Digital base fusion system and electronic equipment

Abstract

The present application relates to a digital base fusion system and electronic equipment. The system includes: a data source access module, which is used to obtain raw data from design tools and data sources in different fields; a data abstraction module, which is used to manage raw data and provide a unified data access interface for upper-level modules; a semantic modeling module, which is used to build a unified semantic model; a semantic alignment module, which is used to perform semantic analysis and matching on raw data from different fields; a data fusion module, which is used to fuse semantically aligned data of corresponding types based on different fusion methods to build a unified data view; a knowledge graph construction module, which is used to build a knowledge graph based on the semantic model and the fused data; an application service module, which is used to provide data services and knowledge services for upper-level applications. The present application solves the problem that the existing digital base architecture lacks a unified semantic conversion and expression mechanism, resulting in difficulties in cross-domain data integration.

Description

Technical Field

[0001] The present application relates to the field of data processing, and in particular to a digital base fusion system and electronic equipment. Background Art

[0002] In the field of high-end equipment design and manufacturing, the focus is on solving the problem of system-level data governance in multidisciplinary collaborative development. In the current industry, in the collaborative process of multiple fields such as mechanics, electronics, and software, there are common problems with heterogeneous design tools and fragmented data formats, which have formed a serious data island effect and also led to a large number of redundant copies of the same data entity in different systems. These copies are scattered and stored in independent data sources in each field, and version management is chaotic and consistency maintenance is difficult. Due to the lack of a unified semantic conversion and expression mechanism, the existing digital base architecture faces three core challenges in cross-domain data integration: first, the feature recognition and alignment of multimodal data such as mechanical three-dimensional models, electronic simulation data, and software logic code are difficult; second, the non-linear reference relationship between complex product data is difficult to effectively model and maintain; third, the semantic ambiguity caused by differences in terminology systems and standard specifications in various fields.

[0003] The existing digital base architecture lacks a unified semantic conversion and expression mechanism, which leads to difficulties in cross-domain data integration. No effective solution has been proposed yet. Summary of the invention

[0004] The digital base fusion system and electronic device provided in the embodiments of the present application at least solve the problem that the existing digital base architecture lacks a unified semantic conversion and expression mechanism, which leads to difficulties in cross-domain data integration.

[0005] According to one aspect of an embodiment of the present application, a digital base fusion system is provided, including: a data source access module, used to obtain raw data from design tools and data sources in different fields; a data abstraction module, used to manage raw data and provide a unified data access interface for upper-level modules; a semantic modeling module, used to build a unified semantic model, wherein the semantic model is used to describe and associate raw data from different fields; a semantic alignment module, used to perform semantic analysis and matching on raw data from different fields to identify the same or similar concepts, entities and relationships; a data fusion module, used to fuse semantically aligned data of corresponding types based on different fusion methods to build a unified data view; a knowledge graph construction module, used to build a knowledge graph based on the semantic model and the data fused by the data fusion module, wherein the knowledge graph is used to store and manage cross-domain knowledge; an application service module, used to provide data services and knowledge services for upper-level applications.

[0006] Optionally, the data source access module is used to implement the following functions: access structured data sources based on a database connection pool or data connection technology; store multimodal data based on object storage, wherein the multimodal data includes: documents, pictures and videos; select corresponding open source tools and libraries according to the data type of the multimodal data to pre-process the multimodal data; identify and extract features of the multimodal data based on a deep learning model and a multimodal model.

[0007] Optionally, the data abstraction module is used to implement the following functions: store and manage metadata information of raw data; integrate raw data from different data sources into a virtual data set and link different data sources through data connectors; parse and execute query requests submitted by users through a query engine.

[0008] Optionally, the semantic modeling module is used to implement the following functions: create domain ontologies and cross-domain ontologies, where ontology is a formalized vocabulary used to describe concepts and relationships in a specific domain; establish mapping relationships between ontologies in different domains based on vocabulary matching, structure matching, and instance matching methods to achieve semantic alignment of cross-domain concepts; semantically annotate data from different domains based on the constructed ontology and mapping relationships, and associate them with concepts in the ontology.

[0009] Optionally, the semantic alignment module is used to implement the following functions: identify entities in the original data based on the large model, and link the entities to corresponding concepts in the domain knowledge base; extract the association relationships between entities; identify concepts with the same or similar functions in the original data, and map concepts with the same or similar functions; convert entities and concepts into vectors, and calculate the similarity between different vectors to determine whether different vectors are semantically related.

[0010] Optionally, the data fusion module is used to implement the following functions: when the semantically aligned data needs to be aligned across domains, the domain ontology is used as a framework for data fusion to map data from different data sources to the concepts, relationships and instances of the ontology; when there is no data conflict and data redundancy in the semantically aligned data, data fusion is implemented based on preset rules; when the semantically aligned data cannot be fused using preset rules, data fusion is implemented based on a machine learning algorithm; when logical relationships are stored between the semantically aligned data and real-time calculations are required, dynamic references and real-time calculations between data stored in different storage systems are implemented through a formula engine compatible with EXCEL formulas to eliminate data redundancy and data conflicts; and online real-time calculations or offline batch calculations are implemented through a high-concurrency formula engine.

[0011] Optionally, the knowledge graph construction module is used to implement the following functions: using a graph database to store knowledge in the form of a graph, wherein the nodes of the graph represent the entities of the knowledge and the edges of the graph represent the relationships of the knowledge; providing knowledge query function, reasoning function and analysis function, wherein the analysis function includes: using predefined rules for reasoning; using statistical methods for reasoning; mapping entities, relationships and attributes in the knowledge graph to a high-dimensional vector space, and searching for nodes with semantically similar target entities or relationships in the high-dimensional vector space through vector retrieval; inputting the retrieval results into a large model, and using the large model to integrate or complete the retrieval results.

[0012] Optionally, the application service module is used to implement the following functions: providing data services, where data services include: data query, data analysis, and real-time data stream processing; providing knowledge services, where knowledge services include: knowledge query, knowledge reasoning, and knowledge recommendation; building a low-code service platform to provide visual data services and knowledge service orchestration tools; optimizing the performance of services such as data query and knowledge reasoning, and establishing a service access control mechanism.

[0013] According to another aspect of an embodiment of the present application, an electronic device is provided, including: a processor, and a memory storing a program, wherein the program includes instructions, and when the instructions are executed by the processor, the processor executes the method executed by the above system.

[0014] According to another aspect of an embodiment of the present application, a non-transitory machine-readable medium storing computer instructions is also provided, where the computer instructions are used to enable a computer to execute the method executed by the above system.

