A business demand analysis method and system based on cognitive enhanced industrial identification

Abstract

The application discloses a business demand analysis method and system based on cognitive enhanced industrial identification. The method of the application is as follows: taking each micro management object in an enterprise as an entity, and generating a cognitive enhanced industrial identification for each micro management object; the cognitive enhanced industrial identification comprises a global unique resource locator, a cognitive descriptor, and a digital signature; generating a task instruction according to a user's business demand, and generating an entity context snapshot for each in the business demand; generating an enhanced prompt word according to the task instruction and the entity context snapshot, and submitting the enhanced prompt word to a large language model to generate an executable plan; and executing the executable plan to obtain an intention analysis result of the business demand. The application has a high autonomous operation capability, can analyze user intention, autonomously generate a data processing plan, and perform safe and controllable execution, and greatly accelerates the process of enterprise digital transformation.

Description

Technical Field

[0001] This invention belongs to the field of interdisciplinary technology of industrial internet and artificial intelligence, and relates to a business requirement parsing method and system based on cognitively enhanced industrial identifiers. Background Technology

[0002] Currently, the Industrial Internet Identifier Resolution System has become the infrastructure for connecting data across the entire industrial chain. However, in practical applications, the existing system has the following limitations: First, there are limitations in identifier positioning: the existing system treats identifiers as static pointers with a unique code for each item, lacking semantic descriptive capabilities. This means that business systems, upon receiving a request, cannot directly "understand" the context of an entity through the identifier. This underlying semantic deficiency limits identifiers to passively serving as keywords for retrieval, making it difficult for them to directly participate in and drive complex business logic.

[0003] Second, there is a disconnect in the coding system: the core of existing parsing methods lies in "integration," that is, passively connecting isolated data sources through a large amount of adaptation code. Faced with vague business intentions described in natural language (such as "optimize the energy efficiency of a certain production line"), traditional methods cannot perform autonomous intention decomposition and arrangement. Due to the lack of cognitively enhanced identifiers as semantic hubs, artificial intelligence such as large language models (LLM) cannot directly intervene, and the system cannot autonomously understand business goals and generate optimal data processing plans like a human.

[0004] Third, semantic mapping is inefficient: traditional business requirement parsing mainly relies on manually predefined static mappings (such as ETL or hard-coded interfaces). This "one-off" integration method is essentially a mechanical stacking of machine instructions. Whenever business requirements are slightly adjusted, or new heterogeneous data sources are connected, the parsing code needs to be rewritten. This rigid parsing chain results in extremely low system robustness, and the marginal cost increases sharply with the increase of business complexity.

[0005] Fourth, there is a conflict regarding secure data sharing: Traditional access control (such as XACML) is an external defense system, disconnected from the parsing logic. When faced with AI-generated, cross-system dynamic parsing commands, traditional methods cannot perform real-time, fine-grained "field-level" verification during the parsing process. This lag in security mechanisms creates a serious conflict between pursuing parsing efficiency and ensuring data compliance. Summary of the Invention

[0006] To address the problems existing in the prior art, the purpose of this invention is to provide a business requirement analysis method and system based on cognitively enhanced industrial identification.

[0007] The invention addresses the shortcomings of existing identification systems in terms of semantic carrying capacity, architectural flexibility, and integration efficiency, and provides the following solutions: First, semantic enhancement of the logo is achieved: by injecting the reasoning ability of LLM, the static logo is upgraded into a "cognitive-enhanced industrial logo" with self-description and understandable characteristics, thereby enhancing the intrinsic value of the logo.

[0008] Second, construct a flexible identification paradigm: provide an adaptive micro-identification generation mechanism to replace the traditional rigid standards and achieve the connection between the macro system and the micro-practice of enterprises.

[0009] Third, improve the level of integration automation: leverage the code generation capabilities of LLM to achieve intent-driven dynamic data fusion, replacing high-cost hard-coded integration.

[0010] Fourth, lower the threshold for security management: establish a declarative security policy model based on natural language and simplify the permission configuration process.

