Role-aware multi-digital employee collaboration conflict detection and consistency guarantee method and device

Abstract

The embodiment provides a role-aware multi-digital employee collaboration conflict detection and consistency guarantee method, establishes a role ontology model of multi-digital employees and dynamically binds and calibrates the role ontology model with digital employee instances, intercepts decision output at a digital employee execution gateway, generates a standardized decision intention flow through semantic analysis and encodes the standardized decision intention flow into a decision intention vector, reads the role ontology model, the standardized decision intention flow and the decision intention vector, identifies resource conflicts, time sequence conflicts and semantic conflicts, and distinguishes role perspective differences to generate multi-level conflict detection results, generates conflict arbitration strategies according to business influence and loss reversibility after grading, executes the conflict arbitration strategies, issues arbitration instructions, and verifies the consistency of coordinated decisions, records decision data, constructs a decision link correlation graph, reversely traces a root cause to generate a responsibility attribution result, and according to the attribution result and feedback, updates conflict detection rules and arbitration strategies and returns to the role ontology model.

Description

Technical Field

[0001] This application relates to the fields of multi-agent systems and intelligent enterprise operation technology, specifically to a method and apparatus for detecting and ensuring consistency in multi-digital employee collaboration based on role perception. Background Technology

[0002] Existing multi-digital employee collaboration systems often employ a flat task allocation mechanism, treating each intelligent agent as a homogeneous execution unit. This ignores the differentiated roles that different digital employees play in enterprise operations, leading to a mismatch between task allocation and agent capabilities, and making it difficult to establish responsibility boundaries that align with business logic. Existing conflict detection methods primarily focus on explicit resource competition, such as multiple agents simultaneously writing to the same database field. They lack the ability to identify semantic-level conflicts. When multiple system interfaces with different decision operations contradict each other in terms of business objectives, problems often only surface at the execution level, resulting in a low detection rate for implicit conflicts.

[0003] Existing consistency guarantee mechanisms mostly employ global locks or centralized arbitration. When a digital employee's decision requires coordination, the execution of all related agents is suspended until the conflict is resolved. This approach results in a high proportion of coordination waiting time in high-concurrency business scenarios, making it difficult to meet the real-time requirements of the business. Existing conflict handling relies heavily on fixed priority rules, lacking dynamic awareness of business contexts such as market conditions, business cycles, user intent, and compliance constraints. Rigid rules lead to a significant proportion of conflict handling results deviating from the optimal business solution.

[0004] The more fundamental problem lies in the lack of continuous tracking and tracing capabilities for decision consistency in existing collaborative systems. When business results deviate, it is difficult to trace back to which specific digital employee made a decision inconsistent with the overall goal and when, leading to vague attribution of responsibility and time-consuming problem localization. Existing multi-agent coordination solutions mostly use task dependency graphs to detect circular dependencies to determine the execution order, only addressing task-level execution order issues and failing to address semantic conflict detection based on role responsibilities, dynamic context-aware conflict arbitration, and end-to-end tracking of decision consistency. Therefore, a multi-digital employee collaboration conflict detection and consistency assurance method is needed that combines role ontology modeling, multi-level conflict detection, context-aware arbitration, and decision link tracing. Summary of the Invention

[0005] To address the problems of existing multi-digital employee collaboration methods that ignore role differences, limit conflict detection to explicit resource competition, and rely on global locks or fixed priorities for consistency assurance without decision-making link tracing, this application provides a role-aware multi-digital employee collaboration conflict detection and consistency assurance method and apparatus. It can model the differences in digital employee work through a role ontology model and distinguish role perspectives, identify implicit conflicts through a three-layer progressive detection of resources, time sequence, and semantics, ensure decision consistency with low coordination costs through context-aware arbitration, and support root cause tracing and responsibility attribution through a decision-making link association graph, thereby improving the detection coverage, coordination timeliness, and problem location accuracy of collaborative conflicts.

[0006] To solve at least one of the above problems, this application provides the following technical solution:

[0007] Firstly, this application provides a method for detecting and ensuring consistency in multi-digital employee collaboration based on role awareness, comprising: establishing a role ontology model of multiple digital employees and dynamically binding and continuously calibrating it with digital employee instances; intercepting decision output and performing semantic parsing at the digital employee execution gateway; generating a standardized decision intent flow based on a preset decision semantic element template; encoding the standardized decision intent flow into a decision intent vector through a domain language model; reading the role ontology model, the standardized decision intent flow, and the decision intent vector; and performing conflict detection and consistency assurance based on global resource lock status, role collaboration relationships and business timing constraints, conflict pattern library matching, and vector adversarial analysis. The system identifies resource conflicts, temporal conflicts, and semantic conflicts, distinguishes differences in role perspectives, and generates multi-level conflict detection results. These results are then categorized based on the scope of business impact and the reversibility of losses. A conflict arbitration strategy is generated by combining the categorized results with the business context. The conflict arbitration strategy is executed, issuing arbitration instructions to the involved digital employees and verifying the consistency of decisions after coordination to generate consistency assurance results. Data from decision generation to execution is recorded to construct a decision link association graph. Based on this association graph, root causes are traced back to generate responsibility attribution results when business deviations occur. The conflict detection rules and arbitration strategy are updated based on the attribution results and feedback, and then fed back to the role ontology model.

[0008] Furthermore, it also includes: establishing a role ontology model for each business role based on a preset role ontology structure, including responsibility domains, capability lists, and collaborative relationships; having business experts label the responsibility boundaries and permission constraints of each role; extracting the actual capability range from the historical behavior logs of digital employee instances to generate role profiles; comparing and verifying the capability list of the role profiles with the capability list of the role ontology model to identify capability gaps and capability overflows; binding digital employee instances to the corresponding roles; monitoring the deviation between the behavior distribution of the digital employee instances and the typical distribution of the bound roles; and registering role deviation flags and triggering behavior constraints and role reassessment when the deviation exceeds a preset threshold condition.