[0015] Beneficial effects of the embodiments of the present application: In an embodiment of the present application, a digital base fusion system is provided, including: a data source access module, which is used to obtain raw data from design tools and data sources in different fields; a data abstraction module, which is used to manage raw data and provide a unified data access interface for upper-level modules; a semantic modeling module, which is used to build a unified semantic model, wherein the semantic model is used to describe and associate raw data from different fields; a semantic alignment module, which is used to semantically analyze and match raw data from different fields to identify the same or similar concepts, entities and relationships; a data fusion module, which is used to fuse the semantically aligned data of corresponding types using different fusion methods to build a unified data view; a knowledge graph construction module, which is used to build a knowledge graph based on the semantic model and the data fused by the data fusion module, wherein the knowledge graph is used to store and manage cross-domain knowledge; an application service module, which is used to provide data services and knowledge services for upper-level applications, designs and implements a unified semantic model for describing and associating data from different fields, and uses advanced AI and machine learning technologies to achieve cross-domain semantic alignment, thereby breaking data silos, improving data integration efficiency, enhancing data analysis capabilities, promoting knowledge sharing and reuse, and ultimately improving the innovation efficiency of high-end equipment design. Compared with existing technologies, it improves the accuracy of data integration and provides a dynamic and adaptive data governance paradigm for data management throughout the life cycle of high-end equipment.

[0016] Details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present application, and for ordinary technicians in this field, other embodiments can be obtained based on these drawings without creative work.

[0018] Figure 1 is a data flow diagram of a digital base fusion system according to an embodiment of the present application; Figure 2 is a system architecture diagram of a digital base fusion system according to an embodiment of the present application; Figure 3 Schematic diagram of the structure of the electronic device of this embodiment. DETAILED DESCRIPTION

[0019] Embodiments of the present embodiment will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present embodiment are shown in the accompanying drawings, it should be understood that the present embodiment can be implemented in various forms and should not be construed as being limited to the embodiments set forth herein, which are instead provided for a more thorough and complete understanding of the present embodiment. It should be understood that the drawings and embodiments of the present embodiment are only for exemplary purposes and are not intended to limit the scope of protection of the present embodiment.

[0020] In order to better understand the embodiments of the present application, the technical terms involved in the embodiments of the present application are explained as follows: Digital base architecture refers to a comprehensive set of information technology infrastructure and platforms built to support the digital transformation of enterprises or organizations. This architecture includes not only hardware facilities such as servers, storage devices and network connections, but also software systems, data management capabilities, security measures, and integrated tools and services. It provides stable, efficient and flexible support for upper-level applications, business processes and user interactions.

[0021] Ontology: It is a formalized knowledge expression model used to describe the concepts, relationships and attributes of a certain field. In the digital base fusion architecture for cross-domain semantic weaving of high-end equipment, ontology can be used to establish a unified semantic model, guide the construction of knowledge graphs, and support semantic query and reasoning.

[0022] Semantic Alignment: It refers to the process of matching and mapping data elements with the same semantics from different data sources. The goal of semantic alignment is to eliminate the semantic differences between data sources and provide a basis for data fusion and knowledge graph construction.

[0023] Data fusion refers to the process of integrating data from multiple data sources to form a unified data view. In the digital base fusion architecture for cross-domain semantic weaving of high-end equipment, the data fusion module is responsible for fusing semantically aligned data to build a unified data view.

[0024] Knowledge Graph: A data structure that uses a graph model to represent knowledge, consisting of nodes (representing entities) and edges (representing relationships). In the digital base fusion architecture for cross-domain semantic weaving of high-end equipment, the knowledge graph is used to store and manage cross-domain knowledge and provide functions such as knowledge query, reasoning, and analysis.

[0025] Graph Database: A database specifically used to store and manage graph data. It can efficiently process the query, traversal, and analysis of graph data. In the knowledge graph construction module, the graph database is used to store knowledge graph data.

[0026] In related technologies, the digital base architecture lacks a unified semantic model, has high integration costs, insufficient scalability, and is difficult to adapt to the needs of different fields. Inconsistent terminology and standards in different fields cause semantic mismatches, and data management and analysis also face challenges. The main problems are as follows: 1) Multimodal data recognition challenges: With the increasing diversification of data sources, multimodal data such as images, audio, and text are pouring into data lakes and data platforms. However, current technologies have major flaws in effectively identifying and understanding multimodal data, making it difficult to accurately extract key information from various types of data, making it difficult to smoothly integrate and collaboratively analyze data of different modalities, and severely limiting the in-depth mining of data value.

[0027] 2) Lack of unified semantic model and cross-domain semantic alignment capabilities: Data lakes and data platforms often simply aggregate data from different fields, lacking a unified semantic model to accurately describe and associate these data. For example, "parts" in mechanical design and "components" in electronic design may refer to the same physical entity, but they cannot be effectively associated and uniformly queried in the existing data architecture, which makes cross-domain data integration and analysis difficult, and knowledge sharing and reuse are even more difficult to achieve.

[0028] 3) Numerous data copies and chaotic management: In the process of data storage and circulation, due to the lack of an effective unified management mechanism, a large number of data copies exist. For example, different departments or business systems use the same process standards, but the same process standards may exist in multiple data sources. Different departments or business systems maintain their own data copies, and it is difficult to update the data source or even the field level when updating the data, which greatly reduces the accuracy and reliability of the data.

[0029] 4) Lack of data reference mechanism, difficult to share: Existing technologies lack an effective mechanism for data to reference each other, and each data module is independent of each other, making it difficult to establish an organic connection. This makes it extremely difficult to share data between different business scenarios and departments, making it impossible to fully realize the collaborative value of data, and limiting the overall data-driven decision-making capabilities of the enterprise.

[0030] 5) It is difficult to ensure the real-time, uniqueness and correctness of data: It is difficult to ensure the uniqueness and correctness of data during the data collection, transmission and processing. Data delays, duplicate data and the emergence of erroneous data will mislead data analysis results, thus affecting the accuracy and timeliness of corporate decision-making.

[0031] Among some existing technical solutions, the data lake architecture is a solution for centralized storage of multi-source heterogeneous data, using Schema-on-Read to support the original storage of structured, semi-structured and unstructured data. Typical technologies include Hadoop Distributed File System (HDFS) and cloud object storage, such as Amazon Simple Storage Service (AWS S3) and Azure Data Lake. AWS S3 is an object storage service provided by Amazon AWS, which provides a highly scalable, secure and easy-to-use infrastructure to store and retrieve any amount of data. Azure Data Lake is a service provided by Microsoft Azure cloud platform, designed to support big data analysis workloads, which allows enterprises to store, process and analyze data at any scale. The advantage of data lake is that it can store massive data at low cost and retain the original state of data, avoiding information loss in traditional ETL process. However, the data lake architecture has significant limitations. First, it lacks a unified semantic model, and the association and dependency between data are only described by basic metadata such as directory trees, resulting in the need to manually construct mapping rules for cross-domain data weaving, forming a "semantic black hole". Secondly, data governance capabilities are weak, and data lineage tracing and version control rely on external systems (such as Delta Lake, which is an open source storage layer designed to bring reliability features to the data lake), making it difficult to ensure the consistency of data entities. These problems limit the application of data lakes in complex scenarios such as high-end equipment manufacturing.

[0032] The data middle platform architecture attempts to solve the data island problem by establishing a unified data platform and data service layer, providing data analysis and application support for business departments. The data middle platform usually includes components such as data warehouse, data lake, and data service. However, the data middle platform still faces challenges when dealing with cross-domain semantic weaving, such as difficulty in coordinating terminology and standards in different fields, resulting in high maintenance costs, difficulty in achieving flexible data integration and knowledge sharing, and inability to adapt to the rapid iteration of process knowledge evolution in the field of high-end equipment, and lack of dynamicity.