[0011] The technical solution of this invention is as follows: A business requirement parsing method based on cognitively enhanced industrial identification includes the following steps: Each micro-management object within the enterprise is treated as an entity, and a cognitively enhanced industrial identifier is generated for it; the cognitively enhanced industrial identifier includes a globally unique resource locator, a cognitive descriptor, and a digital signature; Generate task instructions based on the user's business requirements, and generate an entity context snapshot for each of those business requirements; Enhanced prompts are generated based on task instructions and entity context snapshots and submitted to a large language model to generate an executable plan. Execute the executable plan to obtain the intent parsing result of the business requirement.

[0012] Preferably, the cognitive descriptor includes the entity's metadata and a reference pointer to the entity's dynamic context.

[0013] Preferably, the metadata includes entity type, context, and entity status; the reference pointer includes links defined by the entity's associated data pattern, links to the entity's data service interface document, and links to the entity's access control policy; and the globally unique resource locator and cognitive descriptor are asymmetrically encrypted and signed to generate a hash string as a digital signature.

[0014] Preferably, the method for generating an entity context snapshot is as follows: obtaining the entity's metadata from the cognitive descriptor in the entity's cognitively enhanced industrial identifier, and generating an entity profile of the entity; obtaining the entity's deep context, including the entity's data pattern, API call specifications, and applicable access control policies, based on the reference pointer to the entity's dynamic context in the cognitive descriptor; and generating an entity context snapshot of the entity based on the entity profile and deep context.

[0015] Preferably, the method for generating task instructions based on business requirements is as follows: use a large language model to parse the business requirements, identify the user intent, the entities involved, and the constraints, and transform them into structured task instructions.

[0016] Preferably, the executable plan is run in an isolated sandbox environment.

[0017] Preferably, the micro-management object is a physical entity or a virtual entity; the physical entity includes a product, a batch of products, or a stack of products, and the virtual entity is a data object.

[0018] A business requirement parsing system based on cognitively enhanced industrial identification, characterized in that it includes: The identifier generation module is used to treat each micro-management object within an enterprise as an entity and generate a cognitively enhanced industrial identifier for it; the cognitively enhanced industrial identifier includes a globally unique resource locator, a cognitive descriptor, and a digital signature; The intelligent planning layer is used to generate task instructions based on the user's business needs, and to generate an entity context snapshot for each of those business needs; then, it generates enhanced prompts based on the task instructions and entity context snapshots and submits them to a large language model to generate an executable plan. The secure execution module is used to execute the executable plan and obtain the intent parsing result of the business requirement.

[0019] A computing device, characterized in that it comprises: a processor and a memory storing a computer program, wherein the computer program, when run by the processor, executes the method described above.

[0020] A computer-readable storage medium, characterized in that it stores instructions that, when executed on a computer, cause the computer to perform the method described above.

[0021] The advantages of this invention are as follows: This invention endows industrial data systems with a high degree of autonomous operation capabilities, enabling them to interpret user intent, autonomously generate data processing plans, and execute them securely and controllably. The core of traditional technical approaches is "integration," which involves writing extensive adaptation code to connect passive, isolated data sources; its essence remains a mere accumulation of machine instructions. This invention, however, achieves "cognitive fusion," injecting semantics into the data, allowing the system to understand data and business objectives like a human, and dynamically generate optimal solutions, representing a qualitative leap. Specifically, this includes the following points.

[0022] (1) It greatly enhances the depth and breadth of AI’s understanding of the industrial field, fundamentally solving the problem of “cognitive limitations”.

[0023] Semantic Penetration: Traditional identifiers lack embedded semantic information, making them incomprehensible to AI. The "Cognitive Enhanced Industrial Identifier" of this invention embeds core metadata and contextual references, transforming the identifier into a "self-describing" intelligent entity. AI systems can instantly understand the essential attributes, associated data, and interaction methods of industrial entities (such as a piece of equipment or a batch of products) without relying on complex external knowledge graphs, fundamentally improving cognitive efficiency and accuracy.

