Foundational
Course: Probabilistic Models for NLP
Lecture 1:
Motivation for Probablistic Models for NLP and Probability Theory
|
Lecture 2:
Statistics, Maximum-Likelihood, Learning and next-word prediction
- Lecture
Slides (for lectures 1-3)
- Chapter 6 from Jurafsky and Martin
and
- Sections 6.1-6.3 + 9.1 from
Manning and Schutze boek
- Chapter 7 from Tom Mitchell
(Machine Learning) about Bayesian Learning: recommended
|
Lecture 3:
POStagging Using Markov Models, Hidden Markov Models
- Lecture
Slides (for lectures 1-3)
- Read
chapter 8 (Jurafsky and Martin) about POS tagging in general (you
may skip section 8.6)
- On
HMMs: read from chapter 9 (Manning and Schutze) sections 9.1+9.2
+9.3.1+9.3.2
-
Further on evaluation of Taggers: read section 10.6
(Manning and Schutze)
|
Lecture 4:
Probabilistic Modeling of Context-Free Grammars
- Read
the EM
tutorial's section 5, sub-section "Background: Probabilistic Modeling
of CFGs"
- Chapter
9 of the book of Jurafsky and Martin recommended
|
Lecture 5:
Resolving Ambiguities with Probabilistic Context-Free Grammars
- Read
the EM
tutorial's section 5, sub-section "Background: Resolving Ambiguities
with Probabilistic CFGs"
- Chapter
12 of the book of Jurafsky and Martin
recommended
-
Chapters 5 and 8 of Charniak's book recommended
|
|
Advanced
Course: Probabilistic Parsing
Lecture 1: General
Estimation Methods and a CFG's Probability Model
- Read the EM
tutorial's section 2 "Estimation Methods" (without sub-section
"Maximum-Entropy Estimation")
- Read the EM
tutorial's section 5, sub-section "Maximum-Likelihood Estimation
of PCFGs" (the first two theorems only)
|
Lecture 2: Maximum-Likelihood
Estimation of PCFGs and an Excursus to Maximum-Entropy Modeling
- Read the EM
tutorial's section 5, sub-section "Maximum-Likelihood Estimation
of PCFGs" (the central theorem only: treebank grammars are maximum-likelihhod
estimates)
- Read the EM
tutorial's section 2, sub-section "Maximum-Entropy Estimation"
|
Lecture 3: EM Training of PCFGs
- Read
the EM
tutorial's section 3 "The Expectation-Maximization Algorithm"
- Read the EM
tutorial's section 5, sub-section "EM Training of PCFGs"
-
Chapters 5 and 8 of Charniak's book recommended
|
Lecture 4:
Transforms on Phrase-Structure Treebank for More Accurate PCFG Parsing
- Lecture
Slides (a brief coverage of these models with a unifying view)
- Michael Collins
(1997), Proceedings of ACL'97
- Eugene
Charniak(1996). Statistical parsing with a context-free grammar and word
statistics, Proceedings of the Fourteenth National Conference on
AAAI Press/MIT Press, Menlo Park (1997). An abstract
and postscript version
- Eugene Charniak(1996). Tree-bank grammars,
Technical Report CS-96-02, Department of Computer Science, Brown University
(1996). An abstract
and postscript
version
- Eisner, Jason
M. 1996. Three new probabilistic
models for dependency parsing: An exploration. Proceedings of COLING-96,
Copenhagen, August.
- M. Johnson (1999) ``PCFG models of linguistic
tree representations'' Computational Linguistics,
available in Gzipped Postscript
format or Adobe PDF format
|
Lecture 5:
Data-Oriented Parsing
|
|
