A method and system for constructing a workflow

Abstract

The present invention discloses a method and system for constructing a workflow, belonging to the technical field of electronic digital data processing, the method comprising the following steps: obtaining a language or text for constructing a workflow; obtaining the intent of the language or text based on a large-scale language model; matching components according to the intent, the components being used to perform tasks corresponding to the intent; determining whether two components have an interactive association; if so, establishing a transfer rule for the two components; if not, establishing a connection for the components to obtain a first workflow. Based on a large rule language model, the user's intent is identified, and on this basis, by matching the corresponding components and organizing the components into a complete workflow, the adaptability, scalability and automation of the workflow can be improved; it can quickly respond to user requests and can be applied to a variety of different application scenarios.

Description

Technical Field

[0001] The present invention relates to the technical field of electronic digital data processing, and in particular to a method and system for constructing a workflow. Background Art

[0002] Process management and automated processing are becoming increasingly important in modern enterprises. Traditional process management systems often rely on manual operations and preset rules, resulting in low efficiency and susceptibility to human error. With the development of artificial intelligence (AI) technology, intelligent process management solutions are becoming a market focus.

[0003] There are two workflow management technologies currently used:

[0004] Business process management (BPM) systems primarily rely on process models and rule engines to execute automated tasks based on predefined rules. While these systems can improve efficiency to a certain extent, they lack adaptability and flexibility in complex and dynamic business environments. In response to ever-changing business needs and environments, traditional BPM systems' rule engines require frequent updates, resulting in high maintenance costs. Their reliance on pre-set rules prevents them from autonomously learning and optimizing processes, making them difficult to handle complex business logic.

[0005] Robotic process automation (RPA) technology simulates manual operations to perform repetitive tasks, reducing human error and improving efficiency. However, RPA technology is limited by its lack of intelligent decision-making capabilities and poor adaptability to environmental changes. It can only execute predefined tasks and cannot make intelligent decisions based on real-time data. In response to changing business processes, it requires extensive customization, increasing the complexity of implementation and maintenance.

[0006] Therefore, it is necessary to design a workflow construction method based on artificial intelligence and large-scale language model (LLM). Summary of the Invention

[0007] In response to the above-mentioned technical problems existing in the prior art, the present invention provides a method and system for constructing a workflow, which understands the designer's intention based on a large-scale language model and establishes a workflow according to the intention, thereby improving the adaptability, scalability and automation of the workflow.

[0008] The present invention discloses a method for constructing a workflow, comprising the following steps: obtaining a language or text for constructing a workflow; obtaining the intent of the language or text based on a large-scale language model; matching components according to the intent, wherein the components are used to perform tasks corresponding to the intent; determining whether two components have an interactive association; if so, establishing a transfer rule for the two components; if not, establishing a connection for the components to obtain a first workflow.

[0009] Preferably, the method of matching components includes:

[0010] According to the text, relevant knowledge content is obtained in the knowledge base;

[0011] Based on large-scale language models, the intent of the text and knowledge content is obtained and entities are extracted;

[0012] Match the corresponding intent type according to the intent and entity, as well as the intent category standard library;

[0013] According to the intent type, match the corresponding component;

[0014] Obtaining the urgency of the intention based on the text, the urgency keyword vocabulary, and the urgency classification threshold;

[0015] Based on the urgency, the intent types / components are prioritized.

[0016] Preferably, the two components include a first component and a second component, and the method for the two components to interact includes:

[0017] Determining whether the first task of the first component satisfies any one of the following trigger conditions or a combination thereof: a trigger condition based on business rules, a trigger condition based on data status, and a trigger condition based on time;

[0018] If so, a subtask is established according to the first task; and the subtask is executed through the second component according to the intention type, urgency, processing status, and data transmission rules and data conversion mapping relationship.

[0019] Preferably, the method for establishing a connection for a component includes:

[0020] Obtaining a connection template, wherein the connection template includes: a business process template, an approval process template, and an emergency processing process template;

[0021] Based on the intent and its complexity, select the corresponding connection template;

[0022] Determine the order of components based on data dependencies;

[0023] Connect components according to the order of connection templates and components, and establish data transfer mapping between components;

[0024] The data transfer mapping includes: unified data format, data type conversion between groups, data integrity verification, process status tracking, component execution status maintenance and abnormal status processing.

[0025] Preferably, the workflow optimization method includes:

[0026] Conducting inspection, analysis or periodic evaluation of the first workflow; or

[0027] Obtaining the question text of the first workflow;

[0028] According to the intention of the question text and a preset question template, or the results of inspection, analysis or regular evaluation, the first workflow is adjusted to obtain a second workflow.

