Process modeling method for human-machine collaboration and related device

Abstract

The embodiment of the application relates to the technical field of software engineering and artificial intelligence, and discloses a process modeling method for human-machine cooperation and related equipment, the method comprising: constructing a hierarchical recursive process model, the process model comprising a process domain layer, an activity layer and a task layer; using an eight-tuple model to formally describe each task, the eight-tuple model comprising a task identifier, an input artifact set, an execution logic, an output artifact set, a role performer, a constraint set, a cognitive level and a deterministic classification, defining four types of role performers for the process model, including a human role, an agent role, a system role and a hybrid role, and based on the four types of role performers and the eight-tuple model, rendering each activity of the process model as an activity graph on a preset workbench interface through differentiated visual lanes, and in this way, the embodiment of the application effectively realizes modeling of AI tasks under the framework of organizational system constraints and standard specifications in a human-machine cooperation scenario.

Description

Technical Field

[0001] The embodiments of the present invention relate to the fields of software engineering and artificial intelligence technology, specifically to a process modeling method and related equipment for human-computer collaboration. Background Technology

[0002] With the rapid development of artificial intelligence technology, more and more enterprises are introducing AI into their business, forming a new model of human-AI collaboration in task execution. Existing process modeling methods and tools include the following two types: 1. Business Process Model and Notation (BPMN) is the international standard for process modeling widely adopted in the industry. BPMN describes business processes using primitives such as swimlanes, task nodes, event nodes, gateway nodes, and connections. However, BPMN only distinguishes between human participants and automated systems, failing to accurately describe the participation of AI. In actual human-machine collaboration scenarios, executors may include human executors, AI executors, automated system executors, and hybrid human-AI executors, but BPMN lacks the ability to differentiate and visually distinguish these four types of executors. 2. Existing AI task orchestration frameworks (such as LangChain, AutoGen, CrewAI, etc.) mainly focus on the tool invocation and inference chain orchestration of agents within a single task, lacking a process-level governance perspective. These frameworks can describe how an AI agent executes a specific task, but they cannot place AI tasks under the institutional constraints and standard specifications of an organization for systematic governance.

[0003] In view of this, in human-machine collaboration scenarios, modeling AI tasks within the framework of organizational institutional constraints and standard norms has become an issue that needs to be addressed. Summary of the Invention

[0004] In view of the above problems, embodiments of the present invention provide a process modeling method and related equipment for human-machine collaboration, which are used to solve the problems existing in the prior art.

[0005] According to one aspect of the present invention, a process modeling method for human-machine collaboration is provided, the method comprising: A hierarchical recursive process model is constructed, which includes a process domain layer, an activity layer, and a task layer. The process domains in the process domain layer correspond to a management domain or business domain. The activities in the activity layer are the executable business of the process domain. The tasks in the task layer are the underlying work execution units of the activities. Each process domain contains at least one activity, and each activity contains at least one task. Each task is formally described using an octet model, which includes a task identifier, input artifact set, execution logic, output artifact set, executor role, constraint set, cognitive level, and deterministic classification. The execution logic includes a sequentially arranged sequence of steps, available tool references, execution conditions, and completion definition. The cognitive level is quantified and mapped according to the cognitive complexity of the task, including memory level, comprehension level, application level, analysis level, evaluation level, and creation level. The deterministic classification includes determinism, probabilism, and heuristics. Four types of executors are defined for the process model, including human roles, intelligent agent roles, system roles, and hybrid roles. The intelligent agent roles are associated with intelligent agent specification parameters, and the hybrid roles are configured with both intelligent agent specification parameters and human ability descriptions. Based on the four types of role executors and the octet model, each activity of the process model is rendered as an activity diagram on a preset workbench interface through differentiated visual swimlanes, and the corresponding octet model information of each task contained in the activity is displayed on the workbench interface. Each visual swimlane corresponds to one type of role executor.

[0006] In one alternative approach, the constraint set is used to govern and constrain the process model, and the constraint set includes institutional constraints, standard constraints, time constraints, authority constraints, quality constraints, and audit constraints. The institutional constraints are derived from pre-set systems, which include regulatory, policy, and customary systems. The regulatory-level system is associated with the audit constraints and the access constraints, which are generated based on the regulatory-level system. The policy-level institutions are associated with the quality constraints, which are generated based on the policy-level institutions.

[0007] In one alternative approach, the standard layer constraints are categorized by type, including terminology standards, process standards, activity standards, method standards, tool standards, workpiece standards, and improvement standards, with each process domain associated with at least one type of standard.