[0033] The knowledge graph platform builds an entity relationship network based on the Resource Description Framework (RDF) triples, supports query and logical reasoning, and is widely used in scenarios such as semantic search and intelligent recommendation. Its advantage lies in the ability to explicitly express domain knowledge and implement complex logical reasoning through predefined rules. Typical technology stacks include Neo4j, Apache Jena, etc. However, the application of the knowledge graph platform in the field of high-end equipment manufacturing has limitations. First, its knowledge representation is usually based on static ontology libraries, such as Web Ontology Language (OWL), which is difficult to adapt to the real-time process knowledge update requirements generated during the equipment manufacturing process. Secondly, knowledge extraction relies on the Natural Language Processing (NLP) pipeline, and there is semantic loss when processing non-text multimodal data (such as CAD drawings and BOM tables), which makes engineering implementation difficult.

[0034] In order to solve the above problems and the shortcomings of the prior art, a digital base fusion architecture for cross-domain semantic weaving of high-end equipment is constructed in the embodiment of the present application. This architecture solves the key defects of the existing system through three major technical breakthroughs: innovatively establishes a cross-domain unified semantic model, adopts a large language model and deep learning-driven multimodal feature extraction technology (such as three-dimensional model topology analysis, simulation data semantic deconstruction, code logic abstraction, etc.), realizes the feature-level semantic alignment of multimodal data such as machinery, electronics, and software, and overcomes the semantic ambiguity problem caused by the difference in terminology system in the traditional architecture; the first formula dynamic association engine, through static formula expressions to define the nonlinear dependency relationship between data entities, combined with the topological sorting mechanism of directed acyclic graph (DAG), realizes the intelligent cascade calculation and state synchronization of data changes, and solves the problem of complex product data version confusion and redundant copies; designs a cross-domain knowledge graph construction framework based on deep learning, transforms traditional data islands into cognitive networks with semantic interconnection, and makes data reference relationships traceable and computable. Compared with the existing technology, it improves the accuracy of data integration and provides a dynamic and adaptive data governance paradigm for the data management of the entire life cycle of high-end equipment. The following is a detailed description.

[0035] Figure 1 is a data flow diagram of a digital base fusion system according to an embodiment of the present application, such as Figure 1 As shown, the system includes: The data source access module 10 is used to obtain original data from design tools and data sources in different fields.

[0036] like Figure 1As shown, design tools and data sources in different fields can include: Computer-Aided Design (CAD), Electronic Design Automation (EDA), Product Lifecycle Management (PLM), Product Data Management (PDM) and other systems.

[0037] The data abstraction module 11 is used to manage the original data and provide a unified data access interface for the upper layer modules.

[0038] The semantic modeling module 12 is used to build a unified semantic model, wherein the semantic model is used to describe and associate original data from different fields.

[0039] The semantic alignment module 13 is used to perform semantic analysis and matching on raw data from different fields to identify the same or similar concepts, entities and relationships.

[0040] The data fusion module 14 is used to fuse the semantically aligned data of corresponding types based on different fusion methods to construct a unified data view.

[0041] The knowledge graph construction module 15 is used to construct a knowledge graph based on the semantic model and the data fused by the data fusion module, wherein the knowledge graph is used to store and manage cross-domain knowledge.

[0042] The application service module 16 is used to provide data services and knowledge services for upper-layer applications.

[0043] The application service module 16 provides data services and knowledge services for upper-level applications, such as data query, data analysis, knowledge reasoning, knowledge mining, etc.

[0044] The above-mentioned system provided in the embodiment of the present application designs and implements a unified semantic model for describing and associating data from different fields, and uses advanced AI and machine learning technologies to achieve cross-domain semantic alignment, thereby breaking data silos, improving data integration efficiency, enhancing data analysis capabilities, promoting knowledge sharing and reuse, and ultimately improving the innovation efficiency of high-end equipment design.

[0045] According to an optional embodiment of the present application, the data source access module 10 is used to implement the following functions: accessing a structured data source based on a database connection pool or data connection technology; storing multimodal data based on object storage, wherein the multimodal data includes: documents, pictures and videos; selecting corresponding open source tools and libraries according to the data type of the multimodal data to preprocess the multimodal data; identifying and extracting features from the multimodal data based on a deep learning model and a multimodal model.

[0046] Figure 2 is a system architecture diagram of a digital base fusion system according to an embodiment of the present application, such as Figure 2 As shown, the data source access module 10 is located at the bottom layer of the entire system architecture and is responsible for accessing design tools and data sources from different fields, such as CAD, EDA, PLM, PDM and other systems, as well as the enterprise's internal databases, data warehouses, data lakes, etc.

[0047] The acquired raw data is mainly divided into structured data sources and unstructured data.

[0048] For structured data sources, such as relational databases, data warehouses, and data lakes, the following technologies can be used to access them: 1) By establishing a database connection pool, you can reuse database connections and improve data access efficiency.

[0049] 2) Use database connection technologies JDBC (Java Database Connectivity) and ODBC (OpenDatabase Connectivity) to connect to various types of relational databases. Java Database Connectivity, or JDBC for short, is a Java API for executing SQL statements, which allows Java programs to interact with databases. Open Database Connectivity, or ODBC for short, is a standard programming interface designed to allow applications to access a variety of database management systems.

[0050] 3) Data Lake / Data Warehouse Application Programming Interface (API): Many data lake and data warehouse platforms provide APIs that can be used to read and write data.

[0051] For multimodal data sources (such as documents, images, videos, etc.), you can use the following technologies to access them: 1) Object storage can be used to store multimodal data, which facilitates efficient management and access to unstructured data. For example, this can be achieved through cloud object storage services (such as AWS S3, Alibaba Cloud OSS) or local object storage systems (such as MinIO), and combined with NoSQL databases (such as MongoDB, Cassandra) to store metadata to support complex queries.

[0052] 2) Preprocessing of multimodal data: You can select corresponding open source tools and libraries according to the data type, use natural language processing tools (such as NLTK, spaCy) to process text data, use computer vision tools (such as OpenCV, Detectron2) to parse image data, use video processing tools (such as FFmpeg, MoviePy) to decompose video frames and extract audio tracks, and use audio processing tools (such as Librosa, Torchaudio) to extract acoustic features.

[0053] 3) Recognition and feature extraction of multimodal data: Open source deep learning methods and multimodal models can be used. For example, inter-modal interaction can be achieved based on the Transformer architecture (such as BERT and ViT), and the semantic spaces of different modalities can be aligning using contrastive learning methods (such as CLIP). Alternatively, open source multimodal models (such as OFA and LLaVA) can be used to complete tasks such as image and text generation. This can capture the deep relationships of multimodal data and generate a unified representation, ultimately meeting the diverse needs of multimodal feature extraction and data processing.

[0054] Domain-specific design tools and data sources, such as CAD, EDA, PLM, PDM, etc., can be accessed using the following technologies: 1) Dedicated API: Many design tools and data sources provide dedicated APIs through which data can be obtained.

