Form and process generation system based on intelligent agent

Abstract

The invention discloses an agent-based form and process generation system, which is characterized by comprising an intelligent form design module, a process center module and an agent design module, the intelligent form design module is used for analyzing a form design demand input by a user, extracting form demand information and generating a form layout through a component layout engine; the flow center module is used for generating node configuration information based on visual operation of flow nodes and mapping the node configuration information into a flow definition object; a standard process definition file is generated through deployment of a process packaging engine, and execution of a process instance is driven; the agent design module is used for receiving a module design request and matching in a pre-registered model library and a preset business library to obtain a target model and adaptive business knowledge; and generating recommended content. The design process is visualized through dragging operation, design suggestions are automatically given by the AI intelligent agent, and the difficulty and period of enterprise-level application development are simplified.

Description

TECHNICAL FIELD

[0001] The present application relates to form and process low-code development, in particular to a form and process generation system based on intelligent agents. BACKGROUND

[0002] With the deepening of enterprise digital transformation, the demand for rapid construction of business processes and data forms is increasingly urgent. Low-code or no-code platforms aim to quickly build forms and application processes through visual methods to reduce application development barriers and allow non-professional developers to participate in construction to adapt to rapidly changing business needs. Existing platforms still have significant deficiencies in achieving this goal.

[0003] However, existing low-code form and process design platforms use completely encapsulated designers, which are simple to use but generate rigid form structures that are difficult to handle complex business logic and custom requirements, and the generated code is difficult to integrate with existing systems. Another existing platform provides open APIs and native development mode, which is highly flexible, but relies on manual coding of professional process definition files, requiring developers to have professional coding skills, increasing the technical threshold. At the same time, the process design and form design of existing platforms are mutually exclusive, the process node configuration is complex, and there is a lack of intelligent assistance capabilities, requiring users to manually handle business rules and component matching, making it difficult to ensure design efficiency and accuracy. These factors together make it difficult for existing platforms to achieve efficient, secure, and easy-to-use form and process development in complex enterprise scenarios. SUMMARY

[0004] To solve the above problems, the present application proposes a form and process generation system based on intelligent agents, comprising: a form intelligent design module, a process center module, and an intelligent agent design module. The form intelligent design module is configured to analyze user input form design requirements, extract form requirement information, and generate a form layout based on the form requirement information through a component layout engine. The process center module is configured to generate node configuration information based on the user's visual operation on the process nodes in the front-end interface, map the node configuration information to a process definition object, and deploy the process definition object through a process packaging engine to generate a standard process definition file and drive the execution of a process instance. The intelligent agent design module is configured to receive module design requests from the form intelligent design module or the process center module, extract the request design type and request design content, match in a pre-registered model library and a pre-set business library to obtain a target model and adaptive business knowledge, generate recommended content based on the adaptive business knowledge and the request design content through the target model, and return to the request initiation module.

[0005] In one example, the form requirement information includes a field name, a field type, and a verification rule; and the form layout is generated based on the form requirement information by the component layout engine, specifically including: based on the field type, matching in a predefined component library to obtain a corresponding form component; configuring a corresponding component attribute for the form component according to the verification rule, and combining the configured form component based on a preset layout rule to generate the form layout.

[0006] In one example, the node configuration information is generated based on the visual operation of the user on the process node in the front-end interface, specifically including: capturing a drag-and-drop operation and a node connection operation of the user on the process node in the front-end interface; analyzing the drag-and-drop operation and the node connection operation to extract corresponding node operation information; the node operation information includes a node type, a node handler, a coordinate position, and a connection relationship; and combining the node operation information of all process nodes to obtain the node configuration information.

[0007] In one example, the node configuration information is mapped to a process definition object, specifically including: identifying the node type in the node configuration information, creating a process node object corresponding to the node type; configuring the node configuration information into a corresponding attribute of the process node object, establishing a flow path between the process node objects according to the connection relationship between the process nodes; and based on the flow path, assembling all the configured process node objects to generate the process definition object.

