A workflow knowledge visualization method and system based on low-code interaction

Abstract

This invention provides a workflow knowledge visualization method and system based on low-code interaction, belonging to the field of workflow technology. The workflow knowledge visualization method includes: converting vectorized workflow knowledge stored in a knowledge base into executable visual components; constructing a component area, an editing area, and a recommendation area based on the visual components; pulling visual components from the component area to the editing area according to requirements to construct workflow nodes; displaying visual components matching the current operation characteristics of pulling visual components from the component area to the editing area in real time in the recommendation area, facilitating their pulling to the editing area to construct workflow nodes; and editing the workflow in the editing area according to requirements to ensure the workflow meets the specifications. This workflow knowledge visualization method allows workers to easily and quickly build workflows through simple operations such as drag and drop.

Description

Technical Field

[0001] This invention relates to the field of workflow technology, and more specifically to a workflow knowledge visualization method based on low-code interaction. Background Technology

[0002] Existing workflow generation technologies rely on skilled technical personnel, making it difficult for business users to participate directly due to a lack of technical background. Workflow knowledge is stored in non-visual formats such as vectors and text, hindering intuitive understanding and retrieval by business users. Furthermore, the generation process lacks manual intervention and optimization mechanisms, easily leading to workflows that deviate from actual business needs. Therefore, a low-code interactive workflow knowledge visualization method is needed. This method would transform abstract vectorized knowledge from the knowledge base into intuitive node-based graphs and icon-based components through a knowledge visualization module, allowing staff to easily and quickly build workflows through simple drag-and-drop operations. Summary of the Invention

[0003] The purpose of this invention is to provide a workflow knowledge visualization method and system based on low-code interaction. This workflow knowledge visualization method makes it convenient and quick for staff to build workflows through simple operations such as drag and drop.

[0004] To achieve the above objectives, embodiments of the present invention provide a workflow knowledge visualization method based on low-code interaction, the workflow knowledge visualization method comprising: Transform the vectorized workflow knowledge stored in the knowledge base into executable visual components; Based on the aforementioned visualization components, a component area, an editing area, and a recommendation area are constructed. Visual components are pulled from the component area to the editing area as needed to build workflow nodes; Based on the current operation characteristics of pulling visual components from the component area to the editing area, visual components that match the operation characteristics are displayed in real time in the recommendation area, so as to facilitate the pulling of workflow nodes to the editing area; The workflow is edited in the editing area according to requirements to make the workflow meet the requirements; The edited results are synchronized to the knowledge base to complete the update of the knowledge base; Once the editing is complete and the workflow is built, preview the workflow and run a simulation to complete the visual construction of the workflow.

[0005] Optionally, the vectorized workflow knowledge stored in the knowledge base can be converted into executable visual components, including: Obtain the business logic vector from the knowledge base and parse the dimensions in the business logic vector, where each dimension corresponds to a business feature; The weights of business features across each dimension are calculated using an attention mechanism: Formula (1), in, This represents the attention weight of the j-th dimension of the business feature. The attention matrix representing the science department can be obtained by training on historical business logic data. This represents the vector representation of the j-th dimension of the business feature, which is the vectorized result of the business logic for that dimension in the knowledge base. The subscript representing the dimension. Indicates the total number of dimensions; Based on the obtained attention weights, the business features corresponding to those with attention weights greater than a preset threshold are selected as core business features. Based on the obtained core business characteristics, establish a one-to-one correspondence between the core business characteristics and logical graph nodes, and design a standardized mapping dictionary; Based on the execution order information between nodes contained in the business logic vector, it is mapped to directed connections between nodes in the logic graph; Based on the core business characteristics and the mapping rules for the execution order information between nodes, a node-based logical graph is constructed.

[0006] Optionally, a node-based logical graph is constructed, including: Initialize the graph canvas and set the canvas size and coordinate system; Based on the nodes obtained from the parsing, generate corresponding node icons on the canvas; The force-directed layout algorithm is used to optimize the positions between nodes to avoid node overlap. Formula (2), in, This represents the force between node i and node j. Representing ideal coordinates, This represents the actual distance between node i and node j; Based on the execution order and flow conditions between nodes contained in the business logic vector, directed connections between nodes are generated, and flow rules are marked on the connections. The detailed attribute information parsed from the business logic vector is used as the floating tooltip information of the node, and an attribute panel is set below the graph to complete the construction of the node-based logic graph and form a visual node component.

[0007] Optionally, the vectorized workflow knowledge stored in the knowledge base can be converted into executable visual components, including: Obtain the tool attribute vector from the knowledge base and parse the dimensions in the tool attribute vector; Calculate the TF-IDF value of the dimension in the tool attribute vector; The dimensions with the highest TF-IDF values ​​in the tool's attribute vector are selected as the core functional features of the tool. Tool icons are generated by selecting corresponding element combinations from the icon element library based on the core functional features. The generated tool icons and detailed information from the tool attribute vectors are encapsulated to form visual tool components.

