how to implement user ranking in php with a huge number of users

alenoosh · 08-21-2010, 06:30 AM

Hi all

I'm writing a user ranking module for a site. This ranking depends on some criterias and it's possible to set or unset any one of these criterias in order to consider them in calculating the user rank or not. And here's the way I've implemented the ranking calculation :

when I set one or more of the criterias to be considered in ranking , for each user in the system I insert one record for each criteria , for example :

if I have 2 criterias and both are set and consider that I have two users , I'll have :

Ranking table
--------------
username | criteria | to_be_added | score
--------------------------------------------------
user1 | criteria1 | 1 | 0
user1 | criteria2 | 1 | 0
user2 | criteria1 | 1 | 0
user2 | criteria2 | 1 | 0

It means I just set the to_be_added field to 1 for all of them and leave the calculation of score for each criteria for each user to the time the user logins so that to prevent doing all these calculations at once , because there are a huge number of users ...

But there is one problem , if I want to show for example the best user (based on the highest score) , the result can't always be true because some users might not logged in at that time and their score might be zero .

How should I solve this problem ?

Thanks in advance for your help ...

graemef · 08-22-2010, 09:00 PM

You need to normalise your data.

Rule 1: Remove repeating groups of data.

user1 and user 2 are being repeated and so they should be removed from the table. Since the criteria1 and criteria2 scores are specific to the user they will go with the user. Hence you would get:

Code:

username criteria1 criteria2
user1            0         0
user2            0         0

Now if you want to display the criteria2 scores then the following will help:

Code:

SELECT username criteria2 from ranking ORDER BY criteria2 DESC

Without knowing the formula behind the criteria calculations it is difficult to say what would be the best way to calculate them. However, I would expect that the best way would be to do them as a batch job and then each time a user logs in to update it (if appropriate).