GQL datatypes bridged to RDF knowledge graph systems for better interoperability
GQL Datatypes as an RDF Vocabulary
Databases
Summary
Graphs store information in different formats, which can make sharing data tricky. This paper shows how certain data types used in property graphs can be represented using formats used in RDF knowledge graphs, a popular method for linking data on the web. The authors created a simple vocabulary that maps GQL data types to RDF equivalents, helping systems using these two formats work together more smoothly. They also built tools that check and generate data using this vocabulary.
What this means in practice
- •For knowledge graph engineers: Use the vocabulary to exchange data between property graph systems and RDF stores without losing datatype details.
- •For graph database developers: Build validation tools that check data formats across different graph database standards leveraging the RDF vocabulary.
Authors
Dominik Tomaszuk, Daniel Szałaj
Abstract
The ISO GQL standard introduces a rich value-type system for property graphs, including scalar, temporal, byte-string, list, record, and dynamic union datatypes. The vocabulary presented here additionally covers a vector extension. At the same time, RDF-based knowledge graph systems identify literal datatypes by IRIs and interpret typed literals through value spaces, lexical spaces, and lexical-to-value mappings. This paper presents the GQL Datatypes RDF Vocabulary, a lightweight RDF vocabulary that represents selected GQL datatypes and the vector extension as RDF datatype IRIs and describes their hierarchy, aliases, and selected constraints using RDF, RDFS, OWL, and XML Schema Datatypes vocabulary terms. The vocabulary is intended to support interoperability between GQL-oriented property graph systems and RDF-based knowledge graph systems. We describe its namespace and datatype model, summarize its datatype coverage, and show how the accompanying generator and validator use the vocabulary. In the reference implementation, RDF literals retain GQL datatype IRIs; XSD mappings declared in the vocabulary support scalar validation, while lists, records, and vectors are encoded directly as JSON lexical forms of vocabulary-typed literals.