An ontology-driven system and method for bi-directional enhancement of government administration and generation

Abstract

This application provides an ontology-driven system and method for bidirectional enhancement of government governance and generation. The system includes: a government domain ontology module for storing a government semantic knowledge graph composed of multiple ontology nodes and semantic relationship edges between ontology nodes; a semantic mapping module for semantically aligning the fields to be accessed with ontology nodes and outputting confirmed mapping relationships; a rule governance module for reading business rule constraints and compiling them into verification rules, performing data verification according to the verification rules, and treating undefined business rule constraints identified during the verification process as rule missing information; a generation service module for retrieving semantic subgraphs according to government needs, inputting the semantic subgraphs as constraint information into the generation model, generating response content, and outputting an ontology update trigger signal when the semantic features in the government needs exceed the coverage range; and an ontology evolution module for generating ontology update instructions based on received information and triggering dynamic updates.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to an ontology-driven system and method that enhances both government governance and data generation. Background Technology

[0002] Government data is scattered across the business systems of dozens of departments, including civil affairs, taxation, social security, housing and construction, and market supervision. Existing technologies have long faced the following technical challenges in engineering practice: Firstly, semantic heterogeneity: the same business concept has different field names, encoding standards, and value ranges in different systems. For example, there are different meanings for the same word as "legal representative" and "statutory representative", or for the same word as "registered residence" and "permanent address", which makes it impossible to directly compare and share data across departments.

[0003] Secondly, governance lags behind: Existing data governance relies heavily on traditional Extract-Transform-Load (ETL) and Master Data Management (MDM) methods, focusing on field-level cleaning and lacking consistency verification at the business semantic level. Cross-department sharing still requires a lot of manual reconciliation, resulting in high governance costs and lag behind business changes.

[0004] Third, the intelligence is limited: directly applying large language models to government Q&A or document generation generally has risks such as conclusions deviating from legal basis, outdated policies cited, non-compliant expressions, and insufficient accountability, making it difficult to implement in government scenarios.

[0005] Fourth, the separation between governance and generation: In existing technologies, data governance systems and intelligent question answering / generation systems are usually built independently, each maintaining a set of knowledge representations (the former is mostly a field-level mapping table, and the latter is mostly a document index or vector library). This results in the semantic knowledge accumulated during the governance process not being able to effectively serve the generation scenario, and the rule defects discovered in the generation scenario cannot be fed back to optimize the governance, creating a situation where two separate entities exist.

[0006] Existing technologies have not yet provided a method that takes the ontology of the government domain as its core and connects the entire chain of data governance, semantic retrieval, and generative services. Summary of the Invention

[0007] The purpose of this invention is to provide an ontology-driven system and method that enhances both government governance and data generation, thereby addressing the aforementioned problems in the prior art.

[0008] This invention provides an ontology-driven system that enhances both government governance and generative capabilities, comprising: The government affairs domain ontology module is used to store the government affairs semantic knowledge graph; wherein, the government affairs semantic knowledge graph includes multiple sub-ontologies, which are interconnected through preset semantic relationships, and the government affairs semantic knowledge graph is composed of multiple ontology nodes and semantic relationship edges between ontology nodes; The semantic mapping module, connected to the government domain ontology module, is used to acquire heterogeneous data sources, semantically align the fields to be accessed in each data source with the ontology node, generate a mapping relationship from field to ontology node, and output the confirmed mapping relationship to the government domain ontology module. The rule governance module, connected to the government domain ontology module and the semantic mapping module, is used to read business rule constraints from the government domain ontology module, compile the business rule constraints into verification rules, perform data verification on the aligned data output by the semantic mapping module according to the verification rules, output the verification results, and treat business rule constraints that are not defined in the government domain ontology module and are identified during the verification process as rule missing information. The generation service module is connected to the government affairs domain ontology module and the rule governance module. It is used to retrieve semantic subgraphs from the government affairs domain ontology module according to the government affairs requirements input by the user, input the semantic subgraphs as constraint information into the generation model, output the response content generated by the generation model based on the constraint information, and output an ontology update trigger signal when the semantic features in the government affairs requirements exceed the coverage of the government affairs domain ontology module. The ontology evolution module, connected to the government domain ontology module, semantic mapping module, rule governance module, and generation service module, is used to receive at least one of the confirmed mapping relationship, the rule missing information, and the ontology update trigger signal, and generate an ontology update instruction based on the received information and output it to the government domain ontology module to trigger dynamic updates.