[0009] Furthermore, it also includes: deploying an interceptor at the execution gateway to capture the call requests from each digital employee to external systems and the collaborative communication between digital employees; performing semantic parsing on the captured decision output and mapping it to a predefined action ontology; organizing and generating a standardized decision intent flow based on a preset decision semantic element template; encoding the action and object elements of the standardized decision intent flow into a domain language model as a high-dimensional vector, so that decision intents with similar semantics are close in distance in the vector space to generate decision intent vectors, and writing the decision intent vectors with the decision subject identifier as the primary key into the intent cache.

[0010] Furthermore, it also includes: maintaining a global resource lock status table that records the operation status of each business entity and its holder; comparing the target object of a newly arriving decision intent with the global resource lock status table to identify write-write conflicts and read-write conflicts; verifying the decision sequence based on the role collaboration relationship diagram and the time sequence constraints defined in the business process to identify time sequence conflicts; matching the standardized decision intent flow with a conflict pattern library to identify explicit semantic conflicts; calculating the adversarial degree of the decision intent vector in the business effect dimension for decision pairs not covered by the conflict pattern library; and determining that decision pairs with adversarial degree exceeding a preset threshold and pointing to related objects are semantic conflicts after being identified by role perspective differences, generating multi-level conflict detection results.

[0011] Furthermore, it also includes: calculating the severity of each conflict in the multi-level conflict detection results according to preset severity assessment rules, prioritizing the severity according to preset grading standards to generate grading results; collecting business context containing market status and compliance constraints in real time, selecting corresponding strategies from a preset arbitration strategy library based on the priority of the grading results and the business context, and generating conflict arbitration strategies with additional explanations.

[0012] Furthermore, it also includes: issuing arbitration instructions of the corresponding type to the digital employees involved according to the conflict arbitration strategy and receiving execution confirmations returned by each digital employee; triggering a secondary coordination and escalation process for execution confirmations that return rejections; verifying the consistency of the decision combination after the digital employees involved execute the arbitration instructions according to the preset multi-dimensional consistency verification rules to generate a consistency guarantee result; rolling back the relevant decisions and re-entering the conflict detection when the consistency verification fails; executing high-priority role decisions first for time-sensitive conflicts and establishing compensation transactions for rollbackable operations.

[0013] Furthermore, it also includes: recording the full lifecycle data of each decision's generation content, conflict detection results, arbitration processing, and final execution status; establishing association edges between each decision based on preset association categories to construct a decision link association graph; when business indicators exceed the threshold condition, tracing back from the deviation results to the root cause decision nodes along the decision link association graph and calculating the contribution of each node to generate responsibility attribution results; adjusting and supplementing the conflict detection rules and the conflict arbitration strategy execution parameters based on the responsibility attribution results and effect feedback, and then returning the data to the role ontology model after grayscale verification.

[0014] Secondly, this application provides a role-aware multi-digital employee collaboration conflict detection and consistency assurance device, comprising: a role modeling module, used to establish a role ontology model for each business role, including a responsibility domain, capability list, and collaboration relationship, based on a preset role ontology structure, and dynamically bind and continuously calibrate the role ontology model with each digital employee instance; an intent extraction module, used to intercept decision output and collaboration communication at the execution gateway of each digital employee and perform semantic parsing on the intercepted decision output, organize and generate a standardized decision intent flow based on a preset decision semantic element template, and encode it into a decision intent vector through a domain language model; and a conflict detection module, used to read the role ontology model, the standardized decision intent flow, and the decision intent vector, and, based on the full... The system identifies resource conflicts, temporal conflicts, and semantic conflicts by matching resource lock status, role collaboration relationships, business timing constraints, conflict pattern library matching, and the adversarial degree of the decision intent vector. It generates multi-level conflict detection results. An arbitration guarantee module assesses the severity and classifies the multi-level conflict detection results based on the scope of business impact and the reversibility of losses. It then generates a conflict arbitration strategy based on real-time business context, issues arbitration instructions to the relevant digital employees, verifies the global consistency of the coordinated decision, and generates a consistency guarantee result. A tracing and attribution module records the generation, conflict, coordination, and execution data of decisions to construct a decision link association graph. When business results deviate, it traces back to the root cause to generate responsibility attribution results and updates the conflict detection rules and the conflict arbitration strategy based on feedback from the collaboration effect.

[0015] Thirdly, this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the role-aware multi-digital employee collaboration conflict detection and consistency assurance method.

[0016] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the role-aware multi-digital employee collaboration conflict detection and consistency assurance method described above.

[0017] Fifthly, this application provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of the role-aware multi-digital employee collaboration conflict detection and consistency assurance method described above.

[0018] As can be seen from the above technical solution, this application provides a method and device for detecting and ensuring consistency of multi-digital employee collaboration conflicts based on role awareness. It formally models and dynamically binds and calibrates the role responsibilities of digital employees through a role ontology model. It identifies resource, temporal, and semantic conflicts and distinguishes role perspective differences through a three-layer progressive detection of resource lock status, temporal constraints, and vector adversarial degree. It reduces coordination costs while ensuring global consistency through the selection of arbitration strategies based on business context awareness. It generates responsibility attribution results through the construction of decision link association graph and reverse tracing. Based on the responsibility attribution results and effect feedback, it updates the conflict detection rules and conflict arbitration strategies and feeds back to the role ontology model to form a closed loop. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0020] Figure 1 This is a flowchart illustrating the role-aware multi-digital employee collaboration conflict detection and consistency assurance method in the embodiments of this application;

[0021] Figure 2 This is a schematic diagram of the layered architecture and device structure of the multi-digital employee collaboration conflict detection and consistency assurance system in this application embodiment;

[0022] Figure 3 This is a schematic diagram of the digital employee role ontology model structure and instance binding calibration process in the embodiments of this application;

[0023] Figure 4 This is a schematic diagram of the semantic standardization and multi-level conflict detection process for decision intent extraction in this application embodiment;

[0024] Figure 5 This is a schematic diagram illustrating the process of ensuring consistency between context-aware arbitration strategy selection and arbitration execution in an embodiment of this application.

[0025] Figure 6 This is a schematic diagram of the decision link association graph construction, responsibility attribution, and closed-loop iteration process of collaboration rules in the embodiments of this application. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of this application, but not all embodiments.

