[Xapian-devel] [GSoC 2014] About "Clustering of Search Results"

[Chi Liu]

*Olly Betts* March 18, 2014, 5:07
a.m.<http://www.google-melange.com/gsoc/proposal/review/student/google/gsoc2014/liuchi/5629499534213120#c5707702298738688>
 wrote:

>Thanks for your proposal, and sorry for not getting to it sooner. Overall
I like the proposal. >One thing I'm not clear on - are you intending to
base your implementation of the existing branch,

>or to start afresh? And why are you taking that approach?

I am intending to base my implementation of the existing branch. Base my
implementation on the

existing branch will let us understand more clearly what caused the
disappointing efficiency of the old

code. If we have time at last, we could implement a different clustering
algorithm afresh with the

experience of avoiding disappointing efficiency.


I have added this to my proposal.

>Do you have a plan for how to get human judgements to compare the
algorithms?

> We've tried this before (with the snippets project in 2012) but sadly
sending out a request to our

>mailing lists asking people to run through a comparison in a simple web UI
resulted in hardly any

>uptake. This makes me a bit concerned about the number of "human judge
effectiveness" entries

>in your project plan - I think we either need a plan for motivating people
better, or a way to

>compare which doesn't require many judgements.

Massive human judgement is not necessary. What I mean human judgement is to
generate several

test cases artificially to help me to know whether the clustering algorithm
could assign documents to correct

groups. A test case include a query, a list of search results and a group
assignment of these search results.

Then we could measure the effectiveness and modify the code promptly. The
set of test cases don't need

to be very large and I could generate it by myself.


I have added this to my proposal.

>The overlap with courses and exams isn't a big problem. A common
workaround is to start

>coding early during the community bonding period - in your case that would
mean you could even

>take 2 weeks off GSoC for your final exams, and still actually have made
good progress by the

>mid-term. I don't know how busy you are with courses, etc during the
community bonding period though.

Yes, I could start coding early.


>We ask students to submit a patch for Xapian so we can get a better feel
for what their skills and aptitudes are. >If you've already submitted a
patch, could you give us a URL? If not, it's better if the patch is
something

>in an area related to your project, but that's not a firm requirement -
you can either find a bug in the

>tracker, take a look at http://trac.xapian.org/wiki/ProjectIdeas or work
on a first step towards your project.


I am regret I have not noticed this before. And I will start to do this
right now.




2014-03-17 6:13 GMT+08:00 Chi Liu <liuchi09 at gmail.com>:

> Hello,
> I have submitted my proposal on GSoC.
> But I have little idea about the timeline. Many things are difficult to be
> determined.
>
>
> Cheers,
>    Liu Chi
>
>
> 2014-03-11 21:33 GMT+08:00 Olly Betts <olly at survex.com>:
>
> On Tue, Mar 11, 2014 at 10:11:31AM +0800, Chi Liu wrote:
>> > Thank you for your patient explanation about the project. My
>> > understanding about the project "Clustering of Search Results" is that
>> > we mainly focus on processing speed of the existing code.
>>
>> We need something which can cluster larger result sets faster than the
>> current code.  Speeding up the existing code might be the best way to do
>> that, but we could start again.  If we start again, I'd suggest it would
>> be prudent to try to understand why the previous attempt didn't succeed.
>> We don't want to end up repeating that.
>>
>> > By "find new approaches" I mean trying other known clustering
>> algorithms.
>>
>> OK - that's fine then.
>>
>> > What I am concerned is whether the low efficiency is caused by
>> > improper algorithm. I am reading the existing clustering branch code
>> > and have not completely finished yet. I might be able to talk more
>> > about existing code in my application of GSoC. But now, I really can
>> > not comment before fully understanding exiting code.
>>
>> Sure.
>>
>> > My idea about measure clustering effectiveness is that when we trying
>> > other known clustering algorithms, we can use the old clustering
>> > result as a baseline.  If the difference of clustering results is
>> > acceptable and new clustering algorithm has high efficiency, we may
>> > find a better approach. I will give more details about this in my
>> > application of GSoC.
>>
>> Great.
>>
>> Cheers,
>>     Olly
>>
>
>
>
> --
> Chi Liu
> +86-15210624786
> Undergraduate Student
> Team of Search Engine and Web Mining
> School of Electronic Engineering  and Computer Science
> Peking University, Beijing, 100871, P.R.China
>



-- 
Chi Liu
+86-15210624786
Undergraduate Student
Team of Search Engine and Web Mining
School of Electronic Engineering  and Computer Science
Peking University, Beijing, 100871, P.R.China
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.xapian.org/pipermail/xapian-devel/attachments/20140318/ed5d4004/attachment-0001.html>