mahout-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sean Owen <sro...@gmail.com>
Subject Re: Taste-GenericItemBasedRecommender
Date Sun, 13 Dec 2009 18:08:53 GMT
My indices are all over the range 0..Integer.MAX_VALUE since they are
hashed from the actual long item IDs, so I imagine a DenseVector is a
non-starter.

Yeah would appreciate a second set of eyes on it. It's in a state
where it's worth looking at. At least I felt it was complete enough to
finish writing up in a chapter.

On Sun, Dec 13, 2009 at 5:58 PM, Jake Mannix <jake.mannix@gmail.com> wrote:
> It looks like your results should probably be fairly dense in the end,
> right?  So you should be accumulating SparseVectors on a DenseVector output,
> at least...  but this should only be a small speedup, not a lower
> complexity... as you say, you're already taking advantage of sparseness in
> the sums...
>
> I need to dig in and try this myself tonight or tomorrow night...

Mime
View raw message