[0027] Considering the problems of existing multi-digital employee collaboration methods that ignore role differences, limit conflict detection to explicit resource competition, and use global locks or fixed priorities for consistency assurance without decision link tracing, this application provides a role-aware multi-digital employee collaboration conflict detection and consistency assurance method. Through role ontology modeling, multi-level conflict detection, context-aware arbitration, and decision link tracing, it automatically detects and coordinates collaboration conflicts among multi-digital employees.

[0028] This application provides an embodiment of a role-aware multi-digital employee collaboration conflict detection and consistency assurance method, see [link to implementation details]. Figure 1 The method specifically includes the following:

[0029] Step S101: Establish a role ontology model of multiple digital employees and dynamically bind and continuously calibrate it with digital employee instances. Intercept decision output at the digital employee execution gateway and perform semantic parsing. Generate a standardized decision intent flow based on the preset decision semantic element template. Encode the standardized decision intent flow into a decision intent vector through the domain language model.

[0030] In this embodiment, the method runs on a multi-digital employee collaboration platform, such as... Figure 1 As shown, the platform adopts a layered architecture consisting of a role modeling layer, a conflict detection layer, a consistency assurance layer, and a tracing and attribution layer. The role ontology model includes role identifiers, responsibility domains, capability lists, decision-making authority matrices, and collaboration relationships. Figure 5 Each field is organized, and the role identifier is a globally unique code.

[0031] In one feasible implementation, the role ontology model is constructed by combining business expert annotation with historical log extraction. Business experts annotate the boundaries of responsibilities and permission constraints according to operational guidelines. The platform extracts actual tool calls and data sources from the historical behavior logs of digital employee instances to generate role profiles, which are then compared and verified with the expert annotations before being written into the role ontology model library.

[0032] Based on this, role binding is performed on digital employee instances. The actual capability configuration of the instance is compared item by item with the capability list of the bound role. Capability gaps are marked as needing to be filled and the configuration completion process is initiated. Capability overflows are marked as exceeding authority risk and trigger permission convergence. One instance is bound to one primary role and one auxiliary role converged according to the least privilege.

[0033] Optionally, the platform measures role deviation by the divergence between the behavior distribution and the typical behavior distribution of the bound role. When the divergence exceeds a preset threshold, a role deviation flag is registered and behavioral constraints, role reassessment, or hybrid role authorization are triggered. The preset threshold is calibrated based on operational experience and adjusted according to business flexibility.

[0034] In the parallel processing path, interceptors are deployed at each digital employee execution gateway to capture call requests to external systems and collaborative communications between digital employees. The interceptors perform semantic parsing on the captured decision outputs and map the operation types to predefined action ontologies. Decision outputs that fail to be parsed are registered with parsing exception flags and transferred to a manual verification queue.

[0035] In one specific implementation, the parsing results are organized into a seven-tuple structure of subject, role, action, object, parameter, time, and basis to generate a standardized decision intent flow. The subject is identified by a digital employee initiating the decision, the object uses a unified entity identifier, the parameter records quantified values, and the time is accurate to milliseconds. Based on this, the action and object elements are input into the domain language model and encoded into a decision intent vector.

[0036] The standardized decision intent flow is organized with seven fields: subject identifier, role, action code, object identifier, parameter set, timestamp, and basis summary. The decision intent vector is written to the intent cache after being aligned with the decision subject identifier as the primary key. It is then read by timestamp at the conflict detection entry point in step S102. The intent registration code that has timed out is marked with a timeout flag and is represented by a downgrade.

[0037] Step S102: Read the role ontology model, the standardized decision intent flow, and the decision intent vector. Based on the global resource lock state, role collaboration relationship and business timing constraints, conflict pattern library matching, and vector adversarial degree, identify resource conflicts, timing conflicts, and semantic conflicts respectively, and distinguish role perspective differences to generate multi-level conflict detection results. Classify the detection results according to the business impact scope and loss reversibility, and generate a conflict arbitration strategy based on the classification results in combination with the business context.

[0038] The standardized decision intent stream and decision intent vector output in step S101 are read into the conflict detection stage along with the role ontology model. Figure 1 As shown, the decision-making entity identifier links the three entities. When the corresponding role is missing in the role ontology model, the role is registered without an identifier. Conflict detection is performed progressively in three layers: resource, time sequence, and semantics.

[0039] In one specific implementation, resource conflict detection maintains a global resource lock status table. This table uses the business entity identifier as the primary key and records the operation status and holder. The operation status takes four values: idle, reading, writing, and locked. Newly arrived target objects are compared to this table; simultaneous writing is considered a write-write conflict, and a read-write interaction is considered a read-write conflict.

[0040] Based on this, timing conflict detection is performed. Three types of timing constraint rules are established according to the role collaboration relationship diagram and business process: prerequisite dependency, mutual exclusion window, and maximum delay. Prerequisite dependency restricts one type of decision to be completed before another type. Mutual exclusion window restricts two types of decisions from taking effect in the same window. When the decision sequence violates the constraints, it is marked as a timing conflict.

[0041] Semantic conflict detection employs a combination of conflict pattern library matching and vector adversarial detection. Explicit semantic conflicts are identified by matching the current decision intent against the conflict pattern library. For uncovered decision pairs, it is first determined whether they point to related objects. If related, the adversarial strength of the decision intent vector in the business effect dimension is calculated. Decision pairs with adversarial strength exceeding a preset threshold are submitted for further review.

[0042] Optionally, semantic conflict detection distinguishes between differences in role perspectives and conflicts over real objectives. Decisions based on differing directions from benefit and risk perspectives are judged as differences in role perspectives and enter the coordination process. Decisions pointing to the same object and whose actions cancel each other out are judged as conflicts over real objectives. The adversarial threshold is determined based on the subject operating characteristic curves on the labeled dataset.

[0043] The three-layer detection results are combined to generate a multi-layer conflict detection result, which is organized by fields such as conflict identifier, conflict type, roles of the conflicting parties, involved objects, and occurrence time. Based on this, the severity is calculated from the dimensions of business impact scope, loss reversibility, time sensitivity, role weight, and historical patterns, and divided into four priorities: urgent, high, medium, and low.

