GSOC-2012 proposal

36 views
Skip to first unread message

Sheel Sindhu Manohar

unread,
Apr 2, 2012, 7:16:14 AM4/2/12
to twitte...@googlegroups.com
Hey friends i am in final year computer engineering at Jamia Millia Islamia, New Delhi

I have a proposal of project on graphical representation of the tweets on which we worked in my last semester as my minor project. The concept is the using twitter as the source of collecting feedbacks. And storing them in a separate database and do a semantic analysis on the tweets.

After that representing those in the form of graph or the pie chart representation.

For example: Like we having feedback on the song "Kolavari" or we can have movie feedback "rockstar" (I am attaching a file similar like )

We will search out the tweets on it and show the graphical representation on that. Also we can use this in many other places.

Kindly suggest me is it a good idea for proposal. I know the technologies which can used and somewhat algorithms on behind this but due to slow in its nature we failed it to represent it on real time.


--
-------------------------------------------------------------
Sheel Sindhu Manohar ( शील सिंधु मनोहर )

Manager JMILUG
www.jmilug.org  
-------------------------------------------------------------

judgement.png
emoction.png

Vinoth

unread,
Apr 3, 2012, 4:31:51 PM4/3/12
to Twitter GSOC
Hi sheel, its a good idea..but did u search for similar apps cuz I've
been working on similar concepts and i came across a lot of apps which
are vry similar to ur proposal...if u r confident tat ur idea is
unique then go ahead with it...I would also like to join ur project if
so :)

On Apr 2, 12:16 pm, Sheel Sindhu Manohar <tuxophi...@gmail.com> wrote:
> Hey friends i am in final year computer engineering at Jamia Millia
> Islamia, New Delhi
>
> I have a proposal of project on graphical representation of the tweets on
> which we worked in my last semester as my minor project. The concept is the
> *using twitter as the source of collecting feedbacks*. And storing them in
> a separate database and do a semantic analysis on the tweets.
>
> After that representing those in the form of *graph or the pie chart
> representation*.
>
> For example: Like we having feedback on the song "Kolavari" or we can have
> movie feedback "*rockstar*" (I am attaching a file similar like )
>
> We will search out the tweets on it and show the graphical representation
> on that. Also we can use this in many other places.
>
> Kindly suggest me is it a good idea for proposal. I know the technologies
> which can used and somewhat algorithms on behind this but due to slow in
> its nature we failed it to represent it on real time.
>
> --
> -------------------------------------------------------------*
> *Sheel Sindhu Manohar ( शील सिंधु मनोहर )
>
> Manager JMILUG
> *www.jmilug.org*
> -------------------------------------------------------------
>
>  judgement.png
> 103KViewDownload
>
>  emoction.png
> 92KViewDownload

Tushar rao

unread,
Apr 3, 2012, 4:35:24 PM4/3/12
to twitte...@googlegroups.com
Pls do checkout Arhivist tool opensourced with microsoft licence. Most the above mentioned stuff is already been done in one app or the another.

Sheel Sindhu Manohar

unread,
Apr 4, 2012, 8:05:57 AM4/4/12
to twitte...@googlegroups.com
@tushar i will definitely check out the tools but the idea is quit new for all of us that i am using it as feedback analyzer means trying to judge customers feeling through tweets mean its a primary idea just for a Glimpse i also having 2 other reasons which are connected to this only.

Idea 1: Using twitter collection of feeds for professional speakers and making a separate database sheets or slides at the runtime. Means we can connect this feedback analyzer in making a pie chart which can judge quality of speakers while on the basis of collecting feedback on twitter and throwing them directly to application and all happening in runtime.

Above idea till now not mentioned anywhere and can be implemented in 3months easily but little bit of concentration required and I need suggestions so that i can work on it I having this unique idea. Also i worked on it earlier but that was a failed attempt. 
--
-------------------------------------------------------------

Sheel Sindhu Manohar ( शील सिंधु मनोहर )

Manager JMILUG
www.jmilug.org 

Founder Linux Adda
www.linuxadda.org
-------------------------------------------------------------

Tushar rao

unread,
Apr 4, 2012, 2:26:19 PM4/4/12
to twitte...@googlegroups.com
More insights into semantic analysis, what kind of classifier you will be using and bias it towards rating lexicon or manually annotating raw tweets. More insights into this area can be helpful. This kind of product is more significant towards commercial analytical add-ons to already existing tools.
Also  more specifications into economic theory of opinion mining and behavioral science inputs can be helpful while playing with math equations instead of using directly using positive/ negative features. Some relevant papers can be found on Google scholar, do checkout.

Sheel Sindhu Manohar

unread,
Apr 4, 2012, 11:55:21 PM4/4/12
to twitte...@googlegroups.com
yeah exactly opinion mining but should work on both side parallel front-end and Back-end. Earlier I used lexical parser which was slower (as it was JAVA implementation )process we have to find some fast algorithm for this as we would be working on the real time.

Tushar rao

unread,
Apr 5, 2012, 2:26:10 AM4/5/12
to twitte...@googlegroups.com
Only back-end required. Simultaneous front end can worked by easy export. Go for python nltk, much faster than any NLP code in Java.
Reply all
Reply to author
Forward
0 new messages