[0055] 2) Plug-ins / Extensions: Plug-ins or extensions can be developed to identify and extract features from data in a multimodal manner. Although these data are usually in structured or semi-structured form, their complexity and diversity make multimodal methods have significant advantages. For example, design drawings in CAD systems can be regarded as image data. Combined with their accompanying metadata (such as dimensioning, material properties) and text descriptions (such as design instructions), image-text alignment and semantic understanding can be achieved through multimodal models (such as CLIP or OFA); circuit design data generated by EDA tools can be jointly modeled by combining circuit diagrams (images), signal waveforms (timing data) and design documents (text), and deep learning methods (such as neural networks GNN or Transformer) are used to capture design intent and functional characteristics; product life cycle data in PLM / PDM systems can be constructed by integrating structured data (such as BOM tables), unstructured data (such as technical documents, test reports) and images / videos (such as product prototype photos, assembly process records) to build a unified multimodal representation for tasks such as design optimization, quality inspection or knowledge mining. This multimodal processing method can not only make full use of the rich information in the field of high-end equipment, but also discover hidden design rules or potential problems through cross-modal formula engine association, thereby improving the intelligent analysis capabilities of data in specific fields.

[0056] According to another optional embodiment of the present application, the data abstraction module 11 is used to implement the following functions: store and manage metadata information of original data; integrate original data from different data sources into a virtual data set and link different data sources through data connectors; parse and execute query requests submitted by users through a query engine.

[0057] The data abstraction module 11 is responsible for managing the access to all data sources and providing a unified data access interface for the upper-layer modules. This module shields the heterogeneity of the underlying data sources, so that the upper-layer modules do not need to care about the specific type and access method of the data source.

[0058] The data abstraction module 11 mainly implements the following functions: 1) Metadata management: Stores metadata information of all data sources, including data table structure, field definition, data type, data constraints, data relationships, etc. Centralized management of metadata can provide basic capabilities for the data abstraction module and help upper-level modules understand and manage data.

[0059] 2) Data virtualization: Create virtual datasets to integrate data from different data sources into a unified view without the need for physical data migration. Users can access and query data from multiple data sources through virtual datasets as if they were accessing a single data source.

[0060] 3) Data connector: Provides various data connectors for connecting different data sources, such as relational databases, data lakes, NoSQL databases, etc. The data connector encapsulates the access logic of the underlying data source and provides a unified data access interface for the upper-level modules.

[0061] 4) Query Engine: Provides a powerful query engine to parse and execute query requests submitted by users. The query engine can decompose query requests into sub-queries based on metadata information and data connectors, send them to the underlying data source for processing, and return the results to the user in summary.

[0062] As some optional embodiments of the present application, the semantic modeling module 12 is used to implement the following functions: create domain ontologies and cross-domain ontologies, where the ontology is a formalized vocabulary used to describe concepts and relationships in a specific domain; establish mapping relationships between ontologies in different domains based on vocabulary matching, structure matching, and instance matching methods to achieve semantic alignment of cross-domain concepts; semantically annotate data from different domains based on the constructed ontology and mapping relationships, and associate them with concepts in the ontology.

[0063] like Figure 2 As shown, the semantic modeling module 12 is responsible for building a unified semantic model for describing and associating data elements from different fields. This module needs to define cross-domain concepts, entities, relationships, and attributes, and combine the mapping relationship between domain ontology and cross-domain ontology to provide basic support for the semantic processing of multimodal data. On the basis of traditional ontology construction technology, combined with multimodal data recognition technology, semantic modeling can be achieved more efficiently.

[0064] Ontology is a formalized vocabulary used to describe concepts and relationships in a specific domain. Ontology construction technology is used to create domain ontologies and cross-domain ontologies, providing a basis for semantic modeling. Common ontology construction methods include: Top-down approach: Start with the most abstract concept, gradually refine it, and build a complete ontology hierarchy. Bottom-up approach: Extract concepts and relationships from existing data and documents to build an ontology. Hybrid approach: Combine top-down and bottom-up approaches to build an ontology. Ontology construction tools can help create, edit, and manage ontologies.

[0065] Domain ontology construction: For different fields such as mechanics, electronics, and software, domain ontologies are constructed to define concepts, entities, relationships, and attributes within the domain.

[0066] Cross-domain ontology mapping: Establish mapping relationships between ontologies in different domains to achieve semantic alignment of cross-domain concepts.

[0067] Data semantic annotation: Using the constructed ontology and mapping relationships, semantically annotate data elements from different fields and associate them with concepts in the ontology.

[0068] Cross-domain ontology mapping technology: Cross-domain ontology mapping technology is used to establish the correspondence between concepts in different domain ontologies and realize semantic interoperability. Common ontology mapping methods include: vocabulary matching-based method: using the name, synonyms, definition and other information of the concept for matching. Structural matching-based method: using the hierarchy, attributes, relationships and other information of the concept for matching. Instance matching-based method: using the instance data of the concept for matching.

[0069] Semantic annotation technology: Semantic annotation technology is used to associate data elements with concepts in the ontology and give the data a semantic interpretation. Common semantic annotation methods include: Rule-based annotation: Annotation using predefined rules. Machine learning-based annotation: Annotation using machine learning algorithms. Manual annotation: Annotation by humans.

[0070] In the embodiment of the present application, the semantic modeling module 12 constructs domain ontology for different fields such as machinery, electronics, and software to define the core concepts, entities, relationships, and attributes in the domain; at the same time, through cross-domain ontology mapping technology, a semantic alignment relationship between different domain ontologies is established to achieve interoperability of cross-domain concepts. The cross-domain ontology mapping technology combines the characteristics of multimodal data and adopts methods based on vocabulary matching (such as names, synonyms), structure matching (such as hierarchical structures, attribute relationships), and instance matching (such as multimodal instance data) to improve the accuracy and coverage of mapping. In addition, semantic annotation technology gives data semantic interpretation by associating multimodal data elements with concepts in the ontology. Combined with multimodal data recognition technology, semantic annotation can also adopt rule-based methods (using predefined rules to process structured data), machine learning-based methods (using deep learning models such as CLIP and OFA to process unstructured data such as images and texts) and manual annotation methods to ensure the accuracy and consistency of the annotation results. By integrating traditional semantic modeling methods with multimodal data recognition technology, the semantic association and application of multi-domain data can be more comprehensively realized.

[0071] In some optional embodiments of the present application, the semantic alignment module 13 is used to implement the following functions: identify entities in the original data based on the large model, and link the entities to the corresponding concepts in the domain knowledge base; extract the association relationships between entities; identify concepts with the same or similar functions in the original data, and map concepts with the same or similar functions; convert entities and concepts into vectors, and calculate the similarity between different vectors to determine whether different vectors are semantically related.

[0072] like Figure 2As shown, the semantic alignment module 13 realizes semantic analysis and matching of cross-domain high-end equipment data by combining traditional semantic modeling methods, multimodal data recognition technology and the powerful capabilities of large models (such as DeepSeek, etc.), and completes tasks such as entity recognition and linking, relationship extraction, concept mapping and semantic embedding, thereby identifying the same or similar concepts, entities and relationships.