[0024] Automatic Context Aggregation: In traditional solutions, obtaining complete contextual information (such as data schema, API, and security policies) for an entity requires complex manual sorting and hard-coded integration. The "Context Extension Parsing Process" of this invention can automatically track references in identifiers and dynamically construct a complete "entity context snapshot." This enables AI to make decisions and plans based on a comprehensive, real-time, and accurate information view, avoiding errors in decision-making and planning caused by incomplete information.

[0025] (2) It has achieved end-to-end automation from business needs to data value, completely bridging the "structural gap" problem.

[0026] Significantly enhances business responsiveness: In traditional models, a new data analysis requirement often means a development cycle of several days or even weeks (requirements analysis, ETL development, report generation). This invention's "intent-driven architecture" transforms this traditional development cycle into a dynamic, automated, real-time generation process. Users simply need to posed business questions in natural language, and the "generative fusion planning module" automatically generates and executes a data fusion plan, achieving instant response from business intent to data insights, greatly improving enterprise decision-making efficiency and market responsiveness.

[0027] Eliminating Technical Debt and Maintenance Costs: Traditional hard-coded data processing workflows are rigid and fragile; any change in business logic triggers a chain reaction of modifications, accumulating substantial technical debt. This invention, through AI-driven dynamic generation of execution plans, completely eliminates predefined, fixed data processing scripts. The system can automatically adapt to changes in underlying data sources or business logic, exhibiting extremely high elasticity and adaptability, significantly reducing long-term system maintenance costs.

[0028] (3) An "intrinsic security" system with both flexibility and strong control has been constructed, effectively resolving the security and integration challenges in the "technical barriers".

[0029] Trustworthy Source and Controllable Process: Traditional security relies on boundary protection and application-layer access control, resulting in a long chain and numerous risk points. This invention's "trust anchor" mechanism ensures the authenticity and integrity of data from its source. Simultaneously, an independent "secure execution module" dynamically verifies permissions and conducts pre-execution reviews of every line of code generated by the AI ​​in a sandbox isolation environment, ensuring that its behavior strictly adheres to dynamically loaded security policies. This forms a complete, zero-trust security closed loop, from 'trustworthy source' (trust anchor) to 'controllable process' (dynamic verification) and then to 'execution isolation' (sandbox), granting the AI ​​a high degree of autonomy while guaranteeing absolute control over the final result.

[0030] Diminishing marginal costs of integration: Traditional heterogeneous system integration is a repetitive, labor-intensive process with high costs, often handled on a case-by-case basis. This invention encapsulates interface differences within the metadata of a "cognitive descriptor," allowing AI to automatically adapt and generate calling code based on this metadata. This means that for each new data source integrated, the system only needs to maintain its metadata, without rewriting the integration logic. This brings the marginal cost of system integration close to zero, significantly accelerating the process of enterprise digital transformation. Attached Figure Description

[0031] Figure 1 This is a flowchart of the method of the present invention.

[0032] Figure 2 This is a system diagram of the present invention. Detailed Implementation

[0033] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0034] like Figure 1 As shown, an optional embodiment of the present invention provides a business requirement parsing method based on cognitively enhanced industrial identifiers, the steps of which include: Each micro-management object within the enterprise is treated as an entity, and a cognitively enhanced industrial identifier is generated for it; the cognitively enhanced industrial identifier includes a globally unique resource locator, a cognitive descriptor, and a digital signature; Generate task instructions based on the user's business requirements, and generate an entity context snapshot for each of those business requirements; Enhanced prompts are generated based on task instructions and entity context snapshots and submitted to a large language model to generate an executable plan. Execute the executable plan to obtain the intent parsing result of the business requirement.

[0035] This invention provides a method and system for parsing business requirements based on cognitively enhanced industrial identifiers. The core idea of ​​this solution is to use a large language model (LLM) as the cognitive center to cognitively empower traditional industrial internet identifiers, thereby constructing a new type of "cognitively enhanced industrial identifier." Based on this, a system capable of understanding users' natural language intentions, automatically generating and securely executing data fusion tasks can be created.

