lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Edoardo Causarano <edoardo.causar...@gmail.com>
Subject A question on distributed, sharded searches
Date Mon, 15 May 2017 11:38:31 GMT
Hi All,

please bear with me for a couple questions on distributed searches.

I have some code at hand that is scoring results - also from remote servers
- in the application itself. I'd like to investigate how hard it would be
to fold the re-scoring back into Lucene, so that individual shards only
need to transmit the documents that are strictly necessary.

I've looked around Lucene (6.x) and understood that this
https://lucene.apache.org/core/6_5_0/core/org/apache/lucene/search/TopDocs.html#merge-org.apache.lucene.search.Sort-int-org.apache.lucene.search.TopFieldDocs:A
operation would help in this case. Or would one use
https://lucene.apache.org/core/6_5_0/core/org/apache/lucene/search/TwoPhaseIterator.html
to provide proxies for iterators of remote results?

Can anyone confirm and possibly point out a nice example, blog, or
presentation deck or source where this is done?


TIA,
Edo

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message