HEG: Hierarchical Embedded Graphs for Scalable Path Queries

This system accelerates shortest-path queries over large-scale knowledge graphs using hierarchical summarization, community-aware embeddings, and distributed A*.

📊 Performance

• Sublinear query time
• Parallel A* outperforms NetworkX for long paths

🧱 Components

• Node2vec embeddings within Louvain/METIS communities
• Bitmask-based boundary node indexing
• Multi-level abstraction + descent-based path planning

🔬 Use Cases

• Knowledge graph search
• Scalable fact validation
• Dynamic graph exploration

🛠 Setup

pip install networkx faiss-cpu

📝 Example

from heg_search import query_path
query_path("protein_kinase", "DNA_repair")

📄 Paper

See "Theoretical Justification for a Hierarchical Embedded Graph Approach to Shortest Path Abstraction and Knowledge Validation" (2024)