[0044] In one feasible implementation, a conflict arbitration strategy is generated based on the hierarchical results, taking into account the real-time business context. The business context collects market status, business cycle, user intent, compliance constraints, and resource status. The arbitration strategy library pre-sets five types of templates: role priority, timing adjustment, parameter compromise, scope isolation, and escalation to manual arbitration. The rule engine selects a strategy and writes it into the arbitration queue.

[0045] Step S103: Execute the conflict arbitration strategy to issue arbitration instructions to the digital employees involved and verify the consistency of the coordinated decision to generate a consistency guarantee result. Record the data from decision generation to execution to construct a decision link association graph. Based on the association graph, trace back the root cause when there is a business deviation to generate a responsibility attribution result. Based on the attribution result and feedback, update the conflict detection rules and arbitration strategy and return them to the role ontology model.

[0046] The conflict arbitration strategy output in step S102 is read into the consistency guarantee stage, such as... Figure 1 As shown. In accordance with the aforementioned conflict arbitration strategy, arbitration orders are issued to the digital employees involved. Arbitration orders are of four types: decision taking effect, decision postponement, decision amendment, and decision revocation. Decision postponement includes an estimated recovery time.

[0047] In one feasible implementation, the digital employee being arbitrated returns an execution confirmation upon receiving the instruction. The execution confirmation has three states: acceptance, partial acceptance, and rejection. An execution confirmation that returns rejection triggers a secondary coordination or escalation process. The secondary coordination process re-reads the business context to select alternative strategies, while the escalation process submits the conflict for manual approval.

[0048] Based on this, a consistency check is performed on the decision combination after executing the arbitration instruction. The consistency check verifies the uniqueness of resource writes, the legality of the timing, the semantic consistency, and the compliance of permissions in sequence. The uniqueness of resource writes requires that there is only a single writer for the same resource at any given time. If the check passes, a consistency guarantee result is generated; if it fails, the relevant decision is rolled back and the process is re-entered into step S102.

[0049] In one specific implementation, an optimistic execution and fast rollback mechanism is used for time-sensitive urgent conflicts. First, the decision of the high-priority role is executed immediately. Simultaneously, a compensation transaction is established for rollback-capable operations. This compensation transaction records a snapshot of the original state and a rollback entry point. If subsequent checks reveal that adjustments are needed, the compensation transaction is executed to restore the original state.

[0050] Data on decision generation, conflict resolution, coordination, and execution are aggregated and recorded as full lifecycle data according to decision identifiers and written to an append-only decision log. The full lifecycle data encompasses four categories: decision generation records, conflict detection records, arbitration processing records, and final execution records. Each record is organized using a decision identifier, timestamp, and status field. Records that fail to be archived are recorded with an archiving exception identifier.

[0051] When a deviation in business results is triggered, the process traces backward along the decision-making link correlation graph from the deviated business result, such as... Figure 1 As shown in the diagram, the decision-making link association graph uses decisions as nodes and four types of associations—causality, conflict, coordination, and dependence—as edges. It traces and identifies the final execution decisions that directly lead to deviations and calculates the contribution based on the direct, indirect, coordination, and external categories to generate responsibility attribution results.

[0052] Based on the attribution results and feedback on collaboration effectiveness, the conflict detection rules and conflict arbitration strategies are updated. Effectiveness feedback is collected to assess changes in short-term business metrics, long-term business impact, and business personnel satisfaction. Updates are triggered when the false alarm rate exceeds a threshold or satisfaction falls below a threshold. Update methods include parameter adjustments, rule additions, and rule obsolescence. Updated rules are then validated in a gray-scale environment before being reverted to the previous state.

[0053] As can be seen from the above description, the multi-digital employee collaboration conflict detection and consistency assurance method based on role awareness provided in this application can distinguish role perspective differences through role ontology modeling, identify implicit conflicts through resource, temporal and semantic three-layer progressive detection, ensure decision consistency through context-aware arbitration, and support root cause reverse tracing and responsibility attribution through decision link association graph.

[0054] In one embodiment of the role-aware multi-digital employee collaboration conflict detection and consistency assurance method of this application, see [link to relevant documentation]. Figure 3 It can also specifically include the following:

[0055] Step S201: Based on the preset role ontology structure, establish a role ontology model for each business role, including responsibility domains, capability lists and collaboration relationships. Business experts mark the responsibility boundaries and permission constraints of each role, and extract the actual capability range from the historical behavior logs of digital employee instances to generate role profiles.

[0056] Step S202: After comparing and verifying the capability list of the character profile with that of the character body model to identify capability gaps and capability overflows, bind the digital employee instance to the corresponding role, monitor the deviation between the behavior distribution of the digital employee instance and the typical distribution of the bound role, and register the role deviation flag when the deviation exceeds the preset threshold condition and trigger behavior constraints and role reassessment.

[0057] Step S101, the role modeling layer, establishes a role ontology model for each business role, such as... Figure 3 As shown, the role ontology model is organized using role identifiers, responsibility domains, capability lists, decision-making permission matrices, and collaboration relationship diagrams. Role identifiers are globally unique role codes. Responsibility domains use business ontology terminology to describe the business scope. The capability list enumerates the tools that can be invoked, the data sources that can be accessed, and the types of operations that can be performed. The decision-making permission matrix defines the role's read-only, suggestion, approval, and execution permissions for various business objects. The collaboration relationship diagram defines the information provision, decision-making approval, and execution supervision relationships between this role and other roles, as well as their triggering conditions. Business experts mark responsibility boundaries and permission constraints according to operational specifications. The platform extracts the actual capability scope from the historical behavior logs of digital employee instances to generate role profiles.

[0058] The character portrait is compared and verified item by item with the ability list of the character's actual model, such as... Figure 3As shown. The system verifies and identifies capability gaps and overflows. Capability gaps are marked as needing to be filled and the configuration completion process begins. Capability overflows are marked as exceeding authority risk and trigger permission convergence confirmation. Based on this, digital employee instances are bound to corresponding roles. One instance is bound to one primary role and a preset number of auxiliary roles converged with minimum permissions. The system continuously monitors the divergence between the behavioral distribution of the digital employee instance and the typical behavioral distribution of the bound role. When the divergence exceeds a preset threshold, a role deviation marker is registered. This role deviation marker triggers behavioral constraints, role reassessment, or mixed role authorization. The preset threshold is determined based on operational experience. The binding relationship table uses the instance identifier as the primary key for step S102 to access at the read entry point.