[0029] Preferably, the inspection, analysis and periodic evaluation include:

[0030] Process integrity check, performance bottleneck identification, exception handling coverage, business rule adaptability assessment, processing efficiency data analysis, error rate statistical analysis, user satisfaction analysis, and resource utilization analysis;

[0031] Based on the results of the inspection, analysis or periodic evaluation, components, parameters and the first workflow are optimized.

[0032] Preferably, the problem text includes: process diagnosis problems, rule improvement problems and performance optimization problems,

[0033] Methods for adapting the first workflow based on the question text include:

[0034] Collect question text based on any of the following methods or their combination: structured question templates, custom questions, and option-based question confirmation;

[0035] Based on the question text, obtaining question intent and question parameters;

[0036] Adjust workflow rules, processing parameters, and component configurations based on the stated problem intent and problem parameters.

[0037] Preferably, the workflow execution method includes:

[0038] Get the user's input text;

[0039] Based on a large-scale language model, the intent of the input text is obtained and the intent type is matched;

[0040] Obtaining entity parameters of the input text and generating a first task;

[0041] Obtaining the urgency of the input text and setting the priority of the first task according to the urgency;

[0042] According to the intent type, matching the corresponding first component;

[0043] executing the first task by the first component according to the priority;

[0044] Determining whether the first task is interactively associated with the second component;

[0045] If so, generating a subtask based on the first task, and calling the second component to execute the subtask;

[0046] Generate reply language according to the first component and the situation of calling the second component;

[0047] The reply language is returned to the user.

[0048] The present invention also provides a system for implementing the above method, comprising a first input module, an intention recognition module, a component matching module, an interaction module, and a connection module;

[0049] The first input module is used to obtain language or text for constructing a workflow;

[0050] The intention recognition module is used to obtain the intention of the language or text based on a large-scale language model;

[0051] The component matching module is used to match components according to the intent, and the components are used to execute the intent;

[0052] The interaction module is used to establish a transfer rule for two components when there is an interactive association between the two components;

[0053] The connection module is used to establish a connection for the components to obtain a first workflow.

[0054] Preferably, the system further comprises an optimization module, a second input module and a question-answering module;

[0055] The optimization module is used to check, analyze or regularly evaluate the first workflow; or obtain the problem text of the first workflow;

[0056] and adjusting the first workflow to obtain a second workflow based on the intent of the question text and a preset question template, or the results of inspection, analysis, or regular evaluation;

[0057] The second input module is used to obtain user input text;

[0058] The question-answering module is used to generate a reply language based on the input text and the workflow, and return the reply language to the user;

[0059] The architecture of the system includes: workflow definition layer, task scheduling layer, parsing layer, engine layer and storage layer;

[0060] The workflow definition layer includes a code module, a branch module, a loop module, an interaction module, an integration module, a knowledge base module, and an intent recognition module;

[0061] The task scheduling layer is used to manage task planning and resource allocation of workflows;

[0062] The parsing layer is used to analyze the workflow and build the logical relationship between modules;

[0063] The engine layer includes a variable replacement module, a workflow execution engine, and an action processor;

[0064] The storage layer includes a data model, a consistency check module, a backup module and a recovery module.

[0065] Compared with the existing technology, the beneficial effects of the present invention are: based on the large rule language model, the user's intention is identified, and on this basis, by matching the corresponding components and organizing the components into a complete workflow, the adaptability, scalability and automation of the workflow can be improved; it can quickly respond to user requests and is applicable to a variety of different application scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 It is a flow chart of the method for constructing the workflow of the present invention;

[0067] Figure 2 It is a system logic block diagram of the workflow construction of the present invention;

[0068] Figure 3 It is a system architecture diagram of the present invention. DETAILED DESCRIPTION

[0069] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0070] The present invention is described in further detail below with reference to the accompanying drawings:

[0071] The first aspect of the present invention provides a method for constructing a workflow, such as Figure 1 , including the following steps:

[0072] Step 101: Obtain a language or text for constructing a workflow.

[0073] User input can be text or voice.

[0074] Step 102: Based on a large-scale language model (LLM), obtain the intent of the language or text.

[0075] Step 103: Match one or more components according to the intent.

[0076] Components are used to perform tasks corresponding to the intent; components can also be described as nodes.

[0077] Step 104: Determine whether there is an interactive relationship between the two components.

