CN-121659911-B - Cross-system form intent generation and collaborative interaction method based on intelligent agent

CN121659911BCN 121659911 BCN121659911 BCN 121659911BCN-121659911-B

Abstract

The invention provides a cross-system form intent generation and collaborative interaction method based on an agent, which relates to the technical field of man-machine interaction and comprises the steps of generating standard intent characterization and checking through semantic analysis, determining a target form and a system entity, and extracting pre-filling data; the method comprises the steps of obtaining interface meta-information, constructing structured field knowledge, generating form configuration, rendering a form interface, injecting pre-filling data, constructing a trust chain, verifying identity, determining an authorization range, converting the form data and calling a target interface. The invention can effectively improve the processing efficiency of the cross-system service form and enhance the data security and the user experience.

Inventors

ZHAO LEIZHEN
SUN SHUMENG
Jia Songrui

Assignees

北京亦庄智能城市研究院集团有限公司

Dates

Publication Date: 20260508
Application Date: 20260209

Claims (9)

1. The intelligent agent-based cross-system form intent generation and collaborative interaction method is characterized by comprising the following steps of: Mapping the user natural language expression to a predefined intention semantic knowledge base through semantic analysis, generating a standard intention representation, executing topology path verification in a business entity relation map based on the standard intention representation, determining a target form entity and a target system entity by combining a form transfer mode, and extracting history filling data as a pre-filling data source; obtaining interface meta information of the target system entity, constructing structured field knowledge, generating an agent by representing the input form by the structured field knowledge and the standard intention, establishing field-control mapping through knowledge reasoning, deriving a cascade check rule, and generating a form configuration, wherein the method comprises the following steps: The intelligent agent extracts field verification rules from the structured field knowledge, analyzes the field verification rules to construct an abstract syntax tree, calculates the tree depth and the branch number of the abstract syntax tree, inquires a depth threshold value and a branch threshold value corresponding to each control type in a control feature library, screens target control types with the depth threshold value being larger than the tree depth and the branch threshold value being larger than the branch number, extracts service type identifiers of corresponding fields from standard intention characterization, prioritizes the target control types, and selects the target control type with the highest priority and the fields to establish field-control mapping; The intelligent agent extracts field names from leaf nodes of an abstract syntax tree, a field dependency graph is constructed, the number of edges of each field node in the field dependency graph is calculated, the field nodes with zero number of edges are put into a processing queue and marked as a first level, the field nodes are taken out from the processing queue and corresponding edges are discharged, the number of edges of subsequent field nodes is updated, the subsequent field nodes with zero number of edges are put into the processing queue and marked as a next level, the front order control and the subsequent control of adjacent levels are extracted from a field-control mapping, a propagation path set is constructed, a public front order control is identified, a batch trigger list is configured, a check execution priority is set according to the level to which the subsequent control belongs, and a cascade check rule is generated; encapsulating the field-control mapping and the cascading verification rule to generate a form configuration; Based on the form configuration rendering form interface, injecting the pre-filled data source into a form domain through semantic alignment, and performing real-time verification to obtain complete form data; And carrying out mode conversion on the complete form data based on the structured field knowledge, and carrying the security token to call a target system interface.
2. The method of claim 1, wherein mapping the user natural language expression to the predefined intent semantic knowledge base by semantic parsing, generating the standard intent representation comprises: performing semantic coding on the natural language expression of the user to obtain query semantic expression, performing semantic coding on each intention template in a predefined intention semantic knowledge base to obtain a template semantic expression set, calculating the semantic distance between the query semantic expression and each template semantic expression, and screening out primary intention templates with the semantic distance smaller than a preset distance threshold; Analyzing the dependency relationship of the natural language expression of the user, extracting a core entity and a corresponding dependency relationship chain, acquiring the corresponding entity type and the relationship constraint among the entities, which are marked in advance, of each primary intent template, screening the entity type of each primary intent template corresponding to the type of the core entity, and determining a matching intent template by the dependency relationship chain meeting the relationship constraint among the entities of each primary intent template; extracting modified semantic components except the core entity and the corresponding dependency chain from the natural language expression of the user, acquiring a parameter slot structure defined in the matching intention template, wherein the parameter slot structure comprises semantic type labels of all parameter slots, matching the modified semantic components with the semantic type labels of all parameter slots according to semantic types, filling the modified semantic components into corresponding parameter slots, and generating the standard intention representation.
3. The method of claim 1, wherein performing a topology path check in a traffic entity relationship graph based on the standard intent representation, determining a target form entity and a target system entity in conjunction with a form transfer pattern, extracting historical filling data as a pre-filled data source comprises: Extracting a service entity identifier from the standard intention representation, positioning a corresponding node in a service entity relationship map as a starting point, and traversing along the associated edge to obtain a form node; Reading a dependence path structure defined by each form node, performing topology comparison on an actual path from a starting point to each form node in a business entity relation map and the dependence path structure, reserving form nodes with node sequences and edge sequences of the actual path consistent with the dependence path structure, and determining candidate form nodes; extracting a history form operation record, identifying form pairs with continuous operation, counting the transfer times of each form pair, constructing a form transfer network, searching a transfer path and the transfer times pointing to each candidate form node in the form transfer network, selecting the candidate form node corresponding to the transfer path with the largest transfer times, and determining a target form entity; extracting a system node connected with the target form entity as a target system entity; and searching filling data of the transfer path with the largest transfer times when the transfer path is operated last time from the history form operation record, and extracting the filling data as a pre-filling data source.