[0059] In one embodiment of the role-aware multi-digital employee collaboration conflict detection and consistency assurance method of this application, see [link to relevant documentation]. Figure 4 It can also specifically include the following:

[0060] Step S301: Deploy an interceptor at the execution gateway to capture the call requests of each digital employee to external systems and the collaborative communication between digital employees. Perform semantic parsing on the captured decision output and map it to a predefined action ontology. Organize and generate a standardized decision intent flow according to the preset decision semantic element template.

[0061] Step S302: The action and object elements of the standardized decision intent flow are input into the domain language model and encoded into high-dimensional vectors, so that decision intents with similar semantics are close in distance in the vector space to generate decision intent vectors. The decision intent vectors are written into the intent cache with the decision subject identifier as the primary key.

[0062] Step S101 describes deploying an interceptor at the execution gateway to capture the decision output, such as... Figure 4 As shown, the interceptor captures interface calls, data writes, and message sending requests from each digital employee to external systems, as well as collaborative communications between digital employees. It performs semantic parsing on the captured decision outputs and maps the operation types to predefined action ontologies. The parsing results are organized into a seven-tuple structure: subject, role, action, object, parameter, time, and basis. The subject is the identifier of the digital employee initiating the decision; the object uses a unified entity identifier; parameters record quantified values; and time is accurate to milliseconds. Based on this, a standardized decision intent stream is generated. Decision outputs that fail to be parsed are registered with a parsing exception flag and transferred to a manual verification queue. After the verification result is written back, the parsing process is restarted.

[0063] The standardized decision intent flow's action and object elements are encoded as high-dimensional vectors in the domain language model, such as... Figure 4As shown, encoding generates decision intent vectors by grouping semantically similar decision intents close together in the vector space. These decision intent vectors, with the decision subject identifier as the primary key, are aligned with the standardized decision intent stream by timestamp and written into the intent cache. This provides a vector representation for semantic conflict detection. The intent cache is organized with subject identifier, action code, object identifier, and vector representation fields. Decision intents whose vector encoding times out are registered with a timeout flag and represented using a downgraded representation based on key fields. The standardized decision intent stream and the decision intent vectors are read sequentially by timestamp at the conflict detection entry point in step S401.

[0064] In one embodiment of the role-aware multi-digital employee collaboration conflict detection and consistency assurance method of this application, see [link to relevant documentation]. Figure 4 It can also specifically include the following:

[0065] Step S401: Maintain and record the operation status of each business entity and the global resource lock status table of the holder; compare the target object of the newly arrived decision intention with the global resource lock status table to identify write-write conflicts and read-write conflicts; verify the decision sequence and identify timing conflicts based on the role collaboration relationship diagram and the timing constraints defined in the business process.

[0066] Step S402: Match the standardized decision intent stream with the conflict pattern library to identify explicit semantic conflicts. For decision pairs not covered by the conflict pattern library, calculate the adversarial degree of the decision intent vector in the business effect dimension. Decision pairs with adversarial degree exceeding the preset threshold and pointing to related objects are judged as semantic conflicts after role perspective difference identification, and multi-level conflict detection results are generated.

[0067] After the standardized decision intent flow and decision intent vector in step S301 are ready, resource conflict and timing conflict detection are performed, such as... Figure 4 As shown, resource conflict detection maintains a global resource lock status table with the business entity identifier as the primary key. This table records the operation status and holder, with operation statuses including idle, reading, writing, and locking. Newly arriving decision intentions are compared with the target object in the global resource lock status table to identify write-write conflicts and read-write conflicts. Based on this, the decision sequence is verified according to the role collaboration relationship diagram and the pre-dependencies, mutual exclusion windows, and maximum latency constraints defined in the business process. Decision combinations that violate constraints are marked as timing conflicts.

[0068] Semantic conflict detection matches the standardized decision intent stream with a conflict pattern library, such as... Figure 4As shown. The conflict pattern library identifies explicit semantic conflicts. Decision pairs not covered by the conflict pattern library are first determined by the business entity relationship graph to determine if they point to related objects. If related, the adversarial degree of the decision intent vector in the business effect dimension is calculated. Based on this, decision pairs with adversarial degrees exceeding a preset threshold are identified as having conflicts between role perspective differences and the actual goal. Decisions with different directions given from different perspectives of benefit and risk enter the coordination process. Decisions pointing to the same object and whose actions cancel each other out are judged as semantic conflicts. The three-layer detection results are combined to generate multi-level conflict detection results, which are written into the conflict queue with the conflict identifier as the primary key. The adversarial degree threshold is calibrated based on the receiver operating characteristic curve analysis, and is read at the hierarchical entry point in step S501.

[0069] In one embodiment of the role-aware multi-digital employee collaboration conflict detection and consistency assurance method of this application, see [link to relevant documentation]. Figure 5 It can also specifically include the following:

[0070] Step S501: Calculate the severity of each conflict in the multi-level conflict detection results according to the preset severity assessment rules, and divide the severity into priorities according to the preset grading standards to generate a grading result;

[0071] Step S502: Collect business context containing market status and compliance constraints in real time, select the corresponding strategy from the preset arbitration strategy library according to the priority of the classification result and the business context, and generate a conflict arbitration strategy with additional explanation.

[0072] After the multi-level collision detection results of step S401 are ready, perform severity assessment and grading, such as... Figure 5 As shown, the severity of each conflict is calculated based on the dimensions of business impact scope, loss reversibility, time sensitivity, role weight, and historical patterns. Business impact scope is measured by the number of affected business entities; loss reversibility is differentiated into three categories: financial loss, inventory mismatch, and content release; and time sensitivity is measured by the available time window for resolving the conflict. Based on this, each conflict is classified into four priorities—urgent, high, medium, and low—according to a preset grading standard. The grading threshold is comprehensively determined based on the amount of business impact and response time requirements, and can be adjusted according to the enterprise's risk appetite.

