jenkins-bot has submitted this change and it was merged. ( https://gerrit.wikimedia.org/r/397887 )
Change subject: [search] better tuning with language fallbacks for prefix search
......................................................................
[search] better tuning with language fallbacks for prefix search
Tuned to make sure that important items that match the fallback
languages are close to the top.
Bug: T182136
Change-Id: Ic85f9518e81701d59514936d2ba07b675f6de3ee
---
M repo/config/EntityPrefixSearchProfiles.php
M repo/tests/phpunit/data/entitySearch/search_de-ch-en.expected
M repo/tests/phpunit/data/entitySearch/search_de-ch.expected
M repo/tests/phpunit/data/entitySearch/search_de-ch_strict.expected
M repo/tests/phpunit/data/entitySearch/search_en.expected
M repo/tests/phpunit/data/entitySearch/search_en_strict.expected
M repo/tests/phpunit/data/entitySearch/search_id.expected
M repo/tests/phpunit/data/entitySearch/search_par.expected
M repo/tests/phpunit/data/entitySearch/search_prop.expected
M repo/tests/phpunit/data/entitySearch/search_url.expected
M repo/tests/phpunit/data/entitySearch/search_zh-de-ch.expected
M repo/tests/phpunit/data/entitySearch/search_zh.expected
12 files changed, 84 insertions(+), 84 deletions(-)
Approvals:
Smalyshev: Looks good to me, approved
jenkins-bot: Verified
diff --git a/repo/config/EntityPrefixSearchProfiles.php
b/repo/config/EntityPrefixSearchProfiles.php
index 34deca9..be757c3 100644
--- a/repo/config/EntityPrefixSearchProfiles.php
+++ b/repo/config/EntityPrefixSearchProfiles.php
@@ -12,16 +12,16 @@
* fallback-discount - multiplier for each following fallback
*/
return [
- // FIXME: right now these weights are completely arbitrary. We need
- // to do some work to validate them.
+ // FIXME: manually tuned, next step is to put in place a golden corpus
of
+ // graded queries and provide metrics to evaluate the quality
objectively.
'default' => [
'any' => 0.001,
'lang-exact' => 2,
- 'lang-folded' => 1.8,
+ 'lang-folded' => 1.6,
'lang-prefix' => 1.1,
- 'fallback-exact' => 1,
- 'fallback-folded' => 0.9,
- 'fallback-prefix' => 0.8,
- 'fallback-discount' => 0.1,
+ 'fallback-exact' => 1.9,
+ 'fallback-folded' => 1.3,
+ 'fallback-prefix' => 0.4,
+ 'fallback-discount' => 0.9,
]
];
diff --git a/repo/tests/phpunit/data/entitySearch/search_de-ch-en.expected
b/repo/tests/phpunit/data/entitySearch/search_de-ch-en.expected
index 2c222be..780f300 100644
--- a/repo/tests/phpunit/data/entitySearch/search_de-ch-en.expected
+++ b/repo/tests/phpunit/data/entitySearch/search_de-ch-en.expected
@@ -38,7 +38,7 @@
"labels.de-ch.near_match_folded": "Wien"
}
},
- "boost": 1.8
+ "boost": 1.6
}
},
{
@@ -68,7 +68,7 @@
"labels.de.near_match": "Wien"
}
},
- "boost": 0.1
+ "boost": 1.71
}
},
{
@@ -78,7 +78,7 @@
"labels.de.near_match_folded": "Wien"
}
},
- "boost": 0.09
+ "boost": 1.17
}
},
{
@@ -88,7 +88,7 @@
