No.13608395 ViewReplyOriginalReport
A group of people is planning to watch an anime together.
In order to choose one they decided to draw up a list of titles and organize a poll where each person is given 1 point and can spread it however they like between the titles.
Naturally, the anime with the biggest total score wins.
However, if the number of people is high enough, all this is pretty much futile, since the chosen anime is always among the most popular ones.

Then they decided to choose in the same manner, rather not from titles, but from genres. Each title has a fixed set of genres.
And my question is what would be a reasonable mapping from poll results to anime list?

As an example, if there are total of n genres, we can place all titles in the [0..1]^n cube vertices according to their genres. And then normalize poll results so they would fit into the cube and choose the appropriate title by some metric, like the closest one to the results or the one that corresponds to the biggest barycentric coordinate.
Thoughts?