[0073] The classification results, combined with real-time business context, generate a conflict arbitration strategy, such as... Figure 5As shown. The business context collects market status, business cycle, user intent, compliance constraints, and resource status in real time. The arbitration strategy library pre-sets five types of strategy templates: role priority, timing adjustment, parameter compromise, scope isolation, and escalation to manual intervention. Based on this, the decision tree and rule engine select strategies according to conflict priority and business context. Urgent priority conflicts are quickly decided using the role priority strategy; high and medium priority conflicts are evaluated for feasibility and then the best strategy is selected; low priority conflicts are handled using default rules. The selected strategy is accompanied by an explanation including the reasons for selection, contextual factors, and expected impact to generate a conflict arbitration strategy, using the conflict identifier as the primary key for step S601 to read at the execution entry point.

[0074] In one embodiment of the role-aware multi-digital employee collaboration conflict detection and consistency assurance method of this application, see [link to relevant documentation]. Figure 5 It can also specifically include the following:

[0075] Step S601: Issue the corresponding type of arbitration instruction to the digital employees involved according to the conflict arbitration strategy and receive the execution confirmation returned by each digital employee. Trigger the secondary coordination and escalation process for the execution confirmation that returns a rejection.

[0076] Step S602: After executing the arbitration instruction on the digital employees involved, the decision combination is verified for consistency according to the preset multi-dimensional consistency verification rules to generate a consistency guarantee result. If the consistency verification fails, the relevant decision is rolled back and the conflict detection is re-entered. For time-sensitive conflicts, the high-priority role decision is executed first and a compensation transaction is established for the rollback operation.

[0077] After the conflict arbitration strategy in step S501 is ready, the arbitration order is issued and the execution confirmation is collected, such as... Figure 5 As shown. Based on the aforementioned conflict arbitration strategy, instructions for decision implementation, decision postponement, decision modification, and decision revocation are issued to the relevant digital employees. Decision postponement includes an estimated recovery time, decision modification includes parameter adjustment requirements, and decision revocation includes a reason explanation. Accordingly, the system receives acceptance, partial acceptance, and rejection confirmations from each digital employee. Rejection confirmations trigger a secondary coordination or escalation process. Secondary coordination involves rereading the business context to select alternative strategies, while escalation submits the conflict along with the decision recommendations for manual approval.

[0078] Perform consistency checks on the decision combination after executing the arbitration instruction, such as... Figure 5As shown. The consistency check sequentially verifies the uniqueness of resource writes, the legality of timing, semantic consistency, and permission compliance. Resource write uniqueness requires that only a single writer can write the same resource at any given time. Permission compliance requires that all effective decisions are within the scope of the corresponding role's permissions. If the check passes, a consistency guarantee result is generated; if it fails, the relevant decision is rolled back and the conflict detection in step S401 is re-entered. Accordingly, for conflicts with urgent time sensitivity, optimistic execution is adopted, allowing high-priority role decisions to be executed immediately and establishing compensation transactions for rollbackable operations. If subsequent checks find that adjustments are needed, the compensation transactions are executed to restore the original state. The consistency guarantee result uses the conflict identifier as the primary key for step S701 to read at the tracing entry point.

[0079] In one embodiment of the role-aware multi-digital employee collaboration conflict detection and consistency assurance method of this application, see [link to relevant documentation]. Figure 6 It can also specifically include the following:

[0080] Step S701: Record the full lifecycle data of each decision's generation content, conflict detection results, arbitration processing, and final execution status; construct a decision link association graph by establishing association edges between each decision based on preset association categories.

[0081] Step S702: When the business indicators exceed the threshold condition, trace back from the deviation results to the root cause decision nodes along the decision link association graph and calculate the contribution of each node to generate the responsibility attribution results. Based on the responsibility attribution results and effect feedback, adjust the conflict detection rules and the conflict arbitration strategy execution parameters and supplement the rules. After grayscale verification, the data is fed back to the role ontology model.

[0082] After the consistency guarantee result of step S601 is ready, the entire decision-making chain is traced, such as... Figure 6 As shown, the entire lifecycle data of each decision—including its generation content, conflict detection results, arbitration processing, and final execution status—is recorded and written to an append-only decision log. Based on this, a decision link graph is constructed by establishing association edges between decisions according to four types of relationships: causal, conflict, coordination, and dependency. Causal relationships indicate that one decision leads to another, while coordination relationships indicate that a decision is modified by an arbitration instruction. Key decision records generate summaries and are archived according to a preset period. Records that fail to be archived are registered with an archiving anomaly flag and retried.

[0083] When business metrics exceed the threshold, trace back from the deviation results along the decision-making link correlation graph, such as... Figure 6As shown. The final execution decision directly causing the deviation is traced back to its generation basis and coordination process level by level. The contribution of each node to the deviation is calculated according to four categories: direct, indirect, coordinated, and external, generating a responsibility attribution result. Based on this responsibility attribution result and effect feedback, the execution parameters of the conflict detection rules and conflict arbitration strategy are adjusted, rules are added, or rules are discarded. An update is triggered when the false alarm rate exceeds a threshold or the satisfaction rate falls below a threshold. The updated rules are then subjected to gray-scale verification, comparing the effect indicators before and after the update. After successful verification, the data is returned to the role ontology model and conflict detection rule base in step S101, forming a closed loop of detection, coordination, feedback, and update.

[0084] To implement all or part of the method, this application provides an embodiment of an apparatus for implementing the role-aware multi-digital employee collaboration conflict detection and consistency assurance method, see [link to embodiment]. Figure 2 The device specifically includes the following components:

[0085] The role modeling module is used to build a role ontology model for each business role based on a preset role ontology structure, including responsibility domains, capability lists and collaboration relationships, and dynamically binds and continuously calibrates the role ontology model with each digital employee instance.

[0086] The intent extraction module is used to intercept decision output and collaborative communication at the execution gateway of each digital employee and perform semantic parsing on the intercepted decision output. Based on the preset decision semantic element template, it organizes and generates a standardized decision intent flow and encodes it into a decision intent vector through a domain language model.

[0087] The conflict detection module is used to read the role ontology model, the standardized decision intent flow and the decision intent vector, and identify resource conflicts, temporal conflicts and semantic conflicts based on the global resource lock status, role cooperation relationship and business timing constraints, conflict pattern library matching and the adversarial degree of the decision intent vector, and generate multi-level conflict detection results.

