How do I efficiently compare users? - php

Question

I have users who can "like" categories. For instance, we may have 2 users:
John likes apples, oranges, pears
Bob likes apples, oranges, pie, cake
They both like apples, oranges
This isn't an issue with two users, but when I imagine scaling it to thousands of users, with thousands of likes, there will be major efficiency concerns.
I need to be able to compare a user with all other users, and determine which likes they have in common.
I have tried array_intersect, but it does not scale. I need a mysql solution.
How would I efficiently return users who share the same likes, and the likes that are shared?
users
+-------+-------------+------+-----+---------+----------------+
| Field | Type | Null | Key | Default | Extra |
+-------+-------------+------+-----+---------+----------------+
| id | int(11) | NO | PRI | NULL | auto_increment |
| name | varchar(16) | NO | | NULL | |
+-------+-------------+------+-----+---------+----------------+
categories
+-------+-------------+------+-----+---------+----------------+
| Field | Type | Null | Key | Default | Extra |
+-------+-------------+------+-----+---------+----------------+
| id | int(11) | NO | PRI | NULL | auto_increment |
| name | varchar(32) | NO | | NULL | |
+-------+-------------+------+-----+---------+----------------+
likes
+-------------+---------+------+-----+---------+-------+
| Field | Type | Null | Key | Default | Extra |
+-------------+---------+------+-----+---------+-------+
| user_id | int(11) | NO | MUL | NULL | |
| category_id | int(11) | NO | MUL | NULL | |
+-------------+---------+------+-----+---------+-------+

Answer 1

function find_intersect($likes1, $likes2){
sort($likes1);
sort($likes2);
$intersect = array();
$i = 0;
$j = 0;
while ($i < count($likes1) and $j < count($likes2)){
if ($likes1[$i] == $likes2[$j]){
array_push($intersect, $likes1[$i]);
$i++;
$j++;
}
else if ($likes1[$i] < $likes2[$j])
$i++;
else
$j++;
}
return $intersect;
}
Above is what I drummed up and it should be most efficient way of finding intersection of two arrays. I do agree with #DanFarrell though in that MySQL or some database will be way more efficient in managing information when it comes to thousands of users.

Answer 2

I was able to solve my problem with the following:
SELECT user_id, count(category_id) AS count, group_concat(category_id separator "|")
FROM likes
WHERE category_id IN (
SELECT category_id
FROM likes
WHERE user_id=1
)
AND user_id != 1
GROUP BY user_id;
This will return the user id, the amount of categories shared, and the items shared separated by the pipeline character. Obviously it will need joins to get username and category, but for simplicity/readability, I left them out.