[0008] In one example, the process definition object is deployed by the process packaging engine to generate a standard process definition file to drive the execution of a process instance, specifically including: receiving the process definition object by the process packaging engine, serializing the process definition object to obtain a standard process definition file; registering the standard process definition file in a workflow engine to complete process deployment; in response to a business trigger instruction, creating a new process instance according to the deployed standard process definition file, and driving the new process instance by the workflow engine to execute a preset node path in the standard process definition file.

[0009] In an example, the generating the recommended content based on the adaptive business knowledge and the request design content through the target model specifically comprises: generating a design context prompt based on the adaptive business knowledge, splicing the design context prompt and the request design content to obtain a model input prompt; inputting the model input prompt into the target model, vectorizing the model input prompt through the target model to obtain an input prompt vector; performing forward propagation calculation based on the input prompt vector through a neural network of the target model to obtain an original prompt vector output by an output layer of the neural network; and generating the recommended content based on the original prompt vector.

[0010] In an example, the generating the recommended content based on the original prompt vector specifically comprises: decoding the original prompt vector to generate an initial text sequence, extracting a structured configuration element in the initial text sequence; assembling the structured configuration element based on a preset structured format to generate the recommended content.

[0011] In an example, the data security module is configured to receive an operation request and to-be-stored data from the form intelligent design module, the process center module or the agent design module, perform permission verification on the operation request, determine whether to execute the operation request according to a verification result, store the to-be-stored data after encryption conversion of a sensitive field, record a storage log, and generate a system audit log based on the operation request, the verification result and the storage log.

[0012] In an example, the performing permission verification on the operation request and determining whether to execute the operation request according to a verification result specifically comprises: obtaining user identity information and a request operation type in the operation request, matching a corresponding system role and a permission set according to the user identity information, verifying whether the request operation type is contained in the permission set, permitting execution of the operation request when the verification is passed, and rejecting execution of the operation request and returning a permission error information when the verification is not passed.

[0013] In an example, the system integration module is configured to receive a data exchange request from the form intelligent design module or the process center module, call a corresponding preconfigured interface to establish a connection with an external business system according to an interaction type of the data exchange request, and perform conversion between a platform data format and a data format of the external business system to complete data synchronization between the platform and the external business system.

[0014] The form and process generation system based on an agent can bring the following beneficial effects: By introducing the agent technology, the technical threshold of form and process design is significantly reduced, and the development efficiency and design quality are improved. The form intelligent design module can automatically analyze the user's natural language requirements, accurately recommend and configure form components, and realize the rapid design of what you think. The process center module converts complex business processes into executable standard files through visual design and automatic deployment, greatly shortening the cycle from process design to online.

[0015] At the same time, the system builds a comprehensive support and guarantee system. The agent design module provides unified intelligent decision support for various design scenarios to ensure that the output results are consistent with the actual business. The data security module establishes a security protection covering the whole operation process to ensure the security and traceability of sensitive enterprise data. The system integration module connects the data channel with external business systems through standardized interfaces, so that the platform can be flexibly integrated into the existing IT ecosystem of the enterprise, expanding the application value. BRIEF DESCRIPTION OF DRAWINGS

[0016] The drawings described herein are used to provide further understanding of the present application, and form a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application, and do not constitute an improper limitation on the present application. In the drawings: Figure 1 The architecture diagram of the form and process generation system based on the agent in the embodiments of the present application is shown. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical scheme and advantages of the present application clearer, the technical scheme of the present application will be described in detail below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0018] The technical scheme provided by each embodiment of the present application will be described in detail below in combination with the drawings.

[0019] As shown in the drawings, Figure 1 The form and process generation system based on the agent in the embodiments of the present application includes a form intelligent design module, a process center module, an agent design module, a data security module and a system integration module. The form intelligent design module is mainly used for analyzing the form design requirements input by the user, extracting form requirement information, and generating a form layout based on the form requirement information through a component layout engine.

[0020] In specific implementation, the form intelligent design module includes a requirement analysis module, a component recommendation and attribute configuration module, a form preview and optimization module.

[0021] The requirement analysis module receives the form design requirements input by the user in natural language, uses semantic understanding technology such as the BERT pre-training model to analyze the input natural language text, extracts key information, and converts unstructured requirements into structured form requirement information by identifying and extracting key information. Among them, the extracted key information includes but is not limited to: the required field name, the data type corresponding to each field, such as text, number, date, etc., and the verification rules required by the business rules, such as whether it is required, format requirements, numerical range, etc.

