Next Session on 21 Dec 2012 @ 10am

[Jun Ping]

Hi all,

As explained earlier, the next session will be postponed till the 21 Dec 2012 (Fri) at 10am.

The venue of the session will be announced closer to the date.

Hadi will be sharing this paper with us:

A Data-Driven Approach to Question Subjectivity Identification in Community Question Answering Tom Chao Zhou, Xiance Si, Edward Y. Chang, Irwin King and M. R. Lyu. Proceedings of the 26th AAAI Conference on Artificial Intelligence (AAAI-12). Jul 22-26, 2012.

More details of the paper can be found on our paper archiving page at http://wing.comp.nus.edu.sg/~wing.nus/sig/sig_nlp.html

All the best to anyone who is submitting to WWW!

Thanks!

Regards,

Jun Ping

[Weinan Zhang]

Hi All :

As Hadi asked this morning, I just check the definition about "POS guessing".

As [1] described:  Naturally,  when  tagging (POS tagging)  real-word  texts, one  can  expect  to  encounter  words  which  were not  seen  at  the  training  phase (OOV)  and  hence  not  included  into  the  lexicon.  This  is  where  word-Pos guessers (POS guessing)  take  their  place.

As [2] introduced: The task of POS guessing is quite different from traditional POS tagging. Traditional POS tagging involves assigning a single POS tag to a word token, provided that it is known what POS tag this word can take on in principle. This task requires a lexicon that lists possible POS tags for all words. However, unknown words are not in the lexicon, so the task of POS guessing of unknown words involves the guessing of a correct POS for an unknown word from the whole POS set of the current language. Obviously, traditional methods of POS tagging cannot effectively solve the problem of POS guessing of unknown words.

As [3] presented: The quality of OOV term detection and POS guessing directly influences the precision of querying more relevant information, and has become a crucial and challenging issue in IR.

[1] Learning Part-of-Speech Guessing Rules from Lexicon: Extension to Non-Concatenative Operations (Coling 1996) 

[2] A Method for Automatic POS Guessing of Chinese Unknown Words (Coling 2008)

[3] Fusion of Multiple Features and Supervised Learning for Chinese OOV Term Detection and POS Guessing (IJCAI 2011)

The above three papers introduced the definition of POS guessing, the main differences between POS tagging and POS guessing and the application/importance of OOV detection and POS guessing respectively.

For my understanding, POS guessing and OOV detection are two highly related tasks. I think the POS guessing is the specific task of POS tagging which is only towards to the OOV terms in text. 

In addition, the approaches to POS guessing is developing from rule based, statistical based to hybrid based. They mainly depend on the word leading and trailing characters and other morphological features. One can get the details from the above 3 papers.

Regards

Weinan

2012/11/23 Jun Ping <ng.ju...@gmail.com>

--
You received this message because you are subscribed to the NUS WING SIGNLP group. To post to this group, send email to sig...@wing.comp.nus.edu.sg. To unsubscribe from this group, send email to signlp+un...@wing.comp.nus.edu.sg. For more options, visit this group at https://groups.google.com/a/wing.comp.nus.edu.sg/d/forum/signlp?hl=en-US
 
 

[Hadi Amiri]

Thanks Weinan for your explanation. 

I think POS tagging is also about tagging OOV words. In fact, the majority of POS taggers

can deal with this issue by learning POS tag patterns from context (i.e. the POS tag of 

surrounding words). For example, if the tagger learns "VBZ + VBG" will be always followed 

by an "NN", then, in the following sentence, no matter "sausage" is OOV or not the tagger label 

it as NN. In other words, you may replace "sausage" with any unknown word and still expect 

the tagger to work accurately in this context.

My/PRP$ dog/NN also/RB likes/VBZ eating/VBG sausage/NN ./.

In this sense, perhaps these two different names represent the same concept. 

Just some thoughts... Thanks again,

Cheers,

-Hadi.

---

From: Weinan Zhang <weina...@gmail.com>
To: "sig...@wing.comp.nus.edu.sg" <sig...@wing.comp.nus.edu.sg>
Sent: Friday, November 23, 2012 3:57 PM
Subject: Re: [SIGNLP #32] Next Session on 21 Dec 2012 @ 10am

[Jun Ping]

Hi everyone,

  After a break of nearly a month, we will be holding the next SIGNLP session again on Fri 21 Dec 2012 at 10am.

  It will be held in DR5 (AS6 Level 2).

  See you all soon!

Regards,

Jun Ping

[Hadi Amiri]

Dear All,

I am presenting the following paper at our SIGNLP group tomorrow. Here are the slides

and the paper attached. 

A Data-Driven Approach to Question Subjectivity Identification in Community Question Answering Tom Chao Zhou, Xiance Si, Edward Y. Chang, Irwin King and M. R. Lyu. Proceedings of the 26th AAAI Conference on Artificial Intelligence (AAAI-12). Jul 22-26, 2012.

Regards,

-Hadi.