[0008] In an alternative approach, the process model further includes a rule validation engine, and the method further includes: performing consistency or integrity validation on the process model based on the rule validation engine, and outputting the validation results according to error level and warning level. The consistency or integrity verification includes at least the verification of process domain integrity, activity integrity, task input and output integrity, role executor uniqueness, dependency acyclicity, artifact creator uniqueness, initiation event uniqueness, review activity role integrity, evaluation consistency, and role specification configuration integrity. The dependency acyclicity verification is performed using a topological sorting algorithm.

[0009] In one alternative approach, the activity graph further includes at least three types of task nodes, at least six event nodes, at least three types of gateway nodes, and at least three types of connection types. The three task nodes include nodes corresponding to tasks, decisions, and reviews, respectively; The six event nodes include nodes corresponding to start, end, middle, timer, message, and signal, respectively; The three types of gateway nodes include nodes corresponding to exclusive, parallel, and inclusive types, respectively; The three connection types include the connection types corresponding to control flow, data flow, and message flow, respectively; For the event node and the gateway node, a four-level priority inference mechanism is used to determine the corresponding visual swimlane affiliation.

[0010] In an alternative approach, the method further includes: In response to an editing command on the workbench interface for the octet model information, the activity graph is modified based on the editing command.

[0011] In one alternative approach, the agent specification parameters include model name, model version, temperature parameters, maximum number of tokens, list of allowed tools, and output verification rules; the agent specification parameters are also associated with armor configuration and skill list; the armor configuration includes context compression strategy, feedback loop definition, and execution boundary constraints, and the skill list includes modular capability units that the agent role can invoke and their progressive loading conditions.

[0012] According to another aspect of the present invention, a process modeling apparatus for human-machine collaboration is provided, the apparatus comprising: The construction module is used to construct a hierarchical recursive process model. The process model includes a process domain layer, an activity layer, and a task layer. The process domains in the process domain layer correspond to a management domain or business domain. The activities in the activity layer are the executable business of the process domain. The tasks in the task layer are the underlying work execution units of the activities. Each process domain contains at least one activity, and each activity contains at least one task. The task description module is used to formally describe each task using an octet model. The octet model includes a task identifier, input artifact set, execution logic, output artifact set, executor role, constraint set, cognitive level, and deterministic classification. The execution logic includes a sequentially arranged sequence of steps, available tool references, execution conditions, and completion definition. The cognitive level is quantified and mapped according to the cognitive complexity of the task, including memory level, understanding level, application level, analysis level, evaluation level, and creation level. The deterministic classification includes deterministic, probabilistic, and heuristic classifications. The definition module is used to define four types of role executors for the process model, including human role, intelligent agent role, system role and hybrid role. The intelligent agent role is associated with intelligent agent specification parameters, and the hybrid role is configured with both intelligent agent specification parameters and human ability description. The rendering module is used to render each activity of the process model as an activity diagram on a preset workbench interface based on the four types of role executors and the octet model, through differentiated visual swimlanes, and display the corresponding octet model information of each task contained in the activity on the workbench interface. Each visual swimlane corresponds to one type of role executor.

[0013] According to another aspect of the present invention, a computer device is provided, comprising: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other via the communication bus; the memory is used to store at least one executable instruction, wherein the executable instruction causes the processor to perform the method described above.

[0014] According to another aspect of the present invention, a computer-readable storage medium is provided, the storage medium storing at least one executable instruction, which, when executed on a computer device, causes the computer device to perform the method described above.

[0015] This invention constructs a three-layer recursive model at the modeling level, including process domains, activities, and tasks. It defines four types of role executors and proposes an octet model to describe tasks. Finally, based on the four types of role executors and the octet model, each activity of the process model is rendered as an activity graph on a preset workbench interface through differentiated visual swimlanes. In human-computer collaboration scenarios, this effectively enables the modeling of AI tasks within the framework of organizational institutional constraints and standard specifications, resulting in efficient and reliable modeling.

[0016] The above description is merely an overview of the technical solutions of the embodiments of the present invention. In order to better understand the technical means of the embodiments of the present invention and to implement them in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the embodiments of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0017] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 A flowchart illustrating the process modeling method for human-computer collaboration provided in an embodiment of the present invention is shown. Figure 2 It shows Figure 1 A schematic diagram of the process model in the diagram; Figure 3 It shows Figure 1 The activity diagram shown; Figure 4 This diagram illustrates the structure of a process modeling device for human-machine collaboration provided in an embodiment of the present invention. Figure 5 A schematic diagram of the structure of a computer device provided in an embodiment of the present invention is shown. Detailed Implementation

[0018] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0019] Figure 1 The following is a flowchart illustrating the process modeling method for human-computer collaboration provided in an embodiment of the present invention, as shown in the figure. Figure 1 As shown, the method includes the following steps: Step 10: Construct a hierarchical recursive process model. The process model includes a process domain layer, an activity layer, and a task layer. The process domains in the process domain layer correspond to a management domain or business domain. The activities in the activity layer are the executable business of the process domain. The tasks in the task layer are the underlying work execution units of the activities. Each process domain contains at least one activity, and each activity contains at least one task.