[0008] Optionally, the component area uses a grid layout to display all icon-based visual components, categorized by tool type. Each category has collapse / expand buttons for easy searching by business personnel. The component area supports a search function, allowing business personnel to enter tool names or function keywords to filter matching components in real time. The editing area uses a canvas design, allowing business personnel to drag and drop visual components to any position on the canvas. When a business personnel drag two visual components to the editing area and establish a relationship, the system automatically identifies the visual component type and business logic, generating logical connections with arrows. The connections are marked with general flow rules by default, and the flow conditions can be modified by double-clicking the connections. The editing area supports basic operations such as undo, redo, and save, and automatically records every step of the business personnel's operation to provide data support for subsequent real-time recommendations. The recommendation area uses a card layout to display adapted tool components and logical combination schemes in real time.

[0009] Optionally, based on the current operation characteristics of pulling visual components from the component area to the editing area, visual components matching the operation characteristics are displayed in real time in the recommendation area, including: The operational behaviors of business personnel are quantitatively modeled to construct their operational characteristics, which include the types of components dragged and dropped, the order of component dragging and dropping, the established connection relationships, the modified flow conditions, the set parameter configurations, and the search keywords. Assign weights to each operational feature; Based on the current operational characteristics of the aforementioned business personnel, calculate the similarity between their operational characteristics and those of historical users: Formula (3), in, Indicates the operational characteristics of the current business personnel Operational characteristics of historical users Similarity; The top n historical users with similarity greater than a preset threshold are selected as similar users; Tool components and logic combinations that are used more than a preset threshold by similar users in the same operation phase are used as the initial recommendation candidate set.

[0010] Optionally, based on the current operation characteristics of pulling visual components from the component area to the editing area, visual components matching the operation characteristics are displayed in real time in the recommendation area, including: Obtain the recommendation candidate set and calculate the matching score of the tool components and logic combination schemes in the recommendation candidate set: Formula (4), in, Indicates the degree of matching. This indicates the component association rule matching score. This indicates the score for matching the scene adaptation rules. Indicates the weighting coefficient; The matching scores are sorted from highest to lowest, and the top a tool components and the top b logic combinations are displayed in the recommendation area.

[0011] Optionally, the workflow can be edited in the editing area as needed to make it meet the requirements, including: modifying the input and output parameters of nodes in the workflow and adjusting the core functions of the tool, and being able to directly modify the flow rules of the connection annotations, and being able to add branch connections and set multi-branch flow conditions at the same time.

[0012] Optionally, after editing and workflow setup are complete, preview the workflow and run a simulation, including: Generate visual flowcharts using flowchart drawing tools, and color-code the flowcharts according to business scenarios; Weights are assigned to the core features of the nodes in the flowchart to calculate the total weight score of the nodes; Nodes with a comprehensive weight score greater than a preset threshold and that meet business rules are selected as key nodes and marked with a special style. The connections in the flowchart are marked with different colors according to the flow logic to complete the visual preview of the workflow; After the visual preview is complete, generate the simulation test data and input it into the workflow for simulation execution; Based on the results of the simulation, we will perform combined optimization and adjustments.

[0013] On the other hand, the present invention also provides a workflow knowledge visualization system based on low-code interaction, the workflow knowledge visualization system including a processor for executing a workflow knowledge visualization method based on low-code interaction as described above.

[0014] Through the above technical solution, this invention provides a workflow knowledge visualization method based on low-code interaction, which converts vectorized workflow knowledge stored in a knowledge base into executable visual components. Based on these visual components, a component area, an editing area, and a recommendation area can be constructed. After partitioning, visual components can be pulled from the component area to the editing area as needed, thereby constructing workflow nodes. During the construction process, based on the current operation characteristics of pulling visual components from the component area to the editing area, visual components matching these operation characteristics are displayed in real time in the recommendation area, facilitating the pulling of visual components from the recommendation area to the editing area to construct workflow nodes. When constructing a workflow, it can be edited in the editing area as needed to ensure it meets requirements. After editing, the results can be synchronized to the knowledge base, updating it. After editing and workflow construction, the workflow can be previewed and simulated, completing the visualization of the workflow. This workflow knowledge visualization method allows workers to easily and quickly construct workflows through simple drag-and-drop operations.

[0015] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0016] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of a workflow knowledge visualization method based on low-code interaction according to an embodiment of the present invention; Figure 2 This is a first flowchart of a workflow knowledge transformation method based on low-code interaction according to an embodiment of the present invention; Figure 3 This is a flowchart illustrating the construction logic graph of a workflow knowledge visualization method based on low-code interaction according to an embodiment of the present invention; Figure 4 This is a second flowchart of a workflow knowledge transformation method based on low-code interactive workflow knowledge visualization according to an embodiment of the present invention. Figure 5 This is a flowchart illustrating the acquisition of an initial recommendation candidate set according to a workflow knowledge visualization method based on low-code interaction, as described in one embodiment of the present invention. Figure 6This is a flowchart of a workflow knowledge visualization method based on low-code interaction according to an embodiment of the present invention, which provides a recommendation scheme. Figure 7 This is a flowchart of the operation simulation of a workflow knowledge visualization method based on low-code interaction according to an embodiment of the present invention. Detailed Implementation

[0017] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0018] In the embodiments of this application, certain software, components, models and other existing solutions in the industry may be mentioned. These should be regarded as exemplary and are only intended to illustrate the feasibility of implementing the technical solution of this application. However, they do not mean that the applicant has used or necessarily used the solution.