[0022] The component recommendation and attribute configuration module supports four types of form components, namely basic components, advanced components, layout components, and business components. Among them, the basic components include eighteen common components such as single-line text, multi-line text, characters, numbers, single selection box, and multiple selection box; the advanced components include eleven advanced components such as sub-form, associated form, organization, personnel selection, embedded web page, and embedded picture; the layout components are mainly six layout components such as grid layout, grid layout, tab page, and folding panel; and the business components are custom components developed by the user through the component development module. Each component is labeled with basic attributes, verification attributes, and advanced attributes. The basic attributes include component name and fill-in prompt; the verification attributes include required and custom verification rules; and the advanced attributes include component id, whether the component can be repeated, and whether the field can be queried. Based on the collaborative filtering algorithm, combined with the user's structured data and historical form design data, such as component selection records in the same business scenario, the optimal component combination is recommended; at the same time, the component basic attributes are automatically configured according to the field type, such as the length limit of the text box component and the date format of the date selector. The user can manually adjust according to the requirements.

[0023] In specific implementation, the component recommendation and attribute configuration module automatically selects and configures form components based on structured form requirement information. Specifically, according to the field type, the most suitable form component is matched from the pre-defined and classified component library. For example, the field type "single selection" will match the single selection box component, and the field type "personnel selection" will match the personnel selector component.

[0024] After matching the basic components, the attributes are automatically configured for the components based on the extracted verification rules. For example, for a field identified as "required", the "required" verification attribute is automatically enabled for the corresponding component; for a field type "mobile phone number", in addition to matching a single-line text input box component, it will also automatically configure a regular expression verification rule for the mobile phone number format.

[0025] In addition, the algorithm of collaborative filtering is combined to analyze the high-frequency component combination in the historical form design under similar business scenarios, and the current component selection is optimized and recommended to improve the accuracy and efficiency of the design.

[0026] After the selection and attribute configuration of the components are completed, the preset layout rules are used to automatically arrange and combine the form components, such as a grid system and a flow layout, to generate an initial form layout.

[0027] The form preview and optimization module generates a form draft, renders a form preview interface through a visual engine, and allows a user to view the form effect in real time. Meanwhile, a form compliance detection algorithm is built in to check the rationality of component combination, such as matching a mobile phone number field with a mobile phone verification component, and to check the completeness of attribute configuration, such as whether a prompt is set for a required field, and to give optimization suggestions. An intelligent agent can design form fields, select corresponding components, supplement component basic configurations, and build a corresponding form according to the actual demand of a user, and the user can optimize the form on this basis.

[0028] The process center module generates node configuration information based on the visual operation of a user on a process node in a front-end interface, maps the node configuration information to a process definition object, deploys the process definition object through a process packaging engine, and generates a standard process definition file to drive the execution of a process instance.

[0029] In a specific implementation, the process center module mainly performs secondary packaging on a flowable workflow engine, packages core interfaces such as process engine initialization, process deployment, and task execution, simplifies the complexity of engine calling, and improves the compatibility of the platform with flowable. The process center module includes simple and advanced modes.

[0030] The simple mode includes six types of core nodes, including a handler node, a copy person node, a conditional branch node, a parallel branch node, a sub-process node, and an executor node. The handler node supports specifying members, specifying posts, self-selecting handlers, the initiator himself, and specifying personnel in a form; in addition to the specified post and the initiator himself, multiple handlers or signatures can be configured; button permissions that can be operated by the handler, such as approval, rejection, return, and transfer, can be configured; in addition, process timeout processing and automatic processing of adjacent nodes can be configured to provide configuration for multiple use scenarios. The copy person node supports email or system message notification. The conditional branch node provides visual configuration of branch conditions. The parallel branch node supports simultaneous execution of multiple branches, and aggregation after the completion of the branches. The sub-process node supports nesting of created processes. The executor node supports custom callback functions and sending notifications to integrated platforms. A user adds nodes through a drag-and-drop method, configures node parameters, and the platform automatically generates a process model, which is suitable for the construction of simple processes.