[0078] If yes, execute step 105 : establish transfer rules and mapping for the two components, and execute step 106 .

[0079] If not, execute step 106: establish a connection and data transfer mapping for the component to obtain a first workflow.

[0080] Step 107: Check, analyze or regularly evaluate the first workflow.

[0081] Step 108: Obtain the question text of the first workflow.

[0082] Step 109: According to the intention of the question text and the preset question template, as well as the results of inspection, analysis or regular evaluation, the first workflow is adjusted to obtain a second workflow.

[0083] Based on the large rule language model, the user's intention is identified. On this basis, by matching the corresponding components and organizing the components into a complete workflow, the adaptability, scalability and automation of the workflow can be improved; it can quickly respond to user requests and is suitable for a variety of different application scenarios.

[0084] In step 103, the method for matching components includes:

[0085] Step 301: Obtain relevant knowledge content in a knowledge base based on the text.

[0086] Knowledge content can be used to supplement the content of the text.

[0087] Step 302: Based on a large-scale language model, the intent of the text and knowledge content is obtained, and entities are extracted.

[0088] Step 303: Match the corresponding intent type according to the intent and entity, and the intent category standard library.

[0089] Step 304: Match corresponding components according to the intent type.

[0090] Step 305: Obtain the urgency of the intention based on the text, the emergency keyword vocabulary and the urgency classification threshold.

[0091] Step 306: Set a priority for the intent type / component based on the urgency.

[0092] In step 105, the two components include a first component and a second component, and the method for the two components to interact includes:

[0093] Step 501: Determine whether the first task of the first component meets any one of the following trigger conditions or a combination thereof: trigger conditions based on business rules, data status, and time.

[0094] If so, execute step 502: create a subtask based on the first task; and call the second component to execute the subtask based on the intention type, urgency, processing status, data transfer rules and data conversion mapping relationship.

[0095] If not satisfied, only the first component is called to complete the first task.

[0096] In step 109, the inspection, analysis, and periodic evaluation include: process integrity inspection, performance bottleneck identification, exception handling coverage, business rule adaptability assessment, processing efficiency data analysis, error rate statistical analysis, user satisfaction analysis, and resource utilization analysis. Based on the results of the inspection, analysis, or periodic evaluation, components, parameters, and the first workflow are optimized to achieve continuous updating and optimization of the workflow.

[0097] The problem text includes: process diagnosis problem, rule improvement problem and performance optimization problem. The method of adjusting the first workflow according to the problem text includes:

[0098] Step 901: Collect question text based on any one of the following methods or a combination thereof: structured question template, custom question, option question confirmation.

[0099] Step 902: Based on the question text, obtain question intent and question parameters.

[0100] Step 903: Adjust workflow rules, processing parameters, and component configurations according to the question intent and question parameters.

[0101] The first workflow that has been established can be tuned by natural language input, but is not limited thereto, and can also be tuned manually.

[0102] The methods for workflow execution include:

[0103] Step 921: Obtain the user's input text.

[0104] Step 922: Based on the large-scale language model, obtain the intent of the input text and match the intent type.

[0105] Step 923: Obtain entity parameters of the input text and generate a first task.

[0106] Step 924: Obtain the urgency of the input text, and set the priority of the first task according to the urgency.

[0107] Step 925: Match the corresponding first component according to the intent type.

[0108] Step 926: Execute the first task through the first component according to the priority.

[0109] Step 927: Determine whether the first task is interactively associated with the second component.

[0110] If so, execute step 928: generate a subtask according to the first task, and call the second component to execute the subtask.

[0111] If not, execute step 929.

[0112] Step 929: Generate a reply language based on the first component (and the situation of calling the second component). Preferably, the reply language is a natural language.

[0113] Step 930: Return the reply language to the user.

[0114] The second aspect of the present invention provides a system for implementing the above method, such as Figure 2 As shown, it includes a first input module 1, an intention recognition module 2, a component matching module 3, an interaction module 4, and a connection module 5;

[0115] The first input module 1 is used to obtain the language or text for constructing the workflow;

[0116] The intention recognition module 2 is used to obtain the intention of the language or text based on a large-scale language model;

[0117] The component matching module 3 is used to match components according to the intention, and the components are used to execute the intention;

[0118] The interaction module 4 is used to establish a transfer rule for two components when there is an interactive association between the two components;

[0119] The connection module 5 is used to establish a connection between the components to obtain a first workflow.