[0019] Figure 1 This is a flowchart of a workflow knowledge visualization method based on low-code interaction according to an embodiment of the present invention. In this invention, the visualization method may include the following steps: In step S1, the vectorized workflow knowledge stored in the knowledge base is converted into executable visual components.

[0020] In step S2, a component area, an editing area, and a recommendation area are constructed based on the visualization components.

[0021] In step S3, visual components are pulled from the component area to the editing area as needed to build workflow nodes.

[0022] In step S4, based on the current operation characteristics of pulling visual components from the component area to the editing area, visual components that match the operation characteristics are displayed in real time in the recommendation area so as to facilitate pulling them to the editing area to build workflow nodes.

[0023] In step S5, the workflow is edited in the editing area according to requirements to make the workflow meet the requirements.

[0024] In step S6, the editing results are synchronized to the knowledge base to complete the update of the knowledge base.

[0025] In step S7, after editing and workflow setup are completed, the workflow is previewed and simulated to complete the visual setup of the workflow.

[0026] In this invention, when visualizing workflows, vectorized workflow knowledge stored in the knowledge base can be converted into executable visual components. Based on these visual components, a component area, an editing area, and a recommendation area can be constructed. After partitioning, visual components can be pulled from the component area to the editing area as needed, thereby constructing workflow nodes. During the construction process, based on the current operation characteristics of pulling visual components from the component area to the editing area, visual components matching these operation characteristics are displayed in real-time in the recommendation area, facilitating the pulling of visual components from the recommendation area to the editing area to construct workflow nodes. When constructing a workflow, it can be edited in the editing area as needed to ensure it meets requirements. After editing, the results can be synchronized to the knowledge base, updating it. After editing and workflow construction, the workflow can be previewed and simulated, completing the visual construction of the workflow. This workflow knowledge visualization method allows workers to easily and quickly build workflows through simple drag-and-drop operations.

[0027] In one embodiment of the present invention, such as Figure 2 As shown, the first step in workflow knowledge transformation includes: In step S8, the business logic vector in the knowledge base is obtained and the dimensions in the business logic vector are parsed, where each dimension corresponds to a business feature.

[0028] In step S9, the weights of the business features in each dimension are calculated using an attention mechanism: Formula (1), in, This represents the attention weight of the j-th dimension of the business feature. The attention matrix representing the science department can be obtained by training on historical business logic data. This represents the vector representation of the j-th dimension of the business feature, which is the vectorized result of the business logic for that dimension in the knowledge base. The subscript representing the dimension. This represents the total number of dimensions.

[0029] In step S10, based on the obtained attention weights, the business features corresponding to those with attention weights greater than a preset threshold are selected as core business features.

[0030] In step S11, based on the obtained core business characteristics, a one-to-one correspondence between the core business characteristics and the logical graph nodes is established, and a standardized mapping dictionary is designed.

[0031] In step S12, the execution order information between nodes contained in the business logic vector is mapped to directed connections between nodes in the logic graph.

[0032] In step S13, a node-based logical graph is constructed based on the mapping rules of core business characteristics and execution order information between nodes.

[0033] In this invention, the vectorized workflow knowledge stored in the knowledge base can include business logic vectors and tool attribute vectors. When transforming workflow knowledge, the business logic vector in the knowledge base can be obtained and the dimensions in the business logic vector can be parsed, thereby extracting the core business features contained in the vector. Each dimension in the business logic vector can correspond to a business feature, such as process type, number of nodes, execution order, approval conditions, branch logic, exception handling rules, etc. After obtaining the dimensions, the weight of the business features of each dimension can be calculated through the attention mechanism, and its core formula is shown in formula (1). After obtaining the attention weights, the corresponding business features with attention weights greater than a preset threshold can be selected as core business features, such as "approval node", "branch flow conditions", "end node", etc. After obtaining the core business features, a one-to-one correspondence between the core business features and the logical graph nodes can be established, and a standardized mapping dictionary can be designed. For example, "starting feature" is mapped to "starting node," with a circular icon in green; "approval feature" is mapped to "approval node," with a rectangular icon in blue; "decision feature" (corresponding to the branch logic dimension in the vector) is mapped to "decision node," with a diamond icon in yellow; "processing feature" is mapped to "processing node," with a rounded rectangle icon in gray; and "ending feature" is mapped to "ending node," with a circular icon in red. Based on the execution order information between nodes contained in this business logic vector (reflected by the temporal features of the vector dimension), it can be mapped to directed connections between nodes in the logic graph, with the direction of the connection corresponding to the execution order. Flow conditions (obtained by parsing approval conditions, branch rules, and other dimensional features in the vector) can be marked on the connections. Based on the above mapping rules, that is, the mapping rules based on core business features and the execution order information between nodes, a node-based logic graph can be constructed.