[0073] The semantic alignment module 13 is mainly used to implement the following functions: 1) Entity recognition and linking: With the powerful contextual understanding capabilities of large models (such as DeepSeek or BERT), key entities (such as component names, material names, process parameters, etc.) can be efficiently extracted from design documents, manufacturing data, and test reports. These models are able to handle complex text structures and link entities to standardized terms through pre-trained knowledge bases. For example, "gears" and "high-strength alloy steel" can be quickly identified and associated with corresponding concepts in the domain knowledge base. In addition, based on the fine-tuning capabilities of DeepSeek, terminology and expressions can be optimized for specific fields to further improve the accuracy of entity recognition.

[0074] 2) Relationship extraction: Combining rule-based methods (such as the predefined pattern "consisting of..."), supervised learning (such as SVM, CNN), and the context-aware capabilities of large models (such as DeepSeek or other large models based on the Transformer architecture), the relationship between entities can be extracted from the text. For example, from "the engine is composed of a cylinder, a piston, and a connecting rod", the "composition" relationship between "engine" and "cylinder" and "piston" can be extracted. The advantage of a large model is that it can automatically capture implicit relationship patterns, reduce reliance on manual rules, and support the generalization of cross-domain data.

[0075] 3) Concept Mapping: Through vocabulary matching (such as synonym detection), structure matching (such as hierarchy and attribute comparison), instance matching (such as case data analysis), and large-model-based embedding learning methods (such as vector representations generated by DeepSeek), semantic alignment between ontologies in the fields of mechanics, electronics, and software can be achieved. For example, "electric motor" and "motor" can be identified as similar concepts through vocabulary matching, while "gear transmission" and "motor drive" can be judged similarly through functional features or instance data. The embedding capability of large models can map complex concepts to high-dimensional vector space, and calculate semantic relevance through cosine similarity, significantly improving the accuracy of cross-domain concept alignment.

[0076] 4) Semantic embedding and similarity calculation: Use large models (such as DeepSeek, BERT or other open source series) to generate high-quality semantic vector representations and convert text, entities and concepts into a unified vector space. For example, the semantic vectors of "gear" and "bearing" can evaluate their functional association through cosine similarity. The context sensitivity and multimodal processing capabilities of large models (such as combining images, text and structured data) enable them to perform well in complex scenarios and better capture the semantic connections between multimodal data.

[0077] In addition, with the help of deep learning technology (such as CNN, RNN or Transformer) and the pre-training advantages of large models, semantic analysis capabilities can be further improved. For example, DeepSeek is used to encode the "gearbox design specifications" in the mechanical field and the "motor control system design documents" in the electronic field, and the text similarity is calculated to discover the commonalities of design goals and constraints, thereby supporting cross-domain semantic alignment. By integrating multimodal data recognition technology and the capabilities of large models, images, texts and other forms of data can be analyzed more comprehensively, enhancing the coverage and accuracy of semantic modeling, while reducing the need for manual intervention and improving the level of automation.

[0078] In other optional embodiments of the present application, the data fusion module 14 is used to implement the following functions: when the semantically aligned data needs to be aligned across domains, the domain ontology is used as a framework for data fusion to map data from different data sources to the concepts, relationships and instances of the ontology; when there is no data conflict and data redundancy in the semantically aligned data, data fusion is implemented based on preset rules; when the semantically aligned data cannot be fused using preset rules, data fusion is implemented based on a machine learning algorithm; when logical relationships are stored between the semantically aligned data and real-time calculations are required, dynamic references and real-time calculations between data stored in different storage systems are implemented through a formula engine compatible with EXCEL formulas to eliminate data redundancy and data conflicts; and online real-time calculations or offline batch calculations are implemented through a high-concurrency formula engine.

[0079] The data fusion module 14 is responsible for fusing the semantically aligned data and building a unified data view. This module needs to design data fusion strategies and algorithms, handle data conflicts and redundancies, and ensure data consistency and integrity. The data fusion module 14 is mainly used to achieve the following functions: 1) Ontology-based fusion: Use domain ontology as a framework for data fusion to map data from different data sources to the concepts, relationships, and instances of the ontology. Fusion of gear data (such as size, material, precision, etc.) from different design software to the instance of the "gear" ontology concept. Ontology mapping: Map the schema of the data source to the domain ontology. Instance matching: Identify the same instance from different data sources. Attribute value fusion: Fusion of attribute values ​​of the same instance to handle data conflicts and redundancy.

[0080] The ontology-based fusion method mainly performs data fusion on structured, semi-structured and unstructured data. It is applicable to cross-domain data alignment and semantic modeling. This fusion method emphasizes semantic consistency and is suitable for complex fields, but requires pre-defined ontology.

[0081] 2) Rule-based fusion: Use predefined rules to fuse data, such as giving priority to data with higher accuracy, taking average values, and data with the latest timestamp. Fusion of engine temperature data collected from different sensors, giving priority to sensor data with higher accuracy. Define data fusion rules, such as data priority, data cleaning rules, data conversion rules, etc.

[0082] The rule-based fusion method is mainly used for data fusion of structured, semi-structured and simple unstructured data. It is applicable to data with clear data conflicts and redundancy problems that can be solved by rules. This fusion method is characterized by simple implementation and high efficiency, but low flexibility and difficulty in processing complex semantics.

[0083] 3) Fusion based on machine learning: Use machine learning algorithms to learn data fusion patterns, such as using clustering algorithms to identify identical instances and using regression algorithms to predict missing values. Use machine learning algorithms to predict the fatigue life of a certain gear. The algorithm can fuse data from multiple sources such as design data, manufacturing data, and test data. Through data preprocessing: Clean, transform, and extract features from data. Through model training: Use labeled data or unsupervised learning methods to train data fusion models. Through model evaluation: Evaluate the accuracy and efficiency of data fusion models.

[0084] The fusion method based on machine learning is mainly used for data fusion of unstructured, semi-structured and large-scale heterogeneous data. It is applicable to scenarios where the data pattern is complex and difficult to describe with rules. This fusion method is characterized by strong flexibility and the ability to handle complex relationships, but it has high requirements for data quality and annotation.

[0085] 4) Fusion based on formula engine: Integrate multi-source heterogeneous data into a unified view, and realize dynamic reference and real-time calculation between data through formula expressions. This method can effectively eliminate data redundancy and data conflicts, and can greatly reduce the frequency of conflict detection and conflict resolution, and also ensure the consistency and uniqueness of data. At the same time, it supports efficient calculation in large-scale real-time or offline scenarios through a high-concurrency formula calculation engine.

[0086] Data consistency and uniqueness: The reference relationship between formulas avoids the problem of multiple copies of the same data. For example, the design parameters, manufacturing processes, and test results of high-end equipment can be dynamically associated through formulas to ensure that all references always point to the latest values, improving the accuracy and consistency of data. Formula engine: The formula engine compatible with Excel formulas lowers the user threshold. Users can define complex data logic through intuitive formula syntax without learning complex programming languages. This design significantly improves the applicability of formulas, especially in cross-domain collaboration. The high-concurrency computing engine supports two scenarios: Online real-time computing: Applicable to scenarios that require user input or dynamic adjustment (such as real-time monitoring of device status).

