Dear Lazyweb::Math,

Jul 24, 2006 13:31

At lunch today we were discussing the assignment of a ranking (1-10) to objects and we were trying to come up with a different system that might take into account more than a simple one dimensional scale. What we are wondering is whether, given the results of a number of queries to an individual of whether Object A is better than Object B for all ( Read more... )

Leave a comment

Comments 7

mtbg July 24 2006, 21:03:06 UTC
I assume there's a concrete problem behind this. Care to share?

Reply

jcmdev0 July 24 2006, 21:11:32 UTC
"How can we come up with a good metric space for hotness?"

I was a mostly casual observer. However, I did make the suggestion of the pairwise comparisons possibly leading to something more useful than a one-dimensional scale. Obviously a superficial question.

I think the conversation stemmed from a comment made about the average hotness of girls at Comic-Con vs the average hotness of girls in San Diego.

Reply

csn July 25 2006, 05:43:23 UTC
hahahaha...

Reply


jcmdev0 July 24 2006, 23:29:35 UTC
I think we could just use the pairwise comparison for voting, rather than trying to construct dimensions.

Reply


purplepsilon July 25 2006, 00:39:09 UTC
Nontransitivity is going to break the deal if you are trying to develop any notion of a partial ordering in the strict mathematical sense. I'd imagine the structure you are picturing is something along the lines of a digraph (when you think about it, it's a looser form of a poset's Hasse diagram), but transitivity seems like a necessary quality to make any ordering (partial or not) "useful ( ... )

Reply

purplepsilon July 25 2006, 00:45:29 UTC
Just remembering Rock Paper Scissors, if that is your application, a digraph would be the way to encode it. Not sure how you'd go about generating the comparison system.

Reply


macdaddyfrosh July 26 2006, 19:11:40 UTC
My gut said "that sounds like a machine learning problem", but I haven't got a clue how to encode it...

Reply


Leave a comment

Up