Building a Lucene quality BM25 index using Python and SQLite
3 by dmezzetti | 0 comments on Hacker News.
The following article discusses a fast and accurate BM25 implementation in pure Python. It uses NumPy to do efficient calculations outside the GIL, the arrays package to use minimal memory on stored terms and SQLite/caching to offload less frequently used term frequency arrays to disk. https://ift.tt/4PQmMs2

Post a Comment

Previous Post Next Post