Supply chain business information query method and device, computer device and medium

Abstract

The application relates to a supply chain business information query method and device, computer equipment and a medium. The method comprises the following steps: inputting a query request of supply chain business information into a large language model for semantic analysis to obtain a business query intention and a business parameter; performing matching processing on the business query intention and the business parameter based on a pre-constructed supply chain knowledge graph to obtain a target business logic rule; performing retrieval in a supply chain business database according to the target business logic rule, the business parameter and the real-time business state data associated with the business parameter; determining supply chain business information based on the real-time business state data and the target business logic rule; generating structured traceability information based on the supply chain business information and the target business logic rule; and inputting the structured traceability information and the query request into the large language model to obtain a supply chain business query text. The method can improve the accuracy of the obtained supply chain business information.

Description

Technical Field

[0001] This application relates to the field of power grid technology, and in particular to a method, apparatus, computer equipment, and medium for querying supply chain business information. Background Technology

[0002] As the digitalization of supply chain operations increases, supply chain business scenarios become increasingly complex, and business personnel's needs for querying supply chain-related business information become more refined and precise.

[0003] To retrieve supply chain-related business information, the main technology used is vector retrieval augmented generation. When a user inputs a query request for supply chain-related business information, the similarity between the query text carried in the query request and the slice text in the knowledge base is calculated. The most relevant slice texts are retrieved and used as the context of the supply chain-related business information to be input into the large language model, which then outputs the supply chain-related business information.

[0004] However, the accuracy of supply chain-related business information retrieved using this technology may be low. Summary of the Invention

[0005] Therefore, it is necessary to provide a method, apparatus, computer equipment, and medium for querying supply chain business information to address the aforementioned technical problems.

[0006] Firstly, this application provides a method for querying supply chain business information, including:

[0007] The query request for supply chain business information is input into the large language model for semantic parsing to obtain the business query intent and business parameters.

[0008] Based on a pre-built supply chain knowledge graph, the business query intent and business parameters are matched to obtain the target business logic rules.

[0009] The system retrieves real-time business status data associated with the business parameters by searching the target business logic rules, business parameters, and supply chain business database. Based on the real-time business status data and the target business logic rules, the system determines the supply chain business information.

[0010] Based on supply chain business information and target business logic rules, structured traceability information containing supply chain business information is generated. The structured traceability information and query requests are input into the large language model to obtain supply chain business query text.

[0011] In one embodiment, based on a pre-built supply chain knowledge graph, business query intents and business parameters are matched to obtain target business logic rules, including:

[0012] Based on the business scenario classification identifiers of the business query intent, the corresponding matching scenario root nodes are searched in the scenario tag index of the supply chain knowledge graph to obtain a list of business scenario root nodes.

[0013] Based on the core business entity name of the business parameters, the corresponding standardized entity node identifier is searched in the entity attribute index of the supply chain knowledge graph to obtain the business entity node sequence.

[0014] The boundaries of the restricted search subgraph are obtained by identifying and processing the business entity node sequence, the supply chain knowledge graph and the corresponding knowledge graph structure.

[0015] Path retrieval is performed based on the boundaries of the restricted search subgraph, the list of root nodes of the business scenario, and the supply chain knowledge graph to obtain a set of candidate logical paths;

[0016] The candidate logical path set is validated and orchestrated to obtain the target business logic rules.

[0017] In one embodiment, permission verification and orchestration are performed on the candidate logical path set to obtain the target business logic rules, including:

[0018] Based on the access control policies of supply chain business users, node reachability verification is performed on each path in the candidate logical path set to obtain the standard logical path set.

[0019] Based on the relation edge types and node attributes contained in each path in the standard logical path set, the corresponding logical execution units are matched in the metadata attribute library of the supply chain knowledge graph to obtain the unit mapping relationship table.

[0020] Based on the unit mapping relationship table and the node traversal order in the standard logical path set, the logical execution units are connected and arranged according to the business logic relationship to construct an executable logical arrangement instruction set that includes the unit execution order, parameter passing rules and logical branch conditions.

[0021] Based on the rule attributes associated with the matched logical execution units in the executable logic orchestration instruction set and the supply chain knowledge graph, the system basis node identifier, business constraints and risk threshold parameters referenced by each logical execution unit are extracted from the executable logic orchestration instruction set to obtain the target business logic rules.

[0022] In one embodiment, the target business logic rule includes computational constraints; based on real-time business status data and the target business logic rule, supply chain business information is determined, including:

[0023] A standardized business dataset is obtained by semantic matching and mapping of data patterns based on field identifiers and calculation constraints in real-time business status data.

[0024] Based on the logical dependencies in the computational constraints, the execution order and data flow path of each business rule in the supply chain knowledge graph are analyzed, and a directed acyclic graph execution plan containing input nodes, computation nodes and output nodes is constructed to obtain a deterministic computation execution topology.

[0025] Based on the deterministic computation execution topology, context initialization is performed in an isolated sandbox environment to obtain an isolated empty computation framework, and a standardized business dataset is added to the input node in the isolated empty computation framework to obtain the computation environment to be executed;

[0026] Based on the logical execution units in the computing environment and computing constraints to be executed, the business logic processing units corresponding to each computing node are instantiated to obtain the bound logical execution instances.

[0027] Deterministic execution and result encapsulation are performed based on logical execution instances, deterministic computation execution topology, and compliance thresholds and risk warning rules in computational constraints to obtain supply chain business information.

[0028] In one embodiment, deterministic execution and result encapsulation are performed based on logical execution instances, deterministic computation execution topology, and compliance thresholds and risk warning rules in computational constraints to obtain supply chain business information, including:

[0029] Based on the traversal order corresponding to the deterministic computation execution topology, business logic operations are performed on each logical execution instance in sequence to obtain a sequence of intermediate operation results;

[0030] Based on the intermediate operation result sequence and the compliance threshold and risk warning rules in the calculation constraints, logical branch judgments are made on the business indicators to be verified in the supply chain business information to obtain a business status judgment mark set.

[0031] According to the preset result aggregation strategy, the business status judgment mark set and intermediate calculation result sequence are aggregated to obtain preliminary results containing specific numerical conclusions, status descriptions and anomaly indicators.

[0032] Based on the preliminary results of the result-oriented approach and the node execution logs in the deterministic calculation execution topology, the rule identifiers, data source fields, and calculation paths hit by each logical operation are extracted and encapsulated to obtain supply chain business information.

[0033] In one embodiment, the target business logic rule includes rule metadata; based on the supply chain business information and the target business logic rule, structured traceability information containing the supply chain business information is generated, including:

[0034] Based on the execution session identifier in the supply chain business information, the log entries in the pre-built computation execution log storage area are searched to obtain the original execution trajectory sequence containing the input and output parameters and call sequence of each logical operation step, and the data access path identifier and rule call record are extracted from the original execution trajectory sequence.

[0035] A reverse mapping query is performed based on the rule call records and the rule index table of the supply chain knowledge graph to obtain the hit rule nodes; and the rule version identifier and business logic description text are extracted from the hit rule nodes to obtain a set of hit rule nodes containing standard version basis and interpretable logic text.

[0036] A reverse mapping query is performed based on the data access path identifier and the data entity index table of the supply chain knowledge graph to obtain the source data nodes; and the data field names and measurement unit standards of the source data nodes are parsed to obtain a set of source data nodes containing standardized data semantics and unified measurement benchmarks.

[0037] Based on the set of hit rule nodes, the set of source data nodes, the business query intent, and the supply chain business information, a graph is dynamically constructed to obtain structured traceability information.

[0038] In one embodiment, a graph is dynamically constructed based on the hit rule node set, the source data node set, the business query intent, and the supply chain business information to obtain structured traceability information, including:

[0039] Based on the hit rule node set, source data node set, business query intent and supply chain business information, the graph structure object is initialized and the core logic node is instantiated to build a blank topology framework.

[0040] Based on the intent mapping relationship in the blank topology framework and rule metadata, a directed semantic edge is established in the graph structure object from the core logical node of the business query intent to the core logical node of the target business logic rule, thus obtaining the intent rule association path;

[0041] Based on the data dependency time sequence and intent rule association path in the original execution trajectory sequence, a directed data input edge is established in the graph structure object, pointing from the core logical node of the source business data to the core logical node of the target business logic rule, to obtain the rule data association path;

[0042] Based on the generation logic and rule data association path of supply chain business information, a directed derivation output edge is established in the graph structure object, pointing from the core logic node of the target business logic rule to the core logic node of the supply chain business information. The calculated values ​​and status markers are written into the attribute fields of the directed derivation output edge to obtain a connected subgraph containing a complete causal chain.

[0043] Based on the node topology and edge attribute fields in the connected subgraph, the format is converted according to the preset graph structure serialization template to obtain a logical interpretation graph containing complete supply chain business information, and the logical interpretation graph is identified as structured traceability information.

[0044] Secondly, this application also provides a supply chain business information query device, including:

[0045] The semantic parsing module is used to input query requests for supply chain business information into the large language model for semantic parsing to obtain the business query intent and business parameters;

[0046] The matching processing module is used to match business query intents and business parameters based on a pre-built supply chain knowledge graph to obtain target business logic rules.

[0047] The retrieval module is used to retrieve real-time business status data associated with business parameters based on target business logic rules, business parameters, and the supply chain business database, and to determine supply chain business information based on the real-time business status data and target business logic rules.

[0048] The processing module is used to generate structured traceability information containing supply chain business information based on supply chain business information and target business logic rules, and input the structured traceability information and query requests into the large language model to obtain supply chain business query text.

[0049] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method in any of the embodiments of the first aspect described above.

[0050] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method in any of the embodiments of the first aspect described above.

[0051] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the method in any of the embodiments of the first aspect described above.

[0052] The aforementioned supply chain business information query method, apparatus, computer equipment, and medium include: inputting a query request for supply chain business information into a large language model for semantic parsing to obtain the business query intent and business parameters; performing matching processing on the business query intent and business parameters based on a pre-constructed supply chain knowledge graph to obtain target business logic rules; retrieving real-time business status data associated with the business parameters based on the target business logic rules, business parameters, and the supply chain business database; determining supply chain business information based on the real-time business status data and the target business logic rules; generating structured traceability information containing supply chain business information based on the supply chain business information and the target business logic rules; and inputting both the structured traceability information and the query request into the large language model to obtain the supply chain business query text. The aforementioned method inputs supply chain business information query requests into a large language model for deep semantic parsing, accurately decomposing the business query intent and standardized business parameters. This avoids the limitations of shallow matching based on simple vector text similarity in traditional technologies. Furthermore, by performing association matching on the business query intent and business parameters based on the supply chain knowledge graph, it accurately identifies the target business logic rules suitable for this query, extracting the underlying business calculation logic and constraints from the source. This overcomes the problem that vector retrieval cannot uncover implicit business rules and logical relationships. Simultaneously, based on the target business logic rules and business parameters, it conducts targeted retrieval of the supply chain business database to obtain matching real-time business status data. Finally, it strictly calculates and outputs supply chain business information based on real-time business data and target business logic rules, using standardized business rules as rigid constraints throughout the entire numerical calculation process. The method improves the accuracy of acquired supply chain business information by avoiding numerical deviations and process judgment errors caused by the unconstrained autonomous generation of large models. Furthermore, it generates structured traceability information containing supply chain business information based on supply chain business information and target business logic rules. This ensures complete retention of the entire process of data retrieval, parameter substitution, and rule calculation, supplementing the traceability credentials required for compliance verification. The original query request and the structured traceability information carrying complete business, traceability, and rule information are input into the large language model, providing it with clear and structured business logic, computational data, and traceability basis as generation constraints. This significantly suppresses model illusions, enhances compliance rigor, and ultimately outputs reliable and accurate supply chain business query text, ensuring the business needs for refined supply chain business queries and reliable decision support. Attached Figure Description

[0053] Figure 1 This is a flowchart illustrating a supply chain business information query method in one embodiment of this application;

[0054] Figure 2 This is a flowchart illustrating the matching process between business query intent and business parameters in another embodiment of this application;

[0055] Figure 3 This is a flowchart illustrating the process of implementing permission verification and orchestration for a candidate logical path set in another embodiment of this application;

[0056] Figure 4 This is a flowchart illustrating the process of determining supply chain business information based on real-time business status data and target business logic rules in another embodiment of this application;

[0057] Figure 5 This is a flowchart illustrating the process of deterministic execution and result encapsulation based on logical execution instances, deterministic computation execution topology, and compliance thresholds and risk warning rules in computational constraints, as described in another embodiment of this application.

