We are pleased to announce our next CUNY-NLP seminar with an invited talk
by Dr. Sam Brody from Columbia University.

Time: 2pm-3pm, Friday, October 9
Place: Room 4422, CUNY Graduate Center, 365 Fifth Ave (34str&35str).
Speaker: Dr. Sam Brody (Columbia)

Title: Unsupervised Methods for Word Sense Disambiguation

For many NLP tasks, supervised machine learning has become the
methodology of choice. Supervised methods offer the advantages of a
wide range of pre-existing tools and high accuracy when sufficient
amounts of annotated data are available.

However, the supervised approach has many disadvantages. From a
pragmatic perspective, its requirement of manually annotated data is
extremely restrictive and does not allow for easy transfer across
domains or for (even minor) modifications of the task definition. From
an academic perspective, it tends to limit the directions of research to
areas and theories where annotated data is available (e.g. reliance on
WordNet, Penn. Treebank, etc.).

Such is the case with the problem of automatic Word Sense Disambiguation
(WSD). In this talk, I will discuss the issue of supervised vs.
unsupervised learning in NLP, using Word Sense Disambiguation (WSD) as a
case study. I will explore the main reasons for the performance
advantage of supervised WSD, and present unsupervised methods that
directly address these issues and manage to reduce the performance gap.

About the Speaker:

Sam Brody is currently a post-doctoral research scientist at the
Department of Biomedical Informatics at the Columbia University Medical
Center. His interests include natural language processing, unsupervised
machine learning, information extraction, and relation detection. Dr.
Brody earned his Ph.D. in Informatics from the University of Edinburgh,
on the topic of "Unsupervised Word Sense Disambiguation." He holds a
B.Sc. and M.Sc. in Computer Science from Hebrew University, Jerusalem.