Multi-dimensional intelligent classification method and device for traffic industry facing all-scenario adaptation

Abstract

The application relates to the technical field of intelligent data processing and intelligent traffic all-scene application in digital management, and provides a traffic industry dynamic multi-dimensional intelligent classification method and device for all-scene adaptation, which comprises the following steps: a multi-dimensional classification system model is constructed with entities, attributes and relationships as three orthogonal classification axes; object definitions are made for the entities, attributes and relationships; according to the object definitions, information of the entities, attributes and relationships is extracted from multi-source data of traffic industry all-business scenes to form standardized entity libraries, attribute libraries and relationship libraries; taking the entities in the entity libraries as cores, attributes in the attribute libraries are dynamically associated according to predefined rules to configure metadata centered on the entities; and based on business scenes, multiple metadata are associated by using the relationships in the relationship libraries to generate metadata models corresponding to the business scenes. The application realizes efficient interoperation and fusion management of traffic industry information in a whole life cycle and under all-business scenes.

Description

Technical Field

[0001] This invention relates to the field of intelligent data processing and intelligent transportation full-scenario application technology in digital management, and in particular to a dynamic multi-dimensional intelligent classification method and device for the transportation industry that is adaptable to all scenarios. Background Technology

[0002] With the rapid development and widespread application of BIM, IoT, big data, and AI technologies, the transportation industry is undergoing a profound digital and information-based transformation. The complex data management of the transportation system is also facing the challenge of heterogeneous data integration. This change permeates all stages of the entire lifecycle, from design and planning, construction, operation and maintenance, upgrading and renovation, to travel services, making data a key driver of industry development. However, the accelerated digitalization process has led to a rapid increase in data volume, placing higher demands on data production, management, and application, especially the integration of multi-source data, which has become a pressing challenge for current data management. How to achieve seamless sharing and efficient interoperability of heterogeneous data acquired through various collection methods across different application scenarios throughout the entire lifecycle of infrastructure is a key issue currently facing us.

[0003] As a complex system spanning multiple scenarios, businesses, and domains, the transportation system generates diverse data sources, including design drawings and models, construction monitoring data, operational equipment status data, and travel management and control. This presents challenges to data interconnectivity and also limits the further development and application of emerging technologies such as artificial intelligence in this field. Scientific classification and a unified coding system are fundamental to solving these problems. A unified and flexible classification system facilitates standardized data management, reduces management complexity, and provides strong support for accurate data retrieval, efficient analysis, and AI development. Currently, although there are standards both domestically and internationally, such as the "Classification and Coding Standard for Building Information Modeling" (GB / T51269-2017) and the "Unified Standard for Application of Highway Engineering Information Modeling" (JTG / T2420-2021), there are significant differences in classification methods, hierarchical divisions, and coding rules. For example, the "bearing system" of bridges is classified as an independent entity in highway standards, while it falls under the "substructure" subcategory in railway engineering standards; the "span combination" attribute is a geometric attribute in highway standards, but falls under the category of technical attributes in international BIM standards. Such inconsistencies make it difficult for heterogeneous data to interact, and cross-system collaboration requires the establishment of a large number of manually mapped rules.

[0004] The current state of traffic information classification and data management is inadequate. Despite the existence of UniFormat, OmniClass, and the aforementioned domestic and international standards, significant differences exist in their classification methods and data expansion frameworks, leading to widespread data silos, poor interoperability, and hindering data collaboration and industry data aggregation. Different standards have varying data classification and coding rules, making it difficult for the same data to be shared across different systems. While data can be managed relatively well within the same system, complex mapping relationships are still required when crossing stages or systems, increasing data application costs and reducing processing efficiency. Furthermore, data classification often focuses on business objects, lacking a holistic consideration of the data, making it difficult to effectively manage and mine large amounts of unstructured data such as text, images, and videos, thus failing to fully realize their business value. Traditional classification standards use static tree structures, which are ill-suited to the rapid iteration of new facilities and technologies. For example, for emerging technologies such as "intelligent sensing road signs" and "photovoltaic pavements," existing standards lack clear classification paths, requiring a 3-6 month standardization process for manual expansion, lagging behind technological development. In artificial intelligence applications, data fragmentation leads to more than 130% of the effort required for model training being spent on entity alignment and attribute normalization, which restricts the efficiency of AI model development for tasks such as disease identification and load prediction. Summary of the Invention

[0005] To address the aforementioned problems in the prior art, this invention provides a dynamic multidimensional intelligent classification method and device for the transportation industry that is adaptable to all scenarios, enabling efficient interoperability of transportation industry information throughout the entire lifecycle and across all scenarios.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: In a first aspect, the present invention provides a dynamic multidimensional intelligent classification method for the transportation industry that is adaptable to all scenarios, including: Construct a multi-dimensional classification system model with entities, attributes, and relationships as three orthogonal classification axes; Define the entity, the attribute, and the relationship as an object; Based on the object definition, information about the entity, the attribute, and the relationship is extracted from multi-source data across all business scenarios in the transportation industry to form a standardized entity library, attribute library, and relationship library. Using entities in the entity library as the core, attributes in the attribute library are dynamically associated according to predefined rules to form entity-centric metadata. Based on the business scenario, multiple metadata are associated using the relationships in the relational database to generate a metadata model corresponding to the business scenario.

[0007] The beneficial effects of this invention are as follows: By constructing a multi-dimensional classification system with entities, attributes, and relationships as three orthogonal classification axes and standardizing the definition of various objects, this invention achieves automatic extraction, normalization, and construction of a unified entity library, attribute library, and relationship library from multi-source data. This enables dynamic configuration of metadata with entities as the core and rapid generation of scenario-based models based on business scenarios and relationship chains, thereby achieving efficient interoperability and integrated management of transportation industry information throughout its entire lifecycle and across all business scenarios.

[0008] Optionally, it includes: Select the starting entity based on the business scenario; Based on the scenario requirements of the business scenario, a predefined relationship type in the relationship database is determined, and other entities related to the starting entity are queried and associated from the entity database using the foreign key corresponding to the relationship type as the association basis, to obtain a set of associated entities; Based on logical analysis, the core entity is determined from the set of related entities; Using the foreign key as a chain, the starting entity, the core entity, and other related entities are linked, and attributes from the attribute library are dynamically matched for each entity, thereby generating a metadata model corresponding to the business scenario.

