both being dimensionless indicators after normalization, with values ​​ranging from [0,1]. R=0 indicates almost no risk reduction, and R=1 indicates maximum risk reduction in the current scenario. B=0 indicates negligible negative impact on business processing time, failure rate, and service level, while B=1 indicates significant negative impact on the business. Let α be the weighting parameter for the importance of risk and return, where 0 < α < 1. A larger α indicates a greater bias towards safety and a higher emphasis on risk reduction, while a smaller α indicates a greater bias towards business continuity and a higher emphasis on minimizing negative business impact.

[0045] Under the above definition, the comprehensive score S of the candidate permission intervention measures in this embodiment can be expressed as: Where S is a dimensionless comprehensive evaluation index, with a value range of [0,1]. When R is high and B is low, S is close to 1, corresponding to the preferred measure of "good risk reduction effect and small business impact"; when R is low and B is high, S is close to 0, corresponding to the inferior measure of "poor risk reduction effect and large business impact"; in other cases, it is between 0 and 1. Since R and B are both normalized dimensionless indices, and α is a dimensionless weighting coefficient, the above formula is self-consistent in terms of dimensions and is easy to calculate and sort directly in the system.

[0046] In practical implementation, it is not necessary to strictly limit it to the above linear weighted formula. This invention can also use methods such as hierarchical analysis and multi-objective optimization to construct other forms of comprehensive scoring functions, calculate and rank the comprehensive scores of candidate measures or combinations of measures, and prioritize the measures with higher comprehensive scores as part of the final permission adjustment strategy.

[0047] In terms of policy expression, this embodiment supports multiple levels of permission adjustment policies: subject-level, session-level, resource-level, and collaborative behavior path-level. Subject-level policies are used to adjust the overall permission configuration of a specific access subject, such as changing a subject from a "high-trust state" to a "cautious state" and restricting its access scope. Session-level policies are used to constrain access frequency, request source, or authentication strength within the validity period of the current access session, such as adding multi-factor authentication to the current session or limiting the number of exports within the session. Resource-level policies are used to set differentiated access control requirements for data resources with different sensitivity levels, such as only allowing access to highly sensitive data during working hours and on specific terminals. Collaborative behavior path-level policies are used to partially or completely block collaborative behavior paths that are determined to be high-risk, such as prohibiting the "batch export" operation on a high-risk collaborative edge, but retaining its "query" permission.

[0048] The generated permission adjustment policies are distributed to permission control points such as authentication gateways, business application permission modules, and interface gateways through a unified policy distribution mechanism. The permission control points then perform real-time access filtering and allow / deny processing when access requests arrive.

[0049] After the permission adjustment policy is issued and executed by each permission control point, this embodiment continues to collect and analyze the policy execution results and subsequent access behavior. On the one hand, the system records the blocked access requests and the corresponding policy hit status, and determines whether the blocking is a false alarm or a real security event that was successfully intercepted, based on the subsequent audit results. On the other hand, the system monitors whether high-risk collaborative behavior paths and similar paths reappear after the policy takes effect, and whether the behavior patterns of the relevant access subjects change.

[0050] Based on the above data collection results, this embodiment performs a closed-loop update of the dynamic trust chain network and the collaborative malicious behavior identification rules: For high-risk collaborative behavior paths that are identified as false alarms after manual verification, the system reduces the weight of abnormal features in the identification rules of the path, increases the weight of features related to normal business behavior, and improves the trust status of relevant collaborative edges in the path, so as to reduce the probability of similar behaviors being judged as high-risk in subsequent identification processes. For actual security events that are successfully blocked during policy execution, the system increases the weight of the corresponding abnormal features in the identification rules and keeps the trust status of relevant collaborative edges in a cautious or high-risk state, so as to more sensitively intercept similar collaborative behavior paths in subsequent identification.

[0051] In one alternative implementation, feature weights can be updated recursively. For example, the new weight can be set as a weighted average of the old weight and the current sample feedback result. Simultaneously, upper and lower limits for the weights can be set to prevent excessive changes in weights due to a single misjudgment. For trust status updates, the maximum allowable increase or decrease within each time period can be specified, and changes in trust status can be linked to audit confirmation results. This ensures that changes in trust status reflect changes in security risks in a timely manner while avoiding excessively frequent fluctuations.

[0052] Through the aforementioned closed-loop feedback mechanism, this embodiment achieves a self-circulating process of "identification—decision-control—evaluation—re-identification," enabling the trust status and collaborative malicious behavior identification rules in the dynamic trust chain network to continuously and adaptively optimize based on the latest business operation conditions. While ensuring the normal operation and service quality of government approval processes, this invention can gradually improve the ability to intercept abnormal collaborative behaviors and the refinement of access control.