[0088] The arbitration guarantee module is used to assess the severity and classify the multi-level conflict detection results based on the scope of business impact and the reversibility of loss, generate a conflict arbitration strategy in combination with the real-time business context, issue arbitration instructions to the digital employees involved, verify the global consistency of the coordinated decision, and generate a consistency guarantee result.

[0089] The attribution tracking module is used to record data on decision generation, conflict, coordination, and execution to construct a decision link association graph. When business results deviate, it traces back to the root cause to generate responsibility attribution results and updates the conflict detection rules and the conflict arbitration strategy based on the feedback of collaboration effects.

[0090] The role modeling module establishes a role ontology model for each business role based on role identifiers, responsibility domains, capability lists, decision-making authority matrices, and collaboration relationship diagrams. Business experts annotate responsibility boundaries and authority constraints, and extract actual capability ranges from the historical behavior logs of digital employee instances to generate role profiles. After comparing and verifying the role profiles with the capability lists, digital employee instances are bound to the corresponding roles. The module monitors the divergence between the instance behavior distribution and the typical behavior distribution of the bound roles, registering role deviation flags when the divergence exceeds a preset threshold. The intent extraction module deploys interceptors at the execution gateway of each digital employee to capture call requests and collaborative communications to external systems. It performs semantic parsing on the captured decision outputs and generates a standardized decision intent flow by organizing them into seven tuples based on subject, role, action, object, parameter, time, and basis. The action and object elements are input into the domain language model and encoded into decision intent vectors.

[0091] The conflict detection module reads the role ontology model, the standardized decision intent flow, and the decision intent vector. It identifies write-write and read-write conflicts by comparing the target object with the global resource lock status table. It verifies the decision sequence using the role collaboration relationship diagram and the pre-dependencies, mutual exclusion windows, and maximum latency constraints defined in the business process to identify temporal conflicts. It matches the conflict pattern library with the adversarial degree of the decision intent vector in the business effect dimension to identify semantic conflicts and distinguish differences in role perspectives. The three-layer detection results are combined to generate a multi-level conflict detection result. The arbitration guarantee module classifies the multi-level conflict detection result based on the business impact scope, loss reversibility, time sensitivity, role weight, and historical patterns. It combines the business context of market status, business cycle, user intent, compliance constraints, and resource status to select strategies from the arbitration strategy library to generate a conflict arbitration strategy.

[0092] The arbitration guarantee module further issues arbitration instructions for decision effectiveness, suspension, modification, and revocation to the relevant digital employees and receives execution confirmation. After executing the arbitration instructions, it verifies the uniqueness of resource entries, temporal legality, semantic consistency, and permission compliance of the decision combination, generating consistency guarantee results. For time-sensitive conflicts, it adopts optimistic execution and establishes compensatory transactions for rollbackable operations. The attribution tracking module records the generation, conflict, coordination, and execution data of decisions and constructs a decision link association graph based on causal, conflict, coordination, and dependency relationships. When business results deviate, it traces the root cause back along the decision link association graph and calculates the contribution according to direct, indirect, coordination, and external categories to generate responsibility attribution results. Based on the responsibility attribution results and effect feedback, it updates the conflict detection rules and the conflict arbitration strategy before feeding back to the role ontology model.

[0093] As can be seen from the above description, the role-aware multi-digital employee collaboration conflict detection and consistency assurance device provided in this application embodiment can automatically detect and ensure consistency of collaboration conflicts among multiple digital employees through the collaboration of five modules: role modeling, intent extraction, conflict detection, arbitration assurance, and tracking attribution.

[0094] This invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-described method.

[0095] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0096] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described method.

[0097] In this embodiment, the role responsibilities of digital employees are formally modeled and dynamically bound and calibrated through a role ontology model. Resource lock status, temporal constraints and vector adversarial degree are detected and identified in a three-layer progressive manner to identify resource, temporal and semantic conflicts and distinguish role perspective differences. The arbitration strategy selected by business context awareness reduces coordination costs while ensuring global consistency. The responsibility attribution result is generated by constructing a decision link association graph and tracing back. The conflict detection rules and conflict arbitration strategy are updated based on the responsibility attribution result and effect feedback and flow back to the role ontology model to form a closed loop.

[0098] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for conflict detection and consistency assurance in multi-digital employee collaboration based on role awareness, characterized in that, The method includes: Establish a role ontology model for multiple digital employees and dynamically bind and continuously calibrate it with digital employee instances. Intercept decision output and perform semantic parsing at the digital employee execution gateway. Generate a standardized decision intent flow based on a preset decision semantic element template. Encode the standardized decision intent flow into a decision intent vector through a domain language model. Read the role ontology model, the standardized decision intent flow and the decision intent vector, and identify resource conflicts, temporal conflicts and semantic conflicts based on the global resource lock state, role cooperation relationship and business time constraints, conflict pattern library matching and vector adversarial degree, and generate multi-level conflict detection results by distinguishing role perspective differences. Classify the detection results according to the business impact scope and loss reversibility, and generate a conflict arbitration strategy based on the classification results in combination with the business context. The conflict arbitration strategy is executed to issue arbitration instructions to the digital employees involved and verify the consistency of the decision after coordination to generate a consistency guarantee result. The data from decision generation to execution is recorded to construct a decision link association graph. Based on the association graph, the root cause is traced back when there is a business deviation to generate a responsibility attribution result. Based on the attribution result and feedback, the conflict detection rules and arbitration strategy are updated and fed back to the role ontology model.

2. The method for detecting and ensuring consistency in multi-digital employee collaboration based on role awareness according to claim 1, characterized in that, The process of establishing a role ontology model for multiple digital employees and dynamically binding and continuously calibrating it with digital employee instances includes: Based on the preset role ontology structure, a role ontology model containing responsibility domains, capability lists and collaboration relationships is established for each business role. Business experts mark the responsibility boundaries and permission constraints of each role, and the actual capability range is extracted from the historical behavior logs of digital employee instances to generate role profiles. After comparing and verifying the capability list of the character profile with that of the character ontology model to identify capability gaps and capability overflows, the digital employee instance is bound to the corresponding role. The deviation between the behavior distribution of the digital employee instance and the typical distribution of the bound role is monitored. When the deviation exceeds a preset threshold, a role deviation flag is registered and behavioral constraints and role reassessment are triggered.