[0031] The advanced mode implementation process is specifically: on the basis of the simple mode, new start events (such as timing start, message trigger start, etc.), task nodes (such as user task, service task, script task, etc.), gateways (such as exclusive gateway, parallel gateway, inclusive gateway, etc.), end events (such as normal end, abnormal end, etc.), intermediate events (such as timer intermediate event, message intermediate event, etc.), boundary events (such as error boundary event, compensation boundary event, etc.) and other nodes are added. Support complex connection relationship configuration between nodes (such as loop process, event triggered process, etc.), support import through XML file, meet the demand of complex business scenario.

[0032] It should be noted that the advanced mode has high flexibility and high requirement for the user. The user can design intelligent agent based on the process to realize the automatic design of the process. First, based on the built-in optional large model in the platform, the user's natural language process requirement is parsed, and the process node, node order, trigger condition and other information are extracted. Based on the parsed process information, the flowable XML generation interface is called, and the process XML file is automatically generated according to the flowable specification, including process definition, node configuration, flow rule and other contents. At the same time, the built-in XML verification engine checks the correctness of the XML syntax and the rationality of the process logic. The generated XML file supports direct import into the process center, and the platform automatically completes the process deployment, so that the user does not need to manually configure, realizes the process online quickly.

[0033] Specifically, the user adds process nodes by dragging and dropping, and establishes the connection relationship between nodes by connecting operation. Real-time capture these front-end interface operations, parse the semantic information behind the operation, and generate structured node configuration information.

[0034]

[0034] Identify the type of the dragged node, and then record the position coordinates of the node on the canvas. For each node, receive the parameters configured by the user through the attribute panel, such as specifying the handler for the "approval node" and setting the condition expression for the "condition branch node", encapsulate the type, coordinates, parameters of all nodes and the connection relationship between nodes into complete node configuration information.

[0035] Further, the process packaging engine receives the node configuration information and maps it to an internal process definition object that can be processed. Specifically, the node type in the node configuration information is identified, and the corresponding process node object is instantiated, and the specific parameters in the node configuration information, such as the handler and the condition expression, are configured into the attributes of the corresponding process node object. The process packaging engine establishes a clear flow path between the process node objects according to the connection relationship recorded in the node configuration information, and assembles and integrates all the configured process node objects and their flow paths to generate a unified and complete process definition object.

[0036] The process packaging engine deploys the built process definition object. Specifically, the process definition object is serialized into a standard process definition file conforming to an industry standard, and the standard process definition file is registered into a process library of the system, completing deployment of the process template.

[0037] When a business event triggers process initiation, the process packaging engine creates and initializes a new process instance according to the deployed standard process definition file. The new process instance proceeds according to the preset node path and rules in the process definition file, automatically completes task allocation, state jumping and other operations, until the process instance ends.

[0038] The intelligent agent design module is configured to receive a module design request from the form intelligent design module or the process center module, extract a request design type and a request design content, match in a pre-registered model library and a preset business library to obtain a target model and adaptive business knowledge, generate recommended content based on the adaptive business knowledge and the request design content through the target model, and return to the request initiation module.

[0039] Specifically, the user can add process nodes and establish connection relationships between the nodes by a drag-and-drop manner. Real-time capture of these visual operations parses the types of the nodes (such as user tasks, gateways, service tasks), coordinate positions on the canvas and flow paths between the nodes. Based on this information, the system creates corresponding configuration objects for each node, records detailed parameters including a processing person assignment manner (such as a specified member or a specified post), a judgment expression of a condition branch, a button permission of a task operation and the like, and finally combines to generate complete node configuration information.

[0040] In a specific implementation, the intelligent agent design module includes a model management module, a knowledge base module and a prompt word module.

[0041] The model management module supports registration of multiple types of AI models, including large language models, component recommendation models, process logic verification models and the like, and each model needs to upload basic information (such as a model name, a version number, a development framework, applicable scenarios and the like), input and output types. The models are classified and archived according to application scenarios, for example, “form demand analysis class”, “process XML generation class” and “knowledge retrieval matching class”, and a label is set for each type of model, so as to facilitate automatic selection of adaptive models according to business scenarios.