[0120] More specifically, the system further includes an optimization module 6, a second input module 7, and a question-answering module 8;

[0121] The optimization module 6 is used to check, analyze or regularly evaluate the first workflow; or obtain the problem text of the first workflow;

[0122] and adjusting the first workflow to obtain a second workflow based on the intent of the question text and a preset question template, or the results of inspection, analysis, or regular evaluation;

[0123] The second input module 7 is used to obtain the user's input text;

[0124] The question-answering module 8 is configured to generate a reply language based on the input text and the workflow, and return the reply language to the user.

[0125] like Figure 3 The architecture of the system includes: a workflow definition layer 11, a task scheduling layer 12, a parsing layer 13, an engine layer 14 and a storage layer 15.

[0126] The workflow definition layer 11 includes a code module, a branch module, a loop module, an interaction module, an integration module, a knowledge base module, an intent recognition module and an intent classification module.

[0127] Code module (Code): used to define the code snippets that need to be executed, currently supporting the commonly used Python / Javascript languages. Branch module Branch: implements conditional branch logic. Loop module Loop: processes loop structures. Interaction module Webhook: HTTP-based callback function, allowing lightweight, event-driven communication between two application programming interfaces (APIs) for interaction with external systems. Integration module Integration: integrates external services and APIs. Knowledge base module Knowledge: stores user knowledge data based on large model vectorization, retrieves text content related to user questions from the knowledge base, and can be used as context for downstream LLM and intent classification. Intent recognition module Llm: calls advanced large language models to answer questions or process user natural language. Intent classification module Intent: LLM intent recognition classification. By defining the classification description, the question classifier can infer the matching classification based on the user input and output the classification result.

[0128] In a specific embodiment, the workflow structure is defined using JSON format. The definition code is as follows:

[0129] { "id": "d60ea71b-84a0-98ef-5546-f7ba724d9d8d",

[0130] "name": "startTrigger",

[0131] "actionType": "TRIGGER",

[0132] "type": "WEBHOOK",

[0133] "layout": {

[0134] "position": {

[0135] "x": "223",

[0136] "y": "323"

[0137] },

[0138] "width": "111",

[0139] "hight": "222"

[0140] },

[0141] "data": {

[0142] "type": "",

[0143] "inputs": {

[0144] "p1": "{{question}}",

[0145] "p2": "{{step_13}}"

[0146] }}}

[0147] Among them, id: This is the unique identifier of the node / component. name: The name of the node, indicating that this is a start trigger. actionType: The action type of the node. type: "WEBHOOK" - The node type is Webhook, which means that it triggers the workflow through HTTP requests. layout: The layout information of the node on the interface. position indicates the horizontal and vertical position of the node on the interface; width and height represent the width and height of the node respectively; data: The user-configured data information stored by the node. type: The type of the node. inputs: defines the input parameters of the node.

[0148] The workflow definition layer also provides a graphical interface, allowing users to design workflows by dragging and dropping, ultimately arranging the data of each node into a complete workflow data protocol. The workflow definition layer can implement a version control system, supporting workflow history tracking and rollback.

[0149] The task scheduling layer 12 is used to manage task planning and resource allocation of workflows.

[0150] Key features include: once-only tasks. Repeating jobs: tasks that are executed periodically. Implementation: Uses the distributed task scheduling framework PowerJob to manage task execution. Implements a priority queue to ensure critical tasks are prioritized. A dynamic resource allocation algorithm automatically adjusts task execution resources based on system load.

[0151] The parsing layer 13 is used to analyze workflows and establish logical relationships between components. The core module is the ParseNodeRelationService, which is responsible for analyzing workflow definitions and establishing logical relationships between nodes. Implementation: Node relationships are represented using a directed acyclic graph (DAG). A topological sorting algorithm is implemented to determine the node execution order. A dependency matrix is ​​constructed to optimize parallel execution strategies.

[0152] The engine layer 14 includes a variable replacement module, a workflow execution engine, and an action processor. The variable replacement module dynamically replaces user-entered variables in the workflow with real data before node execution, ensuring correct data transfer between nodes. Implementation: A custom parsing service, VariableService, uses regular expressions for pattern matching and replacement, supporting complex nested variable structures.

[0153] The FlowExecutor is the main execution engine for workflows. It consists of the following components: a) ExecutionState: maintains the workflow execution state and stores key information during the workflow execution. b) ExecutionOutput: manages the output of execution results.

[0154] Action handlers include several specialized handlers, such as: IntegrationActionHandler: Handles integration actions; BranchActionHandler: Handles branch logic; CodeActionHandler: Executes code snippets; LoopActionHandler: Manages loop operations; KnowledgeActionHandler: Handles LLM vectorized knowledge base operations; LlmActionHandler: Handles LLM natural language dialogue operations; and IntentActionHandler: Manages intelligent recognition of user intent and intelligently distributes data flows.

