I am a lecturer in the Department of Philosophy at King's College London.
I am also an honorary Research Fellow in the Department of Computer Science in Royal Holloway University of London.
I am President of the ACL Special Interest Group on Natural Language Learning (SIGNLL) which organizes the CoNLL conferences, and I am chair of the steering committee of the International Conference on Grammatical Inference, which organizes the ICGI conferences.
A very brief research narrative: the majority of the papers I have written are to do with unsupervised learning of natural language, and its relevance to first language acquisition. I have approached this both theoretically and practically: trying to define what a good definition of learnability is, trying to prove that you can learn languages according to various models of learnability, designing algorithms, and writing computer programs that can learn models of language both from synthetic and natural examples.
In the last few years I have been looking at formal results. If a learning algorithm works at all, then it should be possible to identify a class of languages, and some conditions such that you can prove that the learner will acquire that class under those conditions.
I have published some papers on learning regular languages and finite automata, which are summarised on this page.
In the last few years I have been working almost exclusively on context free and context sensitive grammatical inference using techniques from distributional learning ; this work is summarised on this page.
I won the Omphalos competition, and the Tenjinno competition, which were two grammatical inference competitions in learning context free grammars and transductions, respectively.
Recently I have been working on learning context free and context sensitive languages, using some ideas from structuralist linguistics. A paper on this won the best paper award at CoNLL 2010. The paper is available here. The paper gives an algebraic treatment -- the slides have lots of diagrams that may help -- they are here.
Before arriving here I was a post-doc at ISSCO, at the University of Geneva working on the IM2 Project, a large Swiss project on Multimodal Information Management.
My first degree was in Mathematics from Trinity College, Cambridge. My Ph.D. is from the University of Sussex.
I am working on a project called Statistical Models of Grammaticality (SMOG) which is based an King's College, London.
I taught a course at LOT 2012: the course page is here.
William Sakas and I organised a workshop on Psychocomputational Models of Language Acquisition at the LSA annual meeting in Portland Oregon on Thursday, January 5th, 2012. Further details are here. This was collocated with the Input and Syntactic Acquisition Workshop 2012 organised by Lisa Pearl and Jon Sprouse.
Shalom Lappin and I taught a course at the LSA Summer Institute, in Boulder, July 7 to August 2, 2011.
I was co-editor of a special issue of the Journal of Machine Learning Research on Grammar Induction, Representation of Language and Language Learning.
I taught a course at ESSLLI 2010. This was called Learnable representations of languages. The web page for this course is here.
I was co-editor of a special issue of the journal Research on Language and Computation. The CFP can be found here.
I was one of the organizers of a workshop at NIPS 2009: Grammar Induction, Representation of Language and Language Learning.
I was co-chair of ICGI 2008, and of CoNLL 2008.
With Nick Chater, I organised a workshop on cognitive science and machine learning in London 21-22 June 2007. The website for that is here. The videos are online here.
Grammatical Inference With String Kernels: GISK Project Page.
Code for POS induction here from the paper at EACL 2003 on distributional and morphological learning.
Grammatical Inference with String Kernels (ECML 2006, ML 2010). Code is here and the datasets we used are here.
Look on my publications page for copies of any of these, or for abstracts on this page.
Some videos of research talks are available on the web.
My tutorial at the NIPS 2009 workshop: (slightly rushed!)
Learnable Representations for Natural Language
A talk at ICGI 2008:
A polynomial algorithm for the inference of context free languages
Department of Computer Science,Royal Holloway University of London
Surrey TW20 0EX
tel.: (44) 01784 443430
fax: (44) 01784 439786
Last modified January 2013.