"labels.de.prefix": "Wien"
}
},
- "boost": 0.08
+ "boost": 0.36
}
},
{
@@ -98,7 +98,7 @@
"labels.en.near_match": "Wien"
}
},
- "boost": 0.01
+ "boost": 1.539
}
},
{
@@ -108,7 +108,7 @@
"labels.en.near_match_folded": "Wien"
}
},
- "boost": 0.009
+ "boost": 1.053
}
},
{
@@ -118,7 +118,7 @@
"labels.en.prefix": "Wien"
}
},
- "boost": 0.008
+ "boost": 0.324
}
}
]
diff --git a/repo/tests/phpunit/data/entitySearch/search_de-ch.expected
b/repo/tests/phpunit/data/entitySearch/search_de-ch.expected
index a2450b1..274bd26 100644
--- a/repo/tests/phpunit/data/entitySearch/search_de-ch.expected
+++ b/repo/tests/phpunit/data/entitySearch/search_de-ch.expected
@@ -38,7 +38,7 @@
"labels.de-ch.near_match_folded": "Wien"
}
},
- "boost": 1.8
+ "boost": 1.6
}
},
{
@@ -68,7 +68,7 @@
"labels.de.near_match": "Wien"
}
},
- "boost": 0.1
+ "boost": 1.71
}
},
{
@@ -78,7 +78,7 @@
"labels.de.near_match_folded": "Wien"
}
},
- "boost": 0.09
+ "boost": 1.17
}
},
{
@@ -88,7 +88,7 @@
"labels.de.prefix": "Wien"
}
},
- "boost": 0.08
+ "boost": 0.36
}
},
{
@@ -98,7 +98,7 @@
"labels.en.near_match": "Wien"
}
},
- "boost": 0.01
+ "boost": 1.539
}
},
{
@@ -108,7 +108,7 @@
"labels.en.near_match_folded": "Wien"
}
},
- "boost": 0.009
+ "boost": 1.053
}
},
{
@@ -118,7 +118,7 @@
"labels.en.prefix": "Wien"
}
},
- "boost": 0.008
+ "boost": 0.324
}
}
]
diff --git a/repo/tests/phpunit/data/entitySearch/search_de-ch_strict.expected
b/repo/tests/phpunit/data/entitySearch/search_de-ch_strict.expected
index d46535c..c3358c4 100644
--- a/repo/tests/phpunit/data/entitySearch/search_de-ch_strict.expected
+++ b/repo/tests/phpunit/data/entitySearch/search_de-ch_strict.expected
@@ -38,7 +38,7 @@
"labels.de-ch.near_match_folded": "Wien"
}
},
- "boost": 1.8
+ "boost": 1.6
}
},
{
diff --git a/repo/tests/phpunit/data/entitySearch/search_en.expected
b/repo/tests/phpunit/data/entitySearch/search_en.expected
index 48c02b3..67a82dc 100644
--- a/repo/tests/phpunit/data/entitySearch/search_en.expected
+++ b/repo/tests/phpunit/data/entitySearch/search_en.expected
@@ -38,7 +38,7 @@
"labels.en.near_match_folded": "Duck"
}
},
- "boost": 1.8
+ "boost": 1.6
}
},
{
diff --git a/repo/tests/phpunit/data/entitySearch/search_en_strict.expected
b/repo/tests/phpunit/data/entitySearch/search_en_strict.expected
index dcfa947..6a26f77 100644
--- a/repo/tests/phpunit/data/entitySearch/search_en_strict.expected
+++ b/repo/tests/phpunit/data/entitySearch/search_en_strict.expected
@@ -38,7 +38,7 @@
"labels.en.near_match_folded": "Duck"
}
},
- "boost": 1.8
+ "boost": 1.6
}
},
{
diff --git a/repo/tests/phpunit/data/entitySearch/search_id.expected
b/repo/tests/phpunit/data/entitySearch/search_id.expected
index 5f3e4e1..011ed0e 100644
--- a/repo/tests/phpunit/data/entitySearch/search_id.expected
+++ b/repo/tests/phpunit/data/entitySearch/search_id.expected
@@ -38,7 +38,7 @@
"labels.en.near_match_folded": "q42"
}
},
- "boost": 1.8