4. The method of claim 1, wherein obtaining interface meta-information of the target system entity, constructing structured field knowledge comprises: Detecting whether the target system entity supports a unified interface abstract protocol, and when the detection result is that the target system entity supports the unified interface abstract protocol, sending a meta-information query request corresponding to the protocol to the target system entity, receiving interface description data returned by the target system entity, and extracting field identification, field type and field constraint rules from the interface description data to obtain interface meta-information; When the detection result is that the interface specification document corresponding to the target system entity is not supported, text segmentation is carried out on the interface specification document, paragraphs describing interface parameters in each text segment are identified, text descriptions of parameter names, parameter data types and parameter value constraints are extracted from the paragraphs, the parameter names are mapped into field identifications, the parameter data types are mapped into field types, the text descriptions of the parameter value constraints are converted into field constraint rules, and interface meta-information is determined; And carrying out structural organization on each field identifier, each field type and each field constraint rule in the interface meta-information, establishing a mapping relation from the field identifier to the field type and each field constraint rule, and generating structural field knowledge.
5. The method of claim 1, wherein extracting the precursor controls and the subsequent controls of adjacent levels from the field-control map, constructing a propagation path set, identifying common precursor controls and configuring a batch trigger list, setting a check execution priority according to the level to which the subsequent controls belong, and generating the cascade check rule comprises: Constructing a propagation path set based on the layers of all field nodes in the field dependency graph, traversing from the field node of the first layer to the field node of the next layer along the outgoing edge, and recording the traversed field node sequence as a propagation path; Identifying a plurality of propagation paths starting from field nodes of the same preamble level, determining a control corresponding to the field nodes of the same preamble level as a public preamble control, extracting follow-up controls corresponding to termination field nodes of the plurality of propagation paths, configuring a batch trigger list for a value change event of the public preamble control, wherein the batch trigger list comprises check trigger interfaces of all follow-up controls, and generating batch cascading rules for triggering check of a plurality of follow-up controls by a single value change event; extracting the hierarchy to which each subsequent control corresponding field node belongs in the batch trigger list, and setting a verification execution priority for each subsequent control according to the hierarchy, wherein the hierarchy to which the subsequent control belongs is in direct proportion to the verification execution priority; and packaging the batch cascading rules and the verification execution priority, and generating a final cascading verification rule.
6. The method of claim 1, wherein constructing a trust chain attestation and verifying identity based on a business entity relationship graph, determining an operational authorization scope through a rights inference model, generating a security token comprises: Extracting an authentication dependence path of the target system entity from the service entity relationship graph, collecting trust certificates of all associated system entities along the authentication dependence path, combining the trust certificates with user certificate information to construct a trust chain evidence, initiating cross-domain identity verification to the target system entity, receiving authentication response and analyzing to obtain a user identity attribute set; Inputting the standard intention characterization and field semantic features in the complete form data into a permission reasoning model, wherein the permission reasoning model is obtained based on operation history track learning in the business entity relation graph, reasoning and outputting a minimum permission set matched with the user identity attribute set, and determining an operation authorization range; and generating a data abstract for the complete form data, packaging the data abstract, the hash value of the trust chain evidence and the operation authorization range into a token payload, and signing the token payload to generate a security token.
7. An agent-based cross-system form intent generation and collaboration interaction system for implementing the method of any of the preceding claims 1-6, comprising: The semantic understanding module is used for mapping the user natural language expression to a predefined intention semantic knowledge base through semantic analysis, generating a standard intention representation, executing topology path verification in a business entity relation map based on the standard intention representation, determining a target form entity and a target system entity by combining a form transfer mode, and extracting history filling data as a pre-filling data source; The form construction module is used for acquiring interface meta information of the target system entity, constructing structured field knowledge, generating an intelligent agent by the structured field knowledge and the standard intention representation input form, establishing field-control mapping through knowledge reasoning, deriving a cascading check rule, and generating form configuration; The interface rendering module is used for rendering a form interface based on the form configuration, injecting the pre-filled data source into a form field through semantic alignment, and executing real-time verification to obtain complete form data; The data interaction module is used for constructing trust chain evidence and verifying identity based on the relation graph of the service entity, determining an operation authorization range through the authority reasoning model, generating a security token, and carrying out mode conversion on the complete form data based on the structured field knowledge, carrying the security token and calling a target system interface.
8. An electronic device, comprising: A processor; A memory for storing processor-executable instructions; Wherein the processor is configured to invoke the instructions stored in the memory to perform the method of any of claims 1 to 6.
9. A computer readable storage medium having stored thereon computer program instructions, which when executed by a processor, implement the method of any of claims 1 to 6.

