Language Technology Seminar Series

Title: Application of Hidden Markov Models for Semantic Role Labelling

Speaker: Phil Blunsom

Location: ICT Building, L2.06

Date: Friday 18 June 2004

Time: 1-2pm

Abstract:

The Hidden Markov Model (HMM) is a popular statistical tool for modelling a wide range of time series data. In the context of NLP, HMMs have been applied with great success to problems such as part-of-speech tagging and noun phrase chunking. In this seminar I will introduce and define HMMs and describe efficient methods for addressing the three classic tasks of evaluation, decoding and learning. I will then describe the application of HMMs to the task of Semantic Role Labelling (SRL) and compare my models with those presented in this years Conference on Computational Natural Language Learning (CoNLL) Shared Task. SRL is concerned with identifying and labelling the relationship between a syntactic constituent and a verb predicate, typical semantic arguments are Agent, Patient etc. SRL is a key task for generalising Information Extraction systems beyond task specific and hand coded template filling methods.

Bio:

Phil Blunsom is a Master's student in the Language Technology Group in CSSE. His research is in the area of statistical models for semantic role annotation.

Disclaimer: This page, its contents and style, are the responsibility of the author and do not necessarily represent the views, policies or opinions of The University of Melbourne.