+ "boost": 1.6
}
},
{
diff --git a/repo/tests/phpunit/data/entitySearch/search_par.expected
b/repo/tests/phpunit/data/entitySearch/search_par.expected
index 4601e74..a9aaf4e 100644
--- a/repo/tests/phpunit/data/entitySearch/search_par.expected
+++ b/repo/tests/phpunit/data/entitySearch/search_par.expected
@@ -38,7 +38,7 @@
"labels.en.near_match_folded": "(p128) "
}
},
- "boost": 1.8
+ "boost": 1.6
}
},
{
diff --git a/repo/tests/phpunit/data/entitySearch/search_prop.expected
b/repo/tests/phpunit/data/entitySearch/search_prop.expected
index 5951e7d..ba3c0fc 100644
--- a/repo/tests/phpunit/data/entitySearch/search_prop.expected
+++ b/repo/tests/phpunit/data/entitySearch/search_prop.expected
@@ -38,7 +38,7 @@
"labels.en.near_match_folded": "p42"
}
},
- "boost": 1.8
+ "boost": 1.6
}
},
{
diff --git a/repo/tests/phpunit/data/entitySearch/search_url.expected
b/repo/tests/phpunit/data/entitySearch/search_url.expected
index 38fa992..149efa8 100644
--- a/repo/tests/phpunit/data/entitySearch/search_url.expected
+++ b/repo/tests/phpunit/data/entitySearch/search_url.expected
@@ -38,7 +38,7 @@
"labels.en.near_match_folded": "https:\/\/www.wikidata.org\/wiki\/Q56"
}
},
- "boost": 1.8
+ "boost": 1.6
}
},
{
diff --git a/repo/tests/phpunit/data/entitySearch/search_zh-de-ch.expected
b/repo/tests/phpunit/data/entitySearch/search_zh-de-ch.expected
index cc4045c..9f5c28e 100644
--- a/repo/tests/phpunit/data/entitySearch/search_zh-de-ch.expected
+++ b/repo/tests/phpunit/data/entitySearch/search_zh-de-ch.expected
@@ -38,7 +38,7 @@
"labels.zh.near_match_folded": "\u7236\u985e"
}
},
- "boost": 1.8
+ "boost": 1.6
}
},
{
@@ -68,7 +68,7 @@
"labels.zh-hans.near_match": "\u7236\u985e"
}
},
- "boost": 0.1
+ "boost": 1.71
}
},
{
@@ -78,7 +78,7 @@
"labels.zh-hans.near_match_folded": "\u7236\u985e"
}
},
- "boost": 0.09
+ "boost": 1.17
}
},
{
@@ -88,7 +88,7 @@
"labels.zh-hans.prefix": "\u7236\u985e"
}
},
- "boost": 0.08
+ "boost": 0.36
}
},
{
@@ -98,7 +98,7 @@
"labels.zh-hant.near_match": "\u7236\u985e"
}
},
- "boost": 0.01
+ "boost": 1.539
}
},
{
@@ -108,7 +108,7 @@
"labels.zh-hant.near_match_folded": "\u7236\u985e"
}
},
- "boost": 0.009
+ "boost": 1.053
}
},
{
@@ -118,7 +118,7 @@
"labels.zh-hant.prefix": "\u7236\u985e"
}
},
- "boost": 0.008
+ "boost": 0.324
}
},
{
@@ -128,7 +128,7 @@
"labels.zh-cn.near_match": "\u7236\u985e"
}
},
- "boost": 0.001
+ "boost": 1.3851
}
},
{
@@ -138,7 +138,7 @@
"labels.zh-cn.near_match_folded": "\u7236\u985e"
}
},
- "boost": 0.0009
+ "boost": 0.9477
}
},
{
@@ -148,7 +148,7 @@
"labels.zh-cn.prefix": "\u7236\u985e"
}
},
- "boost": 0.0008
+ "boost": 0.2916
}
},
{
@@ -158,7 +158,7 @@
"labels.zh-tw.near_match": "\u7236\u985e"
}
},
- "boost": 0.0001
+ "boost": 1.24659
}
},
{
@@ -168,7 +168,7 @@
"labels.zh-tw.near_match_folded": "\u7236\u985e"
}
},
- "boost": 9.0e-5
+ "boost": 0.85293
}
},
{
@@ -178,7 +178,7 @@
"labels.zh-tw.prefix": "\u7236\u985e"