[0155] The engine layer 14 also includes multiple functional modules: CODE: Executes custom code. Branch: Implements conditional branching. Loop: Implements loop logic. Knowledge: Vector retrieval allows users to quickly find user structured data. LLM: Integrates a large language model for natural language processing tasks. Intent: Inherits the large language model to enable the model to automatically identify user intent and automatically distribute tasks.

[0156] Implementation: Design action handlers using the chain of responsibility model to improve system scalability. Use a state machine model to manage workflow execution states and ensure consistent state transitions. Implement an event-driven architecture to improve system responsiveness and concurrent processing capabilities. Integrate monitoring and logging systems to track workflow execution in real time.

[0157] The storage layer 15 includes a data model, a consistency check module, a backup module, and a recovery module. PostgreSQL is used as the primary data storage system. Implementation method: Design an optimized data model to support efficient workflow data storage and retrieval. Implement a partitioned table strategy to improve large-scale data processing performance. Use a transaction mechanism to ensure data consistency. Implement a regular backup and recovery mechanism to ensure data security.

[0158] The present invention supports intelligent task decomposition, dynamic workflow generation, intelligent error handling and self-healing, and performance optimization. Intelligent task decomposition: Utilize the LLM model to automatically decompose complex tasks into executable subtasks; implement an adaptive task granularity adjustment algorithm to dynamically optimize the task decomposition strategy based on the execution effect. Dynamic workflow generation: Based on historical execution data and the current context, use machine learning algorithms to dynamically generate and optimize workflows; implement a workflow template library to support rapid reuse and customization. Intelligent error handling and self-healing: Design an error detection system based on a rule engine; implement an automatic retry mechanism, and intelligently select a recovery strategy based on the error type. Performance optimization: The workflow preheating mechanism can be used to load commonly used components in advance; use caching technology to optimize data access performance; implement an intelligent resource scheduling algorithm to dynamically allocate computing resources based on node complexity.

[0159] This invention uses natural language to generate workflows, allowing users to create and modify workflows using natural language descriptions, significantly lowering the barrier to entry and improving system accessibility. This includes natural language understanding, workflow mapping, context understanding and maintenance, natural language generation, workflow optimization suggestions, interactive workflow tuning, and more.

[0160] Natural Language Understanding: Parses user natural language input to extract key information and intent. Implementation: Abstracts the workflow node data model; uses the advanced natural language processing model GPT to learn the data model; identifies user intent to accurately capture user goals and needs; generates node data based on user-provided information. Key parameters: Model type: GPT-4O (for fast processing); intent classification accuracy: >95%; entity recognition F1 score: >0.9.

[0161] Workflow Mapping: Maps the results of natural language understanding to specific workflow structures and components. Implementation: Designs a rule library for mapping semantics to workflow components; implements a node-relationship-based reasoning mechanism to handle complex logical relationships; and uses models to evaluate the accuracy of node data and node relationship mappings. Key Technologies: Component Relationship Construction: Uses Postgres to store workflow components and relationships; Reasoning Algorithm: Employs a rule-based reasoning approach; Mapping Accuracy: >90%.

[0162] Context Understanding and Maintenance: Maintain user intent and workflow context across multiple conversations. Implementation: Implement a conversation state tracker to record historical user input and system responses. Use prompts to address user context illusions. Key Features: Context window size: 128k bytes of information stored over the last five conversations; conflict resolution accuracy: >85%.

[0163] Natural Language Generation: Converts system actions and feedback into natural language for user interaction. Implementation: Uses a template-based generation method to ensure output consistency; combines it with the GPT model to provide more natural and fluent responses; and manages multi-turn conversations to maintain conversational coherence and contextual understanding. Key metrics: Response generation time: <500ms; User satisfaction (assessed through feedback): >4.5 / 5.

[0164] Workflow Optimization Suggestions: Provide workflow optimization suggestions based on the user's natural language description. Implementation: Analyze the task objectives and constraints described by the user; utilize historical workflow data and a GPT model to generate optimization suggestions. Key Metrics: Suggestion Adoption Rate: >70%; Workflow Execution Efficiency Improvement: Average >20%.

