Traqula: Providing a foundation for the evolving SPARQL Ecosystem through modular Parsing, Transformation and Generation

Jitse De Smet

Ruben Taelman

ESWC 2026 Resource Track supporting slides for Traqula Resource, 11 May 2026

Traqula: Providing a foundation for the evolving SPARQL Ecosystem through modular Parsing, Transformation and Generation

Ghent University – imec – IDLab, Belgium

Research Foundation - Flanders

Paper at:
traqula-resource.
jitsedesmet.be
Available on npm, TypeScript and JavaScript, ESM and CJS
Managed by the Comunica Association, already used within Comunica

Overview

One Query — Many Languages

SELECT ?person ?gmt ?certain
WHERE {
  ?r rdf:reifies <<( ?person ex:birthDate ?dateTime )>> ;
     ex:certain ?certain .
  BIND(ADJUST(?dateTime, "-PT0H"^^xsd:dayTimeDuration) as ?gmt))
  FILTER (?certain > 0.5 && MONTH(?gmt) = 1 ) .
}
A SPARQL 1.2 federated query targeting four endpoints with different SPARQL versions and extensions

Bridging is essential

Three key challenges

Query Evaluation

A SPARQL 1.2 query may fail on one endpoint and work on another

SPARQL 1.2 query
→ sent to SPARQL 1.0 endpoint
Error

Tooling

Linters, editors, and formatters assume a single SPARQL version

IDE, linter, language server
→ built for SPARQL 1.1
breaks on SPARQL 1.2 syntax

Maintainability

Multi-version support
→ messy, hard-to-maintain code

if SPARQL_1_1 ...
else if SPARQL 1.2

→ duplicate code
Fragile codebase

Overview

Demystifying parsers

Demystifying parsers

Demystifying parsers

Demystifying parsers

Demystifying parsers

Demystifying parsers

Transform across query string, AST, and algebra

Transform across query string, AST, and algebra

Transform across query string, AST, and algebra

Flexibility through rule composition

Overview

~10x faster than JavaScript/ TS tools

Language and parse targets affect speed

Already in production

Conclusion

Problem

SPARQL language heterogeneity is a growing challenge for query evaluation, tooling, and maintainability

Solution

Traqula: flexible · round-trippable · web-native · language-agnostic

Impact

Foundation for a unifying framework across languages and dialects. Ready for linters, formatters, LSPs, and future query languages

Also check out our demo
compose-parser-engine.demo.jitsedesmet.be

© 2026 Jitse De Smet Creative Commons Attribution 4.0, unless otherwise indicated.

Source