Hello,

I've been working on a project for extending LevenshteinAutomata and
I'd like to know if it would be useful to add it to Lucene.

I've implemented the 'backwards dictionary' technique (see [1],
section 6) for speeding up approximate search. This technique allows
us to narrow down the search and, therefore, reduce the running time
(at the expense of using more memory).

I implemented it quite some time ago using an older version of Lucene,
so I need to revisit the code. However, the implementation was
relatively simple and it didn't require major changes to the core
classes. I can share the code in a public repository and iterate on
it, while I make it compatible for new Lucene APIs, add benchmarks,
and more unit tests.

Ideally, I'd like to contribute to Lucene, either as part of core,
suggest or a different module.

What do you think?

[1] https://www.cis.uni-muenchen.de/download/publikationen/fastapproxsearch.pdf

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to