[0009] This invention provides an ontology-driven method for bidirectional enhancement of government governance and generation, comprising: The government affairs semantic knowledge graph is stored through a government affairs domain ontology module; wherein, the government affairs semantic knowledge graph includes multiple sub-ontologies, which are interconnected through preset semantic relationships, and the government affairs semantic knowledge graph is composed of multiple ontology nodes and semantic relationship edges between ontology nodes; Heterogeneous data sources are obtained through the semantic mapping module, and the fields to be accessed in each data source are semantically aligned with the ontology node to generate the mapping relationship from field to ontology node. The confirmed mapping relationship is then output to the government domain ontology module. The rule governance module reads business rule constraints from the government domain ontology module, compiles the business rule constraints into verification rules, performs data verification on the aligned data output by the semantic mapping module according to the verification rules, outputs the verification results, and treats business rule constraints that are not defined in the government domain ontology module as rule missing information. The generation service module retrieves semantic subgraphs from the government domain ontology module based on the government needs input by the user, inputs the semantic subgraphs as constraint information into the generation model, outputs the response content generated by the generation model based on the constraint information, and outputs an ontology update trigger signal when the semantic features in the government needs exceed the coverage of the government domain ontology module. The ontology evolution module receives at least one of the confirmed mapping relationship, the rule missing information, and the ontology update trigger signal, and generates an ontology update instruction based on the received information and outputs it to the government affairs ontology module to trigger dynamic updates.

[0010] This invention also provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements the steps of the above-described ontology-driven method for bidirectional enhancement of government governance and generation.

[0011] This invention also provides a computer-readable storage medium storing an information transmission implementation program, which, when executed by a processor, implements the steps of the above-described ontology-driven method for bidirectional enhancement of government governance and generation.

[0012] The embodiments of this invention can include the following beneficial effects: These embodiments use a government ontology as a unified semantic foundation, achieving semantic layer unification of cross-departmental data through ontology node association, thus resolving the problem of semantic conflicts at the field level in heterogeneous data sources; they automatically compile business rule ontology into executable verification rules, transforming manual reconciliation into ontology reasoning tasks, reducing governance costs; the generated model outputs content constrained by ontology subgraphs and associated with ontology nodes, ensuring the compliance, interpretability, and accountability of the generated results; governance and generation share the same ontology instance, and the mapping relationships and verification rules accumulated in governance can be directly reused in generation scenarios, while the knowledge gaps identified in generation flow back to optimize the ontology, forming a continuously evolving closed loop of bidirectional reinforcement. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in one or more embodiments of this specification or in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 This is a schematic diagram of an ontology-driven system that enhances both government governance and data generation according to an embodiment of the present invention. Figure 2 This is a flowchart illustrating the overall process of an embodiment of the present invention. Figure 3 This is a schematic diagram of the system functional architecture according to an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the automatic compilation from the business rule sub-body to the executable verification rule in an embodiment of the present invention; Figure 5 This is a flowchart of the ontology-driven method for bidirectional enhancement of government governance and generation according to an embodiment of the present invention. Detailed Implementation

[0015] To enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the technical solutions in one or more embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this specification, and not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of this document.

[0016] System Implementation Examples According to embodiments of the present invention, an ontology-driven system that enhances both government governance and generative capabilities is provided. Figure 1 This is a schematic diagram of an ontology-driven system for bidirectional enhancement of government governance and generation, as described in an embodiment of the present invention. Figure 1 As shown, the ontology-driven system for bidirectional enhancement of government governance and generation according to an embodiment of the present invention specifically includes: The government affairs domain ontology module 10 is used to store a government affairs semantic knowledge graph; wherein, the government affairs semantic knowledge graph includes multiple sub-ontologies, which are interconnected through preset semantic relationships, and the government affairs semantic knowledge graph is composed of multiple ontology nodes and semantic relationship edges between ontology nodes; the government affairs semantic knowledge graph is formally represented using the OWL2 specification; The government semantic knowledge graph includes subject sub-ontology, item sub-ontology, process sub-ontology, coding system sub-ontology and business rule sub-ontology; Each business rule in the business rule sub-ontology adopts a triple structure, including constraint type, target concept, and parameter list; the constraint type includes value range constraint, foreign key reference constraint, business dependency constraint, temporal consistency constraint, uniqueness constraint, and format constraint.

[0017] Semantic mapping module 12, connected to the government domain ontology module, is used to acquire heterogeneous data sources, semantically align the fields to be accessed in each data source with the ontology nodes, generate a mapping relationship from fields to ontology nodes, and output the confirmed mapping relationship to the government domain ontology module; specifically used for: The mapping confidence of the field to be accessed to the ontology node is calculated by weighting the similarity of field name, data distribution, sample value pattern, and upstream and downstream business relationship. The mapping confidence level is compared with the confirmation threshold. Mapping relationships that are greater than or equal to the confirmation threshold are automatically confirmed, while mapping relationships that are less than the confirmation threshold are manually confirmed. The confirmed mapping relationship is written back to the government domain ontology module in the form of mapping assertion triples; wherein, the mapping assertion triples include the source field, the ontology node to which it is mapped, the confidence level, and the confirmation time; When a new data source is connected, a match is made between the new field and the previously written-back mapping assertion triple. If a match is found, the corresponding mapping relationship is reused directly.

[0018] The rule governance module 14, connected to the government domain ontology module and the semantic mapping module, is used to read business rule constraints from the government domain ontology module, compile the business rule constraints into verification rules, perform data verification on the aligned data output by the semantic mapping module according to the verification rules, output the verification result, and treat business rule constraints identified during the verification process that are not defined in the government domain ontology module as rule missing information; specifically used for: Read the business rule constraints from the business rule sub-ontology in the government domain ontology module, and automatically compile the business rule constraints into validation rules; Data verification is performed on the aligned data output by the semantic mapping module according to the verification rules to identify abnormal data; Based on the aforementioned government semantic knowledge graph, the detected abnormal data is inferred, and repair suggestions are output to the ontology evolution module. The data that passes the verification is output to the generation service module as verified data, and the business rule constraints that are not defined in the government affairs ontology module and are identified during the verification process are output to the ontology evolution module as rule missing information.