[0020] like Figure 2 As shown, the first layer is the process area layer, which is the top-level business management unit; the second layer is the activity layer; and the third layer is the task layer.

[0021] For example, in a software engineering field, taking the configuration management process domain as an example, this process domain includes five activities: code commit, build testing, change approval, version release, and configuration audit. These five activities comprise a total of 25 tasks.

[0022] In another financial field, taking the credit risk control process domain as an example, this domain includes four activities: customer data collection (containing three tasks, of which "intelligent extraction and verification of customer data" is performed by an intelligent agent role, responsible for extracting key information from various document images and cross-verifying it with credit data), credit score calculation (containing two tasks, of which "multi-dimensional credit model calculation" is performed by a system role), risk approval (containing three tasks, of which "AI preliminary review opinion generation" is performed by a hybrid role, and "final approval decision" is performed by a human role), and post-loan monitoring. This embodiment demonstrates the differentiated modeling capabilities of the present invention for four types of roles in a financial scenario.

[0023] Step 20: Formal description of each task is performed using an octet model. The octet model includes a task identifier, input artifact set, execution logic, output artifact set, executor role, constraint set, cognitive level, and deterministic classification. The execution logic includes a sequentially arranged sequence of steps, available tool references, execution conditions, and completion definition. The cognitive level is quantified and mapped according to the cognitive complexity of the task, including memory level, understanding level, application level, analysis level, evaluation level, and creation level. The deterministic classification includes determinism, probabilism, and heuristics.

[0024] The octet model can comprehensively describe a task from eight dimensions. The Task ID is a globally unique string identifier, which can use a hierarchical naming convention of "process domain.activity number.task number", such as "CM.1.3". The input artifact set contains at least one input artifact. Each input artifact consists of three attributes: artifact identifier, artifact description, and data format (e.g., Git Diff, JSON, Markdown, PDF). The sequentially arranged sequence of steps in the execution logic is the specific operational flow of the task execution. It clarifies the steps required to complete the task, arranged in a chronological or logical order. For example, the execution logic of a code conflict detection task includes three explicit steps: obtaining differences, analyzing conflicts, and generating a report. This provides clear action guidelines for the executor. Available tool references include the tool name and version. Execution conditions are the prerequisites that must be met for a task to be started. They clarify when and under what circumstances the task can begin execution, linking a single task to the context of the entire process. This ensures that the task is triggered only at the correct time and after the correct prerequisites are met, guaranteeing the logical order and consistency of the process. The completion definition refers to the objective output standards and acceptance criteria that a task must meet to be successfully completed. It provides a clear, unambiguous, and verifiable acceptance standard for task execution results, eliminating ambiguities in task completion and forming the foundation for quality control and reliable inter-task delivery. The output artifact set contains at least one output artifact, and each output artifact can only have one creator throughout the entire process model, ensuring traceability of artifact origins. Each task is associated with a role executor.

[0025] The constraint set is used to govern and constrain the process model. The constraint set includes institutional constraints, standard constraints, time constraints, authority constraints, quality constraints, and audit constraints. Institutional constraints are derived from pre-defined systems, such as those within an organization or enterprise. The institutional layer categorizes the institutional environment faced by the organization or enterprise into three levels: regulatory, policy, and customary. Regulatory constraints originate from mandatory requirements of laws and regulations; policy constraints originate from internal rules and regulations, such as enterprise code review policies; and customary constraints originate from industry best practices, such as code naming conventions. Standard constraints are categorized by type, including terminology standards, process standards, activity standards, methodological standards, tool standards, artifact standards, and improvement standards. Each process domain is associated with at least one type of standard. When constructing any process area, it is essential to explicitly specify the specific standards it adheres to, derived from the constraints of the standards layer. This constitutes an independent governance knowledge base. For example, for a software development process area, this might involve associated terminology standards (such as explicitly defining terms like iteration and release); process standards (such as adhering to the Agile Scrum framework); and tool standards (such as specifying the use of Git for version control). This association mechanism enables the systematic implementation and assurance of macro-level governance requirements down to specific business domains, ensuring consistency and compliance in process modeling. Time constraints include execution time windows or deadlines; permission constraints include authorization levels or access permissions; quality constraints include quality standards or acceptance criteria; and audit constraints include audit logging requirements.