[0058] Figure 6 This is a flowchart illustrating the process of generating structured traceability information containing supply chain business information based on supply chain business information and target business logic rules in another embodiment of this application.

[0059] Figure 7 This is a schematic diagram illustrating the process of dynamically constructing a graph based on a set of hit rule nodes, a set of source data nodes, business query intent, and supply chain business information, as described in another embodiment of this application.

[0060] Figure 8 This is a structural block diagram of a supply chain business information query device in one embodiment of this application;

[0061] Figure 9 This is an internal structural diagram of a computer device in one embodiment of this application. Detailed Implementation

[0062] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0063] The supply chain business information query method provided in this application embodiment can be applied to computer devices and is suitable for intelligent question-and-answer scenarios in supply chain business. The computer device can be, but is not limited to, various personal computers, laptops, smartphones, tablets, and servers. The server can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0064] In one exemplary embodiment, such as Figure 1 As shown, a method for querying supply chain business information is provided. Taking the application of this method to computer equipment as an example, the method includes the following steps:

[0065] Step S100: Input the query request for supply chain business information into the large language model for semantic parsing to obtain the business query intent and business parameters.

[0066] The query requests for the aforementioned supply chain business information can be entered by the user in real time or automatically triggered by computer equipment.

[0067] Specifically, computer equipment can obtain query requests for supply chain business information and call a pre-trained large language model specifically for the supply chain domain. After inputting the query request into the large language model for semantic parsing, it outputs the business query intent and business parameters.

[0068] It should be noted that during the parsing process, the large language model can perform semantic understanding on a word-by-word basis on the text content of the query request, identifying core keywords, business-related expressions, logical relationships, and query constraints. It removes redundant information, modal particles, and invalid expressions to clarify the user's core query purpose and thus obtain the business query intent. Simultaneously, it extracts specific business information directly related to the business query intent, which can be used for subsequent retrieval matching and data calculation—i.e., business parameters—from the query request. In this embodiment, the aforementioned query request can be a natural language query request.

[0069] Optionally, the business query intent can be understood as the specific purpose that the user wants to achieve through the query request, pointing to the category of supply chain business information that needs to be obtained, such as querying the inventory status of any batch of goods, the fulfillment progress of any order, the procurement cost of any type of material, or risk warning information in the supply chain.

[0070] Meanwhile, each query request corresponds to a unique core business query intent; and the business parameters are specific supply chain business data information that supports the realization of the business query intent, and their types correspond to the business query intent. For example, when the business query intent is the inventory status of any batch of goods, the corresponding business parameters include the name, code, storage warehouse number, query time node, etc. of the batch of goods; when the business query intent is the fulfillment progress of any order, the corresponding business parameters include the order number, order signing date, supplier name, purchased material name, etc.

[0071] If the user's query request is: "Query the actual inventory quantity of raw material with number KC20260509001 in warehouse number CC003 on XX / XX / 20XX", then the query request will be input into the large language model for semantic parsing to identify the core keywords as "raw material", "number KC20260509001", "warehouse number CC003", "XX / XX / 20XX", and "actual inventory quantity", clarifying that the business query intent is "to obtain the actual inventory quantity of a specific number of raw material in a specific warehouse at a specific time point"; at the same time, the corresponding business parameters will be extracted as: raw material number "KC20260509001", warehouse number "CC003", and query time "XX / XX / 20XX".

[0072] Step S200: Based on the pre-built supply chain knowledge graph, the business query intent and business parameters are matched to obtain the target business logic rules.

[0073] It should be noted that the aforementioned supply chain knowledge graph can be a structured knowledge base for supply chain business, or a structured knowledge graph built based on a graph database of supply chain business.

[0074] Specifically, computer equipment can employ matching algorithms, based on supply chain knowledge graphs, to match business query intents and business parameters to obtain target business logic rules. Optionally, the aforementioned matching algorithms can be, but are not limited to, string matching algorithms, fuzzy similarity matching algorithms, vector semantic matching algorithms, rule-based structured information matching algorithms, and graph structured information matching algorithms.

[0075] Among them, the supply chain knowledge graph is obtained by structuring and sorting out the entire business process and all business elements of the supply chain, which can clearly present the relationship between various business links and business objects; and the supply chain knowledge graph stores the logical rule information corresponding to various supply chain businesses.

[0076] In this embodiment of the application, the computer device can use the business query intent as the core search direction and the business parameters as specific search conditions to perform accurate search and matching in the supply chain knowledge graph to obtain the target business logic rules.

[0077] Optionally, the aforementioned target business logic rules may include at least a data processing flow, computational constraints, and rule metadata; the data processing flow includes the logic and sequence of data acquisition, data verification, and data computation required to complete the supply chain business information query; computational constraints may include the constraints that must be followed when performing deterministic calculations; rule metadata includes data describing the rule's own attributes, such as rule number, applicable scenarios, effective time, and related business processes.

[0078] Step S300: Based on the target business logic rules, business parameters and supply chain business database, retrieve the real-time business status data associated with the business parameters, and determine the supply chain business information based on the real-time business status data and the target business logic rules.

[0079] Optionally, the aforementioned supply chain business database stores data related to all business links of the supply chain. It can be an independent relational database, or it can be a structured knowledge base of the supply chain formed by extracting, cleaning and integrating unstructured and structured data from all business links of the supply chain. The aforementioned supply chain business database can include various business data such as inventory, orders, procurement, logistics and fulfillment, and the data therein can be updated synchronously in real time.

[0080] Specifically, the computer equipment can acquire a pre-trained retrieval model, and then input the target business logic rules, business parameters, and supply chain business database into the retrieval model for retrieval. After retrieval, it outputs real-time business status data related to the business parameters from the supply chain business database. Optionally, the aforementioned retrieval model can be at least one of the following: convolutional neural network model, fully connected neural network model, residual neural network model, recurrent recurrent neural network model, long short-term memory neural network model, etc.

[0081] In practical applications, computer equipment can retrieve data processing flows and business parameters from the target business logic rules within the supply chain business database. Specifically, it determines the search scope within the supply chain business database based on the business parameters, locates relevant data and tables, and then retrieves the required business data according to the data processing flow. This preprocessed business data is then cleaned, and invalid, duplicate, and abnormal data are removed to obtain preprocessed business data. Finally, the preprocessed business data is formatted to meet the requirements of the computational constraints, resulting in real-time business status data. This real-time business data can be the latest business data, and its type corresponds to the business query intent and business parameters.

[0082] Meanwhile, after obtaining real-time business status data, deterministic calculations can be performed on the real-time business status data according to the calculation constraints in the target business logic rules. That is, unambiguous and approximate deterministic numerical calculations and logical judgments are performed, making the calculation process traceable and verifiable, ensuring that the calculation results are accurate and unique, and ultimately obtaining supply chain business information.

[0083] Optionally, the aforementioned computational constraints are all the rules, standards, and limitations that must be followed when performing numerical calculations, logical judgments, and data processing in supply chain operations. These may include compliance thresholds, risk warning rules, logical dependencies, and data patterns. Specifically, the data pattern represents the format, type, and value range of real-time business status data; logical dependencies represent the logical relationships between data points in deterministic calculations, used to clarify the order and requirements of association; compliance thresholds are numerical standards used to determine whether the calculation results meet compliance requirements, such as inventory warning thresholds and procurement cost limits; and risk warning rules can be rules for identifying potential business risks, which will be marked in the supply chain business information when triggered.

[0084] Step S400: Based on supply chain business information and target business logic rules, generate structured traceability information containing supply chain business information, and input the structured traceability information and query request into the large language model to obtain supply chain business query text.

[0085] Specifically, the computer equipment can acquire a pre-trained algorithm model, then input the supply chain business information and target business logic rules into the algorithm model for processing, and output structured traceability information containing supply chain business information. The structured traceability information and query request are then input into a large language model to obtain the supply chain business query text.

[0086] Simultaneously, computer equipment can associate rule metadata in the target business logic rules with supply chain business information, and retrieve pre-stored computational storage logs. It can extract and trace the entire data flow path corresponding to the supply chain business information from the computational storage logs in reverse, so as to fully cover the entire flow process from the real-time business status data retrieval stage, the execution stage of each computational constraint condition, the business parameter substitution calculation stage, to the final output of the target supply chain business information. Then, based on the entire data flow path corresponding to the supply chain business information, it can construct structured traceability information of the supply chain business information.

[0087] Furthermore, both structured traceability information and query requests can be input into a large language model to obtain supply chain business query text. This large language model can generate supply chain business query text that is fluent, logically rigorous, accurate in results, and includes traceability explanations; the large language model can be DeepSeekR1 or Qwen3, etc. Optionally, the supply chain business query text can directly respond to the supply chain business user's query request in natural language form, containing core information about the supply chain business, fully responding to the business query intent, and clearly presenting the key content of the structured traceability information, informing the user of the source and basis of the results.

[0088] The technical solution in this application embodiment inputs the query request for supply chain business information into a large language model for semantic parsing to obtain the business query intent and business parameters. Based on a pre-built supply chain knowledge graph, the business query intent and business parameters are matched to obtain the target business logic rules. Based on the target business logic rules, business parameters, and the supply chain business database, real-time business status data associated with the business parameters is retrieved. Based on the real-time business status data and the target business logic rules, the supply chain business information is determined. Based on the supply chain business information and the target business logic rules, structured traceability information containing the supply chain business information is generated. Both the structured traceability information and the query request are input into the large language model to obtain the supply chain business query text. This method inputs the query request for supply chain business information into a large language model for deep semantic parsing, accurately decomposing the business query intent and standardized business parameters. This avoids the limitations of shallow matching based on simple vector text similarity in traditional technologies. Furthermore, by performing association matching on the business query intent and business parameters based on the supply chain knowledge graph, the target business logic rules suitable for this query are accurately identified, extracting the underlying business computation logic from the source. This method addresses the limitations of vector retrieval in uncovering implicit business rules and logical relationships. It employs a targeted retrieval of the supply chain business database based on target business logic rules and parameters to obtain matching real-time business status data. Supply chain business information is then calculated and output strictly according to this real-time data and target business logic rules. Standardized business rules serve as rigid constraints for numerical calculations and business judgments throughout the process, preventing numerical deviations and process errors caused by the large model's unconstrained generation of data and improving the accuracy of the obtained supply chain business information. Furthermore, the method generates structured traceability information containing supply chain business information based on the supply chain business information and target business logic rules. It fully retains the data retrieval, parameter substitution, and rule calculation process, supplementing the traceability credentials required for compliance verification. The original query request and the structured traceability information carrying complete business, traceability, and rule information are input into the large language model, providing clear and structured business logic, computational data, and traceability basis as generation constraints. This significantly suppresses model illusions, enhances compliance rigor, and ultimately outputs reliable and accurate supply chain business query text, ensuring the business needs for refined supply chain business queries and reliable decision support.

[0089] The following describes the process of matching business query intents and business parameters based on the pre-built supply chain knowledge graph to obtain target business logic rules. In one embodiment, as... Figure 2 As shown, the process in step S200 above can be implemented in the following way:

[0090] Step S210: Based on the business scenario classification identifier of the business query intent, search for the corresponding matching scenario root node in the scenario tag index of the supply chain knowledge graph to obtain a list of business scenario root nodes.

[0091] Specifically, computer equipment can use a matching algorithm to search for the corresponding matching scenario root node in the scenario tag index of the supply chain knowledge graph based on the business scenario classification identifier of the business query intent, and obtain a list of business scenario root nodes.

[0092] In practical applications, computer devices can extract the corresponding business scenario classification identifiers based on the business query intent, and use the business scenario classification identifiers as search keywords to find the corresponding matching scenario root nodes in the scenario tag index of the supply chain knowledge graph, thereby obtaining a list of business scenario root nodes.

[0093] Among them, the above-mentioned business scenario classification identifiers are used to uniquely distinguish the scenario to which the business query intent of the supply chain business belongs. They correspond one-to-one with the scenario tags in the supply chain knowledge graph, and each business scenario classification identifier corresponds to a specific business scenario, such as inventory query, order fulfillment query, procurement cost query, risk warning query, etc. At the same time, the above-mentioned scenario tag index is an index structure that pre-stores all business scenario tags and corresponding scenario root nodes, with each scenario tag corresponding to a scenario root node.

[0094] During the search process, the business scenario classification identifiers are precisely matched with all scenario tags stored in the scenario tag index of the supply chain knowledge graph. Completely matching scenario tags are then selected, and the scenario root node associated with each scenario tag is obtained. All scenario root nodes are then organized into a business scenario root node list. This business scenario root node list contains all sets of scenario root nodes that match the business query intent. If the business query intent includes a single scenario, the business scenario root node list contains only one scenario root node; if the business scenario root node list includes multiple associated scenarios, the business scenario root node list contains multiple corresponding scenario root nodes.

