Skip to main content

Module traversal

graphrag_core::graph

Module traversal 

Source
Expand description

Graph Traversal Algorithms for GraphRAG

This module implements deterministic graph traversal algorithms that don’t require machine learning, following NLP best practices for knowledge graph exploration:

  • BFS (Breadth-First Search): Level-by-level exploration for shortest paths
  • DFS (Depth-First Search): Deep exploration for discovering all paths
  • Ego-Network Extraction: K-hop neighborhoods around entities
  • Multi-Source Path Finding: Simultaneous search from multiple entities
  • Query-Focused Subgraph Extraction: Context-aware subgraph retrieval

These algorithms are essential for the query phase of GraphRAG, enabling:

  • Efficient entity-centric retrieval
  • Relationship path discovery
  • Context-aware information gathering
  • Multi-hop reasoning without neural networks

Structs§

GraphTraversal
Graph traversal system implementing various search algorithms
TraversalConfig
Configuration for graph traversal algorithms
TraversalResult
Result of a graph traversal operation