[0036] 1. Technological Approach: From "Passive Positioning" to "Intent-Driven" The core idea of ​​this solution is "pre-emptive cognition, intent-driven." Pre-emptive cognition: This invention no longer treats data as a passive, undifferentiated stream of bits. Instead, through an innovative identification technology, it endows each entity (device, batch, order, etc.) with the ability to "self-describe" and "associate with context." This "cognition" injected from the data source is the foundation for the entire upper-layer intelligence. Intent-driven: This invention constructs an intelligent system capable of understanding human natural language business intent and automatically transforming it into a secure, executable data operation plan. Guided by the user's ultimate goal, this system dynamically and on-demand orchestrates and executes data flows, rather than relying on pre-defined, rigid data processing chains.

[0037] 2. System Layered Architecture The core architecture of this solution can be precisely summarized as a "1+1+3" model: 1 data foundation + 1 planning core + 3 collaborative modules. This model clearly reveals the hierarchical and dynamically invoked operational logic among the system's components, ensuring that the transformation process from user intent to data value is efficient, secure, and logically rigorous.

[0038] One data foundation: the basic cognitive layer, which is the static data and semantic foundation of the entire system.

[0039] One core planning layer: the intelligent planning layer, which is the core of command and arrangement of the entire dynamic processing flow.

[0040] The system consists of three collaborative modules: the interaction intent module, the cognitive parsing module, and the security execution module. These modules work together around the core planning framework and are responsible for intent parsing, context parsing, and security execution, respectively.

[0041] 2.1 Foundation & Cognition Layer The foundational cognitive layer is the "digital world cornerstone" of the system. This layer is responsible for assigning a unique, machine-understandable "digital identity" to every entity in the physical world and business processes, fundamentally solving the problem of "cognitive limitations." The "cognitive-enhanced industrial identifier" proposed in this invention is a "self-contained, self-descriptive digital object." Its encoding structure is designed to be multi-segmented, containing not only a unique "locator" but also embedded "cognitive descriptors" that can be directly understood by AI and "trust anchors" to ensure security.

[0042] The coding structure of cognitively enhanced industrial identifiers can be designed as a three-part structure, connected by specific delimiters (such as ::): [Globally Unique Resource Locator]::[Cognitive Descriptor in JSON Format]::[Digital Signature].

[0043] Core technology: Generating a Cognitive-Enhanced Industrial Identifier (CEII) for each entity. Each CEII consists of three indivisible parts that together form a complete and inherently secure "digital object": First paragraph: Global Unique Locator Function: To ensure the global uniqueness and resolvability of the identifier, and to achieve compatibility and interoperability with the existing identifier system.

[0044] Implementation: Existing mature identification systems such as Handle, DID, and OID can be directly adopted or made compatible. For example, the Handle prefix can be used to ensure its uniqueness under the national top-level node.

[0045] An entity can be a product, a batch of products, or a stack of products, or it can be a virtual entity such as data.

[0046] Example: 88.198.1 / CNC / SH-A-L3-007 88.198.1: A prefix granted to enterprises by the state. CNC / SH-A-L3-007: An internally assigned code representing "CNC machine tool No. 7 on production line 3 of workshop A in the Shanghai factory".

[0047] The difference from the traditional system: In the traditional system, the identification ends there. But in this invention, it is merely an "identity card number".

[0048] Second paragraph: Cognitive Descriptor Function: This is the core innovation of this invention. It is a lightweight, structured metadata that can be directly read and understood by machines (especially LLMs). This metadata enables the identifier to have self-descriptive capabilities, explaining its type, attributes, and associated operations.

[0049] Implementation: Encoded using compact JSON (or binary formats such as CBOR), with content expandable as needed. Core fields may include: type: Entity type (e.g., equipment / machine tool, product / batch, supplier, order); context: context (e.g., Shanghai factory / workshop 3 / production line A); status: Real-time or near-real-time status (e.g., running, standby, fault); schema_ref: A link to the associated data schema definition of the entity corresponding to the cognitively enhanced industrial identifier; api_ref: A link to the data service interface documentation for the entity corresponding to the cognitively enhanced industrial identifier; policy_ref: A link to the access control policy of the entity corresponding to the cognitively enhanced industrial identifier.