3. The method for detecting and ensuring consistency in multi-digital employee collaboration based on role awareness according to claim 1, characterized in that, The process of intercepting decision output and performing semantic parsing at the digital employee execution gateway, generating a standardized decision intent stream based on a preset decision semantic element template, and encoding the standardized decision intent stream into a decision intent vector through a domain language model includes: An interceptor is deployed at the execution gateway to capture the call requests from each digital employee to external systems and the collaborative communication between digital employees. The captured decision output is semantically parsed and mapped to a predefined action ontology. A standardized decision intent flow is generated based on a preset decision semantic element template. The standardized decision intent flow's action and object elements are input into the domain language model and encoded into high-dimensional vectors. This allows semantically similar decision intents to be generated into decision intent vectors that are close in distance in the vector space. The decision intent vectors are then written into the intent cache with the decision subject identifier as the primary key.

4. The method for detecting and ensuring consistency in multi-digital employee collaboration based on role awareness according to claim 1, characterized in that, The process involves reading the role ontology model, the standardized decision intent flow, and the decision intent vector. Based on the global resource lock state, role collaboration relationships and business timing constraints, conflict pattern library matching, and vector adversarial degree, resource conflicts, timing conflicts, and semantic conflicts are identified, and multi-level conflict detection results are generated by distinguishing differences in role perspectives. These results include: Maintain and record the operation status of each business entity and the global resource lock status table of the holder; compare the target object of the newly arrived decision intention with the global resource lock status table to identify write-write conflicts and read-write conflicts; and verify the decision sequence to identify timing conflicts based on the role collaboration relationship diagram and the timing constraints defined in the business process. The standardized decision intent stream is matched with the conflict pattern library to identify explicit semantic conflicts. For decision pairs not covered by the conflict pattern library, the adversarial degree of the decision intent vector in the business effect dimension is calculated. Decision pairs with adversarial degree exceeding a preset threshold and pointing to related objects are judged as semantic conflicts after role perspective difference identification, and multi-level conflict detection results are generated.

5. The method for detecting and ensuring consistency in multi-digital employee collaboration based on role awareness according to claim 1, characterized in that, The process of classifying the detection results based on the scope of business impact and the reversibility of loss, and generating a conflict arbitration strategy based on the classification results in conjunction with the business context, includes: The severity of each conflict in the multi-level conflict detection results is calculated according to the preset severity assessment rules, and the severity is prioritized according to the preset grading standards to generate a grading result. The system collects business context containing market status and compliance constraints in real time, selects appropriate strategies from a preset arbitration strategy library based on the priority of the classification results and the business context, and generates conflict arbitration strategies with additional explanations.

6. The method for detecting and ensuring consistency in multi-digital employee collaboration based on role awareness according to claim 1, characterized in that, The execution of the conflict arbitration strategy, including issuing arbitration instructions to the relevant digital employees and verifying the consistency of decisions after coordination to generate a consistency guarantee result, includes: According to the aforementioned conflict arbitration strategy, arbitration instructions of the corresponding type are issued to the digital employees involved and execution confirmations are received from each digital employee. Execution confirmations that return rejections trigger a secondary coordination and escalation process. After executing the arbitration instruction on the digital employees involved, the decision combination is verified for consistency according to the preset multi-dimensional consistency verification rules to generate a consistency guarantee result. If the consistency verification fails, the relevant decision is rolled back and the conflict detection is re-entered. For time-sensitive conflicts, the high-priority role decision is executed first and a compensation transaction is established for the rollback operation.

7. The method for detecting and ensuring consistency in multi-digital employee collaboration based on role awareness according to claim 1, characterized in that, The process of recording decision generation and execution data constructs a decision link association graph. Based on this association graph, when business deviations occur, the root cause is traced back to generate a responsibility attribution result. Based on the attribution result and feedback, conflict detection rules and arbitration strategies are updated and fed back to the role ontology model, including: Record the entire lifecycle data of each decision's generation content, conflict detection results, arbitration processing, and final execution status; and construct a decision link association graph by establishing association edges between each decision based on preset association categories. When business indicators exceed the threshold, the decision link association graph is traced back from the deviation results to the root cause decision nodes, and the contribution of each node is calculated to generate the responsibility attribution results. Based on the responsibility attribution results and effect feedback, the conflict detection rules and the conflict arbitration strategy execution parameters are adjusted and the rules are supplemented. After grayscale verification, the data is fed back to the role ontology model.

8. A multi-digital employee collaboration conflict detection and consistency assurance device based on role awareness, characterized in that, The device includes: The role modeling module is used to build a role ontology model for each business role based on a preset role ontology structure, including responsibility domains, capability lists and collaboration relationships, and dynamically binds and continuously calibrates the role ontology model with each digital employee instance. The intent extraction module is used to intercept decision output and collaborative communication at the execution gateway of each digital employee and perform semantic parsing on the intercepted decision output. Based on the preset decision semantic element template, it organizes and generates a standardized decision intent flow and encodes it into a decision intent vector through a domain language model. The conflict detection module is used to read the role ontology model, the standardized decision intent flow and the decision intent vector, and identify resource conflicts, temporal conflicts and semantic conflicts based on the global resource lock status, role cooperation relationship and business timing constraints, conflict pattern library matching and the adversarial degree of the decision intent vector, and generate multi-level conflict detection results. The arbitration guarantee module is used to assess the severity and classify the multi-level conflict detection results based on the scope of business impact and the reversibility of loss, generate a conflict arbitration strategy in combination with the real-time business context, issue arbitration instructions to the digital employees involved, verify the global consistency of the coordinated decision, and generate a consistency guarantee result. The attribution tracking module is used to record data on decision generation, conflict, coordination, and execution to construct a decision link association graph. When business results deviate, it traces back to the root cause to generate responsibility attribution results and updates the conflict detection rules and the conflict arbitration strategy based on the feedback of collaboration effects.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the role-aware multi-digital employee collaboration conflict detection and consistency assurance method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the role-aware multi-digital employee collaboration conflict detection and consistency assurance method according to any one of claims 1 to 7.

Images