Xapian devel - Mar 2016 - Regarding GSoC 2016 project idea

Hi,

If I take up the task of 'clustering the search results' and I am opting
for the new method to cluster the results, then what will be the
approximate amount of data that will be clustered? So appropriate
clustering mechanism can be thought of according to that.

Thanks & regards,
Ainish

On Sat, Mar 19, 2016 at 5:37 PM, James Aylett <james-xapian at
tartarus.org>
wrote:
> On Sat, Mar 19, 2016 at 11:49:14AM +0530, Ainish Dave wrote:
>
> > I am Ainish Dave from Ahmedabad, India. I am currently pursuing my
> masters
> > level degree with background in computer science. I browsed through
the
> > list of organizations on the GSoC program page and Xapian was the one
> which
> > closely matched with my interest and background. I had some queries:
>
> Hi, Ainish -- welcome to Xapian!
>
> > 1) I am interested in the project 'Clustering of Search
Results'. I have
> > followed the guideline and built the system from Git and now currently
> > exploring the code for this particular module. I have a fair idea of
c++
> > and can manage to work with that. I also have prior experience with
> > clustering, 'topical clustering of tweets using Twitter data'
to be
> > precise. So, what other material shall I refer to get hands on this
> > topic/project to move ahead?
>
> It's worth looking at the previous clustering code, but really the
> next thing is for you to use your experience of clustering to propose
> an approach to doing this.
>
> > 2) Since the application period is started, am I a bit late to join
the
> > procedure or is it okay to start off now?
>
> Absolutely fine to start now -- you still have a week (until Friday)
> to get your application finished.
>
> > 3) I am waiting for my proof of enrollment to come by Monday and then
I
> can
> > apply. So is it fine to apply in next week? Primary reason for asking
> this
> > is that some organizations have strict norms of not considering the
> > applications arriving in second week.
>
> I'd start creating your proposal now, so we can give you feedback on
> it. You can create an application in the GSoC system before your proof
> of enrolment is processed (at least, as I understand things).
>
> J
>
> --
>   James Aylett, occasional trouble-maker
>   xapian.org
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.xapian.org/pipermail/xapian-devel/attachments/20160322/f5053145/attachment.html>

On Tue, Mar 22, 2016 at 11:12:12AM +0530, Ainish Dave wrote:
> If I take up the task of 'clustering the search results' and I am
opting
> for the new method to cluster the results, then what will be the
> approximate amount of data that will be clustered? So appropriate
> clustering mechanism can be thought of according to that.
The intention is to use this in 'live search' environments, so you'd
probably want to pull a medium number of documents from the top of the
search. It's a little difficult to say what a good number is, because
it will depend somewhat on how broad the search is. But maybe in the
100..1000 range?

J

-- 
  James Aylett, occasional trouble-maker
  xapian.org

Hello,

First of all, apologies for submitting the proposal this late due to other
academic duties. But, I have included sufficient amount of information
about the approach I am going to use for 'Clustering of Search Results'.
Kindly go through it and I will provide any clarifications needed on the
doubts arising after/while reading the proposal.

Thank you.

Regards,
Ainish

On Tue, Mar 22, 2016 at 6:16 PM, James Aylett <james-xapian at
tartarus.org>
wrote:
> On Tue, Mar 22, 2016 at 11:12:12AM +0530, Ainish Dave wrote:
>
> > If I take up the task of 'clustering the search results' and I
am opting
> > for the new method to cluster the results, then what will be the
> > approximate amount of data that will be clustered? So appropriate
> > clustering mechanism can be thought of according to that.
>
> The intention is to use this in 'live search' environments, so
you'd
> probably want to pull a medium number of documents from the top of the
> search. It's a little difficult to say what a good number is, because
> it will depend somewhat on how broad the search is. But maybe in the
> 100..1000 range?
>
> J
>
> --
>   James Aylett, occasional trouble-maker
>   xapian.org
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.xapian.org/pipermail/xapian-devel/attachments/20160325/5813533d/attachment.html>