Beyond Explicit Edges: Robust Reasoning over Noisy and Sparse Knowledge Graphs
INSES is a dynamic framework that combines LLM-guided navigation with embedding-based similarity expansion to reason over noisy and sparse knowledge graphs. It replaces static graph traversal with query-specific, semantics-aware reasoning by dynamically adding 'virtual edges' when needed. This approach robustly recovers latent links while pruning errors, outperforming existing methods.