[0034] In one embodiment of the present invention, such as Figure 3 As shown, the process of constructing a logical graph may include: In step S14, the map canvas is initialized, and the canvas size and coordinate system are set.

[0035] In step S15, based on the nodes obtained from the parsing, corresponding node icons are generated on the canvas.

[0036] In step S16, the positions between nodes are optimized using a force-directed layout algorithm to avoid node overlap: Formula (2), in, This represents the force between node i and node j. Representing ideal coordinates, This represents the actual distance between node i and node j.

[0037] In step S17, directed connections between nodes are generated based on the execution order and flow conditions between nodes contained in the business logic vector, and flow rules are marked on the connections.

[0038] In step S18, the detailed attribute information parsed from the business logic vector is used as the floating prompt information of the node, and an attribute panel is set below the graph to complete the construction of the node-based logic graph and form a visual node component.

[0039] In this invention, when constructing a node-based logic graph, the graph canvas can be initialized, and its size and coordinate system can be set. Then, based on the parsed node types, corresponding node icons can be generated on the canvas. The node positions of the node icons on the canvas can be automatically optimized using a force-directed layout algorithm to avoid node overlap. Based on the execution order and flow conditions between nodes contained in the business logic vector, directed connections between nodes can be generated, and flow rules can be marked on these connections, such as "Approval Passed → Processing Node" and "Approval Rejected → Starting Node". Detailed attribute information parsed from the business logic vector (such as node name, processing time limit, responsible person, associated tools, etc.) is used as floating tooltips for nodes. When business personnel hover their mouse over the corresponding node, they can view the complete attribute content. At the same time, an attribute panel can be set at the bottom of the graph. After selecting any node, the panel automatically displays the detailed attributes of that node, supporting subsequent editing operations. In this way, the construction of a node-based logic graph can be completed, forming a visual node component.

[0040] In one embodiment of the present invention, such as Figure 4 As shown, the second step in workflow knowledge transformation may include: In step S19, the tool attribute vector in the knowledge base is obtained, and the dimensions in the tool attribute vector are parsed.

[0041] In step S20, the TF-IDF value of the dimension in the tool attribute vector is calculated.

[0042] In step S21, the dimensions with the highest TF-IDF values ​​in the filter tool attribute vector are used as the core functional features of the tool.

[0043] In step S22, tool icons are generated by selecting corresponding element combinations from the icon element library based on the core functional features.

[0044] In step S23, the generated tool icon and the detailed information in the tool attribute vector are encapsulated to form a visual tool component.

[0045] In this invention, when transforming tool attribute vectors in a knowledge base, the tool attribute vectors in the knowledge base can be obtained first, and the dimensions in the tool attribute vectors can be parsed. Each dimension corresponds to the core attributes of the tool, such as tool type, function description, input parameters, output parameters, applicable scenarios, execution efficiency, and permission level. After obtaining the dimensions, the TF-IDF value of each dimension in the tool attribute vector can be calculated, and then the dimensions with the highest TF-IDF values ​​in the tool attribute vector can be selected as the core functional features of the tool. For example, after parsing the attribute vector of a "electricity billing rule tool," the core functional features are "electricity bill calculation," "rate adaptation," and "abnormal data processing." After obtaining the core functional features, corresponding elements can be selected from the icon element library to generate tool icons based on these core functional features. This icon element library can contain basic graphics (circles, rectangles, triangles, etc.) and functional symbols. A color scheme is used (green for calculation tools, blue for verification tools, yellow for statistical tools, etc.). For example, the "Electricity Billing Rules Tool" is characterized by its "calculation" nature, and its icon is a green circle with an embedded calculation symbol "=", with the tool name below. The "Metering Data Verification Tool" is characterized by its "verification" nature, and its icon is a blue rectangle with an embedded verification symbol "√", with the tool name below. After obtaining the generated tool icons, the generated tool icons and detailed information from the tool attribute vector (input / output parameters, applicable scenarios, access permissions, etc.) can be encapsulated to form a visual tool component. The component supports two interaction modes: one is to click the component icon to view the tool's detailed attribute panel; the other is to drag and drop the component into the editing area to automatically load the tool's default parameter configuration interface, eliminating the need for business personnel to manually input basic parameters.

[0046] Business personnel can retrieve visualized knowledge in two ways: 1. Keyword Search: Business personnel enter keywords (such as "approval," "electricity cost calculation," "data verification"), and the system converts the keywords into search vectors. The system calculates the cosine similarity between these keywords and the business logic vectors and tool attribute vectors in the knowledge base. Knowledge with a similarity ≥ 0.7 is selected and displayed as a node-based logic graph or icon-based component. 2. Graph Zooming: Business personnel can zoom in and out of the logic graph using the mouse wheel. This allows for a global overview of the process or focused viewing of detailed information for a specific node. Graph nodes can be dragged and dropped to adjust their positions, facilitating the understanding of knowledge relationships.