[0026] Furthermore, this embodiment also includes an automatic transmission mechanism from institutional classification to the constraint set of tasks. Regulatory-level institutions are associated with audit constraints and authority constraints, which are generated based on the regulatory-level institutions. Policy-level institutions are associated with quality constraints (such as "code coverage not less than 80%)", which are generated based on the policy-level institutions. This achieves automatic transmission of regulatory-level institutions to generate audit constraints and authority constraints, automatic transmission of policy-level institutions to generate quality constraints, and automatic transmission of customary-level institutions to generate advisory constraints. This automatic transmission or association enables traceability of authority constraints, quality constraints, and audit constraints, transforming the theoretical engineering of institutional classification into automatically executable constraint rules. This eliminates the inefficiency and risk of omissions associated with manually configuring constraints one by one, and further achieves traceability and explainability from micro-task execution to macro-governance requirements. This enhances the management compliance, controllability, and continuous optimization capabilities of the process model, providing a guarantee for efficient process modeling.

[0027] Regarding cognitive levels, this invention proposes a quantitative mapping method for cognitive complexity: drawing on the hierarchical thinking of the cognitive taxonomy framework in cognitive educational psychology, this invention establishes mapping rules from shallow to deep task cognitive complexity and quantifies them: Level 1 corresponds to the memory level (i.e., recalling facts), Level 2 corresponds to the comprehension level (i.e., explaining concepts), Level 3 corresponds to the application level (i.e., using rules), Level 4 corresponds to the analysis level (i.e., decomposing relationships), Level 5 corresponds to the evaluation level (i.e., making judgments), and Level 6 corresponds to the creation level (i.e., generating solutions). It should be noted that this invention does not directly apply the hierarchical thinking of this cognitive taxonomy framework, but rather establishes quantitative mapping rules for automated task decision-making based on its hierarchical thinking. This transformation provides a structured and quantifiable basis for determining whether a task is suitable and how it should be assigned to an agent. For example, in a human-machine collaborative process, it is necessary to decide whether a task should be assigned to a human, an agent, or a hybrid role. The cognitive level of the task is a core quantitative indicator for making this decision. For example, low cognitive levels (such as level 1-2 memory and comprehension tasks) may be more suitable for intelligent agents to perform, while high cognitive levels (such as level 5-6 evaluation and creation tasks) may require human intervention or the use of a hybrid role.

[0028] The deterministic classification is divided into three types: deterministic, probabilistic, and heuristic. Deterministic means that the execution result is completely predictable, probabilistic means that the result is statistically predictable, and heuristic means that it relies heavily on experience and judgment.

[0029] Taking the "code conflict detection" task in the field of software engineering as an example, its octet model includes: the input artifact is a code branch difference file in GitDiff format; the execution logic includes three steps (acquiring differences, analyzing conflicts, and generating a report), the tool uses Git version 2.40, and the completion definition is "all conflict points have been identified and classified by severity"; the output artifact is a conflict detection report in JSON format; the executor role is an LLM agent role (e.g., GPT-4, temperature 0.1); the constraint set includes the quality constraint "detection rate ≥ 99%" (its associated enterprise's code quality policy-level system); the cognitive level is level 4 (analysis); and the deterministic classification is deterministic.

[0030] Taking the "AI-generated preliminary review opinion" task in the field of financial risk control as an example, its octet model includes: input artifacts including customer credit reports (JSON format) and credit inquiry results (PDF format); execution logic includes four steps (extracting customer profiles, multi-dimensional risk assessment, generating preliminary review opinions, and labeling risk levels); output artifact is a risk preliminary review opinion (Markdown format); the executor role is a hybrid role; the constraint set includes the audit constraint "record the complete decision basis for each review" (which is associated with regulatory-level systems) and the quality constraint "the deviation of the preliminary review pass rate does not exceed ±5% of the industry benchmark" (which is associated with policy-level systems); the cognitive level is level 5 (evaluation); the deterministic classification is probabilistic.

[0031] Step 30: Define four types of executors for the process model, including human roles, intelligent agent roles, system roles, and hybrid roles. The intelligent agent roles are associated with intelligent agent specification parameters, and the hybrid roles are configured with both intelligent agent specification parameters and human ability descriptions.

[0032] Human roles are performed independently by human employees without the need to configure intelligent agent specifications. Their execution relies on human expertise, experience, and judgment. Typical human roles include senior architects, project managers, and compliance auditors.

[0033] The agent role autonomously executes tasks using an AI model and must be associated with an agent specification parameter set, protective gear configuration, and skill list. The agent specification parameter set includes the model name, model version, temperature parameter (0.0 to 1.0), maximum token count, list of allowed tools, and output verification rules. The protective gear configuration includes a context compression strategy (specifying the rules for dividing the context window into stable prefix and variable suffix regions to utilize the inference engine's caching mechanism), feedback loop definitions (specifying output verification methods and error retry strategies), and execution boundary constraints (transforming task constraints into insurmountable runtime limitations). The skill list lists the modular capability units that the agent can invoke and their progressive loading rules; each skill can connect to external tool services via the model context protocol.

