From 09ee5aa2e8dcede18da0ddf03a905944059b79d7 Mon Sep 17 00:00:00 2001
From: Chris Truter <crisptrutski@users.noreply.github.com>
Date: Mon, 11 Nov 2024 23:35:33 +0200
Subject: [PATCH] Prefer search results with higher view counts (#49842)
---
src/metabase/models/card.clj | 1 +
src/metabase/models/dashboard.clj | 1 +
src/metabase/models/table.clj | 1 +
src/metabase/search/config.clj | 22 +++++++++++++---------
src/metabase/search/postgres/core.clj | 2 ++
src/metabase/search/postgres/index.clj | 1 +
src/metabase/search/postgres/scoring.clj | 9 +++++++++
src/metabase/search/spec.clj | 1 +
test/metabase/search/spec_test.clj | 3 ++-
9 files changed, 31 insertions(+), 10 deletions(-)
diff --git a/src/metabase/models/card.clj b/src/metabase/models/card.clj
index 8d82ac9ef08..a56455cfb55 100644
--- a/src/metabase/models/card.clj
+++ b/src/metabase/models/card.clj
@@ -1024,6 +1024,7 @@
:last-editor-id :r.user_id
:pinned [:> [:coalesce :collection_position [:inline 0]] [:inline 0]]
:verified [:= "verified" :mr.status]
+ :view-count true
:created-at true
:updated-at true}
:search-terms [:name :description]
diff --git a/src/metabase/models/dashboard.clj b/src/metabase/models/dashboard.clj
index 368e34d6a32..0159f645a80 100644
--- a/src/metabase/models/dashboard.clj
+++ b/src/metabase/models/dashboard.clj
@@ -674,6 +674,7 @@
:last-edited-at :r.timestamp
:pinned [:> [:coalesce :collection_position [:inline 0]] [:inline 0]]
:table-id false
+ :view-count true
:created-at true
:updated-at true}
:search-terms [:name :description]
diff --git a/src/metabase/models/table.clj b/src/metabase/models/table.clj
index cd0209783f8..c3cbab97889 100644
--- a/src/metabase/models/table.clj
+++ b/src/metabase/models/table.clj
@@ -309,6 +309,7 @@
;; legacy search uses :active for this, but then has a rule to only ever show active tables
;; so we moved that to the where clause
:archived false
+ :view-count true
:created-at true
:updated-at true}
:search-terms [:name :description :display_name]
diff --git a/src/metabase/search/config.clj b/src/metabase/search/config.clj
index a9ca61f33a1..a60d4bcc14f 100644
--- a/src/metabase/search/config.clj
+++ b/src/metabase/search/config.clj
@@ -39,6 +39,10 @@
"Results in more dashboards than this are all considered to be equally popular."
10)
+(def ^:const view-count-scaling
+ "The larger this value, the longer it will take for the score to approach 1.0. It will never quite reach it."
+ 50)
+
(def ^:const surrounding-match-context
"Show this many words of context before/after matches in long search results"
2)
@@ -77,15 +81,15 @@
(def weights
"Strength of the various scorers. Copied from metabase.search.in-place.scoring, but allowing divergence."
- {:pinned 2 ;; simple field
- :bookmarked 2 ;; join with multi-table entity
- :recency 1.5 ;; date formula
- :dashboard 1 ;; simple field
- :model 0.5 ;; simple field
- :official-collection 2 ;; a field we can calculate
- :verified 2 ;; a simple field
- :text 10 ;; strength of text-scores-weight previously
- })
+ {:pinned 2
+ :bookmarked 2
+ :recency 1.5
+ :dashboard 1
+ :model 0.5
+ :official-collection 2
+ :verified 2
+ :view-count 2
+ :text 10})
(defn model->alias
"Given a model string returns the model alias"
diff --git a/src/metabase/search/postgres/core.clj b/src/metabase/search/postgres/core.clj
index 6d143c81698..a17556eb234 100644
--- a/src/metabase/search/postgres/core.clj
+++ b/src/metabase/search/postgres/core.clj
@@ -72,6 +72,8 @@
(OffsetDateTime/parse s)))
(defn- rehydrate [index-row]
+ ;; Useful for debugging scoring
+ #_ (dissoc index-row :legacy_input :created_at :updated_at :last_edited_at)
(-> (merge
(json/parse-string (:legacy_input index-row) keyword)
(select-keys index-row [:total_score :pinned]))
diff --git a/src/metabase/search/postgres/index.clj b/src/metabase/search/postgres/index.clj
index bfe35def584..2291330f641 100644
--- a/src/metabase/search/postgres/index.clj
+++ b/src/metabase/search/postgres/index.clj
@@ -56,6 +56,7 @@
[:model_rank :int :not-null]
[:pinned :boolean]
[:verified :boolean]
+ [:view_count :int]
;; permission related entities
[:collection_id :int]
[:database_id :int]
diff --git a/src/metabase/search/postgres/scoring.clj b/src/metabase/search/postgres/scoring.clj
index 16d7e09af97..cf390385c47 100644
--- a/src/metabase/search/postgres/scoring.clj
+++ b/src/metabase/search/postgres/scoring.clj
@@ -15,6 +15,14 @@
[column ceiling]
[:least [:/ [:coalesce column [:inline 0]] [:inline (double ceiling)]] [:inline 1]])
+(defn- atan-size
+ "Prefer items whose value is larger, with diminishing gains."
+ [column scaling]
+ ;; 2/PI * tan^-1 (x/N)
+ [:*
+ [:/ [:inline 2] [:pi]]
+ [:atan [:/ [:cast [:coalesce column [:inline 0.0]] :float] [:inline scaling]]]])
+
(defn- inverse-duration
"Score at item based on the duration between two dates, where less is better."
[from-column to-column ceiling-in-days]
@@ -67,6 +75,7 @@
(def ^:private scorers
{:text [:ts_rank :search_vector :query [:inline ts-rank-normalization]]
+ :view-count (atan-size :view_count search.config/view-count-scaling)
:pinned (truthy :pinned)
:bookmarked bookmark-score-expr
:recency (inverse-duration :model_updated_at [:now] search.config/stale-time-in-days)
diff --git a/src/metabase/search/spec.clj b/src/metabase/search/spec.clj
index 74014434e70..969646bb915 100644
--- a/src/metabase/search/spec.clj
+++ b/src/metabase/search/spec.clj
@@ -41,6 +41,7 @@
:last-editor-id
:pinned
:verified
+ :view-count
:updated-at])
(def ^:private default-attrs
diff --git a/test/metabase/search/spec_test.clj b/test/metabase/search/spec_test.clj
index 399521754e0..047852a1893 100644
--- a/test/metabase/search/spec_test.clj
+++ b/test/metabase/search/spec_test.clj
@@ -70,6 +70,7 @@
:name
:query_type
:type
+ :view_count
:created_at
:updated_at},
:where [:= :updated.id :this.id]}},
@@ -99,7 +100,7 @@
{:search-model "table",
:fields
#{:active :description :schema :name :id :db_id :initial_sync_status :display_name
- :visibility_type :created_at :updated_at}
+ :visibility_type :view_count :created_at :updated_at}
:where [:= :updated.id :this.id]}},
:Database #{{:search-model "table", :fields #{:name}, :where [:= :updated.id :this.db_id]}}
:Segment #{{:search-model "segment"
--
GitLab