[0019] The generation service module 16, connected to the government affairs domain ontology module and the rule governance module, is used to retrieve semantic subgraphs from the government affairs domain ontology module based on user-input government affairs requirements, input the semantic subgraphs as constraint information into the generation model, output the response content generated by the generation model based on the constraint information, and output an ontology update trigger signal when the semantic features in the government affairs requirements exceed the coverage of the government affairs domain ontology module; specifically used for: Receive the verified data output by the rule governance module and the government affairs requests input by the user; Retrieve the corresponding semantic subgraph from the government affairs semantic knowledge graph based on the stated government affairs requirements; The semantic subgraph is input as constraint information into the generative model, and the verified data is used as the input context of the generative model. The output is the response content generated by the generative model based on the constraint information and the verified data, and each response content is associated with a corresponding ontology node identifier. The system verifies whether the legal provisions cited in the response exist in the business rule sub-entity and are currently valid. If they do not exist or are not currently valid, the system intercepts and triggers the regeneration of the generation model. When the semantic features in the government affairs requirements exceed the coverage of the government affairs semantic knowledge graph, an ontology update trigger signal is output to the ontology evolution module.

[0020] The ontology evolution module 18, connected to the government domain ontology module, semantic mapping module, rule governance module, and generation service module, is used to receive at least one of the confirmed mapping relationship, the rule missing information, and the ontology update trigger signal, and generate an ontology update instruction based on the received information, then output it to the government domain ontology module to trigger a dynamic update; specifically used for: Receive the confirmed mapping relationship output by the semantic mapping module, and store the mapping relationship in the form of mapping assertion triples into the government semantic knowledge graph; Receive repair suggestions and rule missing information output by the rule governance module, and supplement new rules, update existing rules, add new ontology nodes or new semantic relationship edges to the government semantic knowledge graph according to the repair suggestions or rule missing information; Receive the ontology update trigger signal output by the generation service module, and add new ontology nodes or new semantic relationship edges to the government semantic knowledge graph according to the ontology update trigger signal; When the legal clauses in the business rule sub-entity are changed, the corresponding clauses are marked as repealed or added, and the rule governance module is triggered to automatically compile the updated verification rules.

[0021] The following describes in detail the above-mentioned technical solutions of the present invention in conjunction with the specific circumstances of the bidirectional enhanced ontology-driven system for government governance and generation in the embodiments of the present invention.

[0022] This invention proposes an intelligent processing method for government data that uses a government ontology as a unified semantic foundation to drive the automation of government data governance and generates model outputs based on ontology constraints. This method enables government data to be modeled once and reused in multiple scenarios, and addresses four shortcomings in existing technologies: 1. Solve the semantic heterogeneity problem: Use five types of sub-ontologies to cover who, what to do, how to do it, what to base it on, and how the encoding corresponds, to achieve uniformity at the semantic layer rather than the field layer; 2. Address the issue of lagging governance: Automatically compile business rule ontology into executable verification rules (SHACL / SQL), transforming manual reconciliation into ontology reasoning tasks; 3. Solving the problem of limited intelligence: The model is generated by constraining the ontology subgraph, so that each output is forcibly associated with ontology nodes (legal clauses, service items, process links) to form an evidence chain, thereby ensuring compliance, explainability, and accountability; 4. Solve the problem of separation between governance and generation: Use a unified ontology as a shared knowledge base for governance and generation, so that the mapping relationships, verification rules and ontology nodes accumulated in the governance can be directly reused in the generation scenario, and the rule defects and new requirements identified in the generation process can be fed back to optimize the ontology, forming a closed loop of bidirectional reinforcement.

[0023] This invention uses a government affairs ontology as a unified semantic foundation, connecting the entire chain of data governance, semantic retrieval, and generative services. Specifically, it first constructs a government affairs domain ontology covering five sub-ontologies: subjects, matters, processes, coding systems, and business rules. Then, it performs semantic mapping on multi-source heterogeneous data based on this ontology. Next, it automatically compiles the business rules in the ontology into executable verification rules (SHACL / SQL) to drive data quality governance. Simultaneously, it injects structured hard constraints into a large language model using ontology subgraphs as structural constraints, enabling traceable and compliant government affairs Q&A and document assistance generation. Finally, new knowledge generated during the governance and generation process is fed back to update the ontology, achieving continuous evolution.