[0034] The system role is defined by the traditional automated system performing tasks, and associated system specification parameters, including system name, interface specifications, execution rules, etc.

[0035] Hybrid roles involve humans and intelligent agents collaborating on tasks. Both the intelligent agent's specifications and the human's capabilities must be configured; missing either configuration will trigger a validation error. A typical scenario includes AI generating a code draft, which is then reviewed and corrected by humans.

[0036] Step 40: Based on the four types of role executors and the octet model, render each activity of the process model as an activity diagram on a preset workbench interface through differentiated visual swimlanes, and display the corresponding octet model information of each task contained in the activity on the workbench interface. Each visual swimlane corresponds to one type of role executor.

[0037] like Figure 3 As shown, each activity in the process model is rendered as an activity graph containing role swimlanes. Swimlanes are horizontal containers; the left side is the label area displaying the role's executor name, and the right side is the content area. Four types of roles correspond to four different swimlane border styles: human roles use solid borders, conveying a stable human execution path; intelligent agent roles use dashed borders, conveying a flexible AI execution path; system roles use dotted-dash borders, conveying a rule-driven system path; and hybrid roles use dotted-dash borders, conveying human-machine dual-track collaboration.

[0038] Furthermore, the activity graph also includes at least three types of task nodes, at least six types of event nodes, at least three types of gateway nodes, and at least three types of connection types; The three task nodes include nodes corresponding to tasks, decisions, and reviews, respectively; The six event nodes include nodes corresponding to start, end, middle, timer, message, and signal, respectively; The three types of gateway nodes include nodes corresponding to exclusive, parallel, and inclusive types, respectively; The three connection types include the connection types corresponding to control flow, data flow, and message flow, respectively; For event nodes and gateway nodes that are not directly associated with role executors, a four-level priority inference mechanism is used to determine the corresponding visual swimlane affiliation. Specifically, the first priority is to check the explicit swimlane affiliation label of the node (which is automatically recorded when the user manually operates); the second priority is to find the role of the directly connected task node; the third priority is to expand to a two-hop search range; and the fourth priority is to default to the swimlane of the first ranked role.

[0039] In an optional embodiment, based on the above embodiments, the method further includes: in response to an editing instruction on the octet model information on the workbench interface, modifying the activity graph based on the editing instruction.

[0040] When a task node in the activity graph is clicked, the corresponding octet model information is displayed on the workbench interface. The octet model information can be edited, and after the editing is confirmed, the activity graph changes accordingly, enabling flexible adjustment of the activity graph.

[0041] In an optional embodiment, the process model further includes a rule validation engine, and the method further includes: performing consistency or integrity validation on the process model based on the rule validation engine, and outputting the validation results according to error level and warning level; The consistency or integrity verification includes at least the verification of process domain integrity, activity integrity, task input and output integrity, role executor uniqueness, dependency acyclicity, artifact creator uniqueness, initiation event uniqueness, review activity role integrity, evaluation consistency, and role specification configuration integrity. The dependency acyclicity verification is performed using a topological sorting algorithm.

[0042] Among these requirements, error-level verification results must be corrected, and warning-level verification results are recommended for correction. Process domain integrity requires that the process domain includes at least one activity; activity integrity requires that an activity includes at least one task; task input / output integrity requires that a task includes at least one input and at least one output; role executor uniqueness requires that the task execution has a unique role; directed acyclic dependency is achieved using a topological sorting algorithm, specifically by constructing a directed graph of task dependencies, initializing a queue of nodes with an in-degree of zero, cyclically popping nodes and decreasing the in-degree of successors, and determining a cycle if the remaining nodes after sorting are greater than 0; artifact creator uniqueness requires that artifacts have a unique creator; starting events are unique; review activity role integrity requires that review activities include at least two different roles; evaluation consistency requires that the scoring dimension values ​​are integers within a predetermined range, and that the evaluation level is consistent with the total score; role specification configuration integrity requires that agent roles have complete specification configurations (including agent specification parameters, protective gear configuration, and skill list; missing any configuration item triggers an error); hybrid roles must have dual configurations; subprocesses cannot self-reference; and the standard layer must be associated with at least one process domain. Through the above rule verification engine, real-time automated verification of the process model is achieved.

[0043] For example, in a simulated comparison scenario, a software engineering process model containing 5 process domains and 25 tasks was used to simulate 3 senior engineers performing manual consistency reviews, which were then automatically validated by the rule validation engine of this embodiment of the invention. The results are as follows: the manual review took an average of about 1 hour and 52 minutes, with an average detection rate of 88% (omission rate of about 12%); the rule validation engine of this embodiment of the invention completed the review within 78 milliseconds, detecting all 11 violations, with a detection rate of 100%, including one task dependency loop violation (two tasks forming an indirect circular dependency) that is easily overlooked in manual reviews.