[0009] As described above, by clearly defining the modeling process of using foreign keys as the chain, identifying core entities, and dynamically matching attributes, the system achieves intelligent model assembly from static data association to business objectives, significantly improving the fit and practicality of the metadata model with complex business scenarios.

[0010] Optionally, it also includes: The multi-dimensional classification system model and the metadata model are dynamically updated.

[0011] Optionally, dynamically updating the multi-dimensional classification system model and the metadata model includes: An incremental learning algorithm is used to continuously monitor and analyze the data source to obtain the change requirements of objects in the entity database, attribute database and relation database. The change requirements include the addition of objects, attribute changes or association conflicts. When determining whether the change request meets the difference triggering condition, if so, the change object is reviewed and the entity library, attribute library or relation library is updated according to the review result; otherwise, the entity library, attribute library or relation library is updated directly according to the change request. Based on the updated entity database, attribute database, and relation database, reconfigure the corresponding metadata and update the corresponding metadata model; Based on the updated content of the entity library and the attribute library, the preset mapping rules between them and at least one external classification standard are updated synchronously to maintain the consistency of data conversion interoperability between the multi-dimensional classification system model and the external classification standard.

[0012] As described above, by introducing incremental learning and difference triggering mechanisms, the classification system is equipped with the ability to proactively perceive, intelligently review, and continuously evolve, which solves the fundamental problem that traditional classification standards lag behind technological development and achieves rapid compatibility with new facilities and technologies.

[0013] Optionally, the mapping rules are defined and maintained through a rule matrix, which records the mapping relationship and data conversion logic between the objects in the entity library and the attribute library and the corresponding entries in the external classification standard.

[0014] Optionally, the multi-dimensional classification system model includes sequentially cooperating: The data primitive layer is used to process the input raw data and output entities, attributes, and relationships to the metadata kernel layer; The metadata kernel layer is constructed using entities, attributes, and relationships as three orthogonal classification axes. The metadata model layer is used to assemble the metadata output by the metadata kernel layer to form the corresponding metadata model; The basic data layer is used to provide raw data input to the data primitive layer.

[0015] Optionally, the step of defining the entity, the attribute, and the relationship as objects includes: Configure constraints for each defined object, which are used to automatically verify the compliance of associations, matching and combination between objects in subsequent steps.

[0016] Optionally, the step of extracting information about the entity, the attribute, and the relationship from multi-source data across all business scenarios in the transportation industry, based on the object definition, to form a standardized entity library, attribute library, and relationship library, includes: Identify original descriptive information related to entities, attributes, and relationships from multi-source data across all business scenarios in the transportation industry; Based on the original description information, natural language processing technology is used to parse and generate candidate entity objects, attribute objects, and relation objects; Using a semantic similarity model, candidate objects whose semantic similarity reaches a similarity threshold are merged and aggregated. Each aggregated independent object is assigned a unique standardized code, and according to its object type, it is classified and entered into the entity library, attribute library, or relation library respectively.

[0017] As described above, the standardized process of aggregation coding enables the automated construction of unstructured text into a high-quality structured knowledge base, providing a reliable and consistent data foundation for upper-level data fusion and intelligent applications.

[0018] Optionally, after the merging and aggregation and before the classification and entry, the following steps are included: Each individual object after aggregation undergoes human-machine collaborative review. Based on the review results, its standardized code and library category are ultimately determined. The human-machine collaborative review provides semantic conflict prompts and confidence assessments for human decision-making.

[0019] Secondly, the present invention provides a dynamic multi-dimensional intelligent classification device for the transportation industry that is adaptable to all scenarios, comprising: The system construction module is used to build a multi-dimensional classification system model with entities, attributes, and relationships as three orthogonal classification axes. The object definition module is used to define the entity, the attribute, and the relationship as objects; The standard database building module is used to extract information about the entities, attributes, and relationships from multi-source data across all business scenarios in the transportation industry, based on the object definition, to form a standardized entity database, attribute database, and relationship database. The metadata configuration module is used to dynamically associate attributes in the attribute library with entities in the entity library as the core, and configure metadata centered on entities according to predefined rules. The scenario modeling module is used to associate multiple metadata with the relationships in the relational database based on the business scenario, and generate a metadata model corresponding to the business scenario.

[0020] The technical effects of the dynamic multidimensional intelligent classification device for the transportation industry that is adaptable to all scenarios provided in the second aspect are described in the relevant description of the dynamic multidimensional intelligent classification method for the transportation industry that is adaptable to all scenarios provided in the first aspect. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the main process of the dynamic multidimensional intelligent classification method for the transportation industry that is adaptable to all scenarios according to an embodiment of the present invention. Figure 2 This is a schematic diagram of the overall process of the dynamic multidimensional intelligent classification method for the transportation industry that is adaptable to all scenarios, according to an embodiment of the present invention. Figure 3 This is a three-axis schematic diagram of the multi-dimensional classification system model involved in the embodiments of the present invention; Figure 4 This is a schematic diagram of the architecture of the multi-dimensional classification system model involved in an embodiment of the present invention; Figure 5This is a schematic diagram of the main interface of the transportation infrastructure entity / attribute extraction system involved in this embodiment of the invention; Figure 6 This is a schematic diagram of the document upload interface in the transportation infrastructure entity / attribute extraction system according to an embodiment of the present invention; Figure 7 This is a schematic diagram of the information extraction interface in the transportation infrastructure entity / attribute extraction system according to an embodiment of the present invention; Figure 8 This is a schematic diagram of the review information in the transportation infrastructure entity / attribute extraction system according to an embodiment of the present invention; Figure 9 This is a schematic diagram of the entity list extracted from the transportation infrastructure entity / attribute extraction system according to an embodiment of the present invention; Figure 10 This is a schematic diagram of the result attribute list extracted by the transportation infrastructure entity / attribute extraction system according to an embodiment of the present invention; Figure 11 This is a schematic diagram of the entity attribute management system in the transportation infrastructure entity / attribute management system according to an embodiment of the present invention; Figure 12 This is a schematic diagram of the structure of a dynamic multi-dimensional intelligent classification device for the transportation industry that is adaptable to all scenarios, according to an embodiment of the present invention. Figure 13 This is a schematic diagram of the structure of a computer device according to an embodiment of the present invention.

