The main benefit of a vector database, however, is that it can index the data by proximity, speeding up clustering, partitioning, or nearest neighbor kinds of queries.

"for one _dimensionality reduction_ technique" i should say

Hmm, probably? I haven't done a whole lotta work in this space recently. But there's inherent difficulty in visualizing vectors in very high dimensional spaces (100s or 1000s, ofttimes). Check out "principal component analysis" for one technique

Howdy blusky. I always forget this place exists (and am slowly but surely forgetting about Twitter) 🙃

@justingarrison.com Graph databases store relationships (directed edges) between things (vertices). May be schemaless. Vector databases are optimized to efficiently store and lookup embeddings, projections of objects between high dimensional vector spaces. Lossy. Very different beasts 😊