Next meeting I will present a paper from Gildea and Jurafsky about Automatic Labeling of Semantic Roles. Their system, a Shallow Semantic Interpreter based on statistical classifiers, labels the constituents of an input sentence with semantic roles. The system is trained on roughly 50,000 hand-annotated sentences from the FrameNet corpus.

In the presentation I will focus more particularly on the set of various lexical and syntactic features, which are extracted from the training corpus and used to compute probabilities.

Note: The paper is available at http://www.cs.rochester.edu/~gildea/gildea-cl02.pdf

Reference:
Gildea & Jurafsky, 2002, Automatic Labeling of Semantic Roles, Computational Linguistics, Vol. 28, No 3, pp.245--288.

Please send comments or queries about this web site to jpprost@ics.mq.edu.au