Lucene for Approximate Nearest-Neighbors Search on Arbitrary Dense Vectors · HackerPapers discussionshttps://hackerpaper.kaipfister.de/p/1910.10208/Collected HN comments and submissions linking this paper.enMon, 02 Nov 2026 11:16:22 GMTLucene for Approximate Nearest-Neighbors Search on Arbitrary Dense Vectors — mentioned by binarymaxhttps://news.ycombinator.com/item?id=21357950urn:hackerpapers:mention:1910.10208:21357950Fri, 25 Oct 2019 18:41:53 GMT<p>I&#x27;ve been working on using BERT for search, for research and training development, with not so great results. Note the quote &quot;when it comes to ranking results, BERT will help Search better understand one in 10 searches&quot;. This is because of the &quot;keywordese&quot; point they noted earlier in the article. Most searches are 1 or 2 words - there isn&#x27;t enough to grab onto for meaningful ranking with short queries and a similarity function for longer text documents. Also, try keeping the systems afloat to handle search like this. BERT is not practical to use for search results by anyone without the scale…</p><p><a href="https://hackerpaper.kaipfister.de/p/1910.10208/">Paper and discussion history</a>. A mention is not an endorsement.</p>cs.IR