Module spatiotemporal 

Spatiotemporal memory organization for efficient episodic retrieval

This module implements Phase 3 (Spatiotemporal Memory Organization) features from the research integration plan, including:

• Context-aware embeddings with task-specific adaptation
• Hierarchical spatiotemporal indexing (future)
• Coarse-to-fine retrieval strategy (future)
• MMR-based diversity maximization (future)

Phase 3 Goals

• Hierarchical Indexing: Domain → Task Type → Temporal clusters
• Coarse-to-Fine Retrieval: Multi-level search for speed and accuracy
• Diversity Maximization: MMR algorithm to avoid redundant results
• Context-Aware Embeddings: Task-specific embedding adaptation

Current Status

• ✅ Context-aware embeddings (Task 4.1, 4.2, 4.3)
• ✅ Diversity maximization (Task 3.1, 3.2) - COMPLETE
• ⏳ Hierarchical indexing (Task 1.1, 1.2, 1.3)
• ⏳ Coarse-to-fine retrieval (Task 2.1, 2.2, 2.3)

Re-exports

pub use diversity::DiversityMaximizer;

pub use diversity::ScoredEpisode;

pub use embeddings::ContextAwareEmbeddings;

pub use embeddings::ContrastivePair;

pub use embeddings::TaskAdapter;

pub use index::SpatiotemporalIndex;

pub use index::domain_index::DomainIndex;

pub use retriever::HierarchicalRetriever;

pub use retriever::HierarchicalScore;

pub use retriever::RetrievalQuery;

pub use types::TaskTypeIndex;

pub use types::TemporalCluster;

pub use types::TemporalGranularity;

Modules

diversity

Diversity Maximization using Maximal Marginal Relevance (MMR)

embeddings

Context-aware embeddings with task-specific adaptation

index

Core implementation of the spatiotemporal hierarchical index.

retriever

Hierarchical Retrieval Implementation

types

Core types for the spatiotemporal hierarchical index.