Events & News

US-based Genii to license intelligent word semantics software developed by CSIDM

SEATTLE, USA. Jan 31 2013 — Genii announces today that it has signed
a partnership agreement with the China-Singapore Institute of Digital
Media (CSIDM) of the National University of Singapore (NUS) to enhance
the game-play of VocabGenii, the company’s popular online vocabulary
learning game. Genii will license an intelligent word semantics
software developed at CSIDM to improve the vocabulary selection for
VocabGenii.

The CSIDM software is created by Min-Yen Kan, associate professor of
CSIDM, and his team. It estimates the difficulty of guessing a word
from one of its given definitions. The software uses statistics about
the word and its definition drawn from public sources to generate its
output.

For more details, please see:
http://www.geniicompany.com/2013-01-31-genii-nus-partnership.html

NUS Team First in Shared Task on Grammatical Error Correction

A team from NUS led by Associate Professor Ng Hwee Tou and comprising his NGS PhD student Daniel Dahlmeier and CS Depart-ment research assistant Eric Ng emerged as the best performing team in the HOO (Helping Our Own) 2012 shared task. The goal of the shared task was to come up with innovative algorithms that can automatically correct errors involving determiner and preposition errors made by English language learners.

Fourteen teams from academia and industry, including well-known institutions like Cambridge University and the Educational Testing Service, participated in the shared task. The teams were given a set of student essays with error annotations to develop their algorithms. Afterwards, the teams were given a blind test set of essays without error annotations and their grammatical error correction algorithms need to automatically correct all determiner or preposition errors in the test essays.

The system submitted by the NUS team achieved the highest score for determiner correction, preposition correction, and the overall correction task among all 85 systems submitted by the four-teen teams. The good result was sweetened by the fact that the NUS team only submitted a single system, and it gave the best scores. The success in the shared task shows that the innovative natural language processing technology developed at NUS can have practical impact on real-world problems.

Featured in NUS Newsletter

Wishing You Happy Chinese New Year

CSIDM researcher recognized as distinguished scientist by ACM

We are pleased to share that Prof Xu Changsheng of our CSIDM had been identified by the Association of Computing Machinery (ACM) as one of the Distinguished Members in 2012 for their individual contributions and their singular impacts on the dynamic computing field. In particular, Prof Xu was recognized as one of the 2012 Distinguished Scientists.

Congratulations, Prof Xu!

Details of the ACM announcement can be found in this link – http://www.acm.org/press-room/news-releases/2012/distinguished-2012

Team’s system came out top in International Evaluation Exercise

In June 2012, a representative team lead by Prof Ng Hwee Tou participated in an international evaluation exercise, called the HOO 2012 shared task, on error detection and correction of determiners and prepositions. The team had emerged as the best team among 14 participating teams (which include Educational Testing Service, Cambridge University, and the University of Illinois at Urbana-Champaign). The team’s system was the best system among 85 systems submitted to the shared task. Our research on this topic had also appeared in five papers published in premier conferences in natural language processing. In addition, the development team had carried out evaluation of commercial software, and our system outperformed three commercial grammar checkers on correction of articles and prepositions.

16-17 July 2012 [Industry Outreach] CSIDM technologies strive to help industry players break down the language barrier

CSIDM recently participated in the International Research Centres (IRC) Research Exhibition and Industry Networking Event held on 16-17 July 2012 at the Shaw Foundation Alumni House, NUS.  The event was jointly organized by the NUS Enterprise (NUSE) and NUS Interactive and Digital Media Institute (IDMI).

The objectives of this 2-day event were two-folds mainly, to showcase innovative research carried out by various IRCs as well as to catalyze commercialization of research by building stronger links with the industry.   This event is supported by the Interactive Digital Media Programme Office (IDMPO) of the Media Development Authority of Singapore (MDA).

