Tom Kwiatkowski
Email : t.m. my last name (no spaces) @ sms.ed.ac.uk

School of Informatics
Informatics Forum
10 Crichton Street
Edinburgh EH8 9AB
Phone : +44 (0) 131 650 4414

Research interests

I am a final year PhD student in ILCC at the University of Edinburgh, supervised by Mark Steedman and Sharon Goldwater. I am also working closely with Luke Zettlemoyer.

My current research is in CCG parsing with semantics. We are building systems that can learn CCG grammars from natural language sentences paired with compositional representations of their meaning. The aim of this work is two-fold: to create general purpose methods of semantic parsing for question answering and dialogue systems; and to provide a basis for modelling a child's acquisition of syntax.

More generally, I am interested in the use of machine learning techniques for structural learning. This is something that we often cover in our machine learning for natural language processing reading group.

Publications

Lexical Generalization in CCG Grammar Induction for Semantic Parsing. Tom Kwiatkowski, Luke Zettlemoyer, Sharon Goldwater and Mark Steedman. Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP), Edinburgh, UK, 2011

Inducing probabilistic CCG grammars from logical form with higher-order unification. Tom Kwiatkowski, Luke Zettlemoyer, Sharon Goldwater and Mark Steedman. Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP), Cambridge, MA, 2010

Computational Grammar Acquisition from CHILDEs data using a probabilistic parsing model. Tom Kwiatkowski, Sharon Goldwater and Mark Steedman. Psychocomputational Models of Human Language Acquisition (PsychoCompLA), Amsterdam, 2009.

This is a funding review poster summarising some of our work on modelling language acquisition using a Bayesian model trained on CHILDEs data.

Resources

eveTrainPairs contains the child-directed utterances paired with meaning represntations used for our 2012 EACL paper.

UBL is the code and data used for our 2010 EMNLP paper.