[0087] Offline batch computing: Suitable for large-scale data analysis tasks (such as historical data backtracking or predictive model training).

[0088] For example, in the field of high-end equipment, different systems store scattered data: Table A: stores the design parameters of the gear, including the field module.

[0089] Table B: stores the manufacturing data of gears, including the fields number of teeth (teeth_count) and material (material_strength).

[0090] Table C: stores gear test data, including the fatigue life field (fatigue_life).

[0091] In order to evaluate the comprehensive performance of a gear, these data need to be dynamically linked through formulas and their key indicators calculated.

[0092] The following logic can be defined through the formula engine: Find the number of teeth in manufacturing data: Use the LookUp function to find the number of teeth that meet the conditions from table B: teeth_count = LookUp(B, material_strength>500, teeth_count) This formula means: find the records with material strength greater than 500 in table B and extract the corresponding number of teeth. Calculate the theoretical load-bearing capacity of a gear: According to the gear module and number of teeth, calculate its theoretical load capacity: theoretical_load_capacity = First(A).module * teeth_count * 0.8 This formula combines the module from Table A with the number of teeth calculated in the previous step. Estimation of actual fatigue life: Use the test data in Table C, combined with the theoretical load capacity, to estimate the actual fatigue life of the gear: actual_fatigue_life = First(C).fatigue_life * (theoretical_load_capacity / 1000) This formula dynamically references the fatigue life in Table C and the theoretical load-bearing capacity calculated in the previous step.

[0093] LookUp() refers to a function in the formula engine compatible with Excel formulas that is used to find data that meets the conditions from a certain table.

[0094] First() refers to a function in the formula engine compatible with Excel formulas that is used to obtain specified data from a table. For example, in the above text, First(A).module is used to obtain the module of the gear from table A; First(C).fatigue_life is used to obtain the fatigue life of the gear from table C.

[0095] Advantages of the formula engine: Eliminate redundancy: All data is dynamically referenced through formulas to avoid duplicate storage. For example, the number of teeth only needs to be stored in table B, and other systems can obtain it in real time through formulas.

[0096] Real-time updates: When manufacturing data in Table B changes, all formulas that rely on that data are automatically recalculated, ensuring that the results are always up to date.

[0097] Cross-domain collaboration: Mechanical design engineers, manufacturing process engineers, and test engineers can collaborate through a unified formula engine without the need to manually synchronize data.

[0098] In the digital base fusion architecture for cross-domain semantic weaving of high-end equipment, we can combine technologies such as ontology-based fusion, rule-based fusion, and machine learning-based fusion, and finally integrate all data in a formula-based expression manner to build a unified data view.

[0099] The formula engine-based fusion method is mainly used for data fusion of structured, real-time dynamic data and cross-system reference data. It is applicable to scenarios where there is a clear mathematical or logical relationship between data and real-time calculation is required. This fusion method is characterized by strong dynamics and intuitive formula expression, but it has limited support for complex semantic relationships.

[0100] According to some optional embodiments of the present application, the knowledge graph construction module 15 is used to implement the following functions: using a graph database to store knowledge in the form of a graph, wherein the nodes of the graph represent entities of the knowledge and the edges of the graph represent relationships of the knowledge; providing knowledge query function, reasoning function and analysis function.

[0101] The knowledge graph construction module 15 is responsible for constructing a knowledge graph based on the semantic model and the fused data for storing and managing cross-domain knowledge. This module needs to use technologies such as graph databases to store data in the form of graphs and provide functions such as knowledge query, reasoning and analysis. The analysis functions include: reasoning using predefined rules; reasoning using statistical methods; mapping entities, relationships and attributes in the knowledge graph to a high-dimensional vector space, and searching for nodes with semantically similar semantics to the target entity or relationship in the high-dimensional vector space through vector retrieval; inputting the retrieval results into a large model, and using the large model to integrate or complete the retrieval results.

[0102] The knowledge graph construction module 15 is mainly used to achieve the following functions: 1) Knowledge extraction: Extracting knowledge elements such as entities, relationships, and attributes from multi-source heterogeneous data.

[0103] 2) Entity extraction: Identify and extract concepts that represent objective existence in the text, such as "gear", "bearing", "engine", etc.

[0104] 3) Relationship extraction: Identify semantic relationships between entities, such as the "meshing" relationship between "gear" and "bearing", the "composition" relationship between "engine" and "cylinder", etc.

[0105] 4) Attribute extraction: Extract the attributes and attribute values ​​of an entity, such as the "module", "number of teeth", "material" and other attributes of a "gear".

[0106] 5) Knowledge fusion: Align and merge knowledge from different data sources, eliminate data conflicts and redundancies, and build a unified knowledge base.

[0107] 6) Entity Alignment: Identify identical entities from different data sources, such as matching a “gear” entity from CAD software with a “gear” entity from a product document.

[0108] 7) Relationship Alignment: Identify identical relationships from different data sources, such as matching a “composed of” relationship from a design document with an “assembly” relationship from a manufacturing BOM.

[0109] 8) Attribute value fusion: Fusion of attribute values ​​of the same entity to handle data conflicts and redundancies, such as fusion of the "modulus" attribute value of the "gear" entity from different data sources.

[0110] 9) Knowledge storage: Use technologies such as graph databases to store knowledge in the form of graphs, where nodes represent entities, edges represent relationships, and attributes serve as properties of nodes or edges.

[0111] 10) Graph database selection: Choose a suitable graph database based on factors such as data scale, performance requirements, and functional requirements, such as Neo4j, Nebula, JanusGraph, OrientDB, etc.

[0112] 11) Schema design: Design the schema of the knowledge graph, define entity types, relationship types, attribute types, etc.

[0113] 12) Data import: Import the extracted and fused knowledge into the graph database.

[0114] 13) Knowledge query: Provides a flexible knowledge query interface and supports graph query languages ​​such as SPARQL, making it convenient for users to quickly retrieve and obtain the required knowledge.

[0115] 14) Knowledge reasoning: Use reasoning engines to reason about knowledge, mine implicit knowledge and discover new knowledge.

[0116] 15) Rule-based reasoning: Reasoning using predefined rules, such as "if A is part of B and B is part of C, then A is part of C".

[0117] 16) Statistical-based reasoning: Use statistical methods to perform reasoning, such as inferring the relationship between entities based on their co-occurrence frequency.

[0118] 17) Vector-based multimodal data reasoning: Combined with the vector function support in the graph database, efficient similarity retrieval can be achieved. By mapping the entities, relations, and attributes in the knowledge graph to a high-dimensional vector space (such as using large models such as BERT and CLIP to generate embedded vectors), nodes with semantic similarity to the target entity or relation can be quickly retrieved in the graph database. For example, in the field of high-end equipment, vector retrieval can be used to find parts similar to "gears" in terms of function or design parameters, and combined with their associated multimodal data (such as design drawings, manufacturing process instructions) for comprehensive analysis. The retrieval results can be further input into large models (such as DeepSeek), and their powerful context understanding and generation capabilities can be used to integrate, complete, or generate new knowledge. For example, DeepSeek can generate an optimized gear design based on the design parameters of similar parts retrieved, or infer potential process improvement directions based on historical data. This vector-based multimodal data reasoning method not only improves the semantic expression ability of the knowledge graph, but also provides strong support for the deep mining and intelligent application of cross-modal data.