[0095] For example, if the obtained business query intent is "to obtain the actual inventory quantity of a raw material with a specific number in a specific warehouse at a specific time node", the business scenario is an inventory query scenario, and the corresponding business scenario classification identifier is "KC-CX-001" (KC-CX represents inventory query, and 001 is the unique code for the scenario). Then, the business scenario classification identifier "KC-CX-001" is extracted and searched in the scenario tag index of the supply chain knowledge graph. The scenario tag index stores the corresponding scenario tag "Inventory Query Scenario", and the scenario root node associated with this scenario tag is the "Inventory Management Root Node" (the top-level node of the inventory query scenario, associated with all entity nodes and rule nodes related to inventory management). After the search is completed, the scenario root nodes are organized into a business scenario root node list: [Inventory Management Root Node].

[0096] Step S220: Based on the core business entity name of the business parameters, search for the corresponding standardized entity node identifier in the entity attribute index of the supply chain knowledge graph to obtain the business entity node sequence.

[0097] Among them, computer equipment can use a matching algorithm to find the corresponding standardized entity node identifier in the entity attribute index of the supply chain knowledge graph based on the core business entity name of the business parameters, and obtain the business entity node sequence.

[0098] Simultaneously, the computer equipment can extract the corresponding core business entity names based on business parameters and use these names as search keywords to find the corresponding standardized entity node identifiers in the entity attribute index of the supply chain knowledge graph. All found standardized entity node identifiers are then organized according to the logical order of the business parameters to generate a business entity node sequence. During the search process, the core business entity name is precisely matched against all business entity names stored in the entity attribute index. If a completely matching name exists, the corresponding standardized entity node identifier from the entity attribute index is retrieved; otherwise, if aliases or abbreviations exist, a synonym match is performed between the core business entity name and all business entity names stored in the entity attribute index to ensure accurate identification of the corresponding standardized entity node identifier.

[0099] Optionally, the aforementioned core business entity name is the name representing the core business object in the supply chain in the business parameters, including raw material name, warehouse name, order name, supplier name, etc.; the aforementioned entity attribute index is an index structure in the supply chain knowledge graph used to map the core business entity name to standardized graph nodes; the aforementioned standardized entity node identifier is a fixed identifier that uniquely marks the business entity node in the supply chain knowledge graph.

[0100] For example, if the obtained business parameters are: raw material number "KC20260509001", warehouse number "CC003", and query time "20XX / XX / XX", the extracted core business entity names are "KC20260509001 raw material" and "CC003 warehouse". Using these two names as search keywords, the entity attribute index is searched for: "KC20260509001 raw material", which yields the standardized entity node identifier "SY-0001" (SY represents the raw material entity, and 0001 is the unique entity code); and "CC003 warehouse" yields the standardized entity node identifier "CK-003" (CK represents the warehouse entity, and 003 is the unique entity code). Finally, according to the logical association order of the core business entities in the business parameters (first raw material, then warehouse), these two standardized entity node identifiers are arranged to form a business entity node sequence: [SY-0001, CK-003].

[0101] Step S230: Based on the business entity node sequence, supply chain knowledge graph and corresponding knowledge graph structure, identification processing is performed to obtain the boundary of the restricted search subgraph.

[0102] Specifically, the computer equipment can employ information recognition algorithms to identify and process the business entity node sequence, supply chain knowledge graph, and corresponding knowledge graph structure to obtain the boundary of the restricted search subgraph. Optionally, the aforementioned information recognition algorithm can be a text recognition algorithm, a semantic recognition algorithm, an image recognition algorithm, etc.

[0103] In practical applications, computer equipment can select the business entity node that is most directly related to the business query intent and occupies a core position from the supply chain knowledge graph as the starting entity node. If there are multiple business entity nodes with the same position, the first business entity node in the business entity node sequence is taken as the starting entity node. With the starting entity node as the core, the domain and associated node range of the starting entity node are identified in the supply chain knowledge graph, and the boundary of the business domain subgraph that meets the preset boundary conditions is determined, i.e., the boundary of the restricted search subgraph.

[0104] Optionally, the aforementioned starting entity node may include a subgraph structure containing all related nodes (entity nodes, relationship nodes, rule nodes, etc.) directly or indirectly associated with the starting entity node, corresponding to a clearly defined supply chain business domain. The aforementioned boundary conditions may include the range of subsequent retrieval logic rule nodes; performing retrieval within this range effectively excludes irrelevant nodes, thereby improving retrieval efficiency.

[0105] For example, if the obtained business entity node sequence is "[SY-0001, CK-003]", then the starting entity node is located based on the relationship between nodes in the supply chain knowledge graph: the business query intent is to obtain the inventory quantity of a specific raw material in a specific warehouse, and the raw material is the core object of the query. Therefore, the first business entity node "SY-0001" (the standardized entity node identifier corresponding to raw material KC20260509001) in the business entity node sequence is located as the starting entity node; at the same time, based on the starting entity node "SY-0001", the business domain subgraph to which it belongs is identified as follows: the starting entity node belongs to the raw material inventory management domain, and the subgraph includes all related nodes such as warehouse nodes, inventory data nodes, and inventory rule nodes related to the raw material; further, with SY-0001 as the core, the scope of all warehouse nodes, inventory data nodes, inventory rule nodes and related relationship nodes is associated to determine the boundary of the business domain subgraph, that is, the boundary of the restricted search subgraph.

[0106] Step S240: Path retrieval is performed based on the boundary of the restricted search subgraph, the list of root nodes of the business scenario, and the supply chain knowledge graph to obtain a set of candidate logical paths.

[0107] Specifically, the computer equipment can employ retrieval algorithms to perform path retrieval based on the boundaries of the restricted search subgraph, the list of root nodes in the business scenario, and the supply chain knowledge graph, thereby obtaining a set of candidate logical paths. Optionally, the aforementioned retrieval algorithms can be string matching, semantic information comparison, text matching, etc.

[0108] In this embodiment, the retrieval direction (only retrieving paths related to the scenario root node) can be determined within the bounded search subgraph based on the scenario root node in the business scenario root node list. All associated directed edges of the starting entity node are traversed in the supply chain knowledge graph, and the nodes connected by each edge are tracked. For any node, it is determined whether the node is a potential rule node: if so, the path from the starting entity node to that node is determined as a candidate logical path; if not, the associated directed edges of that node are continued to be tracked until a potential rule node is found or the bounded search subgraph boundary is exceeded. All retrieved reachable potential rule nodes are determined as a set of candidate logical paths. Here, a reachable potential rule node refers to a complete path in the supply chain knowledge graph that starts from the starting entity node and can be directly or indirectly connected to a potential rule node through one or more directed edges.

[0109] Among them, the aforementioned potential rule nodes are the core nodes in the supply chain knowledge graph that store business logic rules. Each potential rule node corresponds to at least one business logic rule and is associated with related entity nodes through directed edges.

[0110] For example, if the business scenario root node list is “[Inventory Management Root Node]”, the restricted search subgraph boundary is “the range of all warehouse nodes, inventory data nodes, inventory rule nodes and related relationship nodes with SY-0001 as the core”, and the starting entity node is “SY-0001”, then the computer device will take “SY-0001” as the starting point and determine the search direction as the inventory management related path within the restricted search subgraph boundary according to the “Inventory Management Root Node”. Specifically, all associated directed edges of “SY-0001” are traversed: the first directed edge connects to the warehouse node “CK-003”, which is a non-rule node. Continuing to trace its associated directed edges, it is found that “CK-003” connects to the inventory rule node “GZ-001” (GZ represents a rule node, and 001 is the unique code of the rule node). The path is recorded as: SY-0001→CK-003→GZ-001; the second directed edge directly connects to the inventory rule node “GZ-001”. The path is recorded as: SY-0001→GZ-001; these two reachable potential rule nodes are constructed into a candidate logical path set: [SY-0001→CK-003→GZ-001, SY-0001→GZ-001].

[0111] Step S250: Perform permission verification and orchestration on the candidate logical path set to obtain the target business logic rules.

[0112] In practical applications, computer devices can acquire pre-trained algorithm models, then input the candidate logical path set into the algorithm model for permission verification and orchestration, and finally output the target business logic rules.

[0113] The technical solution in this application embodiment, based on the business scenario classification identifier of the business query intent, searches for the corresponding matching scenario root node in the scenario tag index of the supply chain knowledge graph to obtain a list of business scenario root nodes. Based on the core business entity name of the business parameters, it searches for the corresponding standardized entity node identifier in the entity attribute index of the supply chain knowledge graph to obtain a sequence of business entity nodes. Recognition processing is performed based on the business entity node sequence, the supply chain knowledge graph, and the corresponding knowledge graph structure to obtain the boundary of the restricted search subgraph. Path retrieval is performed based on the boundary of the restricted search subgraph, the list of business scenario root nodes, and the supply chain knowledge graph to obtain a set of candidate logical paths. Permission verification and orchestration are then performed on the candidate logical path set to obtain the target business logic rule. The above method implements a hierarchical progressive retrieval logic, quickly matching scenarios and entity nodes through multi-level indexes, progressively compressing the retrieval scope of the supply chain knowledge graph, and avoiding the high computational consumption caused by full graph traversal. Simultaneously, the above method unifies the semantics of business entities and adds a permission verification mechanism, which not only improves the retrieval efficiency and matching accuracy of business logic rules but also ensures the permission compliance of business rule calls. Finally, it quickly outputs standardized target business logic rules that are adapted to the current query scenario and business entity and can directly participate in calculations.

[0114] The following describes the process of permission verification and orchestration for the aforementioned candidate logical path set. In one embodiment, as follows... Figure 3 As shown, the process in step S250 above can be implemented in the following way:

[0115] Step S251: Based on the access control policy of supply chain business users, perform node reachability verification on each path in the candidate logical path set to obtain the standard logical path set.

[0116] Optionally, the access control policy mentioned above can be a pre-defined set of rules used to regulate the access permissions of different supply chain business users to supply chain knowledge graph nodes, that is, to limit the scope of supply chain knowledge graph nodes that business users of different positions and different permission levels can access and the scope of business operations they can perform.

[0117] The computer equipment can perform matching, comparison, and analysis on each path in the candidate logical path set based on the access control policies of supply chain business users, in order to achieve node reachability verification and obtain a standard logical path set. The standard logical path set may include at least one candidate logical path from the candidate logical path set.

[0118] In this embodiment, the computer device can determine the corresponding access control policy based on the supply chain business user, and then perform node reachability verification on each path in the candidate logical path set according to the determined access control policy. This involves checking each candidate logical path in the candidate logical path set to determine whether all nodes on the candidate logical path (including the starting entity node, intermediate associated nodes, and potential rule nodes) are within the access permission range of the supply chain business user, and simultaneously verifying whether the supply chain business user has permission to access the potential rule nodes from the starting entity node through the candidate logical path. If all nodes on any candidate logical path are within the access permission range of the supply chain business user, and the connection relationship between all nodes meets the access control policy requirements, then the verification passes, and the candidate logical path is retained. If at least one node on any candidate logical path is not within the access permission range of the supply chain business user, or the connection relationship between all nodes does not allow the supply chain business user to access, then the verification fails, and the candidate logical path is eliminated. All candidate logical paths that pass verification are combined to form a compliant logical path set, i.e., a standard logical path set.

[0119] For example, if the candidate logical path set is “[SY-0001→CK-003→GZ-001, SY-0001→GZ-001]”, and the supply chain business user is an inventory management personnel with the user ID “KC-YW-001” and the corresponding permission level is inventory management level 2 permission, then the node reachability verification of the two candidate logical paths in the candidate logical path set can be described as follows: The first candidate logical path is “SY-0001→CK-003→GZ-001”, and the nodes on this candidate logical path are SY-0001 (raw material entity node) and CK-003 (warehouse entity node). The first candidate logical path, "SY-0001→GZ-001", consists of two nodes related to inventory management. The directed edges between these nodes meet the access permission requirements of inventory management personnel, and the verification passes. The second candidate logical path, "SY-0001→GZ-001", also has nodes related to inventory management, and the directed edges between these nodes meet the access permission requirements, thus passing the verification. These two verified candidate logical paths are then combined into a standard logical path set, specifically "[SY-0001→CK-003→GZ-001, SY-0001→GZ-001]". The access control policy stipulates that supply chain business users with this access level can only access inventory management-related nodes and candidate logical paths, and are prohibited from accessing nodes in other areas such as procurement and logistics.

