lattix 0.1.0

Graph/KG substrate: triples + PageRank + random walks
Documentation

lattix

Knowledge graph construction, analysis, and embedding inference.

Dual-licensed under MIT or Apache-2.0.

use lattix_core::{KnowledgeGraph, Triple};

let mut kg = KnowledgeGraph::new();
kg.add_triple(Triple::new("Apple", "founded_by", "Steve Jobs"));

// Find path: Apple -> founded_by -> Steve Jobs
if let Some(path) = kg.find_path("Apple", "Steve Jobs") {
    println!("Path: {} hops", path.len());
}

Crates

Crate Purpose
lattix-core Graph storage, PageRank, Random Walks
lattix-nn Node2Vec, GNN layers
lattix-kge Knowledge Graph Embeddings (TransE, RotatE)
lattix-temporal Time-aware graph queries
lattix-cli Command-line interface
lattix-py Python bindings

Embedding Backends

  • BoxE: Containment embeddings (via subsume)
  • Hyperbolic: Hierarchy embeddings (via hyp)
  • ONNX: TransE, RotatE, ComplEx inference