CS 674
Natural Language Processing
Spring 2007
MW 11:15-12:05
Thurston 202
Mats Rooth
mr249@cornell.edu
Morrill 203A
Notices and description
There is a page with
notices.
The course is a graduate-level introduction to NLP, covering
computational morphology, syntactic formalisms and parsing, markup and
evaluation methodology, statistical models, and laboratory methods.
(Students who are uncertain whether they have enough computational and
mathematical preparation should consult me; special arrangements can
be made. Introduction to Computational Linguistics, which is offered
in Fall 2007, is more suitable for students with little computational
preparation.)
Textbooks
- Speech and Language Processing: an Introduction to Speech
Recognition, Computational Linguistics, and Natural Language
Processing. Daniel Jurafsky and James H. Martin.
- Finite State Morphology. Kenneth R. Beesley and Lauri Karttunen.
CSLI Publications.
- Foundations of Statistical Natural Language Processing.
Christopher Manning and Hinrich Schutze. MIT Press.
The 2nd edition of Jurafsky and Martin is not available yet.
We will use preprints of chapters from the new edition
together with chapters from the 1st edition
(
draft chapters ).
Course requirements
- Short assignments and written critiques of readings (20%)
- Oral presentations of readings and research papers
You will do one each. Topics may connect with your
project. (20%)
- Final project (50%)
- Class and laboratory participation (10%)
Topics
- Morphology and finite state transducers
- Tree syntax of natural language;
Context free and other tree grammars
- Parsing tree syntax
- N-grams
- Word classes and tagging
- Probabilistic tree models
- Computational morphology/phonology
- Sequence models
- Features and unification
- Semantics
- Computational lexical semantics
Organization
Assignments and some materials will be posted and
submitted through CMS.
Academic Integrity
Your conduct in the course is governed by CS
and Cornell policies on academic integrity.
These particular policies apply:
- Except for lab problems, assignments must be your own work. This does not exclude
discussion of points of interpretation or general strategy.
- Work on lab problems may be joint; submit your own
result and credit the people you worked with.
- Any resources (e.g. publications and NLP toolkits) used in
assignments, projects, or presentations must be credited.