(copying the user list in case it is of use to others) ... or BooleanTanimotoCoefficientSimilarity if you are using BooleanPrefUsers.
On Thu, Apr 23, 2009 at 8:20 AM, Sean Owen <[email protected]> wrote: > Yes, you want TanimotoCoefficientSimilarity -- how does that work for you? > > On Thu, Apr 23, 2009 at 8:15 AM, 郑楠 <[email protected]> wrote: >> Hi, Sean, >> Recently, I have a problem when dealing with user similarity. I found that >> in the annotation you wrote: the pearson correlation implementation is >> equivalent to the cosine measure correlation since the data it receives is >> assumed to be centered. However, pearson correlation cannot be used for >> binary ratings. Does it mean that if I want to cope with binary ratings, I >> should only use spearman correlation, that is, I cannot >> usecosine correlation with taste? >> Best wishes! >> Alice >