[0120] Step S252: Based on the relation edge types and node attributes contained in each path in the standard logical path set, match the corresponding logical execution units in the metadata attribute library of the supply chain knowledge graph to obtain the unit mapping relationship table.

[0121] The computer equipment can use a matching algorithm to match the corresponding logical execution unit in the metadata attribute library of the supply chain knowledge graph based on the relation edge type and node attributes contained in each path in the standard logical path set, and obtain the unit mapping relationship table.

[0122] Simultaneously, computer equipment can extract the relationship edge types and node attributes contained in each standard logical path in the supply chain knowledge graph based on each standard logical path in the standard logical path set. Among them, the relationship edge type refers to the category of association represented by the directed edge connecting two nodes, including computational relationships, constraint relationships, dependency relationships, and verification relationships; node attributes refer to the unique characteristic information of each node, including material type attributes, cycle calculation attributes, compliance verification attributes, and risk warning attributes, etc.

[0123] Furthermore, using each relationship edge type and its corresponding node attribute as matching keywords, a search is performed in the metadata attribute database of the supply chain knowledge graph to find the logical execution unit that perfectly corresponds to the relationship edge type and node attribute. This metadata attribute database refers to a database within the supply chain knowledge graph specifically used to store all node attributes, relationship edge types, and corresponding logical execution units. For multiple relationship edges and nodes within the same standard logical path, the corresponding logical execution units are matched separately, establishing an association between each standard logical path and its corresponding logical execution unit. All standard logical paths, relationship edge types, node attributes, and matched logical execution units are then combined to form a unit mapping table. This logical execution unit is a standardized functional unit capable of performing specific supply chain business logic operations, and each logical execution unit corresponds to a specific relationship edge type and node attribute, enabling it to implement specific business logic processing functions.

[0124] For example, if the standard logical path set is “[SY-0001→CK-003→GZ-001, SY-0001→GZ-001]”, the relationship edge type and node attributes of each standard logical path are broken down one by one. Specifically, the first standard logical path “SY-0001→CK-003→GZ-001” contains two relationship edges: the first relationship edge connects SY-0001 (raw material entity node, with attributes of “raw material, number SY-0001, and domain inventory management”) and CK-003 (warehouse entity node, with attributes of “warehouse, number CK-0001”) to CK-003 (warehouse entity node, with attributes of “warehouse, number CK-0001”). 3. The first relationship edge connects CK-003 (warehouse entity node) and GZ-001 (potential rule node, with attributes of "rule node, rule type inventory query, and domain inventory management"), and the relationship edge type is "applicable rule". The second standard logical path "SY-0001→GZ-001" contains a relationship edge that connects SY-0001 (raw material entity node) and GZ-001 (potential rule node), with the relationship edge type being "applicable rule". The node attributes are consistent with the attributes of the corresponding node in the first path.

[0125] Furthermore, using the aforementioned relationship edge types and node attributes as matching keywords, logical execution units are matched in the metadata attribute library of the supply chain knowledge graph: the "stored in" relationship edge type and corresponding entity node attributes are matched to the "inventory data retrieval unit" (function: retrieving inventory-related data of specific raw materials in a specific warehouse); the "applicable rules" relationship edge type and corresponding node attributes are matched to the "inventory rule extraction unit" (function: extracting inventory management logical rules corresponding to a specific entity). The resulting unit mapping relationship table is shown in Table 1 below, as detailed below:

[0126] Table 1 Unit Mapping Relationship Table

[0127]

[0128] Step S253: Based on the unit mapping relationship table and the node traversal order in the standard logical path set, the logical execution units are connected and arranged according to the business logic relationship to construct an executable logical arrangement instruction set that includes the unit execution order, parameter passing rules and logical branch conditions.

[0129] Specifically, the computer device can obtain the corresponding relationships recorded in the unit mapping table, obtain the node traversal order in the standard logical path set, and retrieve the corresponding logical execution unit according to the corresponding relationships recorded in the unit mapping table. Then, according to the node traversal order, the logical execution units are sequentially connected and arranged according to the business logic relationship. That is, the logical execution units corresponding to each standard logical path are arranged in an orderly manner according to the business logic relationship. At the same time, the parameter passing rules and logical branch conditions between each logical execution unit are determined to form an executable logical arrangement instruction set.

[0130] Optionally, the above node traversal order can be understood as the order of nodes traversed from the starting entity node to the reachable potential rule node in the standard logical path; the above business logic relationship can be understood as the inherent correlation between various business operations during the execution of supply chain business, including sequential execution relationships, dependency relationships, and parallel relationships; the above logical orchestration instruction set can include unit execution order, parameter passing rules, and logical branch conditions. Specifically, the above parameter passing rules refer to the principle of passing the output parameters of the previous logical execution unit as the input parameters of the next logical execution unit; the above logical branch conditions determine the execution branch of the logical execution unit based on different business scenarios.

[0131] For example, after obtaining the unit mapping table and the standard logical path set, the node traversal order of each standard logical path in the standard logical path set can be obtained: if the node traversal order of the first standard logical path "SY-0001→CK-003→GZ-001" is "starting entity node SY-0001→intermediate entity node CK-003→potential rule node GZ-001"; and the node traversal order of the second standard logical path "SY-0001→GZ-001" is "starting entity node SY-0001→potential rule node GZ-001", then the node traversal order of the second standard logical path "SY-0001→GZ-001" is "starting entity node SY-0001→potential rule node GZ-001". Rule node GZ-001 is further defined by retrieving the corresponding logical execution unit based on the relationship in the unit mapping table. This unit includes an inventory data retrieval unit and an inventory rule extraction unit. These are then linked and arranged according to the node traversal order and business logic relationships. Specifically: The first standard logical path: The node traversal order determines that the inventory data retrieval unit is executed first (corresponding to the "stored in" relationship between SY-0001 and CK-003), followed by the inventory rule extraction unit (corresponding to the "applicable rule" relationship between CK-003 and GZ-001). The parameter passing rule is that the basic inventory data output by the inventory data retrieval unit serves as the input parameter for the inventory rule extraction unit. The second standard logical path: The node traversal order determines that the inventory rule extraction unit is executed directly (corresponding to the "applicable rule" relationship between SY-0001 and GZ-001), with the raw material number as the input parameter.

[0132] Simultaneously, logical branch conditions are set: if the inventory data retrieval unit fails to retrieve basic inventory data, an exception handling branch is triggered, outputting a "No relevant inventory data found" message; if basic inventory data is retrieved, the inventory rule extraction unit continues execution; if the inventory rule extraction unit fails to extract inventory logic rules, an exception handling branch is triggered, outputting a "No corresponding inventory rule found" message. Ultimately, the constructed executable logic orchestration instruction set can be represented as follows: 1. Path 1 instruction: Input parameters (raw material number SY-0001, warehouse number CK-003) → Execute inventory data retrieval unit → Output basic inventory data → Execute inventory rule extraction unit with basic inventory data as input parameters → Output inventory logic rules; 2. Path 2 instruction: Input parameters (raw material number SY-0001) → Execute inventory rule extraction unit → Output inventory logic rules; 3. Logical branch conditions: If basic inventory data is not retrieved / no inventory logic rule is extracted, an exception handling branch is triggered.

[0133] Step S254: Based on the rule attributes associated with the matched logical execution units in the executable logic orchestration instruction set and the supply chain knowledge graph, extract the system basis node identifier, business constraint conditions and risk threshold parameters referenced by each logical execution unit from the executable logic orchestration instruction set to obtain the target business logic rule.

[0134] Specifically, the computer equipment can employ information extraction algorithms to extract the rule attributes referenced by each logical execution unit from the executable logic orchestration instruction set, based on the rule attributes associated with the matched logical execution units in the supply chain knowledge graph, to obtain the target business logic rules. Optionally, the aforementioned rule attributes may include attribute information associated with each logical execution unit, describing the business rules upon which each logical execution unit is based, including the system basis node identifier, business constraints, risk threshold parameters, etc. Optionally, the aforementioned information extraction algorithm can be a traditional rule-based extraction algorithm, a deep learning-based extraction algorithm, etc.

[0135] In this embodiment, all logical execution units in the executable logic orchestration instruction set and the rule attributes associated with each logical execution unit in the supply chain knowledge graph can be obtained. Each logical execution unit in the executable logic orchestration instruction set is traversed, the rule attributes referenced by the logical execution unit are extracted, and the extracted rule attributes are combined with the unit execution order and parameter passing rules in the executable logic orchestration instruction set to form a data processing flow, calculation constraints and rule metadata. The data processing flow, calculation constraints and rule metadata are then determined as the target business logic rule.

[0136] Referring back to the previous example, the obtained executable logic orchestration instruction set includes an inventory data retrieval unit and an inventory rule extraction unit. The rule attributes associated with these two logic execution units are retrieved from the supply chain knowledge graph. The rule attributes associated with the inventory data retrieval unit can be: institutional basis node identifier (ZD-001), business constraints (the retrieved data must be real-time data from the last 30 days, invalid data is excluded), and risk threshold parameters (none). The rule attributes associated with the inventory rule extraction unit are: institutional basis node identifier (ZD-001), business constraints (the extracted rules must be adapted to the inventory query scenario and comply with inventory management compliance requirements), and risk threshold parameters (inventory warning threshold of 50 units).

[0137] Furthermore, rule attributes referenced by each logical execution unit can be extracted from the executable logic orchestration instruction set, and data processing flow, calculation constraints, and rule metadata can be constructed based on the execution order and parameter passing rules of each logical execution unit. The data processing flow is as follows: 1. Retrieve real-time inventory data for the past 30 days based on raw material number and warehouse number, and remove invalid data; 2. Extract logical rules adapted to the inventory query scenario based on the retrieved inventory data; Calculation constraints are: the retrieved data must be real-time data from the past 30 days; invalid data must be removed; Business constraints are: the extracted rules must comply with inventory management compliance requirements; Risk threshold parameter is: inventory warning threshold of 50 items; Rule metadata is: Rule number: GL2026001; Applicable business scenario: supply chain raw material inventory query; Effective date: XX / XX / 20XX; Related business link: inventory management; System basis node identifier: ZD-001.

[0138] The technical solution in this application embodiment, based on the access control policy of supply chain business users, performs node reachability verification on each path in the candidate logical path set to obtain a standard logical path set. Based on the relationship edge type and node attributes contained in each path in the standard logical path set, it matches the corresponding logical execution unit in the metadata attribute library of the supply chain knowledge graph to obtain a unit mapping relationship table. Based on the unit mapping relationship table and the node traversal order in the standard logical path set, the logical execution units are connected and arranged according to business logic relationships to construct an executable logic orchestration instruction set containing unit execution order, parameter passing rules, and logical branch conditions. Based on the executable logic orchestration instruction set and the rule attributes associated with the matched logical execution units in the supply chain knowledge graph, it starts from the executable logic orchestration... The method extracts the institutional basis node identifiers, business constraints, and risk threshold parameters referenced by each logical execution unit from the instruction set to obtain the target business logic rules. This method can filter paths based on access control policies, eliminating unauthorized or non-compliant paths. Through metadata matching, it automatically transforms the semantics of the supply chain knowledge graph into executable computation units and automatically connects and arranges multiple logical segments according to the node order of the supply chain knowledge graph, eliminating the need for manual arrangement. Furthermore, this method can automatically transform candidate logical paths within the supply chain knowledge graph into standardized, directly executable business rules, with built-in access control and compliance traceability capabilities, reducing manual configuration workload and achieving automatic computation, hierarchical data protection, end-to-end business auditability, and automatic risk management.

[0139] The process of determining supply chain business information based on real-time business status data and target business logic rules is described below. In one embodiment, the target business logic rules include computational constraints; such as... Figure 4 As shown, the process in step S300 above, which determines the supply chain business information based on real-time business status data and target business logic rules, can be achieved in the following way:

[0140] Step S310: Based on the field identifiers in the real-time business status data and the data patterns of the calculation constraints, perform semantic matching and mapping to obtain a standardized business dataset.

[0141] The computer equipment can acquire a pre-trained algorithm model, and then input the field identifiers and data patterns of the calculation constraints in the real-time business status data into the algorithm model for semantic matching and mapping, and output a standardized business dataset.

[0142] In practical applications, computer equipment can use the information matching method to perform semantic matching and mapping based on the field identifiers and calculation constraint data patterns in real-time business status data to obtain a standardized business dataset.

[0143] In this embodiment of the application, the computer device can extract unique field identifiers for all data fields from real-time business status data, extract data patterns from computational constraints, and perform semantic-level precise matching between each field identifier in the real-time business status data and the field definitions specified in the data pattern to establish a mapping relationship between field identifiers and data pattern field definitions.

