Query-Specific Pruning of RML Mappings (Extended Version)

2026-03-27

AI Summary

This work addresses the inefficiency of existing RML-based knowledge graph construction approaches, which fully materialize RDF graphs without considering user query requirements, leading to poor performance in dynamic querying scenarios. To overcome this limitation, the paper introduces— for the first time—a query-aware mechanism into the RML mapping process. By analyzing the satisfiability of SPARQL queries against RML rules, the approach performs query-driven mapping pruning and materializes only the RDF subgraphs necessary to answer the given queries. Integrating a partial materialization strategy, the proposed method significantly reduces both materialization time and graph size while substantially accelerating query response on the GTFS-Madrid benchmark, thereby enabling efficient, on-demand RDF generation.

Abstract

Current approaches for knowledge graph construction with RML focus on full RDF graph materialization without considering user queries. As a result, mapping engines are inefficient in dynamic query environments, materializing large graphs even when only a small subset is needed to answer user queries. In this paper, we formally define satisfiability for SPARQL queries with respect to RDF data obtained via RML mappings and use this property to prune RML mappings for partial RDF graph materialization. Evaluation on the GTFS-Madrid benchmark shows that pruning significantly reduces materialization time, and RDF graph size while also noticeably improving querying time. Thus, enabling existing materialization engines to efficiently support generating RDF graphs in dynamic federated querying environment where user queries change frequently.

Problem

• RML
• SPARQL query
• RDF materialization
• query-specific pruning
• knowledge graph

Innovation

• Query-specific pruning
• RML mappings
• SPARQL satisfiability
• Partial materialization
• Dynamic federated querying