[0119] 18) Knowledge analysis: Provides functions such as knowledge visualization and knowledge statistical analysis to help users understand and gain insight into knowledge.

[0120] 19) Knowledge visualization: Display the knowledge graph in a graphical way to help users intuitively understand the relationship between knowledge.

[0121] 20) Knowledge statistical analysis: Perform statistical analysis on the knowledge graph, such as calculating the degree distribution of entities, the type distribution of relationships, etc.

[0122] According to another optional embodiment of the present application, the application service module 16 is used to implement the following functions: providing data services, wherein the data services include: data query, data analysis, and real-time data stream processing; providing knowledge services, wherein the knowledge services include: knowledge query, knowledge reasoning, and knowledge recommendation; building a low-code service platform to provide visual data services and knowledge service orchestration tools; optimizing the performance of services such as data query and knowledge reasoning, and establishing a service access control mechanism.

[0123] The application service module 16 is responsible for providing data services and knowledge services for upper-layer applications, such as data query, data analysis, knowledge reasoning, knowledge mining, etc. This module needs to design data interfaces and knowledge interfaces to facilitate application system calls.

[0124] The application service module 16 is mainly used to implement the following functions: 1) Design data interface and knowledge interface: Design RESTful API to provide unified interface specifications to facilitate application systems to call data query, data analysis, knowledge reasoning and other services.

[0125] 2) Provide data services: Data query: Provide flexible data query interface, support structured query and unstructured query. Data analysis: Provide data statistical analysis, data mining, data visualization and other functions to support users to find value from data. Real-time data stream processing: For real-time data, such as sensor data, provide real-time data stream processing function to support real-time monitoring and early warning.

[0126] 3) Provide knowledge services: Knowledge query: Provide semantic-based knowledge query, support users to query knowledge graphs through natural language or keywords. Knowledge reasoning: Use reasoning engines to reason about knowledge, mine implicit knowledge and discover new knowledge, such as fault diagnosis, risk prediction, etc. Knowledge recommendation: Recommend relevant knowledge and data based on the user's query history and interests.

[0127] 4) Build a low-code service platform: In order to facilitate the rapid development and deployment of application systems, a low-code service platform can be built to provide visual data services and knowledge service orchestration tools. Users can orchestrate data query, data analysis, knowledge reasoning and other services by dragging and dropping to quickly build application systems.

[0128] 5) Ensure service quality: Performance optimization: Optimize the performance of services such as data query and knowledge reasoning, such as using caching, indexing, distributed computing and other technologies. Security: Establish a service access control mechanism to ensure data security and privacy. Reliability: Ensure the stability and reliability of services, such as using load balancing, fault tolerance mechanisms and other technologies.

[0129] The digital base fusion system proposed in this application can be applied to the following high-end equipment design and manufacturing scenarios: 1) Cross-domain collaborative design: Support engineers from different fields to conduct collaborative design on a unified digital base. For example, mechanical engineers, electronic engineers, and software engineers can jointly design an intelligent robot.

[0130] 2) Product life cycle management: Supports the management of the entire life cycle of products from design, manufacturing, testing, maintenance to scrapping. For example, it can track product design changes, manufacturing processes, usage and maintenance records.

[0131] 3) Intelligent design and optimization: Supports intelligent design and optimization based on knowledge graphs. For example, knowledge graphs can be used to recommend suitable parts and materials to optimize product structure and performance.

[0132] Through the above technical solution, this application effectively solves the problem of lack of unified semantic model and cross-domain semantic alignment capability in the prior art, and achieves the following technical effects: 1) Multimodal data recognition and semantic alignment capabilities are significantly improved By introducing large language models and multimodal feature extraction technologies driven by deep learning (such as 3D model topology analysis and simulation data semantic deconstruction), feature-level semantic alignment of multimodal data such as machinery, electronics, and software is achieved, solving the semantic ambiguity problem caused by differences in terminology systems in traditional architectures. This enables data from different fields to be accurately identified, understood, and associated, greatly improving the ability to integrate and collaboratively analyze multimodal data. 2) Construction of a unified semantic model across domains An innovative cross-domain unified semantic model was established, combining domain ontology with cross-domain ontology mapping technology to solve the semantic ambiguity caused by differences in terminology systems and standard specifications in the fields of machinery, electronics, software, etc. Through vocabulary matching, structure matching, and instance matching, the interoperability of cross-domain concepts was achieved, providing basic support for the semantic processing of multimodal data. 3) Effective solution to data redundancy and version confusion issues The first formula-based dynamic association engine defines the nonlinear dependency between data entities through static formula expressions, and combines the topological sorting mechanism of DAG directed acyclic graph to realize intelligent cascade calculation and state synchronization of data changes. This solves the problem of chaotic versions and redundant copies of complex product data, ensures data consistency and uniqueness, and reduces manual intervention. 4) Knowledge graphs empower intelligent applications A cross-domain knowledge graph construction framework based on deep learning was designed to transform traditional data islands into cognitive networks with semantic interconnection, making data reference relationships traceable and computable. By combining vector retrieval and large model reasoning capabilities, it supports efficient similarity retrieval, implicit knowledge mining, and intelligent design optimization, providing a dynamic and adaptive data governance paradigm for the full life cycle management of high-end equipment.

[0133] The embodiment of the present application also provides an electronic device, comprising: at least one processor; and a memory connected to the at least one processor in communication. The memory stores a computer program executable by the at least one processor, and the computer program is used to enable the electronic device to perform the method of the embodiment of the present application when executed by the at least one processor.

[0134] An embodiment of the present application also provides a non-transitory machine-readable medium storing a computer program, wherein the computer program, when executed by a processor of a computer, is used to cause the computer to execute the method of the embodiment of the present application.

[0135] The embodiment of the present application also provides a computer program product, including a computer program, wherein the computer program, when executed by a processor of a computer, is used to cause the computer to execute the method of the embodiment of the present application.

[0136] refer to Figure 3 , the structural block diagram of the electronic device that can be used as the server or client of the embodiment of the present application will now be described, which is an example of the hardware device that can be applied to various aspects of the present application. The electronic device is intended to represent various forms of digital electronic computer equipment, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices and other similar computing devices. The components shown herein, their connections and relationships, and their functions are only examples, and are not intended to limit the implementation of the present application described and / or required herein.