[0044] This invention constructs a three-layer recursive model at the modeling level, including process domains, activities, and tasks. It defines four types of role executors and proposes an octet model to describe tasks. Finally, based on the four types of role executors and the octet model, each activity of the process model is rendered as an activity graph on a preset workbench interface through differentiated visual swimlanes. In human-computer collaboration scenarios, this effectively enables the modeling of AI tasks within the framework of organizational institutional constraints and standard specifications, resulting in efficient and reliable modeling.

[0045] Figure 4 A schematic diagram of the structure of a process modeling device for human-machine collaboration according to an embodiment of the present invention is shown. Figure 4 As shown, the device includes: Module 41 is used to construct a hierarchical recursive process model. The process model includes a process domain layer, an activity layer, and a task layer. The process domain in the process domain layer corresponds to a management domain or business domain. The activities in the activity layer are the executable business of the process domain. The tasks in the task layer are the underlying work execution units of the activities. Each process domain contains at least one activity, and each activity contains at least one task. The task description module 42 is used to formally describe each task using an octet model. The octet model includes a task identifier, input artifact set, execution logic, output artifact set, executor role, constraint set, cognitive level, and deterministic classification. The execution logic includes a sequentially arranged sequence of steps, available tool references, execution conditions, and completion definition. The cognitive level is quantified and mapped according to the cognitive complexity of the task, including memory level, understanding level, application level, analysis level, evaluation level, and creation level. The deterministic classification includes deterministic, probabilistic, and heuristic classifications. Definition module 43 is used to define four types of role executors for the process model, including human role, intelligent agent role, system role and hybrid role. The intelligent agent role is associated with intelligent agent specification parameters, and the hybrid role is configured with both intelligent agent specification parameters and human ability description. The rendering module 44 is used to render each activity of the process model as an activity diagram on a preset workbench interface based on the four types of role executors and the octet model through differentiated visual swimlanes, and to display the corresponding octet model information of each task contained in the activity on the workbench interface. Each visual swimlane corresponds to one type of role executor.

[0046] The embodiments of the process modeling device for human-machine collaboration are basically the same as the embodiments of the process modeling method for human-machine collaboration described above, and can be referred to the above embodiments.

[0047] Figure 5The diagram shows a structural schematic of an embodiment of the computer device of the present invention. The specific embodiments of the present invention do not limit the specific implementation of the computer device.

[0048] like Figure 5 As shown, the computer device may include: a processor 402, a communications interface 404, a memory 406, and a communications bus 408.

[0049] The processor 402, communication interface 404, and memory 406 communicate with each other via communication bus 408. Communication interface 404 is used to communicate with other computer devices, such as clients or other server network elements. The processor 402 executes program 410, specifically performing the relevant steps described above in the computer device embodiment.

[0050] Specifically, program 410 may include program code, which includes computer-executable instructions.

[0051] Processor 402 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention. The computer device includes one or more processors, which may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0052] Memory 406 is used to store program 410. Memory 406 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0053] Specifically, program 410 can be called by processor 402 to cause the computer device to perform the following operations: A hierarchical recursive process model is constructed, which includes a process domain layer, an activity layer, and a task layer. The process domains in the process domain layer correspond to a management domain or business domain. The activities in the activity layer are the executable business of the process domain. The tasks in the task layer are the underlying work execution units of the activities. Each process domain contains at least one activity, and each activity contains at least one task. Each task is formally described using an octet model, which includes a task identifier, input artifact set, execution logic, output artifact set, executor role, constraint set, cognitive level, and deterministic classification. The execution logic includes a sequentially arranged sequence of steps, available tool references, execution conditions, and completion definition. The cognitive level is quantified and mapped according to the cognitive complexity of the task, including memory level, comprehension level, application level, analysis level, evaluation level, and creation level. The deterministic classification includes determinism, probabilism, and heuristics. Four types of executors are defined for the process model, including human roles, intelligent agent roles, system roles, and hybrid roles. The intelligent agent roles are associated with intelligent agent specification parameters, and the hybrid roles are configured with both intelligent agent specification parameters and human ability descriptions. Based on the four types of role executors and the octet model, each activity of the process model is rendered as an activity diagram on a preset workbench interface through differentiated visual swimlanes, and the corresponding octet model information of each task contained in the activity is displayed on the workbench interface. Each visual swimlane corresponds to one type of role executor.

[0054] In one alternative approach, the constraint set is used to govern and constrain the process model, and the constraint set includes institutional constraints, standard constraints, time constraints, authority constraints, quality constraints, and audit constraints. The institutional constraints are derived from pre-set systems, which include regulatory, policy, and customary systems. The regulatory-level system is associated with the audit constraints and the access constraints, which are generated based on the regulatory-level system. The policy-level institutions are associated with the quality constraints, which are generated based on the policy-level institutions.

