For the purposes of this book (and otherwise too, as far as i know) "Samsara" is a release code name, defined as 0.10 and after. That includes all new code that happened after that, and the code that is still not deprectated (although most of MapReduce code is, by now, as evidenced by MAHOUT-1510). "Samsara" is thus both platform _and_ algorithms that mostly happened since then.
The book divides attention between off-the-shelf algorithms and programming capabilities of the Mahout "Samsara" releases. In this particular edition the scope is shifted towards programming techniques but there are off-the-shelf pieces and tutorials too. (Again, more information is in the announcement mentioned on top.) The undoubtedly new universal directions among Samsara releases are (1) using single coding principles for any new algorithm as opposed to a collection of hand-crafted algorithms; (2) backend-agnositic approach to optimizations of algebraic computations; (3) no MapReduce for the new code. These principles are unifying and used for all of new recommender code, new classification code, new collaborative filtering/dimensionality reduction/ general math etc. etc. Newer recommenders (what Pat is talking about) are running on top of the same platform and are using the same coding techniques discussed in this book and are just as part of the "Samsara" releases (one of the more prominent parts of its off-the-shelf offer, I would say), assuming my defintion of the "Samsara" code word is correct. In fact, as far as i know, since 0.10 I don't think we had any new code that did not use the platform part. -d On Thu, Feb 25, 2016 at 11:30 AM, Pat Ferrel <p...@occamsmachete.com> wrote: > I’m working on something, @Dmitriy mentioned that we might be able to add > to the second edition of the Beyond Mapreduce book, also looking at a free > self-published PDF book. Not sure how much it helps to have paper or a > publisher behind promotion. Any thoughts are welcome > > In the meantime it will come out incrementally in form of blog posts and > docs updates. Doc updates will be on the Mahout site, blog posts will be as > guest posts but will be mirrored here: blog.actionml.com > > PM me if you have leads on possible guest post sites. > > We are starting to get a lot of interesting real world results data, and > some we can make public. So we can talk 1/2 about the math and 1/2 about > practical applications. > > BTW this has led to some possible gen 2 mods and additions to the Mahout > code. > > On Feb 25, 2016, at 11:04 AM, Suneel Marthi <smar...@apache.org> wrote: > > I have talked with @Pat offline about a book on the new Recommenders. > > Definitely calls for another book. > > On Thu, Feb 25, 2016 at 1:22 PM, Ted Dunning <ted.dunn...@gmail.com> > wrote: > > > The project has moved in various ways since MiA was first published, but > > just covering Samsara leaves a lot of recommendation code that needs to > be > > covered. There is room for another book. > > > > > > > > > > > > On Thu, Feb 25, 2016 at 9:32 AM, Suneel Marthi <smar...@apache.org> > wrote: > > > >> The Mahout project has diverged from 'Mahout in Action' since Mahout 0.7 > >> release in 2012. > >> > >> > >> On Thu, Feb 25, 2016 at 12:30 PM, FRANCISCO XAVIER SUMBA TORAL < > >> xavier.sumb...@ucuenca.ec> wrote: > >> > >>> Awesome!! This book it’s mostly about Samsara. I’ll buy it. BTW do you > >>> know if there is an update of Mahout in Action? or Is it worth to buy > >> now? > >>> It has not changed too much mahout or the approach its same with the > >> actual > >>> implementation of mahout. > >>> > >>> > >> > > > >
