mahout-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Florian Leibert <>
Subject Re: is it possible to create vectors from a sharded index?
Date Thu, 16 Jul 2009 17:50:52 GMT
Thanks Ted for your quick response. I
am concerned about having different sets of vectors because I want to
run LDA over the vector outputs and
looking at the code of LDA I'm not sure if I can update this
incrementally... I'll have to take another look but I wanted to
clarify this before running a
large indexing job over my data...

On Thu, Jul 16, 2009 at 10:42 AM, Ted Dunning <> wrote:

> My first inclination is to think that it will just work if you generate
> vectors from each index independently.
> The only failure mode that I can think of would be due to the same term
> labels in vectors possibly being assigned different indexes.
> Jeff, Grant,  will the labels take precedence over the index when loading a
> vector?
> On Thu, Jul 16, 2009 at 9:15 AM, Florian Leibert <> wrote:
> > Hi,
> > I have a sharded Lucene index that spans about 400 GB and am wondering if
> I
> > can create the vectors (via the patch specified in MAHOUT-126) on this
> > sharded index?
> >
> >

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message