}
},
- "boost": 8.0e-5
+ "boost": 0.26244
}
},
{
@@ -188,7 +188,7 @@
"labels.zh-hk.near_match": "\u7236\u985e"
}
},
- "boost": 1.0e-5
+ "boost": 1.121931
}
},
{
@@ -198,7 +198,7 @@
"labels.zh-hk.near_match_folded": "\u7236\u985e"
}
},
- "boost": 9.0e-6
+ "boost": 0.767637
}
},
{
@@ -208,7 +208,7 @@
"labels.zh-hk.prefix": "\u7236\u985e"
}
},
- "boost": 8.0e-6
+ "boost": 0.236196
}
},
{
@@ -218,7 +218,7 @@
"labels.zh-sg.near_match": "\u7236\u985e"
}
},
- "boost": 1.0e-6
+ "boost": 1.0097379
}
},
{
@@ -228,7 +228,7 @@
"labels.zh-sg.near_match_folded": "\u7236\u985e"
}
},
- "boost": 9.0e-7
+ "boost": 0.6908733
}
},
{
@@ -238,7 +238,7 @@
"labels.zh-sg.prefix": "\u7236\u985e"
}
},
- "boost": 8.0e-7
+ "boost": 0.2125764
}
},
{
@@ -248,7 +248,7 @@
"labels.zh-mo.near_match": "\u7236\u985e"
}
},
- "boost": 1.0e-7
+ "boost": 0.90876411
}
},
{
@@ -258,7 +258,7 @@
"labels.zh-mo.near_match_folded": "\u7236\u985e"
}
},
- "boost": 9.0e-8
+ "boost": 0.62178597
}
},
{
@@ -268,7 +268,7 @@
"labels.zh-mo.prefix": "\u7236\u985e"
}
},
- "boost": 8.0e-8
+ "boost": 0.19131876
}
},
{
@@ -278,7 +278,7 @@
"labels.zh-my.near_match": "\u7236\u985e"
}
},
- "boost": 1.0e-8
+ "boost": 0.817887699
}
},
{
@@ -288,7 +288,7 @@
"labels.zh-my.near_match_folded": "\u7236\u985e"
}
},
- "boost": 9.0e-9
+ "boost": 0.559607373
}
},
{
@@ -298,7 +298,7 @@
"labels.zh-my.prefix": "\u7236\u985e"
}
},
- "boost": 8.0e-9
+ "boost": 0.172186884
}
},
{
@@ -308,7 +308,7 @@
"labels.en.near_match": "\u7236\u985e"
}
},
- "boost": 1.0e-9
+ "boost": 0.7360989291
}
},
{
@@ -318,7 +318,7 @@
"labels.en.near_match_folded": "\u7236\u985e"
}
},
- "boost": 9.0e-10
+ "boost": 0.5036466357
}
},
{
@@ -328,7 +328,7 @@
"labels.en.prefix": "\u7236\u985e"
}
},
- "boost": 8.0e-10
+ "boost": 0.1549681956
}
}
]
diff --git a/repo/tests/phpunit/data/entitySearch/search_zh.expected
b/repo/tests/phpunit/data/entitySearch/search_zh.expected
index 605baae..4160d8f 100644
--- a/repo/tests/phpunit/data/entitySearch/search_zh.expected
+++ b/repo/tests/phpunit/data/entitySearch/search_zh.expected
@@ -38,7 +38,7 @@
"labels.zh.near_match_folded": "\u7236\u985e"
}
},
- "boost": 1.8
+ "boost": 1.6
}
},
{
@@ -68,7 +68,7 @@
"labels.zh-hans.near_match": "\u7236\u985e"
}
},
- "boost": 0.1
+ "boost": 1.71
}
},
{
@@ -78,7 +78,7 @@
"labels.zh-hans.near_match_folded": "\u7236\u985e"
}
},
- "boost": 0.09
+ "boost": 1.17
}
},
{
@@ -88,7 +88,7 @@
"labels.zh-hans.prefix": "\u7236\u985e"
}
},
- "boost": 0.08
+ "boost": 0.36
}
},
{
@@ -98,7 +98,7 @@
"labels.zh-hant.near_match": "\u7236\u985e"
}
},
- "boost": 0.01
+ "boost": 1.539
}
},
{
@@ -108,7 +108,7 @@
"labels.zh-hant.near_match_folded": "\u7236\u985e"
}
},
- "boost": 0.009
+ "boost": 1.053
}
},
{
@@ -118,7 +118,7 @@
"labels.zh-hant.prefix": "\u7236\u985e"
}
},
- "boost": 0.008
+ "boost": 0.324
}
},
{
@@ -128,7 +128,7 @@
"labels.zh-cn.near_match": "\u7236\u985e"
