AI Search with HNSW: Navigable Small World Graphs (NSW) |

🔍 Navigable Small World (NSW) graphs in HNSW: ObjectBox Bites Episode 7

🚀 We continue our exploration of the Hierarchical Navigable Small Worlds (HNSW) algorithm. In this episode, we're focusing on the second fundamental technique: Navigable Small Worlds (NSW).

💡 NSW is a proximity graph, connecting vertices (representing elements) with both short- and long-range links. It features a high clustering coefficient and short average path length and supports efficient navigation between vertices.

🎥 In this episode, we discuss:
-How the NSW graph is constructed
-What the vertex's friends list is
-How to search for the closest value in NSW with greedy-search

👍 Stay tuned for the upcoming episode, where we will build on the combined insights from the probability skip lists and NSW to further discuss the HNSW

📚 Sources and Further Reading:
[1] Y. A. Malkov and D. A. Yashunin, IEEE Trans. Pattern. Anal. Mach. Intell. (2020), 42, 4, 824-836
[2] Y. Malkov et al., Approximate Nearest Neighbor Search Small World Approach (2011), International Conference on Information and Communication Technologies & Applications
[3] Y. Malkov et al., Scalable Distributed Algorithm for Approximate Nearest Neighbor Search Problem in High Dimensional General Metric Spaces (2012), Similarity Search and Applications, 132-147
[4] Y. Malkov et al., Approximate nearest neighbor algorithm based on navigable small world graphs (2014), Information Systems, 45, 61-68
#VectorDatabases #HNSW #ObjectBox #ANN #AI #NSW