[0047] In one embodiment of the present invention, the low-code interactive interface for this visual operation can adopt a modular layout concept, with the interface divided into three core modules: a component area, an editing area, and a recommendation area. The component area can use a grid layout to display all icon-based visual components, categorized by tool type (such as calculation, verification, statistics, approval, etc.). Each category has collapse / expand buttons for easy searching by business personnel. The component area supports a search function, allowing business personnel to enter tool names or function keywords to filter matching components in real time.

[0048] The editing area can adopt a canvas-style design, allowing business users to drag and drop visual components to any position on the canvas. When a business user drags two visual components into the editing area and establishes a relationship, the system automatically identifies the type of visual component and the business logic, generating a logical connection with arrows. The connection is labeled with general flow rules by default, and the flow conditions can be modified by double-clicking the connection. The editing area supports basic operations such as undo, redo, and save, and automatically records every step of the business user's operation to provide data support for subsequent real-time recommendations. The recommendation section can use a card-style layout to display adapted tool components and logic combination solutions in real time. The recommendation section is divided into two sub-modules: "Component Recommendation" and "Logic Combination Recommendation," which provide precise recommendations based on the operational behaviors of business users. In one embodiment of the present invention, such as Figure 5 As shown, the process of obtaining the initial recommendation candidate set may include: In step S24, the operational behavior of business personnel is quantitatively modeled to construct the operational characteristics of business personnel. The operational characteristics include the type of dragged components, the order of component dragging, the established connection relationships, the modified flow conditions, the set parameter configurations, and the search keywords.

[0049] In step S25, a weight is assigned to each operational feature.

[0050] In step S26, based on the current user's operational characteristics, the similarity between their operational characteristics and those of historical users is calculated: Formula (3), in, Indicates the operational characteristics of the current business personnel Operational characteristics of historical users The similarity.

[0051] In step S27, the top n historical users with a similarity greater than a preset threshold are selected as similar users.

[0052] In step S28, tool components and logic combination schemes used more than a preset threshold by similar users in the same operation stage are used as the initial recommendation candidate set.

[0053] In this invention, when recommending and displaying visual components in real time, the operational behavior of business personnel can be quantitatively modeled first, thereby constructing the operational characteristics of business personnel. These operational characteristics may include the type of component dragged, the order of component dragging, the established connection relationship, the modified flow conditions, the set parameter configuration, the search keywords, etc. After obtaining the operational characteristics, weights can be assigned to these operational characteristics, such as a weight of 0.3 for dragging component type, a weight of 0.2 for component dragging order, a weight of 0.25 for connection relationship, and a weight of 0.25 for search keywords. The feature weights can be trained through historical operation data to ensure that the core operational behavior has a greater impact on the recommendation results. Based on the current operational characteristics of business personnel, the similarity between the current business personnel and the operational characteristics of historical users can be calculated using formula (3), and then the top n historical users with similarity greater than a preset threshold can be selected as similar users. The tool components and logical combination schemes used by similar users more than a preset threshold in the same operation stage (such as after dragging a certain type of component) are used as the initial recommendation candidate set.

[0054] In one embodiment of the present invention, such as Figure 6 As shown, the process of obtaining a recommendation can include: In step S29, a recommendation candidate set is obtained, and the matching scores of the tool components and logical combination schemes in the recommendation candidate set are calculated: Formula (4), in, Indicates the degree of matching. This indicates the component association rule matching score. This indicates the score for matching the scene adaptation rules. This represents the weighting coefficient.

[0055] In step S30, the obtained matching degrees are sorted from high to low, and the top a tool components and the top b logical combination schemes are selected and displayed in the recommendation area.

[0056] In this invention, the initial recommendation candidate set is filtered and sorted based on a preset business rule base to improve the recommendation accuracy. The business rule base contains two types of core rules: 1. Component association rules: define the association relationship between different types of components (such as "processing node" and "rejection node" after "approval node", and "verification node" and "statistics node" after "calculation node"); 2. Scenario adaptation rules: define the appropriate component combination according to the business scenario (such as electricity billing and government approval) (such as "electricity billing calculation component" and "rate verification component" after "metering data collection component" in the electricity billing scenario). After obtaining the recommendation candidate set, the matching score of the tool components and logical combination schemes in the recommendation candidate set can be calculated by formula (4). Then, the obtained matching degree can be sorted from high to low, and the first a tool components and the first b logical combination schemes can be selected and displayed in the recommendation area.

[0057] The recommendation area supports real-time updates. Every time a business user completes an operation (such as dragging a new component or modifying connection relationships), the system immediately updates the user operation feature vector, re-executes the recommendation algorithm, and updates the recommendation results. For example, when a business user drags the "Metering Data Acquisition Component" to the editing area, the recommendation area immediately recommends related components such as the "Electricity Fee Calculation Component" and the "Data Verification Component," as well as the logical combination scheme of "Acquisition → Calculation → Verification."