[0055] In one alternative approach, the standard layer constraints are categorized by type, including terminology standards, process standards, activity standards, method standards, tool standards, workpiece standards, and improvement standards, with each process domain associated with at least one type of standard.

[0056] In an alternative approach, the process model further includes a rule validation engine, and the method further includes: performing consistency or integrity validation on the process model based on the rule validation engine, and outputting the validation results according to error level and warning level. The consistency or integrity verification includes at least the verification of process domain integrity, activity integrity, task input and output integrity, role executor uniqueness, dependency acyclicity, artifact creator uniqueness, initiation event uniqueness, review activity role integrity, evaluation consistency, and role specification configuration integrity. The dependency acyclicity verification is performed using a topological sorting algorithm.

[0057] In one alternative approach, the activity graph further includes at least three types of task nodes, at least six event nodes, at least three types of gateway nodes, and at least three types of connection types. The three task nodes include nodes corresponding to tasks, decisions, and reviews, respectively; The six event nodes include nodes corresponding to start, end, middle, timer, message, and signal, respectively; The three types of gateway nodes include nodes corresponding to exclusive, parallel, and inclusive types, respectively; The three connection types include the connection types corresponding to control flow, data flow, and message flow, respectively; For the event node and the gateway node, a four-level priority inference mechanism is used to determine the corresponding visual swimlane affiliation.

[0058] In an alternative approach, the method further includes: In response to an editing command on the workbench interface for the octet model information, the activity graph is modified based on the editing command.

[0059] In one alternative approach, the agent specification parameters include model name, model version, temperature parameters, maximum number of tokens, list of allowed tools, and output verification rules; the agent specification parameters are also associated with armor configuration and skill list; the armor configuration includes context compression strategy, feedback loop definition, and execution boundary constraints, and the skill list includes modular capability units that the agent role can invoke and their progressive loading conditions.

[0060] This invention provides a computer-readable storage medium storing at least one executable instruction that, when executed on a computer device, causes the computer device to perform any of the above-described method embodiments.

[0061] This invention provides a computer program that can be invoked by a processor to cause a computer device to execute any of the above-described method embodiments.

[0062] This invention provides a computer program product, which includes a computer program stored on a computer-readable storage medium. The computer program includes program instructions that, when executed on a computer, cause the computer to perform any of the above-described method embodiments.

[0063] The algorithms or displays provided herein are not inherently related to any particular computer, virtual system, or other device. Various general-purpose systems can also be used in conjunction with the teachings herein. The required structure for constructing such systems is apparent from the above description. Furthermore, the embodiments of the present invention are not directed to any particular programming language. It should be understood that the content of the invention described herein can be implemented using various programming languages, and the above description of specific languages ​​is for the purpose of disclosing the best mode of implementation of the invention.

[0064] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0065] Similarly, it should be understood that, in order to streamline the invention and aid in understanding one or more of the various aspects of the invention, features of the embodiments of the invention are sometimes grouped together in a single embodiment, figure, or description thereof in the above description of exemplary embodiments of the invention. However, this disclosure should not be construed as reflecting an intention that the claimed invention requires more features than are expressly recited in each claim.

[0066] Those skilled in the art will understand that modules in the computer device of the embodiments can be adaptively modified and placed in one or more computer devices different from that embodiment. Modules, units, or components in the embodiments can be combined into a single module, unit, or component, and can be divided into multiple sub-modules, sub-units, or sub-components. Except where at least some of such features and / or processes or units are mutually exclusive, any combination can be used to combine all features disclosed in this specification (including the accompanying claims, abstract, and drawings) and all processes or units of any method or computer device so disclosed. Unless expressly stated otherwise, each feature disclosed in this specification (including the accompanying claims, abstract, and drawings) may be replaced by an alternative feature that serves the same, equivalent, or similar purpose.

[0067] It should be noted that the above embodiments are illustrative of the invention and not restrictive, and that those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be construed as limiting the order of execution.

Claims

1. A process modeling method for human-computer collaboration, characterized in that, The method includes: A hierarchical recursive process model is constructed, which includes a process domain layer, an activity layer, and a task layer. The process domains in the process domain layer correspond to a management domain or business domain. The activities in the activity layer are the executable business of the process domain. The tasks in the task layer are the underlying work execution units of the activities. Each process domain contains at least one activity, and each activity contains at least one task. Each task is formally described using an octet model, which includes a task identifier, input artifact set, execution logic, output artifact set, executor role, constraint set, cognitive level, and deterministic classification. The execution logic includes a sequentially arranged sequence of steps, available tool references, execution conditions, and completion definition. The cognitive level is quantified and mapped according to the cognitive complexity of the task, including memory level, comprehension level, application level, analysis level, evaluation level, and creation level. The deterministic classification includes determinism, probabilism, and heuristics. Four types of executors are defined for the process model, including human roles, intelligent agent roles, system roles, and hybrid roles. The intelligent agent roles are associated with intelligent agent specification parameters, and the hybrid roles are configured with both intelligent agent specification parameters and human ability descriptions. Based on the four types of role executors and the octet model, each activity of the process model is rendered as an activity diagram on a preset workbench interface through differentiated visual swimlanes, and the corresponding octet model information of each task contained in the activity is displayed on the workbench interface. Each visual swimlane corresponds to one type of role executor.

