|Date||Feb 05, 2015|
|Title||Abstract Meaning Representation: comparing languages and their representations|
Abstract Meaning Representation is a newly developedformalism for representing meaning, which abstracts from syntax and some otherphenomena but is (still) language-dependent. It has been developed by aconsortium of mostly U.S. universities, with team members including MarthaPalmer, Kevin Knight, Philipp Koehn, Ulf Hermjakob, Kathy McKeown, Nianwen Xueand others. In the talk, the basic facts about the AMR will be presented, andthen comparison will be made for Czech and English as carried out in detail ona small 100-sentence corpus; some examples from Chinese-English comparison willalso be shown. In addition, AMR will be compared to the deep syntacticrepresentation used in the set of Prague Dependency Treebanks, and observationswill be made about the level of abstraction used in these two formalisms. Thereis an ongoing work on possible conversion between the Prague deep syntacticrepresentation and the AMR representation, and the main issues of such aconversion will also be described, as well as plans for future studies andpossible corpus annotation work in the nearest future.
Prof. Jan Hajic is a full professor of ComputationalLinguistics at the Institute of Formal and Applied Linguistics, School ofComputer Science at the Charles University in Prague. His interests covermorphology of inflective languages, machine translation, deep languageunderstanding, and the application of statistical methods in natural languageprocessing in general. He also has an extensive experience in buildingannotated language resources. His work experience includes both industrialresearch (IBM Research, Yorktown Heights, 1991-1993) and academia (CharlesUniversity, Prague and Johns Hopkins University, Baltimore, MD, USA). He hasbeen the PI or Co-PI of several national and international grants and projects,most notably the large Czech Grant Agency grant for the Center of ComputationalLinguistics (2005-2011), several EU projects on Machine Translation(EuroMatrix/Plus, Faust, QTLeap, META-NET; currently HimL, QT21 and CRACKER)and the U.S.-based large ITR project “Malach” (coordinated by the VisualHistory Foundation, Los Angeles, CA, USA). Currently he is head of the largenational infrastructure project “LINDAT/CLARIN” (2010- 2015) and a vicedirector of the Institute. He has about110 cited publications, including a book on computational inflectionalmorphology and tagging.