[0165] Interactive Workflow Tuning: Allows users to gradually adjust and optimize workflows through natural language conversations. Implementation: Design an interactive question-and-answer module to guide users through workflow details; implement real-time workflow visualization to intuitively demonstrate the effects of user adjustments; and employ incremental learning algorithms to continuously refine the system's understanding of user preferences. Key features: Average conversation turns (per workflow optimization): <5; Reduced user operations (compared to traditional interfaces): >60%.

[0166] Interactive workflow tuning allows users to gradually adjust and optimize workflows through natural language conversations. Implementation: Design an interactive question-and-answer module to guide users through workflow details; implement real-time workflow visualization to intuitively demonstrate the effects of user adjustments; and employ incremental learning algorithms to continuously refine the system's understanding of user preferences. Key features: Average conversation turns (per workflow optimization): <5; Reduced user operations (compared to traditional interfaces): >60%.

[0167] This invention offers a high degree of integration and synergy, seamlessly integrating the natural language interface with the workflow definition layer. It incorporates natural language understanding result processing into the parsing layer (ParseNodeRelationService). It also extends FlowExecutor to support dynamic workflow execution based on natural language generation.

[0168] The present invention also implements an intent confirmation mechanism to prevent misoperation. It also sets permission controls to limit the scope of operations that specific users can perform through natural language. It also implements sensitive information filtering to ensure that confidential data is not leaked during natural language interactions.

[0169] In a specific embodiment, an intelligent HR customer service workflow is constructed.

[0170] Objective: Create an intelligent HR customer service workflow to automatically handle employees' frequently asked questions, leave applications, and document requests, improving HR department efficiency and employee satisfaction.

[0171] Here are the steps:

[0172] Step 951, User Input: User (HRBP) input: "Create an HR customer service workflow that can answer common employee questions, process leave requests, provide company policy documentation, and transfer complex questions to human HR. The workflow should be able to identify urgent issues and prioritize them."

[0173] Step 952, Natural Language Understanding: Use the GPT model to identify intent and extract entities; then match relevant components based on the context. Key components identified: Question and Answer System, Leave Processing, Social Security Consultation, Document Provisioning, Emergency Response Identification, and Manual Transfer. Intent Classification Accuracy: 98.2%.

[0174] The component matching process after intent recognition is as follows: pre-configure a second input module to process employee input requests; configure request format recognition rules; set request priority parameters; match the LLM for text analysis, configure an emergency keyword library, and set urgency classification thresholds.

[0175] Intent Classification: Use the CODE component to implement classification logic, establish a standard library of intent categories, and configure the intent recognition confidence threshold. The knowledge base query node connects to the Knowledge component, sets knowledge retrieval rules, and configures similarity matching parameters.

[0176] Step 953: Automatically generate workflow.

[0177] Interaction between components: Design data flow, define standard data structures (including intent type, urgency, processing status, etc.); set data transmission rules; configure data conversion mapping relationships. Trigger condition settings: conditional triggering based on business rules, conditional triggering based on data status, and conditional triggering based on time.

[0178] Automated connection implementation: Pre-configured connection templates include templates for common HR business processes, standard approval processes, and emergency response processes. Dynamic connection generation: Automatically selects connection templates based on intent complexity, determines node order based on data dependencies, and supports automatic generation of conditional branches.

[0179] Data transfer mechanism: Data transfer mapping between nodes: Data transfer association between nodes / components; unified data format definition; automatic data type conversion; data integrity verification. State management: Process state tracking; node execution state maintenance; abnormal state handling.

[0180] Step 954: Interactively optimize the design.

[0181] System optimization suggestion generation and design: analysis based on rule engine: process integrity check rules; performance bottleneck identification rules; exception handling coverage rules; business rule adaptability evaluation; analysis based on historical data; processing efficiency data analysis; error rate statistical analysis; user satisfaction analysis; resource utilization analysis.

[0182] User-guided optimization design: Process diagnosis questions, such as "Is the average processing time for this node reasonable?", "Is it necessary to add additional review steps?", and "Does the existing emergency processing mechanism meet requirements?". Rule improvement questions, such as "Please confirm the time threshold for vacation approvals," "Do we need to set up differentiated processes for different departments?", and "How do we define the priority of emergency situations?". Performance optimization questions, such as "Is it necessary to process multiple approval steps in parallel?", "What automated processing rules should be set up?", and "How can we improve knowledge base query efficiency?"

[0183] Optimize feedback collection and application: Question collection methods include: structured question templates; custom question input; and optional question confirmation. Feedback application mechanisms include: automatic update of process rules; dynamic adjustment of processing parameters; and optimized component configuration.