2. The method according to claim 1, characterized in that, The constraint set is used to govern and constrain the process model, and the constraint set includes institutional constraints, standard constraints, time constraints, authority constraints, quality constraints, and audit constraints. The institutional constraints are derived from pre-set systems, which include regulatory, policy, and customary systems. The regulatory-level system links the audit constraints and the access constraints, which are generated based on the regulatory-level system. The quality constraints are associated with policy-level institutions, and these quality constraints are generated based on the policy-level institutions.

3. The method according to claim 2, characterized in that, The standard layer constraints are categorized by type, including terminology standards, process standards, activity standards, method standards, tool standards, workpiece standards, and improvement standards, with each process domain associated with at least one type of standard.

4. The method according to claim 1, characterized in that, The process model also includes a rule validation engine, and the method further includes: performing consistency or integrity validation on the process model based on the rule validation engine, and outputting the validation results according to error level and warning level. The consistency or integrity verification includes at least the verification of process domain integrity, activity integrity, task input and output integrity, role executor uniqueness, dependency acyclicity, artifact creator uniqueness, initiation event uniqueness, review activity role integrity, evaluation consistency, and role specification configuration integrity. The dependency acyclicity verification is performed using a topological sorting algorithm.

5. The method according to claim 1, characterized in that, The activity graph also includes at least three types of task nodes, at least six event nodes, at least three types of gateway nodes, and at least three types of connection types; The three task nodes include nodes corresponding to tasks, decisions, and reviews, respectively; The six event nodes include nodes corresponding to start, end, middle, timer, message, and signal, respectively; The three types of gateway nodes include nodes corresponding to exclusive, parallel, and inclusive types, respectively; The three connection types include the connection types corresponding to control flow, data flow, and message flow, respectively; For the event node and the gateway node, a four-level priority inference mechanism is used to determine the corresponding visual swimlane affiliation.

6. The method according to claim 1, characterized in that, The method further includes: In response to an editing command on the workbench interface for the octet model information, the activity graph is modified based on the editing command.

7. The method according to claim 1, characterized in that, The agent specification parameters include model name, model version, temperature parameters, maximum number of tokens, list of allowed tools, and output verification rules; the agent specification parameters are also associated with protective gear configuration and skill list; the protective gear configuration includes context compression strategy, feedback loop definition, and execution boundary constraints, and the skill list includes modular capability units that the agent role can call and their progressive loading conditions.

8. A process modeling device for human-machine collaboration, characterized in that, The device includes: The construction module is used to construct a hierarchical recursive process model. The process model includes a process domain layer, an activity layer, and a task layer. The process domains in the process domain layer correspond to a management domain or business domain. The activities in the activity layer are the executable business of the process domain. The tasks in the task layer are the underlying work execution units of the activities. Each process domain contains at least one activity, and each activity contains at least one task. The task description module is used to formally describe each task using an octet model. The octet model includes a task identifier, input artifact set, execution logic, output artifact set, executor role, constraint set, cognitive level, and deterministic classification. The execution logic includes a sequentially arranged sequence of steps, available tool references, execution conditions, and completion definition. The cognitive level is quantified and mapped according to the cognitive complexity of the task, including memory level, understanding level, application level, analysis level, evaluation level, and creation level. The deterministic classification includes deterministic, probabilistic, and heuristic classifications. The definition module is used to define four types of role executors for the process model, including human role, intelligent agent role, system role and hybrid role. The intelligent agent role is associated with intelligent agent specification parameters, and the hybrid role is configured with both intelligent agent specification parameters and human ability description. The rendering module is used to render each activity of the process model as an activity diagram on a preset workbench interface based on the four types of role executors and the octet model, through differentiated visual swimlanes, and display the corresponding octet model information of each task contained in the activity on the workbench interface. Each visual swimlane corresponds to one type of role executor.

9. A computer device, characterized in that, include: The processor, memory, communication interface, and communication bus are provided, wherein the processor, memory, and communication interface communicate with each other via the communication bus. The memory is used to store at least one executable instruction that causes the processor to perform the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The storage medium stores at least one executable instruction, which, when executed on a computer device, causes the computer device to perform the method as described in any one of claims 1-7.

Images