Seminar on Word Sense Disambiguation
The WSD seminar will meet on Mondays and Wednesdays between 1:30 and
3:18 in Jennings 140 during the Winter quarter. Participants are
expected to take active part in presenting WSD papers as well as
producing a programming project within the probabilistic
WSD framework, with a view to a
(potentially class) publication being submitted to the
upcoming NAACL-HLT
Workshop on Semantic Evaluations: Recent Achievements and Future
Directions (submission deadline 8 March 2009).
Read Chapter 7, Word Sense Disambiguation, of Foundations of
Statistical Natural Language Processing By Christopher D. Manning,
Hinrich Schuetze (available
through Google
books online).
Goals for this week are:
- Understand the WSD problem in general
- Understand the problem in relation to the lexicon
- Distinction between supervised and unsupervised approaches
- Evaluation of WSD
- Selecting presentations for following weeks
Assignment Use WordNet to produce a most frequent sense
assignment for the Senseval-3 data provided (in the format required by
the official scorer) and evaluate the resulting algorithm. Programming
language of your choice, program and result to
be emailed to me by
Saturday 17 January at the latest — though a simple task, it
will take time to carry out various text processing. You can select
which WordNet you use, and whether you attempt the lexical sample or
the all words task. Please stick to English!