[0184] Continuous Optimization Mechanism: Regular optimization evaluations include performance indicator analysis, user satisfaction evaluation, and process efficiency evaluation. Dynamic adjustment strategies include automatic optimization based on evaluation results, rule updates based on feedback, and support for manual intervention and adjustment.

[0185] The established workflow performance parameter settings are as follows: 5. LLM uses the GPT-4O model with a response time limit of 3 seconds; the knowledge base uses Milvus, with an index update frequency of once an hour; intent classification uses a fine-tuned GPT model with a classification threshold set to 0.85; emergency identification priority settings are set as follows: security issues > discrimination / harassment complaints > salary issues; the maximum number of concurrent workflow processes is 200 requests / minute.

[0186] Test execution: Using 5,000 real employee request data (including inquiries, leave requests, document requests, etc.), the average first response time was 2.3 seconds, the problem resolution rate (without human intervention) was 87%, the automatic processing rate of leave requests was 95%, the accuracy of document provision was 99%, the accuracy of emergency identification was 99.5%, and the employee satisfaction score (based on feedback) was 4.6 / 5.

[0187] The execution process of a specific request is as follows: a. The employee enters: "I would like to apply for leave from next Wednesday to Friday."; b. The emergency identification node determines it is non-emergency (confidence level 99.8%); c. The intent classification node identifies it as "leave application" (confidence level 97.5%); d. The leave application processing node performs the following actions: checks the employee's leave balance (15 days of annual leave remaining); confirms the requested number of days (3 days); and triggers the supervisor approval process; e. The response generation node creates the reply: "Your 3-day leave application has been received and submitted to your supervisor for approval. Your current annual leave balance is 15 days. We will notify you of the approval result by email within 24 hours. If you have any questions, please feel free to ask."; f. The reply is sent to the employee via a webhook.

[0188] System learning and optimization: Automatically recommend relevant information to HR administrators and add it to the knowledge base; optimize the leave application process and add automatic calendar synchronization function.

[0189] The intelligent HR customer service workflow system demonstrated exceptional performance and efficiency. It successfully automated 87% of employee requests, significantly reducing the workload of the HR team. The system quickly responded to employee inquiries, accurately processed leave requests, and effectively identified and prioritized emergency cases. 95% of leave requests were automatically processed, significantly improving efficiency.

[0190] Of particular note is the system's learning capabilities, which enable it to continuously optimize and expand its functionality. For example, it automatically identifies new topics of concern to employees (such as high-temperature subsidy policies) and proactively recommends updating the knowledge base, significantly improving the system's practicality and adaptability.

[0191] The system achieved an employee satisfaction rating of 4.6 out of 5, demonstrating that it not only improves HR department efficiency but also significantly enhances the employee experience. Employees generally appreciate the system's fast response time (average 2.3 seconds) and high accuracy.

[0192] This invention has the following features: Intelligent process modeling: Automatically converts natural language descriptions into process models through NLP technology. AI-driven task scheduling: Dynamically optimizes task allocation and resource scheduling using machine learning algorithms. Adaptive rule engine: Integrated with LLM, the system autonomously learns and optimizes business rules to improve decision-making accuracy. Real-time monitoring and data analysis: Provides real-time visual monitoring and data analysis capabilities to optimize process execution. Multi-platform integration: Standardized API interfaces enable seamless integration with various enterprise systems.

[0193] Enterprises will experience significant improvements in business process management: Efficiency Improvement: Intelligent task scheduling and resource allocation optimization enable efficient business process operation. Intelligent Decision-Making: An adaptive rules engine automatically optimizes business rules using AI technology, improving the accuracy and timeliness of decision-making. Cost Reduction: Reduce manual intervention and maintenance costs by automating processes through intelligent means. Enhanced Flexibility: The system can adapt to business changes, quickly respond to dynamic demands, and enhance enterprise resilience. Data-Driven Optimization: Through real-time monitoring and data analysis, business processes are continuously optimized to improve overall operational efficiency.

[0194] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A method for constructing a workflow, characterized in that: The following steps are involved: Step 101: Obtaining a language or text for constructing a workflow; Step 102: Obtaining the intent of the language or text based on a large-scale language model; Step 103: matching a corresponding component according to the intent, wherein the component is used to execute a task corresponding to the intent; Step 104: Determine whether there is an interactive association between two components; If yes, execute step 105: establish a transfer rule for the two components, and execute step 106; If not, executing step 106: establishing a connection for the component to obtain a first workflow; The methods of workflow execution include: Get the user's input text; Based on a large-scale language model, the intent of the input text is obtained and the intent type is matched; Obtaining entity parameters of the input text and generating a first task; Obtaining the urgency of the input text, and setting the priority of the first task according to the urgency; According to the intent type, matching a corresponding first component; According to the priority, executing the first task by the first component; Determining whether the first task is interactively associated with the second component; If yes, generate a subtask according to the first task, and call the second component to execute the subtask; Generate reply language according to the first component and the situation of calling the second component; The reply language is returned to the user.

