RMLStreamer supported by RML-view-to-CSV in the performance track of the KGCW Challenge 2024
Keywords: RMLStreamer, RML-view-to-CSV, challenge, knowledge graph construction
TL;DR: The combination of RML-view-to-CSV and RMLStreamer emerges as an efficient approach, showcasing the potential of modular mapping engines that delegate each task to the most suitable framework.
Abstract: This paper presents the results of the performance track of the Knowledge Graph Construction Workshop 2024 Challenge with RMLStreamer, an RML mapping engine that processes all data in a streaming fashion. On mappings without joins, RMLStreamer scales well regarding execution time and CPU usage, while maintaining a constant memory usage. To optimize the processing of the joins, we added RML-view-to-CSV as a first step to our knowledge graph construction pipeline. RML-view-to-CSV is a proof-of-concept implementation for RML Logical Views, i.e. flattened, source format-agnostic views over one or more existing data sources. RML-view-to-CSV can additionally rewrite referencing object maps as logical views, before it materializes the logical views as CSV files. The combination of RML-view-to-CSV and RMLStreamer emerges as an efficient approach, showcasing the potential of modular mapping engines that delegate each task to the most suitable framework.
Submission Number: 9
Loading