Rebuttal for submission 7667 - Rand-NSG

Reviewer 1: We will release our code along with the paper. A few contributions are: (1) We introduce a new pruning strategy parameterised by α that reduces the number of hops from navigating vertex to any other vertex. This provides a greater control over the diameter of the graph which is important for disk search. This is a feature that previous methods like NSG and HNSW lack. The plots above suggest that Rand-NSG indices are competitive with NSG and HNSW.