As one of the major participating IRCs for the event, CSIDM demonstrated several of the key technologies researched and developed by the Centre over the last 3 years.  These technologies are now available for interested industry players to tap into.  Interested industry members can check out this link – http://www.idm.sg/annex-a-csidm/ to find out more about the technologies on offer as well as on-going funding support available for industry members to tap into.

Check out these photos of our students and researchers, hard at work, during the CSIDM technology showcase.

Finally a big thank you to all the unsung heroes behind the CSIDM showcase for the 2-day event

• FlashMob: A social flashcard application
• Tweet Classification

• Personalised video search
• Landmark search

Industry Networking Event on 17 July 2012

Best Paper Award for Associate Professor Ng Hwee Tou and Dr Lu Wei

A paper co-authored by Dr Lu Wei and Associate Professor Ng Hwee Tou has won the Best Paper Award at the 2011 Conference on Empirical Methods in Natural Language Processing (EMNLP 2011). The conference was held in Edinburgh, UK, from 27 to 29 July 2011. EMNLP is a premier conference on natural language processing, bringing together researchers working in the research area from various parts of the world.

The winning paper, titled “A Probabilistic Forest-to-String Model for Language Generation from Typed Lambda Calculus Expressions”, presents the first probabilistic model for generating natural language sentences from typed lambda calculus expressions. Its contributions include a novel forest-to-string generation algorithm and a novel grammar induction algorithm. The new model outperforms previous proposed models when evaluated on common benchmark data sets.

The conference received 628 submissions, out of which 95 were accepted for oral presentation, and 54 for poster presentation. The area chairs of the conference nominated best paper candidates in their areas, and a Best Paper Award Committee comprising leading researchers in the field was formed to select the best paper. The Committee unanimously selected the paper of Dr Lu and Prof Ng for the Best Paper Award of the conference. Dr Lu is an SoC research fellow and a former PhD student of Prof Ng.

Reported in NUS News

International Review Panel Visit

31 May 2011 – 2 June 2011

CSIDM had the pleasure of having Prof. Dan Roth (University of Illinois at Urbana Champaign), Prof. Mari Ostendorf (University of Washington, Seattle) and Dr Zhang Hong-Jiang (Microsoft Distinguished Scientist and Chief Technology Officer, Microsoft Asia-Pacific Research and Development Group) as our distinguished International Review Panel members.  The IRP members were in Singapore from 31 May to 2 June 2011 to review and provide feedback to the 6 research collaboration projects between NUS and CSIDM.

During the course of the visit, the IRP members also shared their knowledge with NUS students and staff at a seminar held on 2 June 2011.

Reference material (Prof Dan Roth’s Slides)

An interpreter in your pocket

An app that translates on the go is being developed in Singapore

WE’VE all fumbled with unwieldy phrasebooks and dictionaries — probably to unintentionally hilarious effect — on our trips overseas. Technology to the rescue.

Imagine saying a Chinese or Malay phrase into your iPhone or iPad, and having the device translate it instantly into English. Imagine how many frustrating conversations with taxi drivers you could avoid on your travels.

When the NUS Speech Translation System app debuts on the iOS platform by the end of the year, you’ll have two professors at the National University of Singapore — Associate Professor Ng Hwee Tou and Assistant Professor Sim Khe Chai — to thank.

The two have been juggling their day jobs as lecturers at the NUS School of Computing while developing this app. Dr Ng was doing research on text translation while Dr Sim
worked on automatic speech recognition. The Speech Translation System seemed a
natural product of their two specialisations. “We’ve basically combined the two technologies together,” said Dr Ng.

They say it will be particularly useful to foreigners visiting Singapore, or Singaporeans visiting countries where Malay and Mandarin are the default modes of communication.

Read Full Article:

http://imcmsimages.mediacorp.sg/CMSFileserver/documents/006/PDF/20110114/1401ITP019.pdf

http://www.todayonline.com/Tech/Mobileapps/EDC110114-0000174/An-interpreter-in-your-pocket