[0024] This invention is implemented through five steps: government ontology construction → multi-source data semantic mapping → ontology rule-driven data governance → OAG government Q&A and document-assisted generation → permission and privacy control. The overall process is as follows: Figure 2 As shown. The system's end-to-end architecture, encompassing governance, retrieval, and generation, is as follows. Figure 3As shown, the automatic compilation of business rules into SHACL / SQL is illustrated below. Figure 4 As shown. The specific content is as follows: Step 1: Construction of Government Ontology Construct a government affairs domain ontology comprising five sub-ontologies as a unified semantic foundation: (1) Subject Sub-entities: Natural persons, legal persons, organizations; (2) Sub-items: Classification system of items such as administrative licensing, administrative confirmation, and public services; (3) Process sub-entities: processing steps, time limits, and material list; (4) Encoding system sub-ontologies: national standards, industry standards, local standards and their mapping relationships; (5) Business rules sub-entity: legal provisions, policy documents, approval rules, and rules are expressed in structured constraints (constraint type, target concept, parameters).

[0025] Among them, the main body sub-entity and the item sub-entity are linked through the processing / application relationship, expressing who can process what items; The item sub-entity and the process sub-entity are related through an inclusion relationship, expressing which steps are required to complete a certain item; The process sub-entity and the business rule sub-entity are linked by constraints, expressing which legal provisions a certain step is subject to; The encoding system sub-ontology provides standardized code mapping for the above-mentioned types of ontology, ensuring that cross-system data can be aligned at the encoding level.

[0026] Each business rule is expressed using a triple structure: constraint type, target concept, and parameter list. Where: Constraint types include: value range constraints, foreign key reference constraints, business dependency constraints, temporal consistency constraints, uniqueness constraints, and format constraints; Target concepts include: specific concepts within the subject / item / process sub-ontology; The parameter list includes: specific parameters of the constraint, such as maximum / minimum length, regular expression, reference target, dependency conditions, etc.

[0027] Furthermore, the ontology is formally represented using the OWL 2 specification to support consistency checks, classification reasoning, and relation inference by inference engines (such as HermiT and Pellet). The ontology serialization format uses RDF / XML or Turtle, facilitating integration with the SHACL validation engine and the SPARQL query engine.

[0028] Step 2: Semantic Mapping of Multi-Source Data For heterogeneous data sources, field-concept alignment relationships are automatically established based on ontology concepts: combining field names, data distribution characteristics, sample value patterns, and upstream and downstream business relationships, candidate mappings are generated and confidence levels are provided. Fields Mapping to ontology concepts The confidence level is: ; The four items are: field name similarity, data distribution similarity, sample value pattern similarity, and upstream and downstream business relationship similarity. (For example, configured as 0.3, 0.2, 0.3, 0.2). For mappings below the confirmation threshold (set to 0.8), manual confirmation is introduced. The confirmation result is fed back into the ontology as mapping knowledge accumulation, so that similar fields can be automatically aligned with high confidence in the future.

[0029] Manually verified mapping relationships are not used only once, but are written back to the ontology as a persistent part of the ontology in the form of mapping assertion triples (source field, mapping to, ontology concept, confidence, verification time). When a new data source is subsequently connected, the system first performs a fast match between the new field and the accumulated mapping assertions (based on field name similarity and context similarity). If a match is found, the historical mapping result is directly reused; if a partial match is found, its confidence is initialized to the weighted value of the historical mapping, thereby reducing the frequency of manual verification and realizing incremental accumulation and reuse of mapping knowledge.

[0030] Step 3: Ontology-based rule-driven data governance The constraints (value range, foreign key reference, business dependency, and temporal consistency) in the business rule sub-ontology are automatically translated into executable verification rules (SHACL / SQL), anomaly identification is performed on the access data, and repair suggestions are given based on ontology reasoning; systemic anomalies generate governance work orders to drive source system rectification.

[0031] Taking the business rule that a legal entity's unified social credit code must be 18 digits and unique as an example, the automatically compiled output is: #Translated into SHACL shape constraints gov: LegalEntityShape a sh: NodeShape; sh: targetClass gov: LegalEntity; sh:property [ sh:path gov:creditCode; sh: minLength 18; sh: maxLength 18; sh: pattern "^[0-9A-HJ-NPQRTUWXY]{18}$"; sh:minCount 1; ] # Translate into equivalent SQL validation statement SELECT credit_code, COUNT(*) FROM legal_entity WHERE LENGTH(credit_code)<>18 OR credit_code !~ '^[0-9A-HJ-NPQRTUWXY]{18}$' GROUP BY credit_code HAVING credit_code IS NOT NULL UNION ALL SELECT credit_code,COUNT(*) FROM legal_entity GROUP BY credit_code HAVING COUNT(*)>1; -- Uniqueness check #Translate pseudocode: for rule in business rule sub-entity: (type, target_concept, params) = parse(rule) shacl=compile_to_SHACL(type, ontology.map(target_concept), params) sql=compile_to_SQL(type, ontology.table(target_concept), params) violations = run(shacl, sql, access data) for v in violations: fix = infer_fix(v, ontology) # Provides repair suggestions based on ontology inference If a systemic anomaly occurs (v): `emit_workorder(v)` # Generate a governance work order. Step 4: OAG Government Affairs Q&A and Document Generation Assistance Using a government affairs ontology as the knowledge framework for enhanced generation: After receiving user questions or document drafting requests, the system first retrieves relevant subjects, matters, processes, and legal subgraphs from the ontology, and injects these subgraphs into the large language model with structured hard constraints. Each conclusion or document paragraph generated by the model is forcibly associated with an ontology node (corresponding legal clause, service item, or process step), forming a traceable chain of evidence. The legal clauses cited in the generated conclusions must exist in the business rule sub-ontology and be in a current valid state (intercepting expired policies); otherwise, regeneration is triggered.