[0144] Optionally, the above field identifiers can be used to mark the fixed name of the meaning, data type, and data unit of each data item in the real-time business status data.

[0145] Specifically, before executing step S310 above, the computer equipment can standardize the units of measurement for numeric fields in the real-time business status data, extract the current units of measurement for the numeric fields, and convert data with different units of measurement according to the standard units of measurement specified in the data model. The conversion process strictly follows industry-standard unit conversion rules. For non-numeric data, the format and type are verified to ensure they meet the data model requirements. This ensures that the results after semantic matching and mapping are summarized with all data that has been normalized to obtain a standardized business dataset with a unified format, standard fields, and consistent units.

[0146] Step S320: Based on the logical dependencies in the computational constraints, the execution order and data flow path of each business rule in the supply chain knowledge graph are analyzed, and a directed acyclic graph execution plan containing input nodes, computation nodes and output nodes is constructed to obtain a deterministic computation execution topology.

[0147] Optionally, the computer device can acquire a pre-trained analytical model, and then input the logical dependencies in the computational constraints and the supply chain knowledge graph into the analytical model. After parsing the execution order and data flow path between various business rules in the supply chain knowledge graph, it outputs a deterministic computation execution topology. Optionally, the above analytical model can be composed of at least one of a convolutional neural network model, a fully connected neural network model, a residual neural network model, etc.

[0148] Specifically, computer equipment can extract logical dependencies from computational constraints and, based on these dependencies, analyze the execution order and data flow direction of all business rules involved in deterministic computation. This yields the dependencies between business rules, their execution order, and their data flow paths. The analysis results are then abstracted into input nodes, computation nodes, and output nodes. These nodes are connected with directed edges according to their logical dependencies, resulting in a computation execution plan presented as a directed acyclic graph (DAG), i.e., a DAG execution plan, or deterministic computation execution topology.

[0149] The above-mentioned input node refers to the starting calculation node used to receive standardized business datasets, the above-mentioned operation node refers to the intermediate processing node used to perform data calculations and logical judgments, and the above-mentioned output node refers to the termination node used to output the final deterministic calculation result.

[0150] Step S330: Based on the deterministic computation execution topology, context initialization is performed in an isolated sandbox environment to obtain an isolated empty computation framework. The standardized business dataset is loaded into the corresponding input node in the isolated empty computation framework according to the node correspondence in the deterministic computation execution topology, thereby binding the data with the computation node and forming a complete computation environment, namely the computation environment to be executed, which includes the deterministic computation execution topology, the initialized computation context, and the input node with loaded data.

[0151] Specifically, the computer device can perform context initialization in a sandbox environment isolated from external systems to obtain an isolated empty computing framework based on the deterministic computing execution topology, and add standardized business datasets to the input nodes in the isolated empty computing framework to obtain the computing environment to be executed.

[0152] The aforementioned sandbox environment can be a pre-built, independent computing environment completely isolated from external systems and other computing tasks. The context includes a set of environment parameters and basic configurations that support the execution of deterministic computing, containing information such as system parameters, rule configurations, data caches, and execution status required for deterministic computing. The initialization process is as follows: first, a computing context instance is created in the sandbox environment; then, the configuration information of each node in the deterministic computing execution topology is loaded (data receiving rules for input nodes, execution parameters for computation nodes, and result storage rules for output nodes); and the data cache space is initialized (to store intermediate and final results in the deterministic computing process). After that, the basic parameters for computing execution (execution timeout and exception handling rules) are set.

[0153] Step S340: Based on the logical execution units in the computing environment and computing constraints to be executed, instantiate the business logic processing units corresponding to each computing node to obtain the bound logical execution instances.

[0154] Specifically, the computer device can call the instantiated model, and then input the logical execution units in the computing environment to be executed and the computing constraints, as well as the business logic processing units corresponding to each computing node, into the instantiated model for processing to obtain the bound logical execution instance.

[0155] In practical applications, based on the computing environment to be executed, the logical execution units in the computing constraints can be called, and each operation node in the deterministic computing execution topology can be extracted one by one. The business logic processing unit and the required logical execution unit corresponding to each operation node can be obtained. Then, the corresponding logical execution units can be extracted from the computing constraints. According to the logical structure of the business logic processing unit, the logical execution units can be combined and bound to the computing context parameters (such as execution precision and timeout) and the data references of the input nodes in the computing environment to be executed. The execution rules of the business logic processing unit can be set, and the execution order and data transmission method of each atomic logic operator can be determined to complete the instantiation of all operation nodes and obtain a set of bound logical execution instances with independent execution capabilities.

[0156] Optionally, the above-mentioned logic execution unit encapsulates at least one atomic logic operator, which can be understood as the smallest computational execution unit constituting the business logic, including addition calculation, threshold judgment, conditional branching, and data verification.

[0157] Step S350: Based on the logical execution instance, the deterministic computation execution topology, and the compliance threshold and risk warning rules in the computation constraints, deterministic execution and result encapsulation are performed to obtain supply chain business information.

[0158] Specifically, computer equipment can acquire a pre-trained algorithm model, and then input the logical execution instance, the deterministic computation execution topology, and the compliance threshold and risk warning rules in the computation constraints into the algorithm model for deterministic execution and result encapsulation, and output supply chain business information.

[0159] The technical solution in this application embodiment performs semantic matching and mapping between field identifiers and computational constraint data patterns in real-time business status data to obtain a standardized business dataset. Based on the logical dependencies in the computational constraints, it parses the execution order and data flow path between various business rules in the supply chain knowledge graph, constructs a directed acyclic graph execution plan containing input nodes, computation nodes, and output nodes, and obtains a deterministic computation execution topology. Based on the deterministic computation execution topology, it performs context initialization in an isolated sandbox environment to obtain an isolated empty computation framework, and adds the standardized business dataset to the input nodes in the isolated empty computation framework to obtain the computation environment to be executed. The logical execution unit in the context and computational constraints instantiates the business logic processing unit corresponding to each computing node to obtain bound logical execution instances. Based on the logical execution instances, the deterministic computation execution topology, and the compliance thresholds and risk warning rules in the computational constraints, deterministic execution and result encapsulation are performed to obtain supply chain business information. The above method can realize accurate, standardized, and secure deterministic computation from real-time business status data to supply chain business information. It effectively solves the problems of lack of clear structured logical guidance, easy numerical calculation deviation, process judgment error, and lax compliance verification in traditional technologies, ensuring the accuracy and reliability of supply chain business information in numerical calculation and process judgment.

[0160] The following describes the process of deterministically executing and encapsulating the results of the above-mentioned logical execution instances, deterministic computation execution topology, and compliance thresholds and risk warning rules in the computational constraints to obtain supply chain business information. In one embodiment, as... Figure 5 As shown, the process in step S350 above can be implemented in the following way:

[0161] Step S351: Based on the traversal order corresponding to the deterministic computation execution topology, perform business logic operations on each logical execution instance in sequence to obtain an intermediate operation result sequence.

[0162] Specifically, the computer device can call the business logic tool, input the traversal order corresponding to the deterministic computation execution topology and each logic execution instance into the business logic tool, perform business logic operations on each logic execution instance in sequence, and return the intermediate operation result sequence.

[0163] The computer equipment can, based on a deterministic computation execution topology, pre-specify the node traversal order and trigger each bound logic execution instance sequentially according to the traversal order. Each instance then executes the corresponding business logic operation according to its bound execution rules, computation context parameters, and data references. In adjacent computation nodes with dependencies, the output of the preceding node directly serves as the input data for the following node (without additional data conversion or transmission). The computation is completed sequentially according to the data flow path, recording the stage data output after each node's execution and arranging all stage data in the order of execution to obtain a sequence of intermediate computation results.

[0164] Optionally, the above node traversal order can be understood as the fixed order in which each operation node is executed from upstream to downstream as specified in the deterministic computation execution topology, which is equal to the execution order of the business rules parsed in step 320 above.

[0165] Step S352: Based on the intermediate calculation result sequence and the compliance threshold and risk warning rules in the calculation constraints, perform logical branch judgment on the business indicators to be verified in the supply chain business information to obtain the business status judgment mark set.

[0166] In practical applications, computer equipment can compare, match, and analyze the business indicators to be verified in the supply chain business information based on the intermediate operation result sequence and the compliance threshold and risk warning rules in the calculation constraints, so as to obtain the business status judgment mark set after completing the logical branch judgment.

[0167] In this embodiment of the application, the computer device can obtain all the business indicators to be verified in the intermediate calculation result sequence, and call the compliance threshold and risk warning rules in the calculation constraint conditions. For each business indicator to be verified, the calculation result corresponding to the business indicator to be verified in the intermediate calculation result sequence is extracted, and the calculation result is compared with the preset judgment standard to complete the logical branch judgment and obtain the business status judgment mark set.

[0168] Specifically, comparing the calculation results with preset judgment criteria involves the following steps: if the calculation result meets the compliance threshold and does not trigger any risk warning rule, the business status judgment mark is determined to be a normal judgment mark; if the calculation result does not meet the compliance threshold or triggers any risk warning rule, the business status judgment mark is determined to be an abnormal judgment mark; if the calculation result value approaches the threshold and is within the risk critical range, the business status judgment mark is determined to be a critical judgment mark. Furthermore, these business status judgment marks are combined to form a business status judgment mark set.

[0169] Step S353: According to a preset result aggregation strategy, the business status determination mark set and the intermediate calculation result sequence are aggregated to obtain a preliminary result containing specific numerical conclusions, status descriptions, and anomaly indicators. The above result aggregation strategy can be a preset combination rule for integrating intermediate calculation results and business status determination marks and outputting standardized results.

[0170] In this embodiment, the computer device can perform the following aggregation process according to a pre-configured result aggregation strategy based on the business status determination mark set and the intermediate operation result sequence: bind each operation result in the intermediate operation result sequence to the business status determination mark that matches the business status determination mark set, construct the corresponding association between the value and the status, classify all associations according to the business indicator type, perform a traversal operation based on each group of associations after classification to identify the business status determination mark in each group of associations, for each group of associations, if an abnormal type determination mark is detected in the group of associations, add the corresponding abnormal identifier and abnormal details to the group of associations, if no abnormal type determination mark is detected in the group of associations, set the abnormal type determination mark to an empty identifier, and further, according to a preset output format, combine the numerical conclusion, status description, abnormal identifier and abnormal details after classification and traversal to generate a preliminary result.

[0171] Optionally, the above result aggregation strategy can be understood as a pre-defined combination rule for integrating classification and traversal results with business status determination tags into a unified format result, which synthesizes business status determination tags with valid values ​​and status information in the intermediate calculation result sequence.

[0172] Among them, the above numerical conclusions can be understood as the calculated values ​​corresponding to the classification and traversal results, the above state descriptions can be understood as the state interpretation text matched by the judgment markers of each business state, and the above anomaly identifiers are used to characterize whether there are anomalies in the current business indicators and the corresponding anomaly types.

[0173] Step S354: Based on the preliminary results of the resultization and the node execution logs in the deterministic calculation execution topology, extract and encapsulate the rule identifiers, data source fields and calculation paths hit by each logical operation link to obtain supply chain business information.

[0174] In practical applications, computer equipment can extract and encapsulate the rule identifiers, data source fields, and calculation paths hit by each logical operation step according to a preset extraction and encapsulation strategy, based on the preliminary results of the resultization and the node execution logs in the deterministic calculation execution topology, to obtain supply chain business information.

[0175] In this embodiment, the computer device can retrieve the node execution logs automatically generated during the operation of the deterministic computation execution topology based on the preliminary results. Using the node execution logs as the original data source for tracing, the device can uniformly extract three types of tracing elements from the node execution logs: rule identifiers, data source fields, and computation paths corresponding to all logical operation steps of this computation. These elements are then encapsulated into full causal link information. Finally, the encapsulated full causal link information is fused and superimposed with the preliminary results to generate supply chain business information that embeds complete tracing links, numerical conclusions, and full causal link information.

[0176] Optionally, the above-mentioned node execution logs may include the execution time, running status, hit rule identifier, data source field, and complete calculation path of each computing node, which can be used to trace the execution process of each logical operation and the underlying original data source.