Description

Cross-system form intent generation and collaborative interaction method based on intelligent agent Technical Field The invention relates to the technical field of man-machine interaction, in particular to a cross-system form intent generation and collaborative interaction method based on an intelligent agent. Background With the deep development of enterprise informatization, the demands of form processing and business collaboration among systems are increasing. Conventional form handling systems typically employ fixed templates and hard-coded rules that are difficult to accommodate for complex and diverse business scenarios. A plurality of business systems often exist in an enterprise, each system is provided with independent form processing logic and a data model, so that similar information needs to be repeatedly input when a user switches between different systems, and the working efficiency is affected. In recent years, development of artificial intelligence technologies such as natural language processing and knowledge graph has brought new possibilities for form processing, so that automatic generation and processing of forms based on user intention are possible. The prior art still has defects and shortcomings in the aspects of cross-system form generation and interaction, the prior form processing system lacks deep understanding capability for natural language expression of users, usually only can identify keywords or preset instructions, cannot accurately capture complex business intentions of the users, causes deviation between the generated form and actual demands of the users, and reduces the working efficiency because the users need to carry out a large amount of modification and adjustment, and the traditional form generation method is usually based on a hard-coded template, lacks support of semantic knowledge and is difficult to dynamically adjust form structures and field configurations according to business contexts. The method has the advantages that the form data among different systems are difficult to communicate, users need to fill in the same information repeatedly during cross-system operation, the operation burden and the error probability are increased, the problem that a security authentication mechanism is imperfect in cross-system form interaction in the prior art is solved, a static authority control strategy is generally adopted, dynamic authorization cannot be carried out according to service scenes and user intention, and the problem that the authority is too large or too small is easily caused. Meanwhile, a trust chain construction mechanism based on a business knowledge graph is lacking, and flexible cross-system data exchange is difficult to realize on the premise of ensuring safety. Disclosure of Invention The embodiment of the invention provides a cross-system form intent generation and collaborative interaction method based on an intelligent agent, which can solve the problems in the prior art. In a first aspect of the embodiment of the present invention, a method for intentionally generating and cooperatively interacting a cross-system form based on an agent is provided, including: in a second aspect of the embodiment of the present invention, there is provided an agent-based cross-system form intentional generation and collaboration interaction system, including: The semantic understanding module is used for mapping the user natural language expression to a predefined intention semantic knowledge base through semantic analysis, generating a standard intention representation, executing topology path verification in a business entity relation map based on the standard intention representation, determining a target form entity and a target system entity by combining a form transfer mode, and extracting history filling data as a pre-filling data source; The form construction module is used for acquiring interface meta information of the target system entity, constructing structured field knowledge, generating an intelligent agent by the structured field knowledge and the standard intention representation input form, establishing field-control mapping through knowledge reasoning, deriving a cascading check rule, and generating form configuration; The interface rendering module is used for rendering a form interface based on the form configuration, injecting the pre-filled data source into a form field through semantic alignment, and executing real-time verification to obtain complete form data; The data interaction module is used for constructing trust chain evidence and verifying identity based on the relation graph of the service entity, determining an operation authorization range through the authority reasoning model, generating a security token, and carrying out mode conversion on the complete form data based on the structured field knowledge, carrying the security token and calling a target system interface. In a third aspect of an embodiment of the present invent