[0032] #1. Retrieving ontology subgraphs based on user needs subgraph = retrieve_subgraph(user question / official document requirement, government affairs ontology) #2. Construct hint words containing structured hard constraints prompt=build_structured_prompt(subgraph) #3. Enter the generation and self-correction loop repeat: draft = LLM.generate(prompt) # Generate a draft violations=[] # Initialize the violation list #4. Verify the legal citations in the generated content item by item. for citation ref in draft: If `ref` is not in the business rule sub-entity or not `ref`, it is currently in effect. violations.add(ref) # Block expired / fictitious regulations #5. Verify whether the conclusion has been linked to the ontology node. if not linked_to_ontology_node (conclusion): violations.add(conclusion) #6. If there are no violations, break out of the loop and return to the successfully generated draft. if violations is empty: break #7. If a violation occurs, append the error feedback to the prompt word to trigger a regeneration. prompt = prompt + feedback (violations) until the maximum number of retries is reached return draft# Each conclusion / paragraph relates to relevant regulations / matter / process nodes Step 5: Permissions and Privacy Controls Based on an ontology-based three-dimensional authorization model of subject-item-field, fine-grained access control is achieved. Sensitive fields are anonymized and disclosure minimized according to the sensitivity levels defined in the ontology, ensuring data compliance during generative service processes. Authorization determination: A user is allowed to see a field within a specific event context if and only if the triple matches the authorization matrix and the field's sensitivity level is not higher than the role's permission level.

[0033] The permissions and privacy control process is as follows: (1) When a user initiates a question-and-answer or document generation request, the system first identifies the user's role (such as a regular service window, an approver, a supervisor, etc.) and the context of the current matter; (2) After the LLM-generated draft passes the verification in step 4, it enters the permission filtering stage: traverse each data field involved in the output and query its sensitivity level (such as public, internal, sensitive and confidential levels) defined in the ontology. (3) For fields with a sensitivity level higher than the user role permission level, perform desensitization according to the desensitization strategy predefined in the ontology (e.g., only the first 6 and last 4 digits of the ID card number are displayed, and the middle is replaced with *; only the first 3 and last 4 digits of the mobile phone number are displayed); for fields that are completely inaccessible, delete or replace them with placeholders. (4) The final output only contains information that the user role is authorized to view, ensuring that the principle of minimizing disclosure is implemented.

[0034] Furthermore, the ontology of this invention is not a static model built all at once, but a dynamic semantic foundation that continuously evolves during the governance and service generation process. The specific evolution path includes: (1) Accumulation of mapping knowledge: The mapping relationship confirmed manually in step 2 is continuously written back to the ontology to form a mapping assertion library of (source system, source field, target concept, confirmation time). When a new data source is connected, the historical mapping experience is automatically inherited. (2) Dynamic rule updates: When policies and regulations change, the corresponding clauses in the business rule sub-entity are automatically marked as repealed, and new currently valid clauses are added. The rule compilation process in step 3 automatically detects the ontology changes and automatically generates updated SHACL / SQL validation rules the next time it is executed, without the need for manual code modification; (3) Generate feedback loop: In step 4, when the LLM identifies that the user’s needs exceed the coverage of the current ontology during the generation process, it triggers an exception log. After manual review, new concepts or relationships can be added to the ontology accordingly. (4) Consistency guarantee: Since governance and generation share the same ontology instance, any update to the ontology made by any of the above evolution paths will simultaneously affect both governance and generation scenarios, thus avoiding the problem of inconsistency between the two sets of knowledge representations.

[0035] The above-mentioned mapping confidence weights, thresholds, SHACL / SQL snippets, and code formats are all exemplary and used to illustrate feasibility.

[0036] The following example, using the semantic alignment of cross-departmental fields for legal representatives / legal persons and the ontology node tracing of a government affairs Q&A, details the implementation steps, parameters, and conditions.

[0037] The market supervision department's system uses the field "legal representative," a certain government service system's historical field uses "legal person representative," and the tax system's field uses "enterprise head." While the three have the same business meaning, their naming differs, and in some systems, "enterprise head" actually refers to the actual business operator (risk of using the same word with different meanings).

[0038] Steps 1-2: Ontology and Mapping In the main sub-ontology, define the concept "gov": legal representative, and calculate the mapping confidence of each field to this concept: Legal Representative: Name Similarity 1.0, Distribution Similarity 0.9, Sample Pattern (Name) 0.95, Business Relationship (Related Legal Entities) 0.9 → High-confidence auto-alignment; Legal Representative: Automatic alignment occurs when the value exceeds the threshold; Company representative: Name similarity was only 0.5, and some samples did not match the legal entity. If the value falls below the confirmation threshold of 0.8, manual confirmation is triggered. After manual verification, two systems are distinguished: those with the same meaning are aligned to the gov system (legal representative), while the actual operating person in charge is defined separately. The confirmation result is fed back into the ontology, storing the mapping knowledge of this field.