[0050] Example of a cognitive descriptor (JSON format): {"type":"device / machine tool", "context":"Shanghai factory / workshop 3 / production line A", "status":"running", "schema_ref":" / schemas / cnc01.json", "policy_ref":" / policies / level3.acl"}.

[0051] The function of the cognitive descriptor: Embedded in a structured format such as JSON, it is the core of achieving "cognitive enhancement." It not only contains the entity's static metadata (type, affiliation), but more importantly, it contains pointers to its dynamic context. These pointers link to external metadata repositories, defining the entity's data schema (what data it has), API interface (how to obtain data), and security policy (who can obtain data). This allows the identifier itself to carry complete access and usage rules, solving the problems of missing context and information silos.

[0052] Third paragraph: Trust Anchor as a digital signature Function: To provide guarantees of the authenticity and integrity of the identifier. To ensure that the identifier and its cognitive descriptor are issued by a legitimate entity and have not been tampered with.

[0053] Implementation: Use the issuer's private key to perform an asymmetric cryptographic signature on the content of [first segment + second segment] (such as using the ECDSA algorithm), and generate a hash string.

[0054] Example of a trust anchor: e7d4c1b9a3f2... (a very long signature hash) Trust Anchor Function: The first two parts of the content are hashed and cryptographically signed using the issuer's private key. This endows the identifier with inherent security, allowing the recipient to verify the authenticity of its origin and the integrity of the content using the public key. This eliminates the risk of data poisoning and identity forgery at the source, laying the foundation for solving security issues in the "technical barrier."

[0055] Summary: The fundamental differences between this and traditional industrial internet identifiers are shown in the table below. 2.2 Dynamic processing flow: uniformly arranged by the planning core. When a business requirement arises, the system will initiate a dynamic processing closed loop, with the intelligent planning layer at its core and three collaborative functional modules working in a unified manner.

[0056] 2.2.1 Core of Planning: Intelligent Planning Layer Positioning: The core module for system planning and orchestration. This layer is crucial for bridging "structural gaps," responsible for translating the business intentions of higher levels into executable data operation plans at the lower levels.

[0057] Core technology: Generative Fusion Planning Input reception: This layer receives two core inputs: a structured "task instruction" from the interaction intent module, and one or more "entity context snapshots" from the cognitive parsing module. One entity context snapshot is generated for each entity; when the task involves multiple entities, multiple entity context snapshots are generated.

[0058] Constructing Augmented Prompts: The intelligent planning layer internally combines task instructions and entity context snapshots according to a preset template to create a highly informative "augmented prompt," which is then submitted to the Large Language Model (LLM). This prompt explicitly tells the LLM: what your role is (e.g., data analyst), what your goal is (task instructions), what tools and resources you have (APIs and data patterns in the context snapshot), and what rules you must follow (security policies), etc.

[0059] Generating an executable plan: In this "information-complete" environment, the task of LLM is no longer open-ended creation, but rather logical reasoning and code generation within a defined scope. It generates a precise, efficient, and secure executable code (such as a Python script) that details how to call APIs, combine data, and perform calculations to complete the task instructions.

[0060] Output plan: The generated code is passed to the secure execution module as a "fusion plan" for execution.

[0061] Value: This layer uses AI's dynamic generation capabilities to completely replace the ETL, stored procedures, and API aggregation logic that traditional IT systems require manual writing and pre-deployment, achieving end-to-end automation from business requirements to data operations.

[0062] 2.2.2 Collaborative Function Module 2.2.2.1 Interaction Intent Module Location: The entry point for user intent parsing in the system. This layer is the entry point for the entire system and is responsible for understanding the user's natural language needs.

[0063] Core technology: Natural Language Understanding (NLU) Capture natural language: Users input vague business requirements via text or voice, such as: "Compare the on-time delivery rate and quality inspection pass rate of part No. 3 from suppliers A and B last month."

