Expand description
Episodic memory retrieval with caching and hybrid search.
This module provides efficient retrieval of episodes with:
- LRU caching and TTL for query results
- BM25 keyword search (via CSM crate, first-tier, no API calls)
- ConceptGraph ontology expansion (via CSM crate, synonym matching)
- HDC hyperdimensional vectors (via CSM crate)
- Cascading retrieval pipeline (WG-131)
- Hierarchical/gist reranking for dense context (WG-118)
Re-exports§
pub use cache::CacheKey;pub use cache::CacheMetrics;pub use cache::DEFAULT_CACHE_TTL;pub use cache::DEFAULT_MAX_ENTRIES;pub use cache::QueryCache;pub use cascade::CascadeConfig;pub use cascade::CascadeResult;pub use cascade::CascadeRetriever;pub use gist::EpisodeGist;pub use gist::GistExtractor;pub use gist::GistScoredItem;pub use gist::HierarchicalReranker;pub use gist::RerankConfig;