[0058] In one embodiment of the present invention, a multi-level manual editing entry can be added to the interactive interface, supporting business personnel to perform full-dimensional intervention and optimization of visualized knowledge, while ensuring the consistency between the edited knowledge and the original data in the knowledge base. Editing the workflow can include modifying the input and output parameters of nodes in the workflow and adjusting the core functions of the tools, and can directly modify the flow rules of connection annotations, and can simultaneously add branch connections and set multi-branch flow conditions.

[0059] Editing entry points can be set at logic graph nodes, visualization tool components, and flow lines: 1. Node editing entry point: Click on any node in the logic graph, and three buttons, "Edit," "Delete," and "Copy," will appear on the edge of the node. Click "Edit" to enter the node attribute editing interface; 2. Component editing entry point: Select a visualization tool component in the editing area, and "Parameter Configuration" and "Function Modification" buttons will appear below the component. Click to enter the component editing interface; 3. Connection editing entry point: Double-click on a flow connection to bring up the flow condition editing box, which supports direct modification of flow rules.

[0060] Specific editing functions include: Node attribute editing: The editing interface includes fields such as basic node information (name, type, processing time limit, responsible person) and business rule descriptions (e.g., approval conditions, processing logic). Business personnel can directly modify the field content. For example, for the "Approval Node," the approval conditions can be modified to "Amount > 10,000 yuan requires department manager approval" and "Amount ≤ 10,000 yuan requires supervisor approval." For the "Processing Node," the processing logic can be adjusted to "Automatically synchronize data to the ERP system" and "Generate an Excel report and push it." Tool component editing: Supports two types of editing operations: 1. Parameter configuration editing: Modify the input and output parameters of the tool (e.g., in the "Electricity Fee Calculation Component," modify parameters such as electricity rate and shared coefficient); 2. Function modification: Adjust the core functions of the tool (e.g., in the "Data Validation Component," add validation rules such as "Data format must be YYYY-MM-DD" and "Value range must be between 0-1000"). Flow rule editing: Supports direct modification of the flow rules of the connection annotations, and also allows adding branch connections and setting multi-branch flow conditions. For example, add a connection after the "Decision Node" and set the flow conditions as "Condition A meets → Node 1", "Condition B meets → Node 2", and "Condition does not meet → Node 3".

[0061] After business personnel complete the editing and click "Save", the knowledge base synchronization module can be automatically started to achieve consistent updates between visualized knowledge and original vectorized knowledge: 1. Text information update: The edited node attributes, component parameters, flow rules and other text information are directly updated to the text storage unit of the knowledge base; 2. Vector data update: A pre-trained text vector generation model (such as the BERT model) is used to convert the edited text information into new vector data, replacing the original business logic vectors and tool attribute vectors in the knowledge base; 3. Version management: A version record is created for each editing operation, recording the editor, editing time and editing content, and supporting rollback to historical versions to avoid knowledge loss due to accidental operation.

[0062] In one embodiment of the present invention, such as Figure 7 As shown, the process of running a simulation may include: In step S31, a visual flowchart is generated using a flowchart drawing tool, and the flowchart is color-coded according to the business scenario.

[0063] In step S32, weights are assigned to the core features of the nodes in the flowchart to calculate the total weight score of the nodes.

[0064] In step S33, nodes with a comprehensive weight score greater than a preset threshold and that meet business rules are selected as key nodes and marked with a special style.

[0065] In step S34, the lines in the flowchart are marked with different colors according to the flow logic to complete the visual preview of the workflow.

[0066] In step S35, after the visualization preview is completed, the simulation test data is generated and entered into the workflow for simulation operation.

[0067] In step S36, combined optimization and adjustment are performed based on the results of the simulation.

[0068] In this invention, after the workflow is built, it can be previewed and simulated to expose potential problems in advance and support business personnel to perform reverse optimization.

[0069] During preview and simulation, flowchart drawing tools (such as Draw.io) can be used to generate high-definition visual flowcharts. The flowcharts are color-coded according to the business scenario (e.g., blue for government approval scenarios and yellow for electricity billing scenarios) to ensure visual clarity. The preview interface supports zooming, panning, and node positioning, allowing business personnel to quickly view the overall process and details. Weights are assigned to the core features of nodes in the flowchart to calculate the total weight score of nodes. Based on the workflow's business logic vector, core features of nodes can be extracted (e.g., whether it is a branch node, whether it contains approval rules, whether it is associated with core tools, whether the processing time limit is strict, etc.). Weights are then assigned to each feature (e.g., branch node weight 0.3, approval node weight 0.25, associated core tool weight 0.25, strict time limit node weight 0.2), and the comprehensive weight score of each node is calculated. Combined with preset business rules (e.g., "both the start and end nodes are key nodes" and "nodes involving core data processing are key nodes"), nodes undergo secondary screening. In this way, nodes with a comprehensive weight score greater than a preset threshold and that meet business rules can be selected as key nodes. For example, in the electricity billing process, the "metering data collection node," "electricity bill calculation node," "approval node," and "result output node" are all key nodes. Confirmed key nodes are marked with special styles, such as bolding the node border, adding a highlighted background (red), and labeling them with the word "key." Connections in the flowchart are marked with different colors according to the flow logic. For example, black lines are used for normal flows, blue lines for branch flows, and red lines for abnormal flows. The flow conditions are clearly marked on the lines, ensuring that business personnel can intuitively identify core processes and flow rules, thus enabling a visual preview of the workflow.

