RDF Stream Production Consumption

Initiative dedicated to defining how RDF streams are produced and consumed.

Notes

There have been different ways of producing and consuming RDF streams. Choices are motivated by different circumstances and requirements, often related to architectural and system design principles.

Producing RDF streams

RDF streams are produced in different ways, depending on the original data source, modeling decisions, and architectural design. Multiple aspects need to be studied in this scope

RDF Stream metadata

• Vocabularies for publishing RDF streams (RDF Stream, endpoint, data model)
• Vocabulaires for structuring RDF stream content (time annotated RDF graphs, triples)
• Vocabulaires for representing RDF stream interactions (WoT, etc)
• Vocabulaires for domain specific RDF streams (SSN, SOSA, Events, etc)

Native RDF generation

RDF messages are produced natively

• Publishing RDF messages continuously and accessible on a URI through polling (e.g., through LDP, etc)
• Publishing RDF messages continuously through subscription (e.g., LDN, etc.)
• Publishing RDF messages on bidirectional channels (e.g., Websockets)

Virtual/Mapping-based RDF generation

• RDF stream data transformation (e.g., using RML mappings)
• Non-RDF streams available (through query rewriting)
• Publishing RDF messages on distributed data stream systems (Kafka, etc.)

Consuming RDF streams

• Polling RDF Messages
• Subscribing to RDF streams
• Bidirectional RDF stream consumption
• Continuous querying over RDF streams
• Complex event processing on RDF streams

References

• Bonte, P., Callé, C., Curé, O. et al. Languages and systems for RDF stream processing, a survey. The VLDB Journal 34, 50 (2025). https://doi.org/10.1007/s00778-025-00927-7