[0177] The technical solution in this application embodiment, based on the traversal order corresponding to the deterministic computation execution topology, sequentially performs business logic operations on each logical execution instance to obtain an intermediate operation result sequence. Based on the intermediate operation result sequence and the compliance threshold and risk warning rules in the calculation constraints, logical branch judgments are performed on the business indicators to be verified in the supply chain business information to obtain a business status judgment mark set. According to the preset result aggregation strategy, the business status judgment mark set and the intermediate operation result sequence are aggregated to obtain a preliminary result containing specific numerical conclusions, status descriptions, and anomaly identifiers. Based on the preliminary result and the node execution logs in the deterministic computation execution topology, the rule identifiers, data source fields, and calculation paths hit by each logical operation link are extracted and encapsulated to obtain the supply chain business information. The above method can realize the accurate conversion, standardized synthesis, and traceable encapsulation from logical execution to supply chain business information, further enhancing the rigor and reliability of deterministic computation, effectively solving the defects of disordered operation process, lack of status identifiers in results, and inability to trace data sources in traditional technologies, ensuring that the generated supply chain business information is not only numerically accurate and clearly defined, but also has complete traceability.

[0178] The following describes the process of generating structured traceability information containing supply chain business information based on the aforementioned supply chain business information and target business logic rules. In one embodiment, the aforementioned target business logic rules include rule metadata; such as... Figure 6 As shown, the process in step S400 above may include:

[0179] Step S410: Based on the execution session identifier in the supply chain business information, search the log entries in the pre-built computation execution log storage area to obtain the original execution trajectory sequence containing the input and output parameters and call sequence of each logical operation step, and extract the data access path identifier and rule call record from the original execution trajectory sequence.

[0180] Specifically, the computer equipment can employ a matching method, based on the execution session identifier carried in the supply chain business information, to search through log entries in a pre-built computation execution log storage area to obtain the original execution trajectory sequence containing the input and output parameters and call sequence of each logical operation step. Then, the matching method is used to extract data access path identifiers and rule call records from the original execution trajectory sequence. Optionally, the above matching method can be fuzzy rule matching, exact matching, distance matching, string matching, etc.

[0181] The computer equipment can use the execution session identifier carried in the supply chain business information as a search keyword to match the execution session identifier in the computing execution log storage area, filter out all log entries under the current computing task, sort the filtered log entries in ascending order according to the call timing information in each log entry, generate a time-ordered original execution trajectory sequence, and then extract two types of core traceability elements from the original execution trajectory sequence, namely data access path identifier and rule call record.

[0182] Optionally, the aforementioned execution session identifier is a unique session identifier assigned to the current computing task; the aforementioned computing execution log storage area refers to a pre-built dedicated storage area that stores the execution logs of all business computing tasks in a structured manner. The aforementioned log entries may include information such as the execution session identifier, execution stage, input parameters, output parameters, call sequence, data access path, and rule call information. Among these, the aforementioned data access path identifier can be understood as a unique path code corresponding to the business data read by each computing stage, used to locate the underlying data storage location and the original data source; the aforementioned rule call record includes information such as the rule identifier, call time, execution status, and rule input parameters for each execution stage calling the business rule.

[0183] Step S420: Perform a reverse mapping query based on the rule call records and the rule index table of the supply chain knowledge graph to obtain the hit rule nodes; and extract the rule version identifier and business logic description text from the hit rule nodes to obtain a set of hit rule nodes containing standard version basis and interpretable logic text.

[0184] Specifically, the computer equipment can use a matching method to perform a reverse mapping query based on the rule call records and the rule index table of the supply chain knowledge graph to obtain the hit rule nodes. Then, the matching method is used to extract the rule version identifier and business logic description text from the hit rule nodes to obtain a set of hit rule nodes containing standard version basis and interpretable logic text.

[0185] In practical applications, computer devices can extract rule identifiers from rule call records as retrieval keys, perform matching queries on the rule index table configured in the supply chain knowledge graph to achieve reverse mapping queries, obtain matching hit rule nodes, and then extract rule version identifiers and business logic description text from the hit rule nodes to generate a set of hit rule nodes containing standard version basis (i.e., compliance version basis) and interpretable logic text.

[0186] The aforementioned rule index table is an index table storing information related to rule nodes in the supply chain knowledge graph. It includes mapping relationships such as rule identifier, rule node identifier, rule version identifier, business logic description text, and rule type. The aforementioned rule nodes represent graph nodes in the supply chain knowledge graph that store business rules, logical constraints, and institutional basis (i.e., institutional basis). Meanwhile, the rule version identifier is the version number associated with the matched rule node, used to ensure the traceability of business rules. The business logic description text is text information that details the logic of the business rule, using natural language combined with business terminology to describe the rule's calculation logic, judgment conditions, and execution process.

[0187] Step S430: Perform a reverse mapping query based on the data access path identifier and the data entity index table of the supply chain knowledge graph to obtain the source data nodes; and parse the data field names and measurement unit standards of the source data nodes to obtain a set of source data nodes containing standardized data semantics and unified measurement benchmarks.

[0188] Specifically, computer equipment can use a matching method to perform reverse mapping queries based on data access path identifiers and the data entity index table of the supply chain knowledge graph to obtain source data nodes. Then, it can perform matching, comparison, analysis, and mapping on the data field names and measurement unit standards of the source data nodes to achieve parsing, thereby obtaining a set of source data nodes containing standardized data semantics and unified measurement benchmarks.

[0189] In this embodiment, the computer device can obtain the extracted data access path identifier and retrieve the built-in data entity index table of the supply chain knowledge graph. Using the data access path identifier as the search keyword, it performs a matching query on the data access path identifier in the data entity index table to achieve reverse mapping query and locate the matching source data node. Furthermore, a standardized parsing process is performed on all matched source data nodes. Specifically, this involves extracting the built-in data field names and unit standards of measurement for each source data node, and parsing the data field names and unit standards of measurement for the source data nodes to obtain a set of source data nodes containing standardized data semantics and a unified measurement benchmark.

[0190] Among them, the data field names and the field identifiers of the standardized business dataset strictly correspond, and the measurement unit standards uniformly follow the measurement specifications stipulated in the data model within the calculation constraints. It should be noted that the above-mentioned source data nodes that have completed the standardized parsing process are combined to generate a source data node set with standardized data semantics and a unified measurement benchmark.

[0191] Step S440: Dynamically construct the graph based on the hit rule node set, source data node set, business query intent, and supply chain business information to obtain structured traceability information.

[0192] Specifically, computer equipment can classify and combine the hit rule node set, source data node set, business query intent and target business result to complete the dynamic construction of the graph and obtain structured traceability information containing supply chain business information.

[0193] In addition, computer equipment can acquire a pre-trained graph construction model, and then input the set of hit rule nodes, the set of source data nodes, the business query intent, and the supply chain business information into the graph construction model for dynamic graph construction, and output structured traceability information. Optionally, the above-mentioned graph construction model can be composed of at least one of the following: convolutional neural network model, fully connected neural network model, residual neural network model, self-attention mechanism, etc.

[0194] In one embodiment, such as Figure 7 As shown, the process in step S440 above can be implemented in the following way:

[0195] Step S441: Based on the hit rule node set, source data node set, business query intent and supply chain business information, initialize the graph structure object and instantiate the core logic node to build a blank topology framework.

[0196] In practical applications, computer equipment can initialize graph structure objects for the hit rule node set, source data node set, business query intent and supply chain business information according to a preset initialization strategy, obtain an initialized graph storage container, and based on the initialized graph storage container, use a preset instantiation strategy to instantiate the core logic nodes required for this traceability to obtain a blank topology framework.

[0197] In this embodiment, the computer device can initialize a graph structure object based on the hit rule node set, the source data node set, the business query intent, and the supply chain business information to obtain an initial graph storage container, and instantiate the core logic nodes required for this traceability based on the initial graph storage container to obtain a blank topology framework.

[0198] Optionally, the above-mentioned graph structure object initialization can be understood as the process of constructing a blank graph storage container for the traceability graph, completing the configuration of the container's basic operating parameters, and building the graph storage carrier; the above-mentioned core logic node instantiation process can be understood as: generating independent node entities with unique identifiers and fixed element types within the initialized graph storage container; sequentially instantiating four types of basic core logic nodes, corresponding to the four core traceability elements of business query intent, target business logic rules, source business data, and supply chain business information; and placing all four types of instantiated core logic nodes into the initialized graph storage container to generate a blank topology framework containing only independent nodes and without element association edges yet to be built.

[0199] Furthermore, the instantiation of the aforementioned core logical nodes can be understood as the creation of node entities with unique identifiers and fixed types from graph structure objects. It should be noted that the four core logical nodes mentioned above are the basic node units constituting the traceability graph, corresponding to four core elements: query requirements, rule basis, data source, and calculation results.

[0200] Step S442: Based on the intent mapping relationship in the blank topology framework and rule metadata, a directed semantic edge is established in the graph structure object, pointing from the core logical node of the business query intent to the core logical node of the target business logic rule, to obtain the intent rule association path.

[0201] Specifically, the computer device can retrieve the intent mapping relationship in the rule metadata based on the blank topology framework. Within the blank topology framework, the core logical node corresponding to the business query intent is used as the starting end of the edge, and the core logical node corresponding to the target business logical rule is used as the ending end of the edge. The pairing rules and connection directions between the core logical nodes are constrained by the intent mapping relationship to generate a directed semantic edge with a unidirectional pointing attribute. Based on the directed semantic edge, the topological association of the two types of core logical nodes is performed to obtain the intent rule association path.

[0202] Optionally, the above intent mapping relationship is a predefined matching correspondence between business query intent and target business logic; the above directed semantic edge is used to represent a unidirectional connection structure representing the subordinate correspondence between user query needs and matching rule basis; the above intent rule association path can intuitively represent all business rule basis matched by the current business query needs, and can realize the source binding of query intent and underlying rules.

[0203] Step S443: Based on the data dependency timing and intent rule association path in the original execution trajectory sequence, a directed data input edge is established in the graph structure object, pointing from the core logical node of the source business data to the core logical node of the target business logic rule, to obtain the rule data association path.

[0204] Specifically, the computer device can extract the data dependency time sequence from the original execution trajectory sequence, take the intent rule association path as the basis, and in the blank topology framework, take the core logical node of the source business data as the starting point and the core logical node of the target business logic rule as the ending point, create a directed data input edge to represent the data flow direction, and associate the source business data and the target business logic rule through the directed data input edge to generate the rule data association path.

[0205] The aforementioned data dependency sequence can be understood as the order in which source business data is input into target business logic rules and participates in calculations, serving as the basis for determining the data flow direction and edge construction order between nodes; the aforementioned directed data input edge refers to the unidirectional topological connection structure used to represent the transmission of source data to business rules and participation in rule operations; the aforementioned rule data association path can intuitively and clearly define all the original data sources on which each business rule depends for logical operations.

[0206] Step S444: Based on the generation logic and rule data association path of supply chain business information, a directed derivation output edge is established in the graph structure object, pointing from the core logic node of the target business logic rule to the core logic node of the supply chain business information. The calculated value and status mark are written into the attribute field of the directed derivation output edge to obtain a connected subgraph containing a complete causal chain.

[0207] The computer equipment can extract the generation logic of supply chain business information. Based on the rule data association path, in the topological framework obtained above, it takes the core logic node of the target business logic rule as the starting point and the core logic node representing the target business result as the ending point, and creates a directed derivation output edge to represent the calculation derivation relationship. The calculated values ​​and business status judgment marks in the supply chain business information are written into the attribute fields of the directed derivation output edge, and the full link connection of the four core logic nodes is completed through three layers of directed edges to generate a connected subgraph containing a complete causal chain.

[0208] Optionally, the aforementioned connected subgraph can be understood as a graph structure in which all nodes are interconnected by directed edges, there are no isolated nodes, and the computation process can be fully traced; the aforementioned supply chain business information generation logic can be understood as the causal logic of the output results after the target business logic rules perform calculations on the source business data; the aforementioned directed derivation output edge refers to a unidirectional connection edge used to represent the output results of rule calculations. The aforementioned attribute fields can be understood as fixed storage bits used to store the descriptive information, numerical information, and state information of the edges.

[0209] Step S445: Based on the node topology and edge attribute fields in the connected subgraph, the format is converted according to the preset graph structure serialization template to obtain a logical interpretation graph containing complete supply chain business information, and the logical interpretation graph is identified as structured traceability information.

[0210] Specifically, computer equipment can perform format conversion according to a preset graph structure serialization template based on the topological relationship of all core logical nodes and the attribute fields of all directed edges in the connected subgraph to generate a logical interpretation graph containing complete supply chain business information, and determine the logical interpretation graph as the final structured traceability information.

[0211] Optionally, the above logical interpretation graph refers to a full-link traceability graph presented in the form of a graph structure that can intuitively show the data source, rule basis, calculation process, and result output; the above topological relationship refers to the positional relationship, connection relationship, and pointing relationship between each core logical node; the above graph structure serialization template refers to a predefined standardized format template used to convert the graph structure into a storable, transmittable, and interpretable format.

