GBase Insights | Integrating a Chinese-Developed Relational Database with Knowledge Graphs
Knowledge Graph is an important branch of artificial intelligence technology. To build a knowledge graph, data must first be obtained, which may include tables, text, databases, audio, video, images, etc. Depending on the type, data can be classified as structured, unstructured, or semi-structured. Information must be extracted from these data to further build a knowledge graph. When receiving data from different sources, knowledge fusion is required—that is, merging entities that represent the same concept and combining datasets from multiple sources into a single dataset, on which basis the corresponding knowledge graph can be constructed.
Internationally, the AI community has widely recognized that the application of knowledge graph technology is a key factor in achieving the breakthrough from “perception” to “cognition” in artificial intelligence. Moreover, the data management system for large-scale knowledge graphs is a critical common technology urgently needed for the infrastructure of next-generation AI.
Knowledge graphs use graph databases as their storage engine. Building on its existing self-developed distributed relational database management system, GBase has studied the integration of China-developed, self-controlled relational database systems with graph databases—seamlessly merging modules such as cluster management, metadata management, storage management, and unified access interfaces. This enables completely transparent knowledge graph querying, management, and applications through a unified, packaged access interface service layer.
1. Key Technical Research
Unified Query Language Interface Architecture for Knowledge Graphs
The functionality supports dialect parsers, standard SQL conversion and execution, proprietary function management, and data type conversion. The GBase system provides unified data access interfaces via JDBC and Python API for both graph query language and SQL query language, adding dialect grammar recognition and support at the SQL engine layer.
Distributed Query Planning
The GBase system parses the unified query language, performs optimization, and generates a query plan. The query plan is composed of unified graph computation operators and can be executed on the graph computation engine to return results.
GBase System Supports Graph Databases
The GBase system sits between application software and foundational database software, serving as a bridge. It receives user commands, analyzes engine information, invokes the appropriate engine, and executes operations on different database systems.
2. Application Scenarios
Integrating domestically developed, self-controlled relational database systems with graph databases provides effective support for business systems across many application scenarios.
Support for Precision Medicine
When applied in smart healthcare scenarios, integrating knowledge graphs into the GBase relational database system significantly improves the precision of smart healthcare tasks, including disease diagnosis and hospitalization cost estimation, offering both clinical and social significance.
Support for Fraud Detection
Through modeling and analyzing massive transaction data, fraudulent groups and other complex fraud patterns can be detected.
Support for Semantic Parsing
By converting natural language into knowledge graphs and storing them in the database, questions posed in natural language can be parsed and reorganized by the semantic parser within an intelligent Q&A system, and possible answers are then retrieved from the knowledge graph to provide to the user.
Summary
Seamlessly integrating the knowledge graph management system into the GBase relational database system, with a demand-oriented research approach, enables horizontal information integration and convenient visualization. It allows users to perform completely transparent knowledge graph queries, management, and applications through a unified, packaged access interface service layer, thus promoting the development of integrated database technology. The integrated system intelligently processes massive amounts of information, leveraging its advantages in model storage and query analysis to serve multiple application scenarios such as intelligent recommendation, intelligent customer service, financial risk control, healthcare, and e-commerce. GBase’s product portfolio and technology research will continue to advance, providing technical support for large-scale knowledge data applications across various industries.