[0039] After the mapping knowledge is accumulated, if a new system is connected and contains the field of enterprise person in charge, the system will automatically retrieve the historical mapping record of the field (confidence level 0.72, which has been manually confirmed as partially consistent). Based on the context (such as the business type of the system being enterprise registration), the system will automatically use the high-confidence mapping to gov: legal representative, without the need for manual confirmation again.

[0040] Step 3: Rule Governance The SHACL / SQL verification compiled in step 3 is performed on the accessed legal entity data. It is found that there are 17 abnormal items and 3 duplicate items in the unified social credit code of a certain batch of data. An abnormal list and repair suggestions (filling in the blanks / deduplication) are automatically generated, and a governance work order is generated for systemic missing data and pushed to the source system.

[0041] Step 4: OAG Question and Answer Source Tracing When a user asks, "What materials and what is the basis for changing the legal representative of a company?" the system retrieves the relevant sub-entities (change registration items), process sub-entities (material list), and business rule sub-entities (corresponding legal clauses) to construct a hard constraint prompt generation model. The model then outputs a response, associating each material with a process node and the basis with a currently valid legal clause node. During generation, the model initially referenced a repealed old regulation, which was intercepted by the current validity check and triggered a regeneration. Ultimately, only the current clause was referenced, forming a traceable chain of evidence.

[0042] In this question-and-answer example, the final structured output (JSON format) is as follows: { The answer states: "The following materials are required for a company to change its legal representative: 1. Company registration (filing) application form; 2. Legal representative information form; 3. Shareholders' meeting resolution…". "evidence_chain": [ {"node_id": "gov: Procedure.MaterialList.001", "node_type": "Process Node", "label": "Bill of Materials"}, {"node_id": "gov: Regulation.CompanyLaw.Article13", "node_type": "Regulations", "label": "Article 13 of the Company Law", "status": "Currently in effect"} {"node_id": "gov: Regulation.MarketRegulation.Order_2023_XX", "node_type": "Regulations", "label": "Article X of the Regulations on the Registration and Management of Market Entities", "status": "Currently in effect"} ] } Each output conclusion carries an evidence_chain array, which the front end uses to present a clickable chain of evidence, allowing users to view the original legal text and procedural basis one by one.

[0043] Step 5: Access Control Desensitization The legal representative's ID number involved in the response was anonymized according to the sensitivity level, with only a portion of the number displayed. Ordinary service window personnel do not have the authority to view the complete number, which complies with the three-dimensional authorization of subject-matter-field.

[0044] In this embodiment, the system identifies the user role as a regular service window, and the ID number is marked as sensitive in the system. The role's permission level is internal. Because the sensitivity level is higher than the permission level, the system automatically desensitizes the ID number from 123456199001011234 to 123456********1234 before returning it. If the user role is a supervisor, and their permission level is confidential, then the complete number can be viewed.

[0045] This embodiment fully reproduces the process of semantic layer alignment to crack homonyms / synonyms with different characters—automatic rule compilation to complete data verification—associating question-and-answer conclusions with legal nodes and intercepting expired policies.

[0046] The confidence scores, SHACL / SQL fragments, thresholds, etc. in the examples are all exemplary and used to illustrate feasibility. In actual applications, they can be flexibly adjusted according to the scenario.

[0047] Method Implementation Examples According to embodiments of the present invention, an ontology-driven method for bidirectional enhancement of government governance and generation is provided. Figure 5 This is a flowchart of the ontology-driven method for bidirectional enhancement of government governance and generation, as described in an embodiment of the present invention. Figure 5 As shown, the ontology-driven method for bidirectional enhancement of government governance and generation according to an embodiment of the present invention specifically includes: Step S501: Store the government semantic knowledge graph through the government domain ontology module; wherein, the government semantic knowledge graph includes multiple sub-ontologies, which are interconnected through preset semantic relationships, and the government semantic knowledge graph is composed of multiple ontology nodes and semantic relationship edges between ontology nodes; the government semantic knowledge graph is formally represented using the OWL2 specification. The government semantic knowledge graph includes subject sub-ontology, item sub-ontology, process sub-ontology, coding system sub-ontology and business rule sub-ontology; Each business rule in the business rule sub-ontology adopts a triple structure, including constraint type, target concept, and parameter list; the constraint type includes value range constraint, foreign key reference constraint, business dependency constraint, temporal consistency constraint, uniqueness constraint, and format constraint.

[0048] Step S502: Obtain heterogeneous data sources through the semantic mapping module, and semantically align the fields to be accessed in each data source with the ontology node to generate a mapping relationship from fields to ontology nodes, and output the confirmed mapping relationship to the government domain ontology module. Step S503: Read business rule constraints from the government domain ontology module through the rule governance module, compile the business rule constraints into verification rules, perform data verification on the aligned data output by the semantic mapping module according to the verification rules, output the verification results, and take the business rule constraints that are not defined in the government domain ontology module identified during the verification process as rule missing information. Step S504: The generation service module retrieves a semantic subgraph from the government affairs domain ontology module based on the government affairs needs input by the user, inputs the semantic subgraph as constraint information into the generation model, outputs the response content generated by the generation model based on the constraint information, and outputs an ontology update trigger signal when the semantic features in the government affairs needs exceed the coverage of the government affairs domain ontology module. Step S505: The ontology evolution module receives at least one of the confirmed mapping relationship, the rule missing information, and the ontology update trigger signal, and generates an ontology update instruction based on the received information and outputs it to the government domain ontology module to trigger dynamic updates.