[0022] Explanation of reference numerals in the attached figures: 1. A dynamic, multi-dimensional intelligent classification device for the transportation industry that adapts to all scenarios; 2. Processor; 3. Memory. Detailed Implementation

[0023] To better understand the above technical solutions, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present invention can be understood more clearly and thoroughly, and that the scope of the present invention can be fully conveyed to those skilled in the art.

[0024] The embodiments of this application are applied to all scenarios in the transportation industry, including travel services, smart parking, traffic control, and integrated traffic management. This embodiment illustrates the scenario of digital management and intelligent operation and maintenance of transportation infrastructure throughout its entire lifecycle. Specifically, it includes design collaboration, construction control, health monitoring, asset management, and renovation decisions for major transportation facilities such as bridges, tunnels, and roads. For example, in tunnel asset management, it is necessary to integrate design drawings in CAD and other formats, building information models in BIM and other formats, construction monitoring time-series data, operation-phase inspection images and videos, and various industry standard and specification documents to achieve accurate mapping and dynamic updates from physical facilities to digital models. Existing technologies mainly rely on static tree-like classification standards, which suffer from problems such as large system differences, lagging updates, and inability to integrate unstructured data, resulting in severe data silos and hindering the deep application of artificial intelligence.

[0025] To this end, in various embodiments of this application, a multi-dimensional classification system model is constructed with entities, attributes, and relationships as three orthogonal classification axes; entities, attributes, and relationships are defined as objects; based on the object definitions, information on entities, attributes, and relationships is extracted from multi-source data across all business scenarios in the transportation industry to form standardized entity, attribute, and relationship libraries; with entities in the entity library as the core, attributes in the attribute library are dynamically associated according to predefined rules to configure entity-centric metadata; based on business scenarios, relationships in the relationship library are used to associate multiple metadata sets to generate a metadata model corresponding to the business scenario. Thus, by constructing a multi-dimensional classification system with entities, attributes, and relationships as three orthogonal classification axes and a unified entity, attribute, and relationship library, this invention enables dynamic configuration of metadata with entities as the core and rapid generation of scenario-based models based on business scenarios and relationship chains, thereby achieving efficient interoperability and integrated management of transportation industry information throughout its entire lifecycle and across all business scenarios.

[0026] The present application will now be described in further detail with reference to the accompanying drawings and embodiments.

[0027] This application provides a dynamic, multi-dimensional intelligent classification method for the transportation industry that is adaptable to all scenarios, such as... Figure 1 As shown, the method includes: Step S101: Construct a multi-dimensional classification system model with entities, attributes, and relationships as three orthogonal classification axes.

[0028] In step S101, as follows Figure 2As shown, data is divided into nine categories: entities, attributes, relationships, metadata, metadata models, business scenarios, master data, structured data, and unstructured data. Entities, attributes, and relationships serve as three classification axes, forming a three-dimensional orthogonal classification system. This system decomposes the semantic kernel of heterogeneous data elements, breaking down traditional tree-based classification methods into independently defined axes for entities, attributes, and relationships. Each axis is independently encoded and mutually exclusive, eliminating semantic overlap in multi-standard classifications. Figure 3 As shown. Based on this, a four-layer architecture is constructed, consisting of a data primitive layer, a metadata kernel layer, a metadata model layer, and a basic data layer, as follows: Figure 4 As shown, the multi-dimensional classification system model in this embodiment includes the following components working in sequence: (a) Data primitive layer, used to process the input raw data and output entities, attributes and relationships to the metadata kernel layer.

[0029] Specifically, the original data is atomically decomposed and its features are extracted to generate metadata elements with independent semantics, including entities, attributes and relationships, which are also the basic elements that make up metadata.

[0030] (b) Metadata kernel layer, which is constructed with entities, attributes and relationships as three orthogonal classification axes.

[0031] Specifically, a three-dimensional orthogonal classification axis modeling method is adopted to construct a standardized metadata framework consisting of entity classification axis, attribute classification axis and relation classification axis.

[0032] (c) Metadata model layer, used to assemble the metadata output by the metadata kernel layer to form the corresponding metadata model.

[0033] (d) Basic data layer, used to provide raw data input for the data primitive layer.

[0034] Step S102: Define the entities, attributes, and relationships as objects.

[0035] This step, based on industry standards, business practices, or basic logic, defines specific parameters for entities, attributes, and relational objects, including names, codes, and categories.

[0036] In this embodiment, the entities, attributes, and relationships are defined as objects, including: Configure constraints for each defined object. These constraints are used to automatically verify the compliance of associations, matching, and combinations between objects in subsequent steps.

[0037] Table 1. Entity Class Definition Matrix

[0038] Table 2. Meaning and Examples of Entity Categories

[0039] Specifically, the definition of entities involves creating an entity description paradigm indexed by a unique classification code, defining the multi-dimensional feature parameter cluster shown in Table 1, which covers key information such as Chinese name, English name, classification code, entity category, relational foreign key, and remarks. The parameter entity category is a preliminary classification of entities based on the basic logic of business practices throughout the entire lifecycle of transportation infrastructure. The meaning and examples of different entity categories are shown in Table 2. Entities can be quickly filtered and retrieved based on their category. Furthermore, entity tags can be customized to further classify entities according to specific circumstances. Relationship foreign keys, as an expression of object association within a specific relational layer, can be specified as entity parameters. Constraints are used to define the associated objects of an entity under special conditions. For example, the main cable only exists when the bridge type is a suspension bridge; this can be recorded in the constraints of the main cable entity object. Version identifiers are used to record the iterative change history of the entity, which is helpful for historical data backtracking.

[0040] Specifically, the definition of attributes includes: a method for establishing attribute classification axes is proposed, and a global attribute expression paradigm is defined as shown in Table 3, covering key information such as Chinese name, English name, classification code, attribute category, and attribute label. Attribute categories are a preliminary division of attributes based on the basic logic of business practices throughout the entire lifecycle of transportation infrastructure, with the meanings and examples of different attribute categories shown in Table 4. Attributes can be quickly filtered and retrieved based on their categories. Furthermore, attribute labels can be customized to further classify attributes according to specific circumstances. Foreign keys, as a way to express object associations within a specific relationship layer, can be specified as attribute parameters. Constraints are used to define the associated objects of attributes under special conditions, supporting automated adjustments after batch establishment of relationships for attributes. Units of measurement refer to all standardized units of measurement available for the attribute, and a default attribute is set. Referenced standards are used to record the source of the attribute, supporting multiple sources.