[0212] The technical solution in this embodiment of the application, based on the execution session identifier in the supply chain business information, searches through log entries in a pre-built computation execution log storage area to obtain the original execution trajectory sequence containing the input and output parameters and call sequence of each logical operation step. Data access path identifiers and rule call records are extracted from the original execution trajectory sequence. A reverse mapping query is performed based on the rule call records and the rule index table of the supply chain knowledge graph to obtain the hit rule nodes. Rule version identifiers and business logic description text are extracted from the hit rule nodes to obtain a set of hit rule nodes containing standard version criteria and interpretable logic text. Based on the data access path identifiers and the data access path identifiers in the supply chain knowledge graph... A reverse mapping query is performed based on the entity index table to obtain the source data nodes. The data field names and unit standards of the source data nodes are then parsed to obtain a set of source data nodes containing standardized data semantics and unified measurement benchmarks. Based on the hit rule node set, the source data node set, the business query intent, and the supply chain business information, a graph is dynamically constructed to obtain structured traceability information. The above method achieves accurate traceability of the generation process of supply chain business information, effectively solving the defects of traditional technologies such as the inability to trace the data source, the inability to clarify the rule basis, and the poor interpretability of supply chain business information. It further enhances the credibility of supply chain business information and provides complete traceability support for subsequent applications.

[0213] To facilitate understanding by those skilled in the art, the supply chain business information query method provided in this application is described in detail below. This method may include:

[0214] Step S601: Input the query request for supply chain business information into the large language model for semantic parsing to obtain the business query intent and business parameters;

[0215] Step S602: Based on the business scenario classification identifier of the business query intent, search for the corresponding matching scenario root node in the scenario tag index of the supply chain knowledge graph to obtain a list of business scenario root nodes.

[0216] Step S603: Based on the core business entity name of the business parameters, search for the corresponding standardized entity node identifier in the entity attribute index of the supply chain knowledge graph to obtain the business entity node sequence.

[0217] Step S604: Based on the business entity node sequence, supply chain knowledge graph and corresponding knowledge graph structure, identification processing is performed to obtain the boundary of the restricted search subgraph;

[0218] Step S605: Based on the boundary of the restricted search subgraph, the list of root nodes of the business scenario, and the supply chain knowledge graph, a path retrieval is performed to obtain a set of candidate logical paths;

[0219] Step S606: Based on the access control policy of supply chain business users, perform node reachability verification on each path in the candidate logical path set to obtain the standard logical path set.

[0220] Step S607: Based on the relation edge type and node attributes contained in each path in the standard logical path set, match the corresponding logical execution unit in the metadata attribute library of the supply chain knowledge graph to obtain the unit mapping relationship table.

[0221] Step S608: Based on the unit mapping relationship table and the node traversal order in the standard logical path set, the logical execution units are connected and arranged according to the business logic relationship to construct an executable logical arrangement instruction set containing the unit execution order, parameter passing rules and logical branch conditions.

[0222] Step S609: Based on the rule attributes associated with the matched logical execution units in the executable logic orchestration instruction set and the supply chain knowledge graph, extract the institutional basis node identifier, business constraints, and risk threshold parameters referenced by each logical execution unit from the executable logic orchestration instruction set to obtain the target business logic rule; the target business logic rule includes computational constraints and rule metadata;

[0223] Step S610: Based on the target business logic rules, business parameters, and supply chain business database, retrieve real-time business status data associated with the business parameters;

[0224] Step S611: Based on the field identifiers in the real-time business status data and the data patterns of the calculation constraints, perform semantic matching and mapping to obtain a standardized business dataset;

[0225] Step S612: Based on the logical dependencies in the computational constraints, the execution order and data flow path of each business rule in the supply chain knowledge graph are analyzed, and a directed acyclic graph execution plan containing input nodes, computation nodes and output nodes is constructed to obtain a deterministic computation execution topology.

[0226] Step S613: Based on the deterministic computation execution topology, context initialization is performed in the isolated sandbox environment to obtain an isolated empty computation framework, and the standardized business dataset is added to the input node in the isolated empty computation framework to obtain the computation environment to be executed;

[0227] Step S614: Based on the logical execution units in the computing environment and computing constraints to be executed, instantiate the business logic processing units corresponding to each computing node to obtain the bound logical execution instances.

[0228] Step S615: Based on the traversal order corresponding to the deterministic computation execution topology, perform business logic operations on each logical execution instance in sequence to obtain an intermediate operation result sequence;

[0229] Step S616: Based on the intermediate calculation result sequence and the compliance threshold and risk warning rules in the calculation constraints, perform logical branch judgment on the business indicators to be verified in the supply chain business information to obtain the business status judgment mark set.

[0230] Step S617: According to the preset result aggregation strategy, aggregate the business status judgment mark set and intermediate calculation result sequence to obtain a preliminary result containing specific numerical conclusions, status descriptions and anomaly indicators.

[0231] Step S618: Based on the preliminary results of the result transformation and the node execution logs in the deterministic calculation execution topology, extract and encapsulate the rule identifiers, data source fields and calculation paths hit by each logical operation link to obtain supply chain business information;

[0232] Step S619: Based on the execution session identifier in the supply chain business information, search the log entries in the pre-built computation execution log storage area to obtain the original execution trajectory sequence containing the input and output parameters and call sequence of each logical operation step, and extract the data access path identifier and rule call record from the original execution trajectory sequence.

[0233] Step S620: Perform a reverse mapping query based on the rule call records and the rule index table of the supply chain knowledge graph to obtain the hit rule nodes; and extract the rule version identifier and business logic description text from the hit rule nodes to obtain a set of hit rule nodes containing standard version basis and interpretable logic text.

[0234] Step S621: Perform a reverse mapping query based on the data access path identifier and the data entity index table of the supply chain knowledge graph to obtain the source data nodes; and parse the data field names and measurement unit standards of the source data nodes to obtain a set of source data nodes containing standardized data semantics and unified measurement benchmarks.

[0235] Step S622: Based on the hit rule node set, source data node set, business query intent and supply chain business information, initialize the graph structure object and instantiate the core logic node to build a blank topology framework;

[0236] Step S623: Based on the intent mapping relationship in the blank topology framework and rule metadata, establish a directed semantic edge in the graph structure object from the core logical node of the business query intent to the core logical node of the target business logic rule, and obtain the intent rule association path.

[0237] Step S624: Based on the data dependency time sequence and intent rule association path in the original execution trajectory sequence, establish a directed data input edge in the graph structure object from the core logical node of the source business data to the core logical node of the target business logic rule, and obtain the rule data association path;

[0238] Step S625: Based on the generation logic and rule data association path of supply chain business information, a directed derivation output edge is established in the graph structure object, pointing from the core logic node of the target business logic rule to the core logic node of the supply chain business information. The calculated value and status mark are written into the attribute field of the directed derivation output edge to obtain a connected subgraph containing a complete causal chain.

[0239] Step S626: Based on the node topology and edge attribute fields in the connected subgraph, the format is converted according to the preset graph structure serialization template to obtain a logical interpretation graph containing complete supply chain business information, and the logical interpretation graph is determined as structured traceability information.

[0240] Step S627: Input the structured traceability information and query request into the large language model to obtain the supply chain business query text.

[0241] It should be noted that the descriptions of steps S601-S627 above can be found in the relevant descriptions in the above embodiments, and their effects are similar, so they will not be repeated here.

[0242] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0243] Based on the same inventive concept, this application also provides a supply chain business information query device for implementing the supply chain business information query method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more embodiments of the supply chain business information query device provided below can be found in the limitations of the supply chain business information query method described above, and will not be repeated here.

[0244] In one embodiment, such asFigure 8 As shown, a supply chain business information query device is provided, including: a semantic parsing module 11, a matching processing module 12, a retrieval module 13, and a processing module 14, wherein:

[0245] Semantic parsing module 11 is used to input the query request for supply chain business information into the large language model for semantic parsing to obtain the business query intent and business parameters;

[0246] The matching processing module 12 is used to match business query intent and business parameters based on a pre-built supply chain knowledge graph to obtain target business logic rules.

[0247] The retrieval module 13 is used to retrieve real-time business status data associated with business parameters based on target business logic rules, business parameters and supply chain business database, and determine supply chain business information based on real-time business status data and target business logic rules.

[0248] Processing module 14 is used to generate structured traceability information containing supply chain business information based on supply chain business information and target business logic rules, and input the structured traceability information and query request into the large language model to obtain supply chain business query text.

[0249] The supply chain business information query device provided in this application embodiment can be used to execute the technical solutions in the above-described supply chain business information query method embodiments of this application. Its implementation principle and technical effect are similar, and will not be repeated here.

[0250] In one embodiment, the matching processing module 12 includes: a first search unit, a second search unit, an identification processing unit, a path retrieval unit, and a verification unit, wherein:

[0251] The first search unit is used to search for the corresponding matching scenario root node in the scenario tag index of the supply chain knowledge graph based on the business scenario classification identifier of the business query intent, and obtain a list of business scenario root nodes.

[0252] The second search unit is used to search for the corresponding standardized entity node identifier in the entity attribute index of the supply chain knowledge graph based on the core business entity name of the business parameters, so as to obtain the business entity node sequence.

[0253] The identification and processing unit is used to perform identification and processing based on the business entity node sequence, supply chain knowledge graph and corresponding knowledge graph structure to obtain the boundary of the restricted search subgraph.

[0254] The path retrieval unit is used to perform path retrieval based on the boundary of the restricted search subgraph, the list of root nodes of the business scenario, and the supply chain knowledge graph to obtain a set of candidate logical paths.

[0255] The verification unit is used to perform permission verification and orchestration on the candidate logical path set to obtain the target business logic rules.

[0256] The supply chain business information query device provided in this application embodiment can be used to execute the technical solutions in the above-described supply chain business information query method embodiments of this application. Its implementation principle and technical effect are similar, and will not be repeated here.

[0257] In one embodiment, the verification unit is specifically used for:

[0258] Based on the access control policies of supply chain business users, node reachability verification is performed on each path in the candidate logical path set to obtain the standard logical path set.

[0259] Based on the relation edge types and node attributes contained in each path in the standard logical path set, the corresponding logical execution units are matched in the metadata attribute library of the supply chain knowledge graph to obtain the unit mapping relationship table.

[0260] Based on the unit mapping relationship table and the node traversal order in the standard logical path set, the logical execution units are connected and arranged according to the business logic relationship to construct an executable logical arrangement instruction set that includes the unit execution order, parameter passing rules and logical branch conditions.

[0261] Based on the rule attributes associated with the matched logical execution units in the executable logic orchestration instruction set and the supply chain knowledge graph, the system basis node identifier, business constraints and risk threshold parameters referenced by each logical execution unit are extracted from the executable logic orchestration instruction set to obtain the target business logic rules.

[0262] The supply chain business information query device provided in this application embodiment can be used to execute the technical solutions in the above-described supply chain business information query method embodiments of this application. Its implementation principle and technical effect are similar, and will not be repeated here.

[0263] In one embodiment, the target business logic rule includes computational constraints; the retrieval module 13 includes: a matching unit, a parsing unit, an adding unit, an instantiation unit, and an execution encapsulation unit, wherein:

[0264] The matching unit is used to perform semantic matching and mapping based on the data patterns of field identifiers and calculation constraints in real-time business status data to obtain a standardized business dataset.

[0265] The parsing unit is used to parse the execution order and data flow path of various business rules in the supply chain knowledge graph based on the logical dependencies in the computational constraints, and to construct a directed acyclic graph execution plan containing input nodes, computation nodes and output nodes to obtain a deterministic computation execution topology.

[0266] Add a unit to perform context initialization in an isolated sandbox environment based on deterministic computation execution topology to obtain an isolated empty computation framework, and add a standardized business dataset to the input node in the isolated empty computation framework to obtain the computation environment to be executed;

[0267] The instantiation unit is used to instantiate the business logic processing unit corresponding to each computing node based on the logical execution unit in the computing environment and computing constraints to be executed, so as to obtain the bound logical execution instance;

[0268] The execution encapsulation unit is used to perform deterministic execution and result encapsulation based on logical execution instances, deterministic computation execution topologies, and compliance thresholds and risk warning rules in computational constraints, thereby obtaining supply chain business information.

[0269] The supply chain business information query device provided in this application embodiment can be used to execute the technical solutions in the above-described supply chain business information query method embodiments of this application. Its implementation principle and technical effect are similar, and will not be repeated here.

