News for hackers. Stuff that matters.
Re:Nearest neighbor indexes for similarity search by PaulHoule on Thursday August 12, 2021 @08:28AM (#28155373)
I worked on a heavily hacked Solr installation that used a neural network that cooked documents down to 50-d vectors which were similarity searched and also used an "information theoretic" similarity function for the ordinary word frequency vectors before that was officially supported.

Performance in terms of indexing time and throughput was awful, but searches were quick enough and performance in terms of "write a paragraph about your invention and find related patents" was fantastic.

There are no comments on this comment yet.