Skip to main content
Hybrid search is how Worlds retrieves the most relevant slice of a World for an agent’s context window. Rather than relying on a single index, it fuses three complementary signals:
SignalTechniqueWhat it captures
SemanticVector embeddings, 1536-dimConceptual meaning
KeywordFTS5 / BM25Exact term matches
Graph contextRDF relationship filtersStructural relationships

Reciprocal Rank Fusion

Results from each signal are merged using Reciprocal Rank Fusion (RRF), an industry-standard algorithm that produces a single unified ranking: score=dD160+rank(d)score = \sum_{d \in D} \frac{1}{60 + rank(d)} This ensures that a result ranked highly by multiple signals surfaces to the top, while noise from any single signal is suppressed. Vector search alone can return semantically similar but factually wrong results. By adding keyword precision and graph-aware structural filters, hybrid search dramatically improves retrieval accuracy for agentic workloads.

Learn more