[0049] The embodiments of the present invention are method embodiments corresponding to the system embodiments described above. The specific operations of each step can be understood by referring to the description of the system embodiments, and will not be repeated here.

[0050] In summary, compared with the prior art, the embodiments of the present invention have the following beneficial effects: 1. The government ontology serves as a cross-departmental semantic translator, covering who, what, how, what, and how the coding corresponds through five sub-ontologies. This solves the governance problem of different meanings for the same word and different words for the same meaning in traditional government data, achieving unification at the semantic layer rather than the field layer.

[0051] 2. Ontology-driven governance automation: Business rule ontology can be directly compiled into SHACL / SQL verification rules, transforming cross-departmental data verification that originally relied on manual reconciliation into ontology reasoning tasks, significantly reducing governance costs and improving consistency. Moreover, rules are automatically recompiled as the ontology is updated, so governance is no longer lagging behind.

[0052] 3. OAG generation service traceability loop: Each output of the generation model is attached to the government ontology node (corresponding to legal provisions, service items or process links), and expired or fictitious legal references are intercepted, fundamentally solving the problem of unclear basis for government generation services, and meeting the basic requirements of government affairs being explainable and accountable.

[0053] 4. Ontology Continuous Evolution Mechanism: New mappings, rules, and semantic associations formed during the data governance and service generation processes are written back to the ontology, enabling continuous iteration of the government semantic foundation, which is different from the traditional ontology method of one-time modeling; and the same ontology drives both governance and generation scenarios simultaneously, avoiding inconsistencies between the two sets of knowledge representations.

[0054] 5. Two-way Enhancement Closed Loop between Governance and Generation: This embodiment of the invention connects the two major scenarios of governance and generation through a unified ontology, forming a positive feedback loop: the field mapping, concept alignment and business rules accumulated in the governance process provide rich semantic knowledge for the ontology, making the generation service more accurate and compliant; new requirements, new concepts and new rule relationships identified in the generation service trigger ontology updates, making subsequent governance more comprehensive and refined. This two-way enhancement mechanism is an architectural advantage that existing technologies do not possess.

[0055] Device Example 1 This invention provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it performs the steps described in the method embodiment.

[0056] Device Example 2 This invention provides a computer-readable storage medium storing an information transmission implementation program, which, when executed by a processor, performs the steps described in the method embodiment.

[0057] The computer-readable storage media described in this embodiment include, but are not limited to, ROM, RAM, disk, or optical disk.

[0058] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An ontology-driven system that enhances both governance and generation, characterized in that, include: The government affairs domain ontology module is used to store the government affairs semantic knowledge graph; wherein, the government affairs semantic knowledge graph includes multiple sub-ontologies, which are interconnected through preset semantic relationships, and the government affairs semantic knowledge graph is composed of multiple ontology nodes and semantic relationship edges between ontology nodes; The semantic mapping module, connected to the government domain ontology module, is used to acquire heterogeneous data sources, semantically align the fields to be accessed in each data source with the ontology node, generate a mapping relationship from field to ontology node, and output the confirmed mapping relationship to the government domain ontology module. The rule governance module, connected to the government domain ontology module and the semantic mapping module, is used to read business rule constraints from the government domain ontology module, compile the business rule constraints into verification rules, perform data verification on the aligned data output by the semantic mapping module according to the verification rules, output the verification results, and treat business rule constraints that are not defined in the government domain ontology module and are identified during the verification process as rule missing information. The generation service module is connected to the government affairs domain ontology module and the rule governance module. It is used to retrieve semantic subgraphs from the government affairs domain ontology module according to the government affairs requirements input by the user, input the semantic subgraphs as constraint information into the generation model, output the response content generated by the generation model based on the constraint information, and output an ontology update trigger signal when the semantic features in the government affairs requirements exceed the coverage of the government affairs domain ontology module. The ontology evolution module, connected to the government domain ontology module, semantic mapping module, rule governance module, and generation service module, is used to receive at least one of the confirmed mapping relationship, the rule missing information, and the ontology update trigger signal, and generate an ontology update instruction based on the received information and output it to the government domain ontology module to trigger dynamic updates.

2. The system according to claim 1, characterized in that, The government affairs semantic knowledge graph is formally represented using the OWL2 specification; The government semantic knowledge graph includes subject sub-ontology, item sub-ontology, process sub-ontology, coding system sub-ontology and business rule sub-ontology; Each business rule in the business rule sub-ontology adopts a triple structure, including constraint type, target concept, and parameter list; the constraint type includes value range constraint, foreign key reference constraint, business dependency constraint, temporal consistency constraint, uniqueness constraint, and format constraint.