[0041] Table 3. Attribute Expression Parameters

[0042] Table 4. Meaning and Examples of Attribute Categories

[0043] Specifically, the definition of relational objects involves proposing a method for establishing relational classification axes and creating a relational description framework as shown in Table 5, covering key information such as Chinese name, English name, classification code, relational object, expression method, and definition. It supports mixed association definitions of composite objects such as entity-entity, entity-attribute, and attribute-attribute. The relational object must clearly specify the type of object on which the relation is established. The expression method of the relation refers to how the relation is implemented in the system, and can adopt a hybrid architecture of regular expressions and dynamic templates to adapt to the relational expression needs of different application scenarios.

[0044] Table 5. Relationship Expression Parameters

[0045] Step S103: Based on the object definition, extract entity, attribute and relationship information from multi-source data of the entire business scenario of the transportation industry to form a standardized entity library, attribute library and relationship library.

[0046] This step involves extracting entities, attributes, and relational objects from existing standards and business documents using manual, semi-automated, or fully automated methods, and then normalizing them based on semantic similarity algorithms to build a standard-neutral basic library.

[0047] Step S104: Using entities in the entity library as the core, dynamically associate attributes in the attribute library according to predefined rules to configure and form entity-centric metadata.

[0048] This step involves building a configuration method for entity-centric metadata that combines manual and automatic methods, based on the established entity database, relation database, and attribute database. It introduces a rule engine to achieve dynamic attribute filtering, uses basic logic such as predefined relations and attributes to automatically filter associated attributes, and presets default units of measurement according to entity type to form a metadata kernel that dynamically changes in different dimensions.

[0049] Specifically, based on entity databases, relation databases, and attribute databases, metadata can be established with entities as the core, attributes as the foundation, and relationships as the anchors. In this step, the association between entities and different attributes is established according to different relation categories. The associated attribute set is automatically loaded according to the relation category and entity type. The associated attributes can also be further refined through entity tags and attribute tags to form metadata for specific entities.

[0050] This system utilizes predefined rules, combining manual creation with artificial intelligence, to create different attribute sets via attribute tags, enabling targeted and rapid association of entity attributes. When the same attribute is associated with different entities, the units of measurement and value ranges may differ. To generate more accurate metadata for different entity objects, the rule engine performs initial screening of units of measurement and value ranges, and AI-based optimization sets the optimal choice within each range as the default value. Simultaneously, the database is continuously monitored; if an attribute in metadata remains empty for an extended period or has insufficient content, the system determines whether to de-associate that attribute, maintaining dynamic consistency with minimal metadata redundancy. Furthermore, manual intervention allows for further adjustment and refinement of metadata. Each metadata adjustment records the modification time, operator, and change details, supporting historical version traversal along a timeline to ensure data traceability.

[0051] Step S105: Based on the business scenario, use the relationships in the relational database to associate multiple metadata to generate a metadata model corresponding to the business scenario.

[0052] This step involves establishing a metadata model adaptation algorithm, defining scenario elements, introducing rules, and using artificial intelligence self-recognition algorithms to link different metadata kernels using foreign keys as a chain, and further dynamically optimizing the metadata kernels based on business logic during this linking process to form a metadata model suitable for the business scenario.

[0053] Therefore, this embodiment realizes the automatic extraction, normalization and construction of a unified entity library, attribute library and relation library from multi-source data, and then enables dynamic configuration of metadata with entities as the core, and rapid generation of scenario-based models based on business scenarios and relation chains, thereby realizing efficient interoperability and integrated management of transportation infrastructure information throughout the entire life cycle and in all business scenarios.

[0054] In one embodiment, step S103 includes: Step S1031: Identify the original descriptive information related to entities, attributes and relationships from multi-source data of the entire business scenario of the transportation industry.

[0055] Step S1032: Based on the original description information, use natural language processing technology to parse and generate candidate entity objects, attribute objects, and relation objects.

[0056] Step S1033: Using a semantic similarity model, merge and aggregate candidate objects whose semantic similarity reaches the similarity threshold.

[0057] Step S1034: Assign a unique standardized code to each aggregated independent object, and classify and enter it into the entity library, attribute library or relation library according to its object type.

[0058] In this embodiment, after merging and aggregation but before categorizing and entering data, the process includes: Each individual object after aggregation undergoes human-machine collaborative review. Based on the review results, its standardized code and library category are ultimately determined. The human-machine collaborative review provides semantic conflict prompts and confidence assessments for human decision-making.

[0059] Based on the steps above, it can be seen that entities in the entity database can be created manually or non-manually. Manual methods involve technicians manually entering entity parameters to create entity objects. Non-manual methods involve using information extraction algorithms to extract structured information from various data, generating a candidate entity set, and then using AI algorithms such as BERT or predefined rules to perform entity similarity verification, achieving semantic aggregation of synonymous entities and simultaneously establishing mapping relationships. Newly created entities need to be matched with the entity database they are added to. Additionally, manual review can be added to provide conflict alerts and determine whether synonymous entities should be merged, ultimately generating a dynamic, unified entity database across standards.

[0060] For attribute databases, attributes can be created manually or semi-manually, similar to entity creation. Natural language processing algorithms can be used to extract attribute objects, and then algorithms like BERT can be used to calculate similarity, determine whether attributes should be merged, fuse multi-source entity data, and simultaneously establish mapping relationships. New attributes can be continuously collected through online learning algorithms and automatically added to the database after comprehensive verification including manual review. Attributes can be set up based on parameters such as attribute classification and attribute tags, or through manual selection, ultimately generating a multi-source data fusion attribute database.

[0061] For a relational database, a relation represents a metadata model dimension. Based on business needs, various relations can be established using multiple methods. For example, there are basic relation types defined based on standards in the engineering field, such as structural composition and temporal sequence; business relations defined based on actual business needs; and relations discovered through the deployment of artificial intelligence algorithms and relation path reasoning, which are then added to a list of relations to be verified. After professional review, relation types and relation chains can be further formed, ultimately generating a relational database that supports business operations.

[0062] Therefore, this embodiment realizes the automated construction of unstructured text into a high-quality structured knowledge base, providing a reliable and consistent data foundation for upper-layer data fusion and intelligent applications.

[0063] In one embodiment, the specific implementation of step S105 in the above embodiments includes: Step S1051: Select the starting entity based on the business scenario.