[0137] like Figure 3 As shown, the electronic device includes a computing unit 301, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 302 or a computer program loaded from a storage unit 308 into a random access memory (RAM) 303. In the RAM 303, various programs and data required for the operation of the electronic device can also be stored. The computing unit 301, the ROM 302, and the RAM 303 are connected to each other via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0138] Multiple components in the electronic device are connected to the I / O interface 305, including: an input unit 306, an output unit 307, a storage unit 308, and a communication unit 309. The input unit 306 can be any type of device that can input information to the electronic device, and the input unit 306 can receive input digital or character information, and generate key signal input related to user settings and / or function control of the electronic device. The output unit 307 can be any type of device that can present information, and can include but is not limited to a display, a speaker, a video / audio output terminal, a vibrator, and / or a printer. The storage unit 308 can include but is not limited to a disk, an optical disk. The communication unit 309 allows the electronic device to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks, and can include but is not limited to a modem, a network card, an infrared communication device, a wireless communication transceiver, and / or a chipset, such as a Bluetooth device, a WiFi device, a WiMax device, a cellular communication device, and / or the like.

[0139] The computing unit 301 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 301 include, but are not limited to, a CPU, a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 301 performs the various methods and processes described above. For example, in some embodiments, the method embodiments of the present application may be implemented as a computer program, which is tangibly contained in a machine-readable medium, such as a storage unit 308. In some embodiments, part or all of the computer program may be loaded and / or installed on an electronic device via a ROM 302 and / or a communication unit 309. In some embodiments, the computing unit 301 may be configured to perform the above method in any other appropriate manner (e.g., by means of firmware).

[0140] The computer program for implementing the method of the embodiment of the present application can be written in any combination of one or more programming languages. These computer programs can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program can be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.

[0141] In the context of the embodiments of the present application, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or apparatus. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable signal medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, or infrared system, device, or device, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0142] It should be noted that the term "including" and its variations used in the embodiments of the present application are open inclusions, that is, "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one other embodiment"; the term "some embodiments" means "at least some embodiments". The modifications of "one" and "multiple" mentioned in the embodiments of the present application are illustrative and not restrictive. Those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0143] The user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in the embodiments of the present application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0144] The various steps described in the method implementation methods provided in the embodiments of the present application can be performed in different orders and / or in parallel. In addition, the method implementation methods may include additional steps and / or omit the steps shown. The scope of protection of the present application is not limited in this respect.

[0145] The term "embodiment" in this specification refers to specific features, structures or characteristics described in conjunction with the embodiment that can be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily mean the same embodiment, nor does it mean that it is mutually exclusive with other embodiments and is independent or optional. The various embodiments in this specification are described in a related manner, and the same or similar parts between the various embodiments refer to each other. In particular, for the device, equipment, and system embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts refer to the partial description of the method embodiment.

[0146] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of patent protection. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the scope of protection of the present application. Therefore, the scope of protection of the present application shall be subject to the attached claims.

Claims

1. A digital base fusion system, characterized in that: include: Data source access module, used to obtain raw data from design tools and data sources in different fields; A data abstraction module is used to manage the original data and provide a unified data access interface for the upper layer modules; A semantic modeling module, used to build a unified semantic model, wherein the semantic model is used to describe and associate the original data from different fields; A semantic alignment module, used for semantically analyzing and matching the raw data from different fields to identify the same or similar concepts, entities and relationships; The data fusion module is used to fuse the semantically aligned data of corresponding types based on different fusion methods to build a unified data view; A knowledge graph construction module, used to construct a knowledge graph based on the semantic model and the data fused by the data fusion module, wherein the knowledge graph is used to store and manage cross-domain knowledge; The application service module is used to provide data services and knowledge services for upper-level applications.

2. The system according to claim 1, characterized in that The data source access module is used to implement the following functions: Access structured data sources based on database connection pool or data connection technology; Storing multimodal data in an object-based storage manner, wherein the multimodal data includes: documents, pictures and videos; Selecting corresponding open source tools and libraries according to the data type of the multimodal data to preprocess the multimodal data; The multimodal data is identified and features are extracted based on a deep learning model and a multimodal model.

3. The system according to claim 1, characterized in that The data abstraction module is used to implement the following functions: Storing and managing metadata information of the original data; integrating said raw data from different data sources into a virtual data set; Connect different data sources through data connectors; The query engine parses and executes the query requests submitted by users.

4. The system according to claim 1, characterized in that The semantic modeling module is used to implement the following functions: Creating domain ontologies and cross-domain ontologies, where the ontology is a formal vocabulary for describing domain-specific concepts and relationships; The mapping relationship between ontologies in different fields is established based on vocabulary matching, structure matching and instance matching methods to achieve semantic alignment of cross-domain concepts. Based on the constructed ontology and the mapping relationship, data from different fields are semantically annotated and associated with concepts in the ontology.

5. The system according to claim 1, characterized in that The semantic alignment module is used to implement the following functions: Identify entities in the raw data based on the big model, and link the entities to corresponding concepts in the domain knowledge base; Extracting the association relationships between the entities; Identify concepts with the same or similar functions in the original data, and map the concepts with the same or similar functions; The entities and the concepts are converted into vectors, and similarities between different vectors are calculated to determine whether the different vectors are semantically related.

6. The system according to claim 1, characterized in that The data fusion module is used to realize the following functions: When semantically aligned data needs to be aligned across domains, domain ontology is used as a framework for data fusion to map data from different data sources to the concepts, relationships, and instances of the ontology. In the case that there is no data conflict or data redundancy in the semantically aligned data, data fusion is achieved based on preset rules; When the semantically aligned data cannot be fused using the preset rules, data fusion is achieved based on a machine learning algorithm; When logical relationships are stored between semantically aligned data and real-time calculations are required, a formula engine compatible with EXCEL formulas is used to implement dynamic references and real-time calculations between data stored in different storage systems to eliminate data redundancy and data conflicts. And realize online real-time calculation or offline batch calculation through high-concurrency formula engine.

7. The system according to claim 1, characterized in that The knowledge graph construction module is used to achieve the following functions: Using a graph database to store knowledge in the form of a graph, wherein the nodes of the graph represent entities of knowledge and the edges of the graph represent relationships of knowledge; Provides knowledge query function, reasoning function and analysis function, wherein the reasoning function includes: Reasoning using predefined rules; Use statistical methods to make inferences; Map entities, relationships and attributes in the knowledge graph to a high-dimensional vector space, and retrieve nodes with semantically similar target entities or relationships in the high-dimensional vector space through vector retrieval; input the retrieval results into a large model, and use the large model to integrate or complete the retrieval results.

8. The system according to claim 1, characterized in that The application service module is used to implement the following functions: Providing data services, wherein the data services include: data query, data analysis and real-time data stream processing; Providing knowledge services, wherein the knowledge services include: knowledge query, knowledge reasoning and knowledge recommendation; Build a low-code service platform to provide visual data services and knowledge service orchestration tools; Optimize the performance of services such as data query and knowledge reasoning, and establish service access control mechanisms.

9. An electronic device, comprising: A processor, and a memory storing a program, wherein the program comprises instructions, and when the instructions are executed by the processor, the processor executes the method executed by the system according to any one of claims 1 to 8.

10. A non-transitory machine-readable medium storing computer instructions, characterized in that: The computer instructions are used to cause the computer to execute the method performed by the system according to any one of claims 1 to 8.

Images