3. The system according to claim 1, characterized in that, The semantic mapping module is specifically used for: The mapping confidence of the field to be accessed to the ontology node is calculated by weighting the similarity of field name, data distribution, sample value pattern, and upstream and downstream business relationship. The mapping confidence level is compared with the confirmation threshold. Mapping relationships that are greater than or equal to the confirmation threshold are automatically confirmed, while mapping relationships that are less than the confirmation threshold are manually confirmed. The confirmed mapping relationship is written back to the government domain ontology module in the form of mapping assertion triples; wherein, the mapping assertion triples include the source field, the ontology node to which it is mapped, the confidence level, and the confirmation time; When a new data source is connected, a match is made between the new field and the previously written-back mapping assertion triples. If a match is found, the corresponding mapping relationship is reused directly.

4. The system according to claim 1, characterized in that, The rule governance module is specifically used for: Read the business rule constraints from the business rule sub-ontology in the government domain ontology module, and automatically compile the business rule constraints into validation rules; Data verification is performed on the aligned data output by the semantic mapping module according to the verification rules to identify abnormal data; Based on the government semantic knowledge graph, the abnormal data detected is inferred, and repair suggestions are output to the ontology evolution module; The data that passes the verification is output to the generation service module as verified data, and the business rule constraints that are not defined in the government affairs ontology module and are identified during the verification process are output to the ontology evolution module as rule missing information.

5. The system according to claim 4, characterized in that, The generation service module is specifically used for: Receive the verified data output by the rule governance module and the government affairs requests input by the user; Retrieve the corresponding semantic subgraph from the government affairs semantic knowledge graph based on the stated government affairs requirements; The semantic subgraph is input as constraint information into the generative model, and the verified data is used as the input context of the generative model. The output is the response content generated by the generative model based on the constraint information and the verified data, and each response content is associated with a corresponding ontology node identifier. The system verifies whether the legal provisions cited in the response exist in the business rule sub-entity and are currently valid. If they do not exist or are not currently valid, the system intercepts and triggers the regeneration of the generation model. When the semantic features in the government affairs requirements exceed the coverage of the government affairs semantic knowledge graph, an ontology update trigger signal is output to the ontology evolution module.

6. The system according to claim 1, characterized in that, The ontology evolution module is specifically used for: Receive the confirmed mapping relationship output by the semantic mapping module, and store the mapping relationship in the form of mapping assertion triples into the government semantic knowledge graph; The system receives repair suggestions and rule missing information output by the rule governance module, and supplements new rules, updates existing rules, adds new ontology nodes or new semantic relationship edges to the government semantic knowledge graph based on the repair suggestions or rule missing information. Receive the ontology update trigger signal output by the generation service module, and add new ontology nodes or new semantic relationship edges to the government semantic knowledge graph according to the ontology update trigger signal; When the legal clauses in the business rule sub-entity are changed, the corresponding clauses are marked as repealed or added, and the rule governance module is triggered to automatically compile the updated verification rules.

7. An ontology-driven method for bidirectional enhancement of government governance and generation, characterized in that, include: The government affairs semantic knowledge graph is stored through a government affairs domain ontology module; wherein, the government affairs semantic knowledge graph includes multiple sub-ontologies, which are interconnected through preset semantic relationships, and the government affairs semantic knowledge graph is composed of multiple ontology nodes and semantic relationship edges between ontology nodes; Heterogeneous data sources are obtained through the semantic mapping module, and the fields to be accessed in each data source are semantically aligned with the ontology node to generate the mapping relationship from field to ontology node. The confirmed mapping relationship is then output to the government domain ontology module. The rule governance module reads business rule constraints from the government domain ontology module, compiles the business rule constraints into verification rules, performs data verification on the aligned data output by the semantic mapping module according to the verification rules, outputs the verification results, and treats business rule constraints that are not defined in the government domain ontology module as rule missing information. The generation service module retrieves semantic subgraphs from the government domain ontology module based on the government needs input by the user, inputs the semantic subgraphs as constraint information into the generation model, outputs the response content generated by the generation model based on the constraint information, and outputs an ontology update trigger signal when the semantic features in the government needs exceed the coverage of the government domain ontology module. The ontology evolution module receives at least one of the confirmed mapping relationship, the rule missing information, and the ontology update trigger signal, and generates an ontology update instruction based on the received information and outputs it to the government affairs ontology module to trigger dynamic updates.

8. The method according to claim 7, characterized in that, The government affairs semantic knowledge graph is formally represented using the OWL2 specification; The government semantic knowledge graph includes subject sub-ontology, item sub-ontology, process sub-ontology, coding system sub-ontology and business rule sub-ontology; Each business rule in the business rule sub-ontology adopts a triple structure, including constraint type, target concept, and parameter list; the constraint type includes value range constraint, foreign key reference constraint, business dependency constraint, temporal consistency constraint, uniqueness constraint, and format constraint.

9. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the ontology-driven method for bidirectional enhancement of governance and generation as described in any one of claims 7-8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores an implementation program for information transmission, which, when executed by a processor, implements the steps of the ontology-driven method for bidirectional enhancement of government governance and generation as described in any one of claims 7-8.

Images