[0064] Metadata models typically have unique starting or core nodes. The selection logic for the starting node is usually determined by the entry entity of a single relationship or business scenario. For example, in a tunnel structure resolution relationship, the "tunnel" entity serves as the starting node. The selection criterion is that there is no superior entity within the relationship, and the starting node should be the initiator of data flow or the root entity of business logic.

[0065] Step S1052: Based on the scenario requirements of the business scenario, determine a predefined relationship type in the relational database, and use the foreign key corresponding to the relationship type as the basis for association to query and associate other entities related to the starting entity from the entity database to obtain a set of associated entities.

[0066] Step S1053: Based on logical analysis, determine the core entity from the set of related entities.

[0067] Core nodes are key entities that bear the primary responsibility for data association. For example, in a tunnel asset management scenario, all asset management objects—that is, engineering entities under structural analysis—are core nodes, supporting the entire asset management business. They can be automatically identified through business importance analysis based on factors such as usage frequency and data volume, or through AI algorithms such as cluster analysis, or can be manually designated.

[0068] Step S1054: Using the foreign key as a chain, associate the starting entity, core entity and other related entities, and dynamically match the attributes from the attribute library for each entity to generate the metadata model corresponding to the business scenario.

[0069] In this model, the model is constructed based on the dependencies of the metadata model, using the foreign keys corresponding to those dependencies. For example, in the structural resolution relationship of a tunnel, the foreign key is the unique code of its parent object, and associations are formed sequentially by matching the unique codes of the parent objects. Specific rules exist for entity association. When constructing the metadata model, association rules for different entities under different relationships can be predefined. These predefined rules can be bound to scenarios, relationships, or entities and triggered when the model is generated. For example, the entity "middle wall" only exists when the tunnel type is a continuous arch tunnel. Therefore, when the tunnel type is unknown or the known tunnel type is a continuous power supply tunnel, the entity "middle wall" is associated; otherwise, no association occurs. Furthermore, the metadata model for specific business scenarios can also be managed as entities, further supporting the construction of larger business logic structure models.

[0070] Thus, this embodiment realizes the intelligent model assembly from static data association to business objectives, significantly improving the fit and practicality of the metadata model with complex business scenarios.

[0071] In one embodiment, the above embodiments further include: Step S106: Dynamically update the multi-dimensional classification system model and the metadata model.

[0072] In this step, based on incremental learning algorithms and industry development research, new entities and attributes are continuously identified, and the attribute database and relation database are updated. A corresponding rule matrix is ​​established based on current standards, and the current standard metadata is digitized to support rapid data conversion. Therefore, a specific implementation of step S106 includes: Step S1061: Using an incremental learning algorithm, continuously monitor and analyze the data source to obtain the change requirements of objects in the entity database, attribute database, and relation database. The change requirements include the addition of objects, attribute changes, or association conflicts.

[0073] Step S1062: If the change request meets the difference triggering condition, then review the change object and update the entity database, attribute database or relation database according to the review result; otherwise, update the entity database, attribute database or relation database directly according to the change request.

[0074] Step S1063: Based on the updated entity database, attribute database, and relation database, reconfigure the corresponding metadata and update the corresponding metadata model; Step S1064: Based on the updated content of the entity library and attribute library, synchronously update the preset mapping rules between them and at least one external classification standard to maintain the consistency of data conversion and interoperability between the multi-dimensional classification system model and the external classification standard.

[0075] In this embodiment, mapping rules are defined and maintained through a rule matrix. The rule matrix records the mapping relationships and data transformation logic between objects in the entity and attribute databases and corresponding entries in external classification standards. Thus, this embodiment constructs an online learning framework and employs an incremental decision tree learning algorithm, such as Hoeffding Tree, to process new online data streams in real time. When the metadata, entities, and attributes of new data differ from existing categories by more than 30%, they automatically enter the classification queue, triggering a classification review process. This allows for continuous updating of entities and attributes in industry standards, research reports, best practices, books, papers, and other texts through artificial intelligence algorithms, constantly enriching the entity and attribute databases and the metadata model. Simultaneously, a mapping matrix is ​​established between the new data and existing standards, systems, and projects, using the metadata of this invention as a medium to construct a multi-source heterogeneous data engine.

[0076] Therefore, the classification system in this embodiment has the ability to proactively perceive, intelligently review, and continuously evolve, solving the fundamental problem that traditional classification standards lag behind technological development and achieving rapid compatibility with new facilities and technologies.

[0077] In the example above, the difference trigger condition is a difference degree greater than 30%. In other embodiments, 25%, 36%, 45%, etc. can be selected.

[0078] Based on the above embodiments, the following example illustrates the tunnel asset management business and related data in transportation infrastructure: Step S101: Construction of a multi-dimensional classification system The construction of a multi-dimensional classification system is the foundation of the entire invention, comprising a data primitive layer, a metadata kernel layer, a metadata model layer, and a basic data layer. In practical implementation, taking tunnel asset management in transportation infrastructure as an example, the multi-dimensional classification system is constructed through the following steps: (1) Data sorting and classification axis definition: use Figure 2 The multi-dimensional classification system model shown divides data into nine categories: entities, attributes, relationships, metadata, metadata models, business scenarios, master data, structured data, and unstructured data. Data is organized and classification axes are defined based on the following standards: Building Information Modeling Classification and Coding Standard (GB / T51269-2017), Unified Standard for Highway Engineering Information Modeling Application (JTG / T2420-2021), Technical Specification for Highway Tunnel Maintenance (JTGH12-2015), and Classification and Coding Standard for Highway Engineering Information Modeling (DB44 / T2490-2024).

[0079] (2) Realization of a three-dimensional orthogonal classification system use Figure 3 The illustrated three-dimensional orthogonal classification system decouples entities, attributes, and relationships into independent classification axes, forming a three-dimensional orthogonal classification system. By decomposing the semantic kernel of heterogeneous data elements, the three-dimensional orthogonal classification system deconstructs traditional tree-based classification methods into independently defined axes for entities, attributes, and relationships. Each axis is independently encoded and mutually exclusive, eliminating semantic overlap in multi-standard classifications. This embodiment uses a tunnel as a transportation infrastructure object, decomposing it at the structural level to identify entities such as the tunnel entrance, lighting facilities, and lining, along with their related attributes.