[0042] The knowledge base module includes a business knowledge base, a form knowledge base, an Excel data knowledge base and a knowledge calling interface.

[0043] Business knowledge base, build knowledge base by business type, each business type contains business rules, field specifications, common form component combinations, etc. Support administrators to manually upload documents (such as PDF, Word format), the platform converts the document content into structured knowledge through OCR technology and text extraction algorithm, supports keyword search and semantic search.

[0044] Form knowledge base, automatically collect form data created in the platform (such as form structure, component configuration, attribute setting, etc.), store by business scenario classification. At the same time, support users to mark high-quality forms as "template", other users can directly reuse the template or modify based on the template, reduce repeated design work. Form knowledge base is updated in real time to ensure data timeliness.

[0045] Excel data knowledge base, after generating data, support users to export to Excel format. The platform has built-in Excel data processing engine, which cleans the exported Excel data (such as removing null values, correcting format errors), classifies (by field type, data source), extracts data features (such as common field combinations, data value range), and converts to knowledge entries stored in the knowledge base.

[0046] Knowledge calling interface, provide knowledge calling interface for form intelligent design module and process design intelligent agent, support real-time access to knowledge base data, improve form design accuracy.

[0047] Prompt word module supports designing prompt word templates by business scenario, covering form design and process design two core scenarios. The template supports custom editing, users can add industry-specific fields or adjust the logic description format, and the edited template can be saved as "personal template" or "team shared template".

[0048] Specifically, the intelligent agent design module continuously listens to and receives module design requests sent from the form intelligent design module or the process center module. After receiving the request, the intelligent agent design module parses the request message and extracts the key parameters. The key parameters mainly include the request design type and the request design content. Among them, the request design type is used to identify the nature of the current design task, such as "form field generation" or "process node design"; the request design content contains the specific description and constraints of the design task.

[0049] Further, based on the extracted request design type and request design content, the agent design module simultaneously performs matching query in the pre-registered model library and the pre-set business library. In the pre-registered model library, AI models suitable for the application scenario are selected according to the request design type, and the most suitable target model is determined by comparing the input-output characteristics of the model with the degree of fit of the current task requirements. In the pre-set business library, the request design content is matched with the knowledge items such as business rules and historical cases in the library through semantic retrieval technology, and the most relevant adaptive business knowledge for the current design task is obtained.

[0050] Further, the agent design module integrates the obtained adaptive business knowledge with the request design content to construct a complete model input prompt, and submits the target model input prompt to the target model for inference calculation. Specifically, the target model first performs vectorization processing on the input prompt to convert it into an input prompt vector, and then performs forward propagation calculation through its internal neural network to finally generate an original prompt vector at the output layer. The original prompt vector is analyzed and processed to generate an initial text sequence, structured configuration elements in the initial text sequence are extracted, and the structured configuration elements are assembled based on a pre-set structured format to generate recommended content which is converted into structured recommended content.

[0051] After generating the recommended content, the agent design module encapsulates it according to a unified response format and returns the result to the source module that initiated the request. For the form intelligent design module, the recommended content returned may be field list and component configuration suggestions; for the process center module, the recommended content returned may be node sequence and flow transfer rules. After receiving the recommended content, the request module can apply it directly or after user confirmation to the specific design.

[0052] The data security module receives operation requests and data to be stored from the form intelligent design module, the process center module or the agent design module, performs permission verification on the operation requests, judges whether to execute the operation requests according to the verification results, and stores the sensitive fields in the data to be stored after encryption conversion, records storage logs, and generates system audit logs based on the operation requests, verification results and storage logs.

[0053] In specific implementation, the data security module adopts a role-based access control model (RBAC model) to assign different roles, organizations and positions to users and set different operation permissions, such as form design permission, process deployment permission, knowledge base modification permission, etc., to ensure data access security. The data security module supports storing sensitive data of a form using an AES-256 encryption algorithm, and encrypting data in a transmission process using an HTTPS protocol to prevent data leakage. The data security module supports recording all user operations, such as form creation, process modification, knowledge base access, etc., including operator, operation time, operation content, etc., and supports log query and audit to facilitate tracing of data operation behavior.