[0053] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0054] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An adaptive data security and access control method oriented towards business processes, characterized in that, include: Model the target business process, determine the process nodes, access subjects, and data resources, and collect node, subject, environment, and data attributes during access to generate access event records; With the accessing entity as the node and delegation, data interaction or collaborative operation between entities as the edge, a dynamic trust chain network is constructed by setting the trust state and updating it with access events. Within a sliding time window, target sensitive data access events are filtered, and multi-subject collaborative behavior paths are constructed around the same sensitive data or the same business instance according to the process node sequence and subject collaboration relationship. Using preset collaborative malicious behavior identification rules, risk scores are calculated based on path frequency, topology, cross-organization and cross-role combinations, and collaboration edge trust status to identify high-risk paths. For high-risk paths, candidate permission intervention measures are generated, including tightening permissions, adding authentication or approval, and restricting access frequency or scope. A permission adjustment algorithm based on a comprehensive assessment of risk-reward and business impact determines the strategy. Collaborative behavior path-level strategies partially or completely block these paths. Partial blocking includes prohibiting batch export of high-risk collaborative edges while retaining queries. The strategy is distributed to the authentication gateway, business application permission module, and interface gateway to filter requests and allow / deny them in real time. The strategy execution results and subsequent access behaviors are collected, recording blocked requests and policy hit rates. The trust status and feature weights of collaborative edges are updated based on audit results. Subsequent identification and permission decisions use the updated trust status and feature weights.

2. The adaptive data security and access control method for business processes according to claim 1, characterized in that, The trust states include high trust state, normal state, cautious state, and high risk state; when constructing a dynamic trust chain network, the trust state is initialized for each cooperative edge; When frequent abnormal access or suspicious behavior is detected in the collaborative operation corresponding to the collaborative edge, the trust status of the corresponding collaborative edge will be adjusted from high trust or normal to cautious or high risk. When the audit confirms that there is no malicious behavior, the trust status of the corresponding collaborative edge will be upgraded from cautious or high risk to normal or high trust. When the audit confirms that there is malicious behavior, the trust status of the corresponding collaborative edge will be downgraded to high risk, and the collaborative operation corresponding to that collaborative edge will be prohibited.

3. The adaptive data security and access control method for business processes according to claim 1, characterized in that, For collaborative edges that have not engaged in collaborative behavior within a preset time period, the trust status is gradually reduced. When the trust status drops below a preset threshold, the collaborative edge is removed from the dynamic trust chain network.

4. The adaptive data security and access control method for business processes according to claim 1, characterized in that, Within each sliding time window, access events related to the target sensitive data are filtered from the access event records, and collaborative behavior paths are constructed based on the order of business process nodes and the collaborative relationships between access subjects; the update results of the dynamic trust chain network are shared between adjacent time windows.

5. The adaptive data security and access control method for business processes according to claim 1, characterized in that, The path topology includes at least one of the following: a multi-hop long chain structure formed within a short time window, a closed-loop structure containing multiple access subjects, and a star-shaped cluster structure centered on the same access subject.

6. The adaptive data security and access control method for business processes according to claim 1, characterized in that, When conducting a business impact assessment of candidate permission intervention measures, the assessment evaluates the impact of the candidate permission intervention measures on the overall processing time, failure rate, and service level of the target business process. Based on the importance of each process node in the business process, a business impact weight is assigned to each process node. The impact of each candidate permission intervention measure on different process nodes is combined with the corresponding business impact weight to determine the business impact level of each candidate permission intervention measure.

7. The adaptive data security and access control method for business processes according to claim 1, characterized in that, The permission adjustment strategy also includes at least one of subject-level, session-level, and resource-level; subject-level policies are used to adjust the overall permission configuration of the accessing subject; session-level policies are used to constrain access frequency, request source, or authentication strength within the validity period of the current access session; resource-level policies are used to set differentiated access control requirements for data resources of preset types or preset sensitivity levels.

8. The adaptive data security and access control method for business processes according to claim 1, characterized in that, The candidate permission intervention measures include at least one or more of the following: setting stricter identity verification methods or adding manual approval steps for access subjects in a cautious or high-risk state; setting an access frequency cap for access subjects whose access frequency exceeds a threshold within a preset time window; setting a single access data volume cap or an export frequency cap for access requests involving highly sensitive data resources in sensitive data. Temporarily prohibit or force manual review of collaborative operations corresponding to collaborative edges that are in a high-risk state.

9. The adaptive data security and access control method for business processes according to claim 1, characterized in that, When updating the trust status and feature weights based on the execution results of the permission adjustment policy, for high-risk collaborative behavior paths that are confirmed by manual verification to be false alarms, the weight of abnormal features in the collaborative malicious behavior identification rules in the collaborative behavior path is reduced, the weight of normal behavior features is increased, and the trust status of the relevant collaborative edges is improved; for actual security events that are successfully prevented during policy execution, the weight of the corresponding abnormal features in the collaborative malicious behavior identification rules is increased, and the trust status of the relevant collaborative edges is kept in a cautious or high-risk state.

10. An adaptive data security and access control system oriented towards business processes, characterized in that, It includes a processor and a memory, the memory storing a computer program, and the processor, when executing the computer program, is configured to implement the steps of the adaptive data security and access control method for business processes as described in any one of claims 1 to 9.

Images