Data Integrity - Noa Amar Roni Natanson

Examples/Topics/matching/similarity_by_actors_Jaccard.sql

@@ -0,0 +1,58 @@
+drop table if exists actors_per_movie;
+
+create table actors_per_movie
+as
+select
+movie_id
+, count(distinct actor_id) as actors_num
+from
+roles
+group by
+movie_id;
+
+
+ALTER TABLE actors_per_movie ADD PRIMARY KEY (movie_id);
+
+
+
+drop table if exists similarity_by_actors;
+
+create table
+similarity_by_actors
+as
+select
+frole.movie_id as fid
+, srole.movie_id as sid
+, count(distinct frole.actor_id) as common_actors_num
+, group_concat(distinct frole.actor_id order by frole.actor_id) as actors
+, max(fam.actors_num) as first_movie_actors_num
+, max(sam.actors_num) as second_movie_actors_num
+,  count(distinct frole.actor_id)/(max(fam.actors_num) + max(sam.actors_num) - count(distinct frole.actor_id)) as Jaccard
+from
+actors_per_movie as fam
+join
+roles as frole
+on
+fam.movie_id = frole.movie_id
+join
+roles as srole
+on
+frole.actor_id = srole.actor_id
+join
+actors_per_movie as sam
+on
+srole.movie_id = sam.movie_id
+where
+frole.movie_id != srole.movie_id
+and
+frole.movie_id < 1000 # NOTE - reduce scope just for a reasonable performance in demonstration. Remove!
+and
+srole.movie_id < 1000
+group by
+fid
+, sid
+having
+count(distinct frole.actor_id) >= 3
+and
+Jaccard > 0.25
+;