Hi, Can someone tell me what was Gaurav Arora's exact contribution in the Clustering Search Results part during GSoC 2014? I guess that will be more helpful in understanding his code. Regards Karthik On Tue, Dec 16, 2014 at 4:06 AM, Olly Betts <olly at survex.com> wrote:> On Mon, Dec 15, 2014 at 06:56:39PM +0530, karthik iyer wrote: >> Could some one tell me some specific expectations from Letor for GSOC >> 2015? > > To quote the project ideas list: > > | The focus for 2015 should be to consolidate the work done so far and > | get it to a state where we can include it in a stable release. The big > | task here is adding automated test cases for the LTR code such that it > | has similarly good test coverage to xapian-core, and fixing the > | problems which doing so will undoubtedly uncover. > > http://trac.xapian.org/wiki/GSoCProjectIdeas#Project:LearningtoRank > > While I wouldn't entirely rule out accepting a proposal which also added > support for something new (e.g. another learning algorithm), that would > really need to be a secondary phase, to be done once the module has been > brought to a releasable state, if time allows. > >> I am currently trying out some learning algorithm >> implementations but need some help with it. Can Parth, James and >> others related to LETOR please tell me the IRC availability for the >> next few days. > > I don't think many of us have fixed "IRC hours". The best thing to do > is ask a question in the channel, and wait for a response. > > Cheers, > Olly
On 30 Dec 2014, at 11:39, karthik iyer <karthikiyer2000 at gmail.com> wrote:> Can someone tell me what was Gaurav Arora's exact contribution in the > Clustering Search Results part during GSoC 2014? I guess that will be > more helpful in understanding his code.Karthik ? Guarav is listed as mentor for that project, but as explained on our page with information for students & potential students, you shouldn?t read too much into that as it?s mostly a Google administrative thing. I?d start in any case with the previous clustering branch (it?s called svn/clustering) in our git tree; you want to start with xapian-core/include/xapian/cluster.h and then the files in xapian-core/docsim (and then work from there out). A good understanding of how Xapian works will be important to understanding what is going on. Once you?ve got up to speed with that, I?d look at the code George wrote during his 2014 project to see if that approach still makes sense. They?re independent implementations, but neither is finished and you may prefer to start again, learning from what they did, rather than building on either of them. J -- James Aylett, occasional trouble-maker xapian.org
hi, Could someone please explain how exactly is a termlist table different from a posting list table? On Wed, Dec 31, 2014 at 2:18 AM, James Aylett <james-xapian at tartarus.org> wrote:> On 30 Dec 2014, at 11:39, karthik iyer <karthikiyer2000 at gmail.com> wrote: > >> Can someone tell me what was Gaurav Arora's exact contribution in the >> Clustering Search Results part during GSoC 2014? I guess that will be >> more helpful in understanding his code. > > Karthik ? Guarav is listed as mentor for that project, but as explained on our page with information for students & potential students, you shouldn?t read too much into that as it?s mostly a Google administrative thing. > > I?d start in any case with the previous clustering branch (it?s called svn/clustering) in our git tree; you want to start with xapian-core/include/xapian/cluster.h and then the files in xapian-core/docsim (and then work from there out). A good understanding of how Xapian works will be important to understanding what is going on. > > Once you?ve got up to speed with that, I?d look at the code George wrote during his 2014 project to see if that approach still makes sense. They?re independent implementations, but neither is finished and you may prefer to start again, learning from what they did, rather than building on either of them. > > J > > -- > James Aylett, occasional trouble-maker > xapian.org >