Question

I have a MySQL database with 3 tables: users, posts, and users_posts. The users-posts relationship is many-to-many. How can I get a list of all pairs of posts and how many users they have in common? It essentially gives me a list of posts and how many users have commented on both. The point is to have data ready to import into network analysis software. The resulting list is called an “edgelist” in network terms, the users in common is an edge weight.

Schema:

users
id
name

posts
id
title
body

users_posts
user_id
post_id

Desired output:

postname1         postname2     users_in_common
Here's a title    Title #2      2
Another post      Title #2      11

I tried searching but didn’t even know the right search terms among many-to-many, join, three tables, count, pairs, shared, etc. Thanks for your help!

Answer 1

My query produces all pairs of posts including those without common users (will have 0 in users_in_common in such case). Try my solution on SQL Fiddle or see the code:

select
    p1.title as postname1,
    p2.title as postname2,
    coalesce(s.users_in_common, 0) as users_in_common
from posts p1
    inner join posts p2 on p1.id < p2.id
    left join (
        select
            up1.post_id as post1_id,
            up2.post_id as post2_id,
            count(*) as users_in_common
        from users_posts up1, users_posts up2
        where up1.user_id = up2.user_id
            and up1.post_id < up2.post_id
        group by up1.post_id, up2.post_id
    ) s
    on (s.post1_id = p1.id and s.post2_id = p2.id)
order by p1.id, p2.id;