Interesting. Last summer I spent several months working on building a high throughput classification system and we only ever used Hadoop for feature extraction (which didn't involve any Mahout code). The Mahout part of the picture involved no Hadoop whatsoever.
On Thu, Jan 6, 2011 at 8:52 AM, Jeff Eastman <[email protected]> wrote: > +1 I'm in agreement with Sean that we are, defacto, Hadoop-based right now > even if our charter is broader. I'm not hung up on the precise wording in > this status report > > -----Original Message----- > From: Sean Owen [mailto:[email protected]] > Sent: Thursday, January 06, 2011 8:36 AM > To: [email protected] > Subject: Re: Apache Board report for January 2011 > > Fixed. > > I suppose I'm trying to communicate some news about what's happened in > the last 3 months, and there hasn't been much concrete, so wanted to > say something about the roadmap and what's taking shape in practice. > > There's no mission to be only Hadoop-based, but, in practice, that's > what it mostly is now. I think it's useful to be able to state that as > a piece of current status. But the wording can be clearer about that > sentiment. Is that the sentiment then? > > On Thu, Jan 6, 2011 at 4:25 PM, Ted Dunning <[email protected]> wrote: > > I think you meant Hadoop here. > > > > Also, I take issue with the idea of Mahout being Hadoop based. I prefer > to > > characterize it as scalable machine learning, taking advantage of Hadoop > > where appropriate. > > > > On Thu, Jan 6, 2011 at 5:44 AM, Sean Owen <[email protected]> wrote: > > > >> Apache *Mahout* 0.20.x-based machine learning, for collaborative > >> filtering, > >> > > >
