mahout-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sean Owen <sro...@gmail.com>
Subject Re: Nearest Neighbour: Similarity search
Date Wed, 23 Mar 2011 17:24:05 GMT
OK, it probably won't quite work then. The item-item similarity
metrics generally assume sparse vectors, and that the values are
something like "ratings". If they're values from different scales, it
won't make much sense.

On Wed, Mar 23, 2011 at 4:27 PM, tog <guillaume.alleon@gmail.com> wrote:
> Ok and what is the underlying algorithm in that job ?
> On my side the vectors are so called feature vectors in image processing so each entry
has no particular meaning. Similarity will be based on the L2 norm
>

Mime
View raw message