[0080] (3) Implementation of the four-layer architecture Based on the three-dimensional orthogonal classification system, such as Figure 4 As shown, a four-layer architecture is constructed, consisting of a data primitive layer, a metadata kernel layer, a metadata model layer, and a basic data layer: Data primitive layer: Generates independent semantic units, such as the "lining" entity and the "material type" attribute.

[0081] Metadata kernel layer: Establish a standardized metadata framework, assign a unique code to each entity, attribute, and relationship, and define constraints, such as "the main cable only exists on suspension bridges".

[0082] Metadata Model Layer: With "tunnel" as the core entity and "structural resolution" as the relation class, it connects sub-entities such as "civil engineering structure," "mechanical and electrical facilities," and "other engineering facilities," as well as component entities such as openings and linings, to form the tunnel structure metadata model.

[0083] Basic data layer: Integrates structured data such as tunnel asset management lists, unstructured data such as inspection images, and semi-structured data such as monitoring time series data, and maps them to the metadata kernel through a data transformation engine to form a unified data resource.

[0084] Step S102: Define entities, attributes, and relationships. (1) Define multidimensional indicators for entities The multi-dimensional metrics for creating entities are defined as shown in Table 1, and the meanings and examples of different entity classes are shown in Table 2. In this embodiment, for example, the central wall only exists when the tunnel type is a continuous arch tunnel, meaning it can be recorded in the constraints of the central wall entity object.

[0085] (2) Define multidimensional indexes of attributes The multi-dimensional metrics for creating attributes are defined as shown in Table 3, and the meanings and examples of different attribute categories are shown in Table 4.

[0086] (3) Define multidimensional indicators of relationships Define the multi-dimensional metrics for creating relationships as shown in Table 5.

[0087] Step S103: Construct entity / attribute / relationship database (1) Constructing relationships This embodiment only uses the hierarchical composition of the tunnel structure as an example to implement the present invention. Therefore, it only considers the relationship types between entities and uses only a single relationship to express this hierarchical relationship. The relationship is created manually and named "structural hierarchy". The specific parameters are shown in Table 6.

[0088] Table 6. Structural Analytical Relationships

[0089] (2) Construct an entity / attribute library In this embodiment, a combination of manual and semi-manual methods is used to extract and create traffic infrastructure entities and attributes. First, based on a large language model and data extraction algorithms, a system is constructed as follows: Figure 5 The system shown is for extracting entities and attributes of transportation infrastructure. It uses a web-based architecture, allowing users to access it via a browser and upload relevant standards, design documents, and other materials for information extraction. Specifically, for example... Figure 6As shown, users can click or drag files to submit documents on the upload interface, supporting PDF, Word, and Excel formats. For example, in this embodiment, a portion of the "Unified Standard for the Application of Highway Engineering Information Modeling" was uploaded.

[0090] After the system receives the document, the user clicks the "Start Extracting Information" button to trigger the information extraction process. For example... Figure 7 As shown, the system automatically parses document content, identifies entity types such as bridges and tunnels, and extracts key attribute information. For bridge entities, extracted attributes include length, width, main span, structural form, construction time, and design unit; for tunnel entities, attributes such as length, width, burial depth, excavation method, lining type, and construction time are extracted. After extraction, the system displays the results in a structured table format and supports user manual verification and re-extraction of the identified content to ensure data accuracy.

[0091] The system then proceeds to the review stage. For example... Figure 8 As shown, the review interface provides a visual display of entity attribute information. Users can view the system's extraction results and manually confirm and correct them using the "Review Comments" input box and the "Confidence Rating" function. Users can choose "Approved" or "Rejected for Modification" based on the recognition accuracy, realizing a human-machine collaborative review mechanism. After the review is completed, the system generates an entity list and an attribute list, as shown in the attached figures. Figure 9 and attached Figure 10 As shown in the figure. The entity list summarizes the identified entity types and the number of their attributes, while the attribute list details the name, data type, unit, and other parameters of each attribute. For example, the length attribute is a numeric type and the unit is meters.

[0092] This embodiment supports manual intervention to supplement the entity library and attribute library. (See attached...) Figure 11 As shown, users can manually add new entities or attributes in the entity attribute management interface, filling in parameters such as Chinese name, English name, classification code, and definition, and manage them through functions such as exporting data and clearing data. Through the above process, the system can batch process multiple standards and design documents. The extracted entity and attribute lists will be deduplicated, normalized, and uniformly coded, and missing parameter values ​​will be supplemented, ultimately forming a complete entity and attribute library.

[0093] The entire process combines automated extraction with manual review, improving the efficiency and reliability of data construction and providing a structured foundation for infrastructure information management. This example extracts data from standards such as the "Unified Standard for Application of Highway Engineering Information Modeling," "Classification and Coding Standard for Highway Engineering Information Modeling," "Technical Specifications for Highway Tunnel Maintenance," and "Classification and Coding of Intelligent Operation and Maintenance Data Information for Bridges, Islandes, and Tunnels," and then constructs the entity database in Table 1 and the attribute database in Table 2 through manual screening and supplementation.

[0094] Step S104: Configure metadata In this embodiment, metadata configuration is a key aspect of tunnel asset management business scenarios, aiming to achieve data standardization and dynamic adaptation through a three-dimensional orthogonal classification system. For example... Figure 2 As shown, metadata configuration is a core step in the invention process. Based on the established entity, attribute, and relational databases, it ensures data interoperability throughout its entire lifecycle. The following uses three engineering entities at different structural levels—tunnels, civil structures, and linings—as examples to illustrate various methods of metadata configuration. Each method supports dynamic expansion and can be optimized through rule engines and AI algorithms, demonstrating the "full-scenario adaptability" advantage of this invention.

[0095] (1) Entity-attribute association hooking method This method is suitable for fine-grained configuration, manually associating entities and attributes one by one to ensure the accuracy and completeness of metadata. Taking the superstructure entity as an example, the configuration process includes: filtering attributes related to the superstructure from the attribute library and defining hierarchical constraints based on the "structure resolution" relationship in the relational database. The resulting metadata is shown in Table 7. Although this method is time-consuming, it is suitable for entity associations with a small number of attributes or special scenarios, ensuring that metadata is not redundant.

[0096] Table 7. Civil Engineering Structure Metadata

[0097] (2) Predefined rule fast association method To improve efficiency, this invention supports batch association of attributes through predefined rules. Specifically, based on parameters such as entity category, name, and attribute category, tag, etc., the system automatically associates a common set of attributes with entities that meet the conditions. For example, all engineering entities share attributes such as name, unique code, parent name, and parent unique code, which are already tagged as "engineering entity" in the attribute library. Taking a tunnel entity as an example, the rule engine automatically loads these attributes to form the metadata shown in Table 8. This method significantly reduces manual intervention and is particularly suitable for standardized, large-scale metadata generation.