[0270] In one embodiment, the execution encapsulation unit includes: a business logic operation subunit, a logic branch judgment subunit, a synthesis subunit, and an extraction encapsulation subunit, wherein:

[0271] The business logic operation subunit is used to perform business logic operations on each logic execution instance in sequence based on the traversal order corresponding to the deterministic computation execution topology, and obtain the intermediate operation result sequence.

[0272] The logical branch judgment subunit is used to make logical branch judgments on the business indicators to be verified in the supply chain business information based on the intermediate operation result sequence and the compliance threshold and risk warning rules in the calculation constraints, so as to obtain the business status judgment mark set.

[0273] The synthesis subunit is used to aggregate the business status judgment mark set and intermediate calculation result sequence according to the preset result aggregation strategy to obtain a preliminary result containing specific numerical conclusions, status descriptions and anomaly indicators.

[0274] The extraction and encapsulation subunit is used to extract and encapsulate the rule identifiers, data source fields, and calculation paths hit by each logical operation step based on the preliminary results and the node execution logs in the deterministic calculation execution topology, in order to obtain supply chain business information.

[0275] The supply chain business information query device provided in this application embodiment can be used to execute the technical solutions in the above-described supply chain business information query method embodiments of this application. Its implementation principle and technical effect are similar, and will not be repeated here.

[0276] In one embodiment, the target business logic rule includes rule metadata; the processing module 14 includes: a third lookup unit, a query unit, a parsing unit, and a construction unit, wherein:

[0277] The third search unit is used to search for log entries in the pre-built computation execution log storage area based on the execution session identifier in the supply chain business information, to obtain the original execution trajectory sequence containing the input and output parameters and call sequence of each logical operation, and to extract the data access path identifier and rule call record from the original execution trajectory sequence.

[0278] The query unit is used to perform reverse mapping queries based on rule call records and the rule index table of the supply chain knowledge graph to obtain the hit rule nodes; and extract the rule version identifier and business logic description text from the hit rule nodes to obtain a set of hit rule nodes containing standard version basis and interpretable logic text.

[0279] The parsing unit is used to perform reverse mapping queries based on the data access path identifier and the data entity index table of the supply chain knowledge graph to obtain the source data nodes; and to parse the data field names and measurement unit standards of the source data nodes to obtain a set of source data nodes containing standardized data semantics and unified measurement benchmarks.

[0280] The construction unit is used to dynamically construct the graph based on the hit rule node set, the source data node set, the business query intent, and the supply chain business information to obtain structured traceability information.

[0281] The supply chain business information query device provided in this application embodiment can be used to execute the technical solutions in the above-described supply chain business information query method embodiments of this application. Its implementation principle and technical effect are similar, and will not be repeated here.

[0282] In one embodiment, the above-mentioned building unit is specifically used for:

[0283] Based on the hit rule node set, source data node set, business query intent and supply chain business information, the graph structure object is initialized and the core logic node is instantiated to build a blank topology framework.

[0284] Based on the intent mapping relationship in the blank topology framework and rule metadata, a directed semantic edge is established in the graph structure object from the core logical node of the business query intent to the core logical node of the target business logic rule, thus obtaining the intent rule association path;

[0285] Based on the data dependency time sequence and intent rule association path in the original execution trajectory sequence, a directed data input edge is established in the graph structure object, pointing from the core logical node of the source business data to the core logical node of the target business logic rule, to obtain the rule data association path;

[0286] Based on the generation logic and rule data association path of supply chain business information, a directed derivation output edge is established in the graph structure object, pointing from the core logic node of the target business logic rule to the core logic node of the supply chain business information. The calculated values ​​and status markers are written into the attribute fields of the directed derivation output edge to obtain a connected subgraph containing a complete causal chain.

[0287] Based on the node topology and edge attribute fields in the connected subgraph, the format is converted according to the preset graph structure serialization template to obtain a logical interpretation graph containing complete supply chain business information, and the logical interpretation graph is identified as structured traceability information.

[0288] The supply chain business information query device provided in this application embodiment can be used to execute the technical solutions in the above-described supply chain business information query method embodiments of this application. Its implementation principle and technical effect are similar, and will not be repeated here.

[0289] Each module in the aforementioned supply chain business information query device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0290] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 9 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores supply chain business query text. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements a supply chain business information query method.

[0291] Those skilled in the art will understand that Figure 9 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0292] In one embodiment, a computer device is also provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the technical solution in the above-described embodiment of the supply chain business information query method of this application. The implementation principle and technical effect are similar, and will not be repeated here.

[0293] In one embodiment, a computer-readable storage medium is also provided, on which a computer program is stored. When the computer program is executed by a processor, it implements the technical solution of the supply chain business information query method described above in this application. Its implementation principle and technical effect are similar, and will not be repeated here.

[0294] In one embodiment, a computer program product is also provided, including a computer program that, when executed by a processor, implements the technical solution of the supply chain business information query method described above in this application. Its implementation principle and technical effects are similar and will not be repeated here.

[0295] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one of relational databases and non-relational databases. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0296] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0297] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for querying supply chain business information, characterized in that, The method includes: The query request for supply chain business information is input into the large language model for semantic parsing to obtain the business query intent and business parameters. Based on a pre-built supply chain knowledge graph, the business query intent and the business parameters are matched to obtain the target business logic rules. The target business logic rules, the business parameters, and the supply chain business database are used to retrieve real-time business status data associated with the business parameters, and the supply chain business information is determined based on the real-time business status data and the target business logic rules. Based on the supply chain business information and the target business logic rules, structured traceability information containing the supply chain business information is generated, and the structured traceability information and the query request are both input into the large language model to obtain the supply chain business query text.

2. The method according to claim 1, characterized in that, The process of matching the business query intent and the business parameters based on the pre-built supply chain knowledge graph to obtain target business logic rules includes: Based on the business scenario classification identifier of the business query intent, the corresponding matching scenario root node is searched in the scenario tag index of the supply chain knowledge graph to obtain a list of business scenario root nodes. Based on the core business entity name of the business parameters, the corresponding standardized entity node identifier is searched in the entity attribute index of the supply chain knowledge graph to obtain the business entity node sequence. Based on the business entity node sequence, the supply chain knowledge graph and the corresponding knowledge graph structure, the bounded search subgraph boundary is obtained through identification processing. Based on the boundary of the restricted search subgraph, the list of root nodes of the business scenario, and the supply chain knowledge graph, a path retrieval is performed to obtain a set of candidate logical paths. The candidate logical path set is subjected to permission verification and orchestration to obtain the target business logic rule.

3. The method according to claim 2, characterized in that, The step of performing permission verification and orchestration on the candidate logical path set to obtain the target business logic rule includes: Based on the access control policies of supply chain business users, node reachability verification is performed on each path in the candidate logical path set to obtain a standard logical path set. Based on the relation edge types and node attributes contained in each path in the standard logical path set, the corresponding logical execution unit is matched in the metadata attribute library of the supply chain knowledge graph to obtain the unit mapping relationship table. Based on the unit mapping relationship table and the node traversal order in the standard logical path set, the logical execution units are connected and arranged according to the business logic relationship to construct an executable logical arrangement instruction set that includes unit execution order, parameter passing rules and logical branch conditions; Based on the executable logic orchestration instruction set and the rule attributes associated with the matched logic execution units in the supply chain knowledge graph, the system basis node identifier, business constraint conditions and risk threshold parameters referenced by each logic execution unit are extracted from the executable logic orchestration instruction set to obtain the target business logic rule.

4. The method according to any one of claims 1-3, characterized in that, The target business logic rules include computational constraints; determining the supply chain business information based on the real-time business status data and the target business logic rules includes: Based on the field identifiers in the real-time business status data and the data patterns of the calculation constraints, semantic matching and mapping are performed to obtain a standardized business dataset. Based on the logical dependencies in the computational constraints, the execution order and data flow path of each business rule in the supply chain knowledge graph are analyzed, and a directed acyclic graph execution plan containing input nodes, computation nodes and output nodes is constructed to obtain a deterministic computational execution topology. Based on the deterministic computation execution topology, context initialization is performed in an isolated sandbox environment to obtain an isolated empty computation framework, and the standardized business dataset is added to the input node in the isolated empty computation framework to obtain the computation environment to be executed; Based on the computational environment to be executed and the logical execution units in the computational constraints, the business logic processing units corresponding to each computational node are instantiated to obtain bound logical execution instances. Based on the logical execution instance, the deterministic computation execution topology, and the compliance threshold and risk warning rules in the computation constraints, deterministic execution and result encapsulation are performed to obtain the supply chain business information.

5. The method according to claim 4, characterized in that, The process of deterministic execution and result encapsulation based on the logical execution instance, the deterministic computation execution topology, and the compliance thresholds and risk warning rules in the computational constraints yields the supply chain business information, including: Based on the traversal order corresponding to the deterministic computation execution topology, business logic operations are performed on each of the logical execution instances in sequence to obtain an intermediate operation result sequence; Based on the intermediate calculation result sequence and the compliance threshold and risk warning rules in the calculation constraints, logical branch judgments are performed on the business indicators to be verified in the supply chain business information to obtain a business status judgment mark set. According to the preset result aggregation strategy, the business status judgment mark set and the intermediate operation result sequence are aggregated to obtain a preliminary result containing specific numerical conclusions, status descriptions and anomaly indicators. Based on the preliminary results and the node execution logs in the deterministic computation execution topology, the rule identifiers, data source fields, and computation paths hit by each logical operation are extracted and encapsulated to obtain the supply chain business information.

6. The method according to any one of claims 1-3, characterized in that, The target business logic rules include rule metadata; the generation of structured traceability information containing the supply chain business information based on the supply chain business information and the target business logic rules includes: Based on the execution session identifier in the supply chain business information, the log entries in the pre-built computation execution log storage area are searched to obtain the original execution trajectory sequence containing the input and output parameters and call sequence of each logical operation step, and the data access path identifier and rule call record are extracted from the original execution trajectory sequence. A reverse mapping query is performed based on the rule call records and the rule index table of the supply chain knowledge graph to obtain the hit rule nodes; and the rule version identifier and business logic description text are extracted from the hit rule nodes to obtain a set of hit rule nodes containing standard version basis and interpretable logic text. A reverse mapping query is performed based on the data access path identifier and the data entity index table of the supply chain knowledge graph to obtain the source data nodes; and the data field names and measurement unit standards of the source data nodes are parsed to obtain a set of source data nodes containing standardized data semantics and unified measurement benchmarks. The structured traceability information is obtained by dynamically constructing a graph based on the hit rule node set, the source data node set, the business query intent, and the supply chain business information.

7. The method according to claim 6, characterized in that, The structured traceability information is obtained by dynamically constructing a graph based on the hit rule node set, the source data node set, the business query intent, and the supply chain business information, including: Based on the hit rule node set, the source data node set, the business query intent, and the supply chain business information, graph structure objects are initialized and core logic nodes are instantiated to construct a blank topology framework. Based on the blank topology framework and the intent mapping relationship in the rule metadata, a directed semantic edge is established in the graph structure object from the core logical node of the business query intent to the core logical node of the target business logic rule, so as to obtain the intent rule association path. Based on the data dependency time sequence in the original execution trajectory sequence and the intention rule association path, a directed data input edge is established in the graph structure object, pointing from the core logical node of the source business data to the core logical node of the target business logic rule, to obtain the rule data association path; Based on the generation logic of supply chain business information and the rule data association path, a directed derivation output edge is established in the graph structure object, pointing from the core logic node of the target business logic rule to the core logic node of the supply chain business information. The calculated value and status mark are written into the attribute field of the directed derivation output edge to obtain a connected subgraph containing a complete causal chain. Based on the node topology and edge attribute fields in the connected subgraph, the format is converted according to the preset graph structure serialization template to obtain a logical interpretation graph containing complete supply chain business information, and the logical interpretation graph is determined as structured traceability information.

8. A supply chain business information query device, characterized in that, The device includes: The semantic parsing module is used to input query requests for supply chain business information into the large language model for semantic parsing to obtain the business query intent and business parameters; The matching processing module is used to match the business query intent and the business parameters based on a pre-built supply chain knowledge graph to obtain the target business logic rules. The retrieval module is used to perform retrieval based on the target business logic rules, the business parameters, and the supply chain business database to obtain real-time business status data associated with the business parameters, and to determine the supply chain business information based on the real-time business status data and the target business logic rules. The processing module is used to generate structured traceability information containing the supply chain business information based on the supply chain business information and the target business logic rules, and input the structured traceability information and the query request into the large language model to obtain the supply chain business query text.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images