A data river is the emerging continuous-extraction paradigm for Knowledge Graphs in which a selective curation mechanism extracts information from sources with established trust in near-real-time, replacing the batch-processing data lake model.

Jason Barnard coined the Data River concept to describe the evolutionary destination of Knowledge Graph processing. Unlike the Data Lake - where information accumulates and waits for periodic batch processing - the Data River flows continuously past an extraction mechanism: the Confidence Curator. This Curator operates like a gold panner sieving gold from a flowing river, selectively extracting relevant information from sources that have achieved Trusted Source status through demonstrated consistency, reliability, and independent corroboration. The critical insight is that the Confidence Curator does not extract from everything. It curates only from sources that have earned trust. Freshness without trust is ignored by the Curator. Trust without freshness leaves an entity stuck in the Data Lake. This is the mechanism that makes Trusted Source status the prerequisite for near-real-time Knowledge Graph presence.

The Data River model has profound strategic implications. As Knowledge Graphs evolve from data lakes toward data rivers, the competitive advantage shifts from grounding optimization (being findable by search) to graph consolidation (being trusted by the Curator). Brands that achieve Trusted Source status gain near-real-time updates to their Entity Graph representation, while competitors remain in the data lake with stale, batch-processed information. This is the infrastructure mechanism behind the Confidence Phase evolution from Phase 1 (grounding-dependent) to Phase 3 (graph-native).

The emerging continuous-extraction paradigm for Knowledge Graphs where a Confidence Curator selectively extracts information from trusted sources in near-real-time. Replaces the Data Lake batch model for entities that achieve Trusted Source status. The mechanism that makes freshness dependent on trust.