Crate chaotic_semantic_memory 

Chaotic Semantic Memory

High-performance memory system using Hyperdimensional Computing (HDC) and chaotic echo-state reservoir dynamics.

Encoding Model

This crate uses HDC (not transformer embeddings) for text-to-vector encoding. encoder::TextEncoder converts text via token-level FNV-1a hashing, positional permutation, and majority-rule bundling into 10240-bit binary hypervectors. Similar tokens in similar positions produce similar vectors. For true semantic similarity (synonyms, paraphrases), inject vectors from an external embedding model via ChaoticSemanticFramework::inject_concept.

Turso Vector Alternative: This crate uses libSQL (local SQLite or remote Turso) for persistence. You can add Turso’s native vector search tables (F32_BLOB with vector_top_k()) alongside the crate’s HDC storage in the same database. The crate manages concepts and associations tables; you manage semantic_vectors for float-vector similarity search.

Concurrency

The framework uses tokio::sync::RwLock internally and is safe to share across async tasks via Arc. SQLite persistence uses WAL mode for concurrent reads. All public APIs are async — do not wrap in block_on inside a Tokio runtime.

WASM Parity

Most APIs are available in WASM. Exceptions:

• Persistence: Replaced with stubs in persistence_wasm
• process_sequence: No Rayon parallelization in WASM

Re-exports

pub use bridge_retrieval::BridgeRetrieval;

pub use bundle::BundleAccumulator;

pub use error::MemoryError;

pub use error::Result;

pub use framework::ChaoticSemanticFramework;

pub use framework_builder::FrameworkBuilder;

pub use hyperdim::HVec10240;

pub use hyperdim::batch_cosine_similarity;

pub use semantic_bridge::BridgeConfig;

pub use semantic_bridge::BridgeHit;

pub use semantic_bridge::CanonicalConcept;

pub use semantic_bridge::ConceptGraph;

pub use semantic_bridge::MemoryPacket;

pub use semantic_bridge::ScoreBreakdown;

pub use singularity::Concept;

pub use singularity::ConceptBuilder;

Modules

bridge_retrieval

Bridge retrieval pipeline for semantic expansion.

bundle

Incremental bundle accumulator for streaming/sliding-window memory.

cli

concept_builder

Concept builder for ergonomic concept construction

encoder

Text-to-Hypervector Encoding using Hyperdimensional Computing (HDC) principles.

error

Error types for chaotic semantic memory

framework

Main framework integrating all components

framework_builder

Framework builder and configuration

graph_traversal

Graph traversal operations on the association graph.

hyperdim

Hyperdimensional computing primitives

metadata_filter

Metadata filtering for similarity search queries.

persistence

Persistence layer using libSQL (SQLite/Turso). Auto-migrations, version retention, FK enabled.

prelude

reservoir

Echo State Network for temporal dynamics.

retrieval

Retrieval modules for hybrid search.

semantic_bridge

Semantic Bridge Layer for Zero-Drift Semantic Generalization.

singularity

Episode-free concept injection

Structs

RetrievalConfig

Configuration for retrieval optimization.

RetrievalStats

Statistics from the last retrieval operation.

Enums

CandidateSource

Source of candidates in reduced-candidate retrieval.

MemoryEvent