This position is part of a project entitled 'Learning to Reason by Exploiting Grounded Text Collections', headed by Ivan Titov. The project is supported by a Google focused research awardand the University of Amsterdam, and close collaboration with Google is anticipated. Initial funding is for three years, and further funding for a fourth year is expected.
The principle goal of the project is to model semantic content of a document specifically to draw inferences about the entities and their relations, explicit or implied, in the text (as needed for question answering or other applications requiring text comprehension). This will be accomplished by learning latent semantic representations, in part, relying on text collections with entities linked to existing knowledge bases such as Freebase, but also on other types of supervision. See recent related work on inducing semantic representations by Ivan Titov and thedescription of the natural language understanding program on the Google research blog.
The University of Amsterdam provides an excellent environment for research in these areas with world-class faculty in the fields of statistical NLP (incl. Rens Bod, Raquel Fernandez, Christof Monz, Khalil Sima’an, Willem Zuidema), machine learning (e.g., Max Welling) and text mining / IR (e.g., Maarten de Rijke).
Starting date: 1 October 2013 (earlier and later starting dates are negotiable).
The PhD candidate should have:
The applications should include:
Completed applications should be sent by 15 July 2013 to applicati...@uva.nl, stating your name and vacancy number in the subject field with a copy (CC) to Ivan Titov (titov at uva.nl). The committee does not guarantee that late or incomplete applications will be considered.
See the formal announcement for details. For further information about the vacancy please contact Ivan Titov (titov at uva.nl).