[0070] After completing the visual preview, simulated test data can be generated, or business personnel can import real business data for simulation.

[0071] During the simulation, each node can be triggered sequentially according to the workflow execution order, recording the execution status (success, failure, blocked), execution time, input and output data, and other information for each node. For key nodes, the simulation focuses on the execution logic of their business rules. For example, the "approval node" simulates the approval process flow under different monetary scenarios, and the "calculation node" simulates the calculation results under different parameter configurations.

[0072] The simulation results output can generate a simulation run report, which includes three parts: 1. Execution result statistics: success rate of each node, average execution time, number of exceptions, etc. 2. Potential problem warning: marking nodes with logical conflicts (such as contradictory flow conditions), nodes with low execution efficiency (time exceeding the threshold), and nodes with abnormal data interaction (input and output data mismatch). 3. Optimization suggestions: recommending optimization solutions based on simulation results (such as replacing inefficient tool components, adjusting the execution order of key nodes, and optimizing flow conditions).

[0073] After reviewing the simulation report, business personnel can directly click on the problematic nodes in the report to jump to the low-code editing interface for targeted adjustments (such as replacing tool components, modifying node attributes, and optimizing workflow rules). After the adjustments are completed, the preview simulation module can be restarted until the workflow simulation execution is error-free and meets the actual business needs. Then, an executable workflow file (supporting multiple formats such as BPMN, XML, and JSON) will be output.

[0074] On the other hand, the present invention can also provide a workflow knowledge visualization system based on low-code interaction. The workflow knowledge visualization system may include a processor for executing a workflow knowledge visualization method based on low-code interaction as described above.

[0075] Through the above technical solution, this invention provides a workflow knowledge visualization method based on low-code interaction, which converts vectorized workflow knowledge stored in a knowledge base into executable visual components. Based on these visual components, a component area, an editing area, and a recommendation area can be constructed. After partitioning, visual components can be pulled from the component area to the editing area as needed, thereby constructing workflow nodes. During the construction process, based on the current operation characteristics of pulling visual components from the component area to the editing area, visual components matching these operation characteristics are displayed in real time in the recommendation area, facilitating the pulling of visual components from the recommendation area to the editing area to construct workflow nodes. When constructing a workflow, it can be edited in the editing area as needed to ensure it meets requirements. After editing, the results can be synchronized to the knowledge base, updating it. After editing and workflow construction, the workflow can be previewed and simulated, completing the visualization of the workflow. This workflow knowledge visualization method allows workers to easily and quickly construct workflows through simple drag-and-drop operations.

[0076] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0077] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0078] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0079] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0080] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0081] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0082] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0083] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0084] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A workflow knowledge visualization method based on low-code interaction, characterized in that, The workflow knowledge visualization method includes: Transform the vectorized workflow knowledge stored in the knowledge base into executable visual components; Based on the aforementioned visualization components, a component area, an editing area, and a recommendation area are constructed. Visual components are pulled from the component area to the editing area as needed to build workflow nodes; Based on the current operation characteristics of pulling visual components from the component area to the editing area, visual components that match the operation characteristics are displayed in real time in the recommendation area, so as to facilitate the pulling of workflow nodes to the editing area; The workflow is edited in the editing area according to requirements to make the workflow meet the requirements; The edited results are synchronized to the knowledge base to complete the update of the knowledge base; Once the editing is complete and the workflow is built, preview the workflow and run a simulation to complete the visual construction of the workflow.

2. The workflow knowledge visualization method according to claim 1, characterized in that, Transform the vectorized workflow knowledge stored in the knowledge base into executable visual components, including: Obtain the business logic vector from the knowledge base and parse the dimensions in the business logic vector, where each dimension corresponds to a business feature; The weights of business features across each dimension are calculated using an attention mechanism: Official (1), in, This represents the attention weight of the j-th dimension of the business feature. The attention matrix representing the science department can be obtained by training on historical business logic data. This represents the vector representation of the j-th dimension of the business feature, which is the vectorized result of the business logic for that dimension in the knowledge base. The subscript indicating the dimension. Indicates the total number of dimensions; Based on the obtained attention weights, the business features corresponding to those with attention weights greater than a preset threshold are selected as core business features. Based on the obtained core business characteristics, establish a one-to-one correspondence between the core business characteristics and logical graph nodes, and design a standardized mapping dictionary; Based on the execution order information between nodes contained in the business logic vector, it is mapped to directed connections between nodes in the logic graph; Based on the core business characteristics and the mapping rules for the execution order information between nodes, a node-based logical graph is constructed.