2. The method according to claim 1, characterized in that The methods for matching components include: According to the text, relevant knowledge content is obtained in a knowledge base; Based on a large-scale language model, the intent of the text and knowledge content is obtained, and entities are extracted; According to the intent and entity, and the intent category standard library, matching the corresponding intent type; According to the intent type, matching corresponding components; Obtaining the urgency of the intention according to the text, the urgency keyword thesaurus and the urgency classification threshold; Based on the urgency, the intent types / components are prioritized.

3. The method according to claim 1, characterized in that The two components include a first component and a second component, and the method for the two components to interact includes: Determine whether the first task of the first component satisfies any of the following trigger conditions or a combination thereof: a trigger condition based on a business rule, a trigger condition based on a data state, and a trigger condition based on time; If satisfied, a subtask is established according to the first task; and the subtask is executed through the second component according to the intent type, urgency, processing status, and data transfer rules and data conversion mapping relationship.

4. The method according to claim 1, characterized in that Methods for establishing connections for components include: Obtaining a connection template, wherein the connection template includes: a business process template, an approval process template, and an emergency processing process template; Based on the intent and its complexity, select the corresponding connection template; Determine the order of components based on data dependencies; Connect each component according to the order of connection template and components, and establish data transfer mapping between components; The data transfer mapping also includes: unified data format, data type conversion between groups, data integrity verification, process status tracking, component execution status maintenance and abnormal status processing.

5. The method according to claim 1, characterized in that Also includes methods for workflow optimization: inspecting, analyzing or periodically evaluating the first workflow; or Obtaining the question text of the first workflow; According to the intention of the question text and a preset question template, or the result of inspection, analysis or regular evaluation, the first workflow is adjusted to obtain a second workflow.

6. The method according to claim 5, characterized in that Such inspections, analyses and periodic assessments include: Process integrity check, performance bottleneck identification, exception handling coverage, business rule adaptability assessment, processing efficiency data analysis, error rate statistical analysis, user satisfaction analysis, and resource utilization analysis; Based on the results of the inspection, analysis or periodic evaluation, components, parameters and first work flows are optimized.

7. The method according to claim 5, characterized in that The problem texts include: process diagnosis problems, rule improvement problems and performance optimization problems. Methods for adjusting the first workflow according to the question text include: Collect question text based on any of the following methods or their combination: structured question templates, custom questions, option-based question confirmation; Based on the question text, obtain question intent and question parameters; Adjust workflow rules, processing parameters, and component configurations based on the problem intent and problem parameters.

8. A system, characterized in that: Used to implement the method according to any one of claims 1 to 7, the system comprises a first input module, an intention recognition module, a component matching module, an interaction module, and a connection module; The first input module is used to obtain the language or text for constructing the workflow; The intention recognition module is used to obtain the intention of the language or text based on a large-scale language model; The component matching module is used to match components according to the intent, and the components are used to execute the intent; The interaction module is used to establish a transfer rule for two components when there is an interactive association between the two components; The connection module is used to establish a connection for the components to obtain a first workflow.

9. The system according to claim 8, characterized in that Also includes an optimization module, a second input module and a question-answering module; The optimization module is used to check, analyze or regularly evaluate the first workflow; or obtain the problem text of the first workflow; and adjusting the first workflow according to the intention of the question text and a preset question template, or the result of inspection, analysis or regular evaluation, to obtain a second workflow; The second input module is used to obtain the user's input text; The question-answering module is used to generate a reply language according to the input text and the workflow, and return the reply language to the user; The architecture of the system includes: workflow definition layer, task scheduling layer, parsing layer, engine layer and storage layer; The workflow definition layer includes a code module, a branch module, a loop module, an interaction module, an integration module, a knowledge base module, and an intent recognition module; The task scheduling layer is used to manage task planning and resource allocation of workflows; The parsing layer is used to analyze the workflow and build the logical relationship between modules; The engine layer includes a variable replacement module, a workflow execution engine, and an action processor; The storage layer includes a data model, a consistency check module, a backup module and a recovery module.

Images