Originally Posted by
radq
The similarities between anime are computed using a lot of user rating data... it works pretty well for the most part (with the exception of really obscure series).
The algorithm basically assumes that a user's tastes are defined by k-dimensional vectors (one for each user) and each anime also has a corresponding k-dimensional vector, and users will tend to like shows which are close to their tastes. It works well for the most part, except for really obscure series.
Could you share which series are giving weird results so I can look into it? Thanks.
Bookmarks