3. The workflow knowledge visualization method according to claim 2, characterized in that, Constructing a node-based logical graph includes: Initialize the graph canvas and set the canvas size and coordinate system; Based on the nodes obtained from the parsing, generate corresponding node icons on the canvas; The force-directed layout algorithm is used to optimize the positions between nodes to avoid node overlap. Official (2), in, This represents the force between node i and node j. Representing ideal coordinates, This represents the actual distance between node i and node j; Based on the execution order and flow conditions between nodes contained in the business logic vector, directed connections between nodes are generated, and flow rules are marked on the connections. The detailed attribute information parsed from the business logic vector is used as the floating tooltip information of the node, and an attribute panel is set below the graph to complete the construction of the node-based logic graph and form a visual node component.

4. The workflow knowledge visualization method according to claim 1, characterized in that, Transform the vectorized workflow knowledge stored in the knowledge base into executable visual components, including: Obtain the tool attribute vector from the knowledge base and parse the dimensions in the tool attribute vector; Calculate the TF-IDF value of the dimension in the tool attribute vector; The dimensions with the highest TF-IDF values ​​in the tool's attribute vector are selected as the core functional features of the tool. Tool icons are generated by selecting corresponding element combinations from the icon element library based on the core functional features. The generated tool icons and detailed information from the tool attribute vectors are encapsulated to form visual tool components.

5. The workflow knowledge visualization method according to claim 1, characterized in that, The component area uses a grid layout to display all icon-based visual components, categorized by tool type. Each category has collapse / expand buttons for easy searching by business personnel. The component area supports a search function, allowing business personnel to enter tool names or function keywords to filter matching components in real time. The editing area uses a canvas design, allowing business personnel to drag and drop visual components to any position on the canvas. When a business personnel drag two visual components to the editing area and establish a relationship, the system automatically identifies the visual component type and business logic, generating a logical connection with arrows. The connection is marked with general flow rules by default, and the flow conditions can be modified by double-clicking the connection. The editing area supports basic operations such as undo, redo, and save, and automatically records every step of the business personnel's operation to provide data support for subsequent real-time recommendations. The recommendation area uses a card layout to display the appropriate tool components and logic combination schemes in real time.

6. The workflow knowledge visualization method according to claim 5, characterized in that, Based on the current operation characteristics of pulling visual components from the component area to the editing area, visual components matching the operation characteristics are displayed in real time in the recommendation area, including: The operational behaviors of business personnel are quantitatively modeled to construct their operational characteristics, which include the types of components dragged and dropped, the order of component dragging and dropping, the established connection relationships, the modified flow conditions, the set parameter configurations, and the search keywords. Assign weights to each operational feature; Based on the current operational characteristics of the aforementioned business personnel, calculate the similarity between their operational characteristics and those of historical users: Official (3), in, Indicates the operational characteristics of the current business personnel Operational characteristics of historical users Similarity; The top n historical users with similarity greater than a preset threshold are selected as similar users; Tool components and logic combinations that are used more than a preset threshold by similar users in the same operation phase are used as the initial recommendation candidate set.

7. The workflow knowledge visualization method according to claim 6, characterized in that, Based on the current operation characteristics of pulling visual components from the component area to the editing area, visual components matching the operation characteristics are displayed in real time in the recommendation area, including: Obtain the recommendation candidate set and calculate the matching score of the tool components and logic combination schemes in the recommendation candidate set: Official (4), in, Indicates the degree of matching. This indicates the component association rule matching score. This indicates the score for matching the scene adaptation rules. Indicates the weighting coefficient; The matching scores are sorted from highest to lowest, and the top a tool components and the top b logic combinations are displayed in the recommendation area.

8. The workflow knowledge visualization method according to claim 1, characterized in that, Edit the workflow in the editing area as needed to make the workflow meet the requirements, including: modifying the input and output parameters of nodes in the workflow and adjusting the core functions of the tool, being able to directly modify the flow rules of the connection annotations, and being able to add branch connections and set multi-branch flow conditions at the same time.

9. The workflow knowledge visualization method according to claim 1, characterized in that, Once editing and workflow setup are complete, preview the workflow and run a simulation, including: Generate visual flowcharts using flowchart drawing tools, and color-code the flowcharts according to business scenarios; Weights are assigned to the core features of the nodes in the flowchart to calculate the total weight score of the nodes; Nodes with a comprehensive weight score greater than a preset threshold and that meet business rules are selected as key nodes and marked with a special style. The connections in the flowchart are marked with different colors according to the flow logic to complete the visual preview of the workflow; After the visual preview is complete, generate the simulation test data and input it into the workflow for simulation execution; Based on the results of the simulation, we will perform combined optimization and adjustments.

10. A workflow knowledge visualization system based on low-code interaction, characterized in that, The workflow knowledge visualization system includes a processor for executing a workflow knowledge visualization method based on low-code interaction as described in any one of claims 1-9.

Images