Noah's ARK

Animals entering the ark.

Edward Hicks' rendition of Noah's ark, and an excuse to violate One Sense Per Discourse.

Noah's ARK^[1] is Noah Smith's informal research group at the Language Technologies Institute, School of Computer Science, Carnegie Mellon University. (The research is formal; the group is informal.) As you may have guessed, our research focuses on problems of ambiguity and uncertainty in natural language processing, including morphology, syntax, semantics, translation, behavioral/social phenomena observed through language.

^[1]The acronym is ambiguous; possible interpretations might include Ambiguity Research Kith or Ambiguity Resolution K. or A. R. Kibbutz. With apologies to the Bible and DAGS.

Lab meetings: Mondays at 11am, NSH 4513

ARK Research in News and Blogs

May 2008: Little Green Footballs, a political blog, happened on some data Tae had on her website, prompting fascinating (to us) speculation about what we were up to. See Noah and William's response here.

January 2008: The New Scientist and Tech Digest blogs commented on Danny and Noah's relative keyboard.

November 2007: Yahoo granted us access to M45, a 4,000-processor supercomputer.

Projects and Resources

RAVINE project to explore news through quotes
DAGEEM code for unsupervised dependency grammar induction
Machine translation resources
CURD: the Carnegie Mellon University Recipe Database

Researchers

Name	Position	Topics	Languages (Spoken and/or Written)	Languages (Researched)	Languages (Hacked In)	Favorite Term of Venery
Shay Cohen	Ph.D. student, LTI	morphosyntactic parsing, unsupervised parsing, parsing algorithms	Hebrew, English, Swedish (lite)	Arabic, Chinese, English, German, Hebrew	C++, C, Perl, Scheme, Java, C#	shoal
Dipanjan Das	Ph.D. student, LTI	sentence-sentence semantic relationships, parsing (RAVINE)	English, Bengali, Hindi, Sanskrit (swalpam)	English, Bengali	Java, C, C++, Perl	bloat
Kevin Gimpel	Ph.D. student, LTI	statistical NLP and translation	English, German, Italian	English, Portuguese, German	Java, C, C++, Matlab	crash
Mohammad Haque	undergraduate researcher, CS
Michael Heilman	Ph.D. student, LTI	question-answer modeling (RAVINE)	English, Japanese (sukoshi)	English, English as a second language	Java, Perl, PHP, Ruby, C++	mob
Dimitry Levin	undergraduate researcher, Computational Finance
André Martins	Ph.D. student, LTI and Universidade Técnica de Lisboa	structured and kernel machine learning, parsing	Portuguese, English, Spanish, German (ein bisschen), French (un petit peu)	Portuguese, English, Spanish	C, C++, Matlab	scurry
Nate Schneider	Ph.D. student, LTI	parsing, grammar learning, semantics; cognitive linguistics	English, Hebrew (כמה), Arabic (قليل), French (un peu)	English, Hebrew	Python, Java, PHP, Scheme, C#, C++	smack
Tae Yano	Ph.D. student, LTI	NLP in the political domain, rich models of structured NL data (e.g., blogs)	Japanese, English, Spanish, French	English	C, C++, Java, Perl, Python	husk
Noah Smith	Assistant Professor, LTI & MLD	(all of the above)	English, French (un peu)	Arabic, Bulgarian, Czech, English, French, German, Hebrew, Korean, Mandarin, Portuguese, Turkish	C++, Perl, Dyna	parade

Alumni

Zach Paine, B.S. '08 (research programmer) → Google
Cari Sisson, M.S. '08 → Lyrix
Dan Tasse, B.S. '08 → Google
Mengqiu Wang, M.S. '07 → Ph.D. student at Stanford

Friends & collaborators

We are very open to collaboration, inside and outside of CMU. Technical correspondence is welcome. Some people we've been talking and working with lately: Pedro Aguiar (Universidade Técnica de Lisboa), Alan Black (LTI), William Cohen (MLD, LTI), Ric Crabbe (U.S. Naval Academy), Jason Eisner (Johns Hopkins University), Mário Figueiredo (Universidade Técnica de Lisboa), Rebecca Hwa (University of Pittsburgh), Shimon Kogan (Tepper School of Business), John Lafferty (CSD, MLD), Alon Lavie (LTI), Teruko Mitamura (LTI), Thuy Linh Nguyen (ISL, LTI), Kemal Oflazer (Sabancı University), Daniel Rashid (LTI), Bryan Routledge (Tepper School of Business), Jacob Sagi (Vanderbilt University), Stu Shulman (University of Pittsburgh), Rob Simmons (CSD), David Smith (Johns Hopkins University), Doug Vail (CSD), Ashish Venugopal (ISL, LTI), Stephan Vogel (ISL, LTI), Mengqiu Wang (Stanford University), Shuly Wintner (University of Haifa), Joy Zhang (ISL, LTI), and Andreas Zollmann (ISL, LTI).

Acknowledgments

Our research is supported in part by the DARPA Computer Science Study Panel program (grant numbers HR-00110110013 and NBCH-1080004), by the National Science Foundation (grant numbers IIS-0713265 and IIS-0836431 and a graduate fellowship to Michael Heilman), by an IBM Faculty Award, by a grant from Google, and by computational resources provided by Yahoo.