}
},
- "boost": 0.001
+ "boost": 1.3851
}
},
{
@@ -138,7 +138,7 @@
"labels.zh-cn.near_match_folded": "\u7236\u985e"
}
},
- "boost": 0.0009
+ "boost": 0.9477
}
},
{
@@ -148,7 +148,7 @@
"labels.zh-cn.prefix": "\u7236\u985e"
}
},
- "boost": 0.0008
+ "boost": 0.2916
}
},
{
@@ -158,7 +158,7 @@
"labels.zh-tw.near_match": "\u7236\u985e"
}
},
- "boost": 0.0001
+ "boost": 1.24659
}
},
{
@@ -168,7 +168,7 @@
"labels.zh-tw.near_match_folded": "\u7236\u985e"
}
},
- "boost": 9.0e-5
+ "boost": 0.85293
}
},
{
@@ -178,7 +178,7 @@
"labels.zh-tw.prefix": "\u7236\u985e"
}
},
- "boost": 8.0e-5
+ "boost": 0.26244
}
},
{
@@ -188,7 +188,7 @@
"labels.zh-hk.near_match": "\u7236\u985e"
}
},
- "boost": 1.0e-5
+ "boost": 1.121931
}
},
{
@@ -198,7 +198,7 @@
"labels.zh-hk.near_match_folded": "\u7236\u985e"
}
},
- "boost": 9.0e-6
+ "boost": 0.767637
}
},
{
@@ -208,7 +208,7 @@
"labels.zh-hk.prefix": "\u7236\u985e"
}
},
- "boost": 8.0e-6
+ "boost": 0.236196
}
},
{
@@ -218,7 +218,7 @@
"labels.zh-sg.near_match": "\u7236\u985e"
}
},
- "boost": 1.0e-6
+ "boost": 1.0097379
}
},
{
@@ -228,7 +228,7 @@
"labels.zh-sg.near_match_folded": "\u7236\u985e"
}
},
- "boost": 9.0e-7
+ "boost": 0.6908733
}
},
{
@@ -238,7 +238,7 @@
"labels.zh-sg.prefix": "\u7236\u985e"
}
},
- "boost": 8.0e-7
+ "boost": 0.2125764
}
},
{
@@ -248,7 +248,7 @@
"labels.zh-mo.near_match": "\u7236\u985e"
}
},
- "boost": 1.0e-7
+ "boost": 0.90876411
}
},
{
@@ -258,7 +258,7 @@
"labels.zh-mo.near_match_folded": "\u7236\u985e"
}
},
- "boost": 9.0e-8
+ "boost": 0.62178597
}
},
{
@@ -268,7 +268,7 @@
"labels.zh-mo.prefix": "\u7236\u985e"
}
},
- "boost": 8.0e-8
+ "boost": 0.19131876
}
},
{
@@ -278,7 +278,7 @@
"labels.zh-my.near_match": "\u7236\u985e"
}
},
- "boost": 1.0e-8
+ "boost": 0.817887699
}
},
{
@@ -288,7 +288,7 @@
"labels.zh-my.near_match_folded": "\u7236\u985e"
}
},
- "boost": 9.0e-9
+ "boost": 0.559607373
}
},
{
@@ -298,7 +298,7 @@
"labels.zh-my.prefix": "\u7236\u985e"
}
},
- "boost": 8.0e-9
+ "boost": 0.172186884
}
},
{
@@ -308,7 +308,7 @@
"labels.en.near_match": "\u7236\u985e"
}
},
- "boost": 1.0e-9
+ "boost": 0.7360989291
}
},
{
@@ -318,7 +318,7 @@
"labels.en.near_match_folded": "\u7236\u985e"
}
},
- "boost": 9.0e-10
+ "boost": 0.5036466357
}
},
{
@@ -328,7 +328,7 @@
"labels.en.prefix": "\u7236\u985e"
}
},
- "boost": 8.0e-10
+ "boost": 0.1549681956
}
}
]
--
To view, visit https://gerrit.wikimedia.org/r/397887
To unsubscribe, visit https://gerrit.wikimedia.org/r/settings
Gerrit-MessageType: merged
Gerrit-Change-Id: Ic85f9518e81701d59514936d2ba07b675f6de3ee
Gerrit-PatchSet: 3
Gerrit-Project: mediawiki/extensions/Wikibase
Gerrit-Branch: master
Gerrit-Owner: DCausse <[email protected]>
Gerrit-Reviewer: Smalyshev <[email protected]>
Gerrit-Reviewer: jenkins-bot <>
_______________________________________________
MediaWiki-commits mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/mediawiki-commits
