I am currently creating a news ranking application wherein I have users connected to one another, and can post news items to a database. Users can vote on said news items within a rating of 10 - 1. 10 giving the item the highest possible vote and in turn this determines the ranking of future news items a user may see once logged in.
Currently my approach at dealing with spammers automatically, involves using a threshold whereby say the average rating of the total number of votes that a user has voted on falls below this threshold. Then the relationship between two users will be automatically deleted.
This approach is okay for individual relationships but, let's say a spammer can be identified very quickly by a number of users. How can I deal with this problem more effectively?