A Brief Review of Modern Shortest Path Algorithms and Graph Optimization Techniques

This article reviews major approaches to computing shortest paths in large-scale graphs, focusing on index-based, embedding-based, and core-periphery algorithms. It highlights how methods like Pruned Landmark Labeling (PLL), hyperbolic embeddings, and GPU-accelerated learning enhance speed and scalability. The piece also contrasts practical “beyond worst-case” approaches with theoretical results, emphasizing how structural assumptions such as core-periphery organization redefine efficiency in real-world networks.