Posted: Sat Jan 07, 2006 12:55 am
-- well a possibility would be:
avgDB = average of all the anime votes in the DB
avg = average of all the permanent anime votes of the users
numvot = numbur of permanent anime votes made by the user
vote = vote by the user
modvote = modified value of the vote
weight = value of a vote
so modvote would be:
modvote = vote * (avgDB / avg )
this would take care of those that make an habit of overrating or underrating everything and it would kill spamvotes since a spam vote would only be a very average vote.
Example:
the average DBvote is 7, i create an account and vote 10 to naruto.
I just voted 7 to naruto. same if i just vote 1
to further refine things we could use the standard error as well but i dont remember the exact formula at the moment.
--if one want to add value to those that have seen much then
KARMA= (avgDB / avg )
weight*modvote
here weight could simply be numvote altought i would find this a bit heavy since for failry unknown anime a single vote by one of us would make for the whole score.
Example: me and deridiot perm vote an anime that just got his 10th user. The others users have seen an average of 100 anime and have voted for all of them ( a generous assumption id say)
I have made 783 perm anime votes so far and deridiot has made 698 perm anime votes our total weight is 1481 while the other 8 users is barely 800. me and deridiot vote alone are worth 65% of this anime vote.
so i would go with something that increase slower as one watch more
PS a ranking by anime watched or perm votes could be funny more then a GB one at least
avgDB = average of all the anime votes in the DB
avg = average of all the permanent anime votes of the users
numvot = numbur of permanent anime votes made by the user
vote = vote by the user
modvote = modified value of the vote
weight = value of a vote
so modvote would be:
modvote = vote * (avgDB / avg )
this would take care of those that make an habit of overrating or underrating everything and it would kill spamvotes since a spam vote would only be a very average vote.
Example:
the average DBvote is 7, i create an account and vote 10 to naruto.
I just voted 7 to naruto. same if i just vote 1
to further refine things we could use the standard error as well but i dont remember the exact formula at the moment.
--if one want to add value to those that have seen much then
KARMA= (avgDB / avg )
weight*modvote
here weight could simply be numvote altought i would find this a bit heavy since for failry unknown anime a single vote by one of us would make for the whole score.
Example: me and deridiot perm vote an anime that just got his 10th user. The others users have seen an average of 100 anime and have voted for all of them ( a generous assumption id say)
I have made 783 perm anime votes so far and deridiot has made 698 perm anime votes our total weight is 1481 while the other 8 users is barely 800. me and deridiot vote alone are worth 65% of this anime vote.
so i would go with something that increase slower as one watch more
PS a ranking by anime watched or perm votes could be funny more then a GB one at least