[0064] Intent recognition and entity extraction: This layer utilizes the NLU capability of LLM to parse the input and accurately identify the user's core intent (comparative analysis), the entities involved (supplier A / B, part number 3), and constraints such as time and indicators.

[0065] Generate structured task instructions: The parsed results are converted into a machine-readable structured JSON object—the "task instruction," for example: { "action": "compare", "targets": ["CEII_Supplier_A", "CEII_Supplier_B"], "metrics": ["on_time_delivery_rate", "quality_pass_rate"], "constraints": {"time_range": "last_month", "part_id": "CEII_Part_3"} } Passing the instruction: This task instruction is passed down to the intelligent planning layer to initiate the subsequent AI planning process.

[0066] 2.2.2.2 Cognitive Analysis Module Positioning: The system's context parsing and extension module. This layer is responsible for processing the Cognitive Enhanced Industrial Identifier (CEII) from the basic cognitive layer into a structured set of context data that can be directly used by the intelligent planning layer.

[0067] Core process: Context-aware dynamic parsing Upon receiving the call request from the intelligent planning layer, this layer performs a deep parsing process for the specified CEII: 1) Syntax decomposition and trust verification: The CEII string is precisely segmented and immediately verified using a trust anchor. Only verified and trusted identifiers can proceed to the next step, ensuring the security of the entire information chain.

[0068] 2) Semantic unpacking and preliminary profile construction: Parse the JSON content of the cognitive descriptor to extract basic information such as the entity's type, location, and status, and form a preliminary entity profile.

[0069] 3) Contextual Association and Deep Expansion: This is the key value-added component of this layer. The parser concurrently tracks all _ref reference pointers in the cognitive descriptor and sends requests to the enterprise's internal metadata center (such as schema library, API gateway, and permission center) to retrieve the entity's current complete data schema, API call specifications, and applicable access control policies in real time.

[0070] 4) Generate a context snapshot: Aggregate all the above information (basic profile + deep context) into a single, structured JSON object – “Entity Context Snapshot”. This snapshot is a complete digital image of the entity at the current moment for the current user, and this entity context snapshot is sent to the intelligent planning layer.

[0071] Value: This layer dynamically resolves and expands a static identifier into a data object containing rich context, thereby avoiding planning errors caused by incomplete information.

[0072] 2.2.2.3 Safe Execution Module Positioning: The system's isolated and secure execution environment. This engine is responsible for receiving and executing the code plan (i.e., the executable plan) generated by the "intelligent planning layer".

[0073] Core functions: Sandbox execution: All AI-generated code runs in an isolated sandbox environment, logically isolated from core system services, preventing any potential malicious or erroneous operations from affecting the stability of the underlying data system.

[0074] Dynamic permission verification: At each step of code execution (such as API calls and data reading), the engine performs real-time, line-by-line permission checks based on the access control policy link (policy_ref) information in the context snapshot passed along with the execution plan. Even if the AI-generated plan itself is reasonable, the operation will be immediately blocked if the current user does not have sufficient permissions.

[0075] Resource and Audit Monitoring: The engine monitors and limits resource consumption (CPU, memory, API call frequency) during code execution to prevent abuse. Simultaneously, all execution actions are recorded in detail, forming an immutable audit log to ensure that every data operation is traceable.

[0076] Value: It provides a strict security and control mechanism for AI-generated code, ensuring that the final output of the entire system is both intelligent and absolutely safe and controllable, perfectly resolving the deep-seated contradiction between security and sharing in the "technical barrier".

[0077] In summary, this technical solution, through its "1+1+3" core planning model, forms an intelligent closed loop from "intent" to "result," uniformly orchestrated by the planning core. The interaction intent module captures user needs and submits them to the planning core; the planning core proactively invokes the cognitive parsing module to obtain context when formulating a plan; finally, the planning core delivers the generated plan to the secure execution module for reliable implementation. This entire process operates on the digital world ontology defined by the basic cognitive layer, systematically solving the core challenges of industrial data applications and achieving unprecedented agility, intelligence, and security in data applications.

