DCausse has uploaded a new change for review. ( https://gerrit.wikimedia.org/r/337572 )
Change subject: Explicitly use BM25 as default for wmf_defaults similarity profile ...................................................................... Explicitly use BM25 as default for wmf_defaults similarity profile The 2.x -> 5.x conversion has some uncertainties aroung similarity profiles and the 'default' similarity. To make things explicit and easy to understand force BM25 as the type for wmf_defaults. Specifically fields in 2.x that specify similarity as default will report in 5.x that their similarity is classic. It appears though that the actual search will still use BM25 (tested in 5.1.2, see https://github.com/elastic/elasticsearch-migration/issues/103#issuecomment-279546227) To ensure that _mapping doesn't report something wrong, and that there arn't other lingering, undetected, bugs with this conversion, we explicitly name BM25 as the similarity config. Change-Id: I32ddeeb18ac93623ff16b3f9b895c925394e3944 (cherry picked from commit 22edc18abdb2f31d1fac76fdf59c4770f75266b9) --- M profiles/SimilarityProfiles.php 1 file changed, 5 insertions(+), 2 deletions(-) git pull ssh://gerrit.wikimedia.org:29418/mediawiki/extensions/CirrusSearch refs/changes/72/337572/1 diff --git a/profiles/SimilarityProfiles.php b/profiles/SimilarityProfiles.php index d3df73f..234deaf 100644 --- a/profiles/SimilarityProfiles.php +++ b/profiles/SimilarityProfiles.php @@ -26,7 +26,8 @@ */ $wgCirrusSearchSimilarityProfiles = [ - // default profile, uses the classic TF/IDF from Lucene + // default profile, uses the classic TF/IDF from Lucene. With ES5 + // this will become BM25. 'default' => [], // BM25 with default values for k and a for all fields 'bm25_with_defaults' => [ @@ -97,6 +98,8 @@ 'wmf_defaults' => [ 'similarity' => [ 'default' => [ + // Although not referenced, this is necessary + // to disable coord 'type' => 'BM25', ], 'arrays' => [ @@ -106,7 +109,7 @@ ], ], 'fields' => [ - '__default__' => 'default', + '__default__' => 'BM25', 'category' => 'arrays', 'heading' => 'arrays', 'redirect.title' => 'arrays', -- To view, visit https://gerrit.wikimedia.org/r/337572 To unsubscribe, visit https://gerrit.wikimedia.org/r/settings Gerrit-MessageType: newchange Gerrit-Change-Id: I32ddeeb18ac93623ff16b3f9b895c925394e3944 Gerrit-PatchSet: 1 Gerrit-Project: mediawiki/extensions/CirrusSearch Gerrit-Branch: wmf/1.29.0-wmf.11 Gerrit-Owner: DCausse <[email protected]> Gerrit-Reviewer: EBernhardson <[email protected]> _______________________________________________ MediaWiki-commits mailing list [email protected] https://lists.wikimedia.org/mailman/listinfo/mediawiki-commits
