Empirical Methods in Natural Language Processing

This course is an introduction to data-driven methods applied to natural language processing. The emphasis is on methods, but we will survey applications such as syntactic parsing, text classification, information extraction, tagging, summarization. The final lectures will deal with statistical machine translation.

Lecturer: Philipp Koehn

TA: Tommy Herbert

Lectures: Monday and Thursday, 5:10pm, changed to: WRB room G.11

Tutorials: Tuesday and Friday, 1pm, AT 4.12
Tutorial group assignments.

Tutorials

Tutorial 1 was discussed on January 22 (Tuesday) and 25 (Friday).
Tutorial 2 will be discussed on February 1 (Friday) and 5 (Tuesday).
Tutorial 3 will be discussed on February 8 (Friday) and 12 (Tuesday).
The project baseline systems will be presented on February 22 (Friday) and 26 (Tuesday).
Tutorial 4 will be discussed on March 4 (Tuesday) and 7 (Friday).
Tutorial 5 will be discussed on March 11 (Tuesday) and 14 (Friday).

Assessment

A single assessment (worth 30%) of the course will be given out late January. You will have to turn in your paper and code at the end of March in class. If you have a problem accessing the data from the web site, it is also available at /home/miles/projects/ner/data-eng/ (English) and /home/miles/projects/ner/data-deu/ (German).

The rest of the marks (70%) will go on the exam. Past exam, solutions.

Syllabus

Exact dates will change and may move around. Topics may shift and change during flight.

No	Date	Topic	Slides	Reference
1	7 Jan	Introduction (I): Words and probability	display \| print	MS chapter 1 K chapter 3
2	10 Jan	Introduction (II): Estimation and information theory	display \| print	MS chapter 2 K chapter 3
3	14 Jan	Language modeling (I): From counts to smoothing	display \| print	MS chapter 6 JM chapter 6 K chapter 7
4	17 Jan	Language modeling (II): Smoothing and back-off	display \| print	MS chapter 6 JM chapter 6 K chapter 7
5	21 Jan	Tagging (I): Part-of-speech tagging with HMM	display \| print	MS chapter 9/10 JM chapter 8
6	25 Jan	Tagging (II): Transformation-Based Learning	display \| print	MS chapter 10
7	28 Jan	Tagging (III): Maximum Entropy Models	display \| print	Ratnaparkhi [1996] Berger et al. [1993]
8	31 Jan	Parsing (I): Context-free grammars and chart parsing	display \| print	JM chapter 9/12
9	4 Feb	Project	display \| print	-
10	7 Feb	Parsing (II): Lexicalised and probabilistic parsing	display \| print	JM chapter 12
11	11 Feb	Word sense disambiguation	display \| print	JM section 17.2, MS chapter 7 Yarowsky [1995]
12	14 Feb	Text categorization and clustering	display \| print	MS chapter 14/16
13	18 Feb	Semantics and discourse	display \| print	Carlson et al. [2001] Pang and Lee [2005]
14	21 Feb	Machine translation (I): Introduction	display \| print	-
15	25 Feb	Machine translation (II): Word-based models and the EM algorithm	display \| print	K chapter 4 Brown et al. [2003]
-	28 Feb	NO CLASS
16	3 Mar	Machine translation (III): Decoding	display \| print	K chapter 6 Koehn [2004]
17	6 Mar	Machine translation (IV): Phrase-based models	display \| print	K chapter 5 Koehn et al. [2003] Och and Ney [2002]
18	10 Mar	Machine translation (V): Syntax-based models	display \| print	K chapter 11 Yamada and Knight [2002] Chiang [2005] Collins et al. [2005]
19	13 Mar	Machine translation (VI): Advanced topics	display \| print	-
20	17 Mar	Review	-	-

MS refers to "Manning and Schütze", JM refers to "Jurafsky and Martin", K to "Koehn", the three textbooks listed below.

References

When possible, online papers will be made available. As for books, the key references are:

Foundations of Statistical Natural Language Processing. Christopher Manning and Hinrich Schütze. Available online.
Speech and Language Processing: An Introduction to Natural Language Processing, Computational Linguistics and Speech Recognition. Daniel Jurafsky and James H. Martin
Statistical Machine Translation. Philipp Koehn. Not yet published, chapter copies will be handed out.

Home : Teaching : Courses