[0078] like Figure 2 As shown, an optional embodiment of the present invention provides a business requirement parsing system based on cognitively enhanced industrial identifiers, characterized in that it includes... The identifier generation module is used to treat each micro-management object within an enterprise as an entity and generate a cognitively enhanced industrial identifier for it; the cognitively enhanced industrial identifier includes a globally unique resource locator, a cognitive descriptor, and a digital signature; The intelligent planning layer is used to generate task instructions based on the user's business needs, and to generate an entity context snapshot for each of those business needs; then, it generates enhanced prompts based on the task instructions and entity context snapshots and submits them to a large language model to generate an executable plan. The secure execution module is used to execute the executable plan and obtain the intent parsing result of the business requirement.

[0079] The present invention provides a computing device, characterized in that it includes: a processor and a memory storing a computer program, wherein the computer program is executed by the processor to perform the above-described method.

[0080] The present invention provides a computer-readable storage medium, characterized in that it stores instructions that, when executed on a computer, cause the computer to perform the method described above.

[0081] The above are preferred embodiments of the present invention. It should be noted that, for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A business requirement parsing method based on cognitively enhanced industrial identification, comprising the following steps: Treat each micro-management object within the enterprise as an entity and generate a cognitively enhanced industrial identifier for it; The cognitive-enhanced industrial identifier includes a globally unique resource locator, a cognitive descriptor, and a digital signature; Generate task instructions based on the user's business requirements, and generate an entity context snapshot for each of those business requirements; Enhanced prompts are generated based on task instructions and entity context snapshots and submitted to a large language model to generate an executable plan. Execute the executable plan to obtain the intent parsing result of the business requirement.

2. The method according to claim 1, characterized in that, The cognitive descriptor includes the entity's metadata and a reference pointer to the entity's dynamic context.

3. The method according to claim 2, characterized in that, The metadata includes entity type, context, and entity status; the reference pointers include links defined by the entity's associated data schema, links to the entity's data service interface documentation, and links to the entity's access control policies. Asymmetric encryption signatures are applied to globally unique resource locators and cognitive descriptors to generate a hash string as a digital signature.

4. The method according to claim 2, characterized in that, The method for generating an entity context snapshot is as follows: obtain the entity's metadata from the cognitive descriptor in the entity's cognitively enhanced industrial identifier, and generate an entity profile for that entity; The entity's deep context is obtained by referencing the entity's dynamic context in the cognitive descriptor, including the entity's data model, API call specifications, and applicable access control policies; based on the entity's entity profile and deep context, an entity context snapshot is generated.

5. The method according to claim 1, characterized in that, The method for generating task instructions based on business requirements is as follows: use a large language model to parse the business requirements, identify the user intent, the entities involved, and the constraints, and transform them into structured task instructions.

6. The method according to claim 1, characterized in that, Run the executable plan in an isolated sandbox environment.

7. The method according to claim 1, characterized in that, The micro-management object is a physical entity or a virtual entity; the physical entity includes a product, a batch of products, or a stack of products, and the virtual entity is a data object.

8. A business requirement parsing system based on cognitively enhanced industrial identification, characterized in that, include The identifier generation module is used to treat each micro-management object within an enterprise as an entity and generate a cognitively enhanced industrial identifier for it; the cognitively enhanced industrial identifier includes a globally unique resource locator, a cognitive descriptor, and a digital signature; The intelligent planning layer is used to generate task instructions based on the user's business needs, and to generate an entity context snapshot for each of those business needs; then, it generates enhanced prompts based on the task instructions and entity context snapshots and submits them to a large language model to generate an executable plan. The secure execution module is used to execute the executable plan and obtain the intent parsing result of the business requirement.

9. A computing device, characterized in that, include: A processor, a memory storing a computer program, wherein the computer program, when executed by the processor, performs the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, A storage instruction that, when executed on a computer, causes the computer to perform the method as described in any one of claims 1 to 7.

Images