reg : QA system update

2 views
Skip to first unread message

Dheeraj Rajagopal

unread,
Jul 27, 2011, 1:51:23 PM7/27/11
to vit...@googlegroups.com, ruchir
this is what we have done so far , 

given a question , i have coded some basic steps : ( we shall improve this further )

1. removed the stop words ( http://en.wikipedia.org/wiki/Stop_words )
2. For now , I am assuming that the other words excluding the stop-words are the keywords
3 . I have done POS tagging for the sentence and extract the POS for the keywords .

I want someone to implement the same in Stanford-NLP library . we need to compare their results . 

the knowledge base we will use is freebase 

I am attaching the freebase paper along with this mail . 

Now , we have to do the following . 

 we need to compare the keywords with the help of the POS tag  and search the knowledge base and come up with something that is on the database . Doesnt matter whether it is right or wrong . we need to evaluate what we are getting and we shall decide about what to do further . 

anyine who would like to do this , may reply to this mail . I will help you with the steps . 
 

--
Regards

Dheeraj

freebase.pdf

Devashish Shankar

unread,
Aug 10, 2011, 7:27:33 AM8/10/11
to VIT LUG
I downloaded the POS tagger library from Stanford NLP

http://nlp.stanford.edu/software/tagger.shtml

It has inbuilt tools for tagging, and its pretty simple to tag words
using those tools.

The main decision that we have to make is which words exactly would be
counted as "keywords".

The library uses the "Pen Treebank tag set" for tagging
http://www.ims.uni-stuttgart.de/projekte/CorpusWorkbench/CQP-HTMLDemo/PennTreebankTS.html

The automatic tagger works in the following manner:
INPUT:
A passenger plane has crashed shortly after take-off from Kyrgyzstan's
capital,

OUTPUT:
A_DT passenger_NN plane_NN has_VBZ crashed_VBN shortly_RB after_IN
take-off_NN from_IN Kyrgyzstan_NNP 's_POS capital_NN ,_,

Using the treebank tag set we can figure out what is what and select
the keywords.

Keywords
---------------
Please advise on what should be the keywords?
Nouns, Adjectives ......


On Jul 27, 10:51 pm, Dheeraj Rajagopal <dheeraj.go...@gmail.com>
wrote:
> this is what we have done so far ,
>
> given a question , i have coded some basic steps : ( we shall improve this
> further )
>
> 1. removed the stop words (http://en.wikipedia.org/wiki/Stop_words)
> 2. For now , I am assuming that the other words excluding the stop-words are
> the keywords
> 3 . I have done POS tagging for the sentence and extract the POS for the
> keywords .
>
> I want someone to implement the same in Stanford-NLP library . we need to
> compare their results .
>
> the knowledge base we will use is freebase <http://www.freebase.com/>
>
> I am attaching the freebase paper along with this mail .
>
> Now , we have to do the following .
>
>  we need to compare the keywords with the help of the POS tag  and search
> the knowledge base and come up with something that is on the database .
> Doesnt matter whether it is right or wrong . we need to evaluate what we are
> getting and we shall decide about what to do further .
>
> anyine who would like to do this , may reply to this mail . I will help you
> with the steps .
>
> --
> Regards
>
> Dheeraj
>
>  freebase.pdf
> 443KViewDownload

Rohit Mishra

unread,
Aug 10, 2011, 2:14:14 PM8/10/11
to vit...@googlegroups.com
I think we will need the question words, nouns, and adjectives. Will we need adverbs ? Can anyone think of a relevant question ? 

Devashish Shankar

unread,
Aug 10, 2011, 3:02:03 PM8/10/11
to VIT LUG
Who is the fastest running athlete?

fastest - adverb but relevant



On Aug 10, 11:14 pm, Rohit Mishra <ro...@rohitmishra.me> wrote:
> I think we will need the question words, nouns, and adjectives. Will we need
> adverbs ? Can anyone think of a relevant question ?
>
> Rohit Mishra
> VIT University
>
> Blog:http://www.rohitmishra.me/blog
> Twitter:http://twitter.com/movingahead
> Facebook:http://www.facebook.com/rohitmishra
>
> On Wed, Aug 10, 2011 at 4:57 PM, Devashish Shankar <
>
>
>
>
>
>
>
> devashish.shan...@gmail.com> wrote:
> > I downloaded the POS tagger library from Stanford NLP
>
> >http://nlp.stanford.edu/software/tagger.shtml
>
> > It has inbuilt tools for tagging, and its pretty simple to tag words
> > using those tools.
>
> > The main decision that we have to make is which words exactly would be
> > counted as "keywords".
>
> > The library uses the "Pen Treebank tag set" for tagging
>
> >http://www.ims.uni-stuttgart.de/projekte/CorpusWorkbench/CQP-HTMLDemo...

Rohit Mishra

unread,
Aug 10, 2011, 3:11:34 PM8/10/11
to vit...@googlegroups.com
Ok, so we need adverbs in that case. 
Reply all
Reply to author
Forward
0 new messages