[0054] Specifically, the data security module continuously receives operation requests and to-be-stored data sent from the form intelligent design module, the process center module or the agent design module. The operation request includes a requested operation type, an operation object and user identity information; the to-be-stored data includes business data that needs to be persisted, which can include sensitive fields.

[0055] The received operation request is subjected to permission verification. First, the user identity information in the operation request is parsed, including a user ID, organization and position information. Then, three-dimensional permission verification is performed based on the user's role, organization and position information, and the permission policy library is queried to verify whether the current user has the permission to execute the operation request. According to the verification result, it is determined whether to execute the operation request: when the verification is passed, the subsequent operation is permitted to be executed; when the verification is not passed, the operation request is rejected and permission error information is returned to the requester.

[0056] For requests that are permitted to be executed and include to-be-stored data, the sensitive fields in the to-be-stored data are subjected to encryption conversion. Specifically, the sensitive fields in the to-be-stored data are identified, and these fields are subjected to encryption conversion using an AES-256 encryption algorithm to generate encrypted data. After encryption is completed, the encrypted data is stored in the system database, and a data storage log containing encryption operation details is recorded.

[0057] The data security module integrates all information generated in the entire security processing process, including original operation requests, permission verification results, data storage logs. Based on this information, complete system audit logs are generated in a unified audit format. The system audit log records the full life cycle information of the operation, including operator, operation time, operation content, permission verification result, data operation result, etc. key information, providing complete data support for system security audit.

[0058] The system integration module receives a data exchange request from the form intelligent design module or the process center module; according to the interaction type of the data exchange request, a corresponding preconfigured interface is called to establish a connection with an external business system, and conversion between a platform data format and a data format of the external business system is performed to complete data synchronization between the platform and the external business system.

[0059] In a specific implementation, the system integration module provides a standardized API interface, supports connection with existing enterprise ERP, CRM, OA and other systems, and realizes data intercommunication.

[0060] Specifically, the system integration module continuously listens to and receives a data exchange request from the form intelligent design module or the process center module. The data exchange request contains specific requirements for data interaction with an external business system, including but not limited to interaction type, data content, target system identifier and other information. The received request is parsed, and key parameters are extracted to provide a basis for subsequent connection establishment and data conversion.

[0061] According to the interaction type and target system identifier in the data exchange request, a corresponding preconfigured interface is called to establish a communication connection with the external business system. Specifically, first, the connection parameters of the target system are queried from the interface configuration library, including access address, authentication information, communication protocol and the like; then, based on these parameters, a connection session is initialized, and a secure handshake with the external business system is completed; finally, the connection state is verified to ensure the availability of the communication channel.

[0062] After a stable communication connection is established, bidirectional conversion between a platform data format and a data format of the external business system is performed. First, the data structure from the platform is parsed, and mapping and recombination are performed according to the data specification of the target system; at the same time, for the data received from the external system, parsing and reconstruction are performed according to the data standard inside the platform.

[0063] Based on the established connection and the converted data format, data synchronization between the platform and the external business system is performed. According to the synchronization direction and synchronization strategy required by the request, data is accurately transmitted to the target system, and feedback data from the external system is received. The system integration module monitors the execution state of the entire synchronization process, records the synchronization log, and returns the execution result to the requester after the synchronization is completed.

[0064] The above only describes the embodiments of the present application and does not limit the present application. For those skilled in the art, the present application can have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the scope of claims of the present application.

Claims

1. A form and process generation system based on intelligent agents, characterized in that, include: Form intelligent design module, process center module, intelligent agent design module; The intelligent form design module is used to parse the form design requirements input by the user, extract the form requirement information, and generate a form layout based on the form requirement information through the component layout engine. The process center module is used to generate node configuration information based on the user's visual operation on process nodes in the front-end interface, and map the node configuration information to process definition objects. The process definition object is deployed through the process encapsulation engine to generate a standard process definition file, which drives the execution of process instances. The intelligent agent design module is used to receive module design requests from the form intelligent design module or the process center module, extract the request design type and request design content, match them in the pre-registered model library and the preset business library to obtain the target model and the adapted business knowledge; and generate recommended content based on the adapted business knowledge and the request design content using the target model, and return it to the request initiation module.