[0098] Table 8. Tunnel Metadata

[0099] (3) AI-assisted association and manual optimization methods To address complex business scenarios, this invention introduces a large AI model for intelligent attribute recommendation. The system automatically generates pre-association schemes based on entity object characteristics, business scenario requirements, and attribute parameter similarity. Taking lining entities as an example, the AI ​​model recommends attributes related to lining. The pre-association results require manual review, and after verification and adjustment, the metadata shown in Table 9 is generated. This method combines the efficiency of AI with the accuracy of human intervention, supports incremental learning, and continuously optimizes metadata quality.

[0100] Table 9. Lining Metadata

[0101] The above three methods can be flexibly combined: first, basic attributes are quickly associated through predefined rules; then, business-specific attributes are added with AI assistance; and finally, manual review and optimization are performed. This hybrid process ensures that metadata is both efficient and accurate. Furthermore, this invention supports multi-scenario adaptation: the same entity can be configured with different metadata models in different business scenarios, thereby supporting flexible application of the same entity database and attribute database across different application scenarios, achieving "one source, multiple uses" of data, and providing a high-quality structured foundation for AI model training.

[0102] Step S105: Construct the metadata model The creation method of the metadata model is basically the same as the configuration method of metadata, using three methods: manual creation, rapid creation based on predefined rules, and AI-assisted creation via an all-in-one machine. The only difference is that objects are replaced by relationships between entities. Different entity objects are linked through foreign keys under specific relationships, forming a clearly structured metadata model. This embodiment creates a metadata model for management objects in the tunnel asset management business scenario, using structural parsing as the unique relationship. Specific metadata model results are shown in Table 10. The same method can be used to establish business process metadata models and other models for asset management business. The application of different metadata models can support data production, management, governance, and application for specific businesses.

[0103] Table 10. Tunnel Metadata Model

[0104] Step S106: Dynamic Expansion and Update of Metadata Model Taking the expansion of tunnel lighting facilities in this embodiment as an example, the dynamic update process is explained in detail: (1) Automatic monitoring and triggering mechanism The system analyzes business data streams in real time using incremental learning frameworks such as the HoeffdingTree algorithm, including unstructured data such as inspection records and equipment ledgers. When the frequency of new entities such as "light poles" exceeds a preset threshold, such as more than 10 times within 30 days, the system automatically identifies the differences between them and the metadata of existing lighting facilities, triggering a dynamic expansion process. This process requires no manual intervention, ensuring timely response to business changes.

[0105] (2) Dynamic addition of entities and attributes To address the need for expansion of tunnel lighting facilities, the system performs the following steps: Entity Extraction and Normalization: "Light Pole" entities are extracted from design documents or inspection reports using natural language processing technology, assigned unique codes, and categorized as "Engineering Entities." The system automatically checks for semantic conflicts, such as distinguishing them from existing "Light Fixture" entities, to avoid duplication.

[0106] Attribute association: Based on a rule engine, it automatically binds identifying attributes such as name and unique code, and recommends technical attributes such as "structural form" and "material type". These attributes are quickly matched using predefined tags such as the "engineering entity" tag set, reducing manual configuration.

[0107] Relationship Assignment: Based on the "Structure Resolution" relationship, the "Light Pole" is associated with the "Lighting Facility" sub-node. The system automatically generates foreign keys for the relationship and updates the metadata model hierarchy.

[0108] (3) Human-machine collaborative review After automatic triggering, the system generates a change plan and submits it to the review interface. Experts confirm the reasonableness of the association using confidence ratings such as "high," "medium," and "low." For example: If it is detected that a "light pole" needs to add a "solar power supply" attribute, its necessity will be manually reviewed. If a conflict arises (such as overlap with existing codes), the system will prompt for manual intervention to ensure data consistency.

[0109] Once the review is approved, the version identifier is automatically updated, and a change log is recorded.

[0110] (4) Synchronization of metadata model with master data Update operations are synchronized to the master database in real time, including: Model Refactoring: The metadata model layer dynamically adjusts its hierarchy, for example, by adding a "light pole" subclass under lighting facilities and associating it with the corresponding attribute set.

[0111] Data synchronization: Master data instances can be updated in batches based on uploaded data or automatic extraction, such as adding light poles to the tunnel asset list and inheriting the associated attributes of the parent facility, such as "unique code of the parent facility".

[0112] In one embodiment, such as Figure 12 As shown, this application also provides a dynamic multi-dimensional intelligent classification device for the transportation industry that is adaptable to all scenarios, including: System construction module 1201 is used to construct a multi-dimensional classification system model with entities, attributes, and relationships as three orthogonal classification axes; Object definition module 1202 is used to define objects for entities, attributes, and relationships; The standard database building module 1203 is used to extract entity, attribute and relationship information from multi-source data of the entire business scenario of the transportation industry according to object definition, and form a standardized entity database, attribute database and relationship database. The metadata configuration module 1204 is used to dynamically associate attributes in the attribute library with entities in the entity library as the core and configure metadata centered on entities according to predefined rules. The scenario modeling module 1205 is used to generate a metadata model corresponding to a business scenario by associating multiple metadata with relationships in a relational database based on the business scenario.

[0113] In one embodiment, such as Figure 13 As shown, this application also provides a computer device 1300, comprising: The communication interface 1301 allows for information exchange with other devices or network nodes.

[0114] One or more processors 1302 are connected to a communication interface 1301 to enable information interaction with other devices or network nodes, and to execute the methods provided by one or more technical solutions in the above embodiments when running computer programs.

[0115] Memory 1303 is used to store computer-readable instructions that can be executed on processor 1302. When executed by one or more processors 1302, the computer-readable instructions perform the steps of the software development method as described in the above embodiments.

[0116] The computer device 1300 in this embodiment only shows 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. A specific computer device may include more or fewer components than shown in the figures, or combine certain components, or have different component arrangements, such as power supplies, input / output interfaces, etc. Furthermore, the computer device 1300 of this embodiment can operate on an operating system stored in the memory 1303, such as Windows Server™, Mac OS X™, Unix™, Linux™, Free BSD™, or similar.

