Main Page
From NLPReading
NLP Reading Group at School of Computing, NUS
Hello! Welcome to the Natural Language Processing Reading Group Wiki at the School of Computing of National University of Singapore.
After hibernating for half a year due to the tight schedules, we have resumed the NLP Reading Group in the coming semester (Semester 1, AY2006-2007). Similar to last year, the NLP reading group will meet biweekly. The NLP reading group and the Machine Learning Reading Group will be on alternate weeks. Please send your suggestions and comments on the topics we should discuss for this semester to the mailing list. Thanks! See you soon in the reading group!
Schedule for Semester I, 2007-2008
| Date | Topic | Discussion Leader | Venue |
|---|---|---|---|
| Sep 6 | Web people search | Yee Fan | Meeting Room 3 (SoC1 05-28) |
| Sep 20 | Machine translation | Hendra | Executive Classroom (SoC1 05-46) |
| Oct 4 | Synchronous grammars | Jin | Meeting Room 6 (AS6 05-10) |
| Oct 18 | Parsing | Yee Seng | Meeting Room 6 (AS6 05-10) |
| Nov 1 | Coreference resolution | Shanheng | Meeting Room 6 (AS6 05-10) |
| Nov 15 | Feature selection | Hai Leong | Meeting Room 6 (AS6 05-10) |
| Nov 29 | Textual entailment | Ziheng | Meeting Room 6 (AS6 05-10) |
| Dec 13 | TBA | TBA | TBA |
29 Nov 2007: Textual entailment
Time: 10:30am
Venue: Meeting Room 6 (AS6 05-10)
For this week's NLP reading group on Recognizing Textual Entailment (RTE), I will be discussing the top RTE system in the second PASCAL RTE challenge. Here is the link to the paper: http://www.cs.biu.ac.il/~dagan/RTE2/Proceedings/14.pdf
For a detailed introduction to current approaches to RTE, here is a tutorial: http://www.cs.biu.ac.il/~dagan/TE-Tutorial-ACL07.ppt
If time allows, I will discuss the top RTE system in this year's PASCAL RTE challenge. The link to the paper: http://acl.ldc.upenn.edu/W/W07/W07-1428.pdf
15 Nov 2007: Feature selection
Time: 10:30am
Venue: Meeting Room 6 (AS6 05-10)
I am supposed to lead a discussion on 15 Nov, on feature selection. I have chosen the following paper: Spectral Feature Selection for Supervised and Unsupervised Learning (ICML 2007) http://www.machinelearning.org/proceedings/icml2007/papers/444.pdf
1 Nov 2007: Coreference resolution
Time: 10:30am
Venue: Meeting Room 6 (AS6 05-10)
The topic will be coreference resolution with semantic knowledge. I have provided a short review (http://www.comp.nus.edu.sg/~zhaoshan/sem_coref_review.pdf). During the reading group, we'll discuss one of the latest papers:
Simone Paolo Ponzetto and Michael Strube. Exploiting Semantic Role Labeling, WordNet and Wikipedia for Coreference Resolution. HLT-NAACL 2006.
18 Oct 2007: Parsing
Time: 10:30am
Venue: Meeting Room 6 (AS6 05-10)
The upcoming NLP Reading Group next week will be on Parsing. Note that my research area is not parsing; however, I'll share what I have learnt so far.
Naturally, I'll cover Collin's Model. His Computational Linguistics journal article (preliminary version) is freely available here: http://people.csail.mit.edu/mcollins/papers/CL2003.ps
However, I think his PhD thesis is clearer. I'll be likely to concentrate on Chapter 3 of his thesis: http://people.csail.mit.edu/mcollins/papers/thesis.ps, which describes all the parameterizations for his models 1, 2, 3.
I might also give a brief review of chart-parsing. I'm providing an algorithm from the book "AI: a modern approach" in case you don't have the book: http://www.comp.nus.edu.sg/~chanys/parsing.pdf Here's an example work-thru of the algo: http://cs.nyu.edu/faculty/davise/ai/chart-ex.txt
In the next few days, I'll source for some more things to discuss.
4 Oct 2007: Synchronous grammars
Time: 10:30am
Venue: Meeting Room 6 (AS6 05-10)
The topic for the upcoming meeting is Synchronous Grammars. The focus of the meeting will be mainly on Synchronous CFG; however, other forms of synchronous grammars will also be briefly introduced.
The format will be tutorial-like except that Q&A and discussions are allowed and welcomed at all times.
Here is the reading list: 1) An introduction to Synchronous Grammars: http://www.isi.edu/~chiang/papers/synchtut.pdf 2) A Hierarchical Phrase-Based Model for Statistical Machine Translation: http://www.isi.edu/~chiang/papers/chiang-acl05.pdf
Please just read this two to get some general knowledge about synchronous grammars. There are also other papers (around 10) which will be briefly introduced on various aspects of synchronous grammars. I will update the list again after the meeting if needed.
20 Sep 2007: Machine translation
Time: 10:30am
Venue: Executive Classroom (SoC1 05-46)
Our big theme for this section is on the decoding part of statistical machine translation (SMT), with the particular focus on integrating language model into syntax-based SMT. The integration would result in a huge search space and the best result is often impossible to obtain. We are going to look at one of several ways to deal with this issue, especially from engineering perspective. Here are the two relevant papers (but probably we are going to discuss only the first paper):
(1) Forest Rescoring: Faster Decoding with Integrated Language Models by Liang Huang; David Chiang [1]
(2) An Efficient Two-Pass Approach to Synchronous-CFG Driven Statistical MT by Ashish Venugopal, Andreas Zollmann, and Stephan Vogel [2]
6 Sep 2007: Web people search
Time: 2pm
Venue: Meeting Room 3 (SoC1 05-28)
The second session will be on Web People Search (WePS). From a NLP perspective, WePS can be seen as a Word Sense Discrimination problem. On the other hand, it can also be seen as an Entity Resolution problem from the database point of view.
(1) Gideon S. Mann and David Yarowsky. Unsupervised Personal Name Disambiguation. Conference on Natural Language Learning (CoNLL), pages 33-40, 2003.
(2a) Dmitri V. Kalashnikov, Stella Chen, Rabia Nuray, Sharad Mehrotra and Naveen Ashish. Disambiguation Algorithm for People Search on the Web. IEEE International Conference on Data Engineering (ICDE), pages 1258-1260, 2007.
(2b) Dmitri V. Kalashnikov and Sharad Mehrotra. Domain-Independent Data Cleaning via Analysis of Entity-Relationship Graph. ACM Transactions on Database Systems (TODS), 31(2), pages 716-767, 2006.
(3) Xiaojun Wan, Jianfeng Gao, Mu Li and Binggong Ding. Person Resolution in Person Search Results: WebHawk. ACM Conference on Information and Knowledge Management (CIKM), pages 163-170, 2005.
Note that (2a) is a short paper that defers most of its details to (2b). We will try to discuss all these papers, but we might skip (3) if we run out of time.
There are many other papers on the topic of WePS. Interested readers can read the SemEval 2007 WePS task description papers and the references therein.
23 Aug 2007: Organizational
Time: 1030am -- 1200noon
Venue: Meeting Room 3 (SoC1 05-28)
The first session will be an organizational meeting. We will discuss what are the topics for this semester. Hopefully it won't take long. As the topics are not fixed yet, we will discuss the best paper of ACL2007.
Yuk Wah Wong; Raymond Mooney: Learning Synchronous Grammars for Semantic Parsing with Lambda Calculus
Our Mailing List at Google Groups
To ensure that you receive prompt updates from the reading group, please subscribe to the mailing list.
