[Xapian-devel] GSoC 14

Olly Betts olly at survex.com
Fri Mar 14 23:08:25 GMT 2014

On Thu, Mar 13, 2014 at 01:31:21AM +0530, Manmohan Chachan wrote:
> I am looking to participate in GSoC 14 and have attached the proposal with
> the mail.

Thanks for your proposal.

> I was unable to submit it earlier because of my mid-semester examination.
> So please review the proposal and suggest any correction if required for
> final submission of proposal.

A couple of procedural notes first:

* Please use our application template as this helps to ensure that
  you cover all the questions we want to know the answers to:

* Your proposal needs to be submitted on https://google-melange.com/
  or else we can't consider it.  Once submitted, you can edit it up
  until the deadline (19:00 UTC on March 21st).

Reading your proposal, I don't think it makes a convincing case that
the ideas you have will actually achieve the aims you claim.  Some
theoretical background to the ideas would help (e.g. published papers).

You might want to look at "Learning to Rank", which uses machine
learning to tune the weightings based on multiple factors.  Xapian
has a "Learning to Rank" framework, but it needs work to be usable
(see our GSoC project ideas list for more information).

You need to make it clear what the mechanics behind your sliding bar
will be - will it cause the search to rerun on the server, will you
try to send the broken down weights to the client and recompute and
reorder results there, or something else?

And you need to think about how this will interact with Xapian's
existing weighting framework, and the optimisations this provides.

Overall, we need to be convinced that you actually have a plan for
how to make this work, that this plan will fit with how Xapian works,
and that this is something you could feasibly implement in the GSoC
timescale.  It's not enough just to have a concept of what it will look
like (a sliding bar in the search UI) and a few high level concepts like
"based on semantic understanding of webpage".

Your timeline granularity is too coarse - the items should last a few
days to a week at most.  It is too hard to judge how long something
will realistically take unless you break it down more than you have.

| Based on semantic understanding of webpage

You need to explain what you have in mind by this.

| Implementing this search with various data such as documents, video,
| images music, online shopping etc. and fixing bug.

Where are you going to get these datasets from?

| Implementing support for external devices

I have no idea what "support for external devices" is meant to mean

You also haven't put anything against the last 10 days of time.

Have you checked out and built the code yet?  Have you tried to fix
any bugs?  If not, http://trac.xapian.org/wiki/GSoC%20Guide has useful
pointers for getting started.  Submitting a patch is the best way to
show us you have the skills required for the project you're proposing.


More information about the Xapian-devel mailing list