2. The form and process generation system based on intelligent agents according to claim 1, characterized in that, The form requirements information includes field names, field types, and validation rules; The step of generating a form layout based on the form requirement information using a component layout engine specifically includes: Based on the field type, a match is made in a predefined component library to obtain the corresponding form component; According to the validation rules, the form components are configured with corresponding component attributes, and the configured form components are combined based on the preset layout rules to generate a form layout.

3. The form and process generation system based on intelligent agents according to claim 1, characterized in that, The process of generating node configuration information based on the user's visual operations on process nodes in the front-end interface specifically includes: Capture the user's drag-and-drop operations on process nodes and the connection operations between nodes on the front-end interface; The drag-and-drop operation and the node connection operation are parsed to extract the corresponding node operation information; the node operation information includes node type, node handler, coordinate position and connection relationship. Combine the node operation information corresponding to all process nodes to obtain node configuration information.

4. The form and process generation system based on intelligent agents according to claim 1, characterized in that, The step of mapping the node configuration information to a process definition object specifically includes: Identify the node type in the node configuration information and create a process node object corresponding to the node type; Configure the node configuration information into the corresponding attributes of the process node object, and establish the flow path between the process node objects according to the connection relationship between the process nodes; Based on the flow path, all configured process node objects are assembled to generate a process definition object.

5. The form and process generation system based on intelligent agents according to claim 4, characterized in that, The process definition object is deployed through the process encapsulation engine to generate a standard process definition file and drive the execution of the process instance, specifically including: The process definition object is received by the process encapsulation engine, and the process definition object is serialized to obtain a standard process definition file; Register the standard process definition file into the workflow engine to complete the process deployment; In response to a business trigger command, a new process instance is created based on the deployed standard process definition file, and the workflow engine drives the new process instance to execute the preset node path in the standard process definition file.

6. The form and process generation system based on intelligent agents according to claim 1, characterized in that, The step of generating recommended content based on the target model, the adapted business knowledge, and the request design content specifically includes: Based on the adapted business knowledge, a design context prompt is generated, and the design context prompt and the requested design content are concatenated to obtain the model input prompt; The model input prompt is input into the target model, and the model input prompt is vectorized using the target model to obtain the input prompt vector; By performing forward propagation calculations based on the input cue vector through the neural network of the target model, the original cue vector output by the output layer of the neural network is obtained; Based on the original suggestion vector, recommended content is generated.

7. The form and process generation system based on intelligent agents according to claim 6, characterized in that, The step of generating recommended content based on the original suggestion vector specifically includes: The original prompt vector is decoded to generate an initial text sequence, and the structured configuration elements in the initial text sequence are extracted. Based on a preset structured format, the structured configuration elements are assembled to generate recommended content.

8. The form and process generation system based on intelligent agents according to claim 1, characterized in that, It also includes a data security module, which is used to receive operation requests and data to be stored from the form intelligent design module, the process center module, or the intelligent agent design module; to perform permission verification on the operation request; and to determine whether to execute the operation request based on the verification result. Sensitive fields in the data to be stored are encrypted and converted before storage, and a storage log is recorded. Based on the operation request, the verification result, and the storage log, a system audit log is generated.

9. A form and process generation system based on intelligent agents according to claim 8, characterized in that, The step of verifying the permission of the operation request and determining whether to execute the operation request based on the verification result specifically includes: Obtain the user identity information and request operation type from the operation request, and match the corresponding system role and permission set based on the user identity information; Verify whether the request operation type is included in the permission set; If the verification passes, the operation request is permitted to be executed; If the verification fails, the operation request will be rejected and a permission error message will be returned.

10. The form and process generation system based on intelligent agents according to claim 1, characterized in that, It also includes a system integration module, which is used to receive data exchange requests from the form intelligent design module or the process center module; according to the interaction type of the data exchange request, it calls the corresponding pre-configured interface to establish a connection with the external business system, and performs the conversion between the platform data format and the data format of the external business system to complete the data synchronization with the external business system.

Images