[0117] In one embodiment, this application also provides a storage medium storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the software development method as described in the above embodiments.

[0118] The computer-readable instructions in the aforementioned computer device 1300 and storage medium may be application programs.

[0119] In addition, the specific descriptions of the technical effects and steps of the above-described embodiments of the dynamic multi-dimensional intelligent classification device for the transportation industry adapted to all scenarios, a computer device 1300, and a storage medium are all based on the relevant descriptions of the embodiments of the dynamic multi-dimensional intelligent classification method for the transportation industry adapted to all scenarios.

[0120] Since the systems / devices described in the above embodiments of the present invention are systems / devices used to implement the methods of the above embodiments of the present invention, those skilled in the art can understand the specific structure and modifications of the systems / devices based on the methods described in the above embodiments of the present invention, and therefore will not be repeated here. All systems / devices used in the methods of the above embodiments of the present invention fall within the scope of protection of the present invention.

[0121] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0122] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (devices), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions.

[0123] It should be noted that any reference numerals placed between parentheses in the claims should not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In claims that enumerate several means, several of these means may be embodied by the same hardware. The use of the terms first, second, third, etc., is merely for convenience of expression and does not indicate any order. These terms can be understood as part of the component names.

[0124] Furthermore, it should be noted that in the description of this specification, the terms "one embodiment," "some embodiments," "embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0125] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the claims should be interpreted to include both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0126] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, then this invention should also include these modifications and variations.

Claims

1. A dynamic multidimensional intelligent classification method for the transportation industry, adaptable to all scenarios, characterized in that: Includes the following steps: Construct a multi-dimensional classification system model with entities, attributes, and relationships as three orthogonal classification axes; Define the entity, the attribute, and the relationship as an object; Based on the object definition, information about the entity, the attribute, and the relationship is extracted from multi-source data across all business scenarios in the transportation industry to form a standardized entity library, attribute library, and relationship library. Using entities in the entity library as the core, attributes in the attribute library are dynamically associated according to predefined rules to form entity-centric metadata. Based on the business scenario, multiple metadata are associated using the relationships in the relational database to generate a metadata model corresponding to the business scenario.

2. The method according to claim 1, characterized in that, The generation of the metadata model corresponding to the business scenario includes: Select the starting entity based on the business scenario; Based on the scenario requirements of the business scenario, a predefined relationship type in the relationship database is determined, and other entities related to the starting entity are queried and associated from the entity database using the foreign key corresponding to the relationship type as the association basis, to obtain a set of associated entities; Based on logical analysis, the core entity is determined from the set of related entities; Using the foreign key as a chain, the starting entity, the core entity, and other related entities are linked, and attributes from the attribute library are dynamically matched for each entity, thereby generating a metadata model corresponding to the business scenario.

3. The method according to claim 1, characterized in that, Also includes: The multi-dimensional classification system model and the metadata model are dynamically updated.

4. The method according to claim 3, characterized in that, The dynamic updating of the multi-dimensional classification system model and the metadata model includes: An incremental learning algorithm is used to continuously monitor and analyze the data source to obtain the change requirements of objects in the entity database, attribute database and relation database. The change requirements include the addition of objects, attribute changes or association conflicts. When determining whether the change request meets the difference triggering condition, if so, the change object is reviewed and the entity library, attribute library or relation library is updated according to the review result; otherwise, the entity library, attribute library or relation library is updated directly according to the change request. Based on the updated entity database, attribute database, and relation database, reconfigure the corresponding metadata and update the corresponding metadata model; Based on the updated content of the entity library and the attribute library, the preset mapping rules between them and at least one external classification standard are updated synchronously to maintain the consistency of data conversion interoperability between the multi-dimensional classification system model and the external classification standard.

5. The method according to claim 4, characterized in that, The mapping rules are defined and maintained through a rule matrix, which records the mapping relationship and data conversion logic between objects in the entity library and attribute library and corresponding entries in the external classification standard.

6. The method according to claim 1, characterized in that, The multi-dimensional classification system model includes the following components working in sequence: The data primitive layer is used to process the input raw data and output entities, attributes, and relationships to the metadata kernel layer; The metadata kernel layer is constructed using entities, attributes, and relationships as three orthogonal classification axes. The metadata model layer is used to assemble the metadata output by the metadata kernel layer to form the corresponding metadata model; The basic data layer is used to provide raw data input to the data primitive layer.

7. The method according to any one of claims 1 to 6, characterized in that, The process of defining objects for the entity, the attribute, and the relationship includes: Configure constraints for each defined object, which are used to automatically verify the compliance of associations, matching and combination between objects in subsequent steps.

8. The method according to any one of claims 1 to 6, characterized in that, The step of extracting information about the entity, the attribute, and the relationship from multi-source data across all business scenarios in the transportation industry, based on the object definition, to form a standardized entity library, attribute library, and relationship library, includes: Identify original descriptive information related to entities, attributes, and relationships from multi-source data across all business scenarios in the transportation industry; Based on the original description information, natural language processing technology is used to parse and generate candidate entity objects, attribute objects, and relation objects; Using a semantic similarity model, candidate objects whose semantic similarity reaches a similarity threshold are merged and aggregated. Each aggregated independent object is assigned a unique standardized code, and according to its object type, it is classified and entered into the entity library, attribute library, or relation library respectively.

9. The method according to claim 8, characterized in that, After the merging and aggregation, and before the classification and entry, the following is included: Each individual object after aggregation undergoes human-machine collaborative review. Based on the review results, its standardized code and library category are ultimately determined. The human-machine collaborative review provides semantic conflict prompts and confidence assessments for human decision-making.

10. A dynamic multi-dimensional intelligent classification device for the transportation industry, adaptable to all scenarios, characterized in that: include: The system construction module is used to build a multi-dimensional classification system model with entities, attributes, and relationships as three orthogonal classification axes. The object definition module is used to define the entity, the attribute, and the relationship as objects; The standard database building module is used to extract information about the entities, attributes, and relationships from multi-source data across all business scenarios in the transportation industry, based on the object definition, to form a standardized entity database, attribute database, and relationship database. The metadata configuration module is used to dynamically associate attributes in the attribute library with entities in the entity library as the core, and configure metadata centered on entities according to predefined rules. The scenario modeling module is used to associate multiple metadata with the relationships in the relational database based on the business scenario, and generate a metadata model corresponding to the business scenario.

Images