CSCI/LING 8570
Natural Language Processing Techniques
Offered every spring.
Information for Spring 2009:
Scheduled meeting time:
2:00-3:15 Tuesday/Thursday and 2:30-3:20 Wednesday.
(The Wednesday session is a joint meeting with the CASPR research group.)
Scheduled to meet in Room 101, Building 1031 (Hardman Hall),
but we will actually meet in the Institute for Artificial Intelligence
(Conference room 110; enter through Room 111, Graduate Studies Research Center).
Instructor: Dr. Michael A. Covington, mc@uga.edu
Course web site: http://www.ai.uga.edu/mc/8570.html
Content: Basic principles of human language; structure of English from a computational point of view; algorithms and techniques for computer understanding of human language.
This is a "hands-on" implementation course for people who already know how to program in Prolog. If you do not know Prolog, it is absolutely impossible to take this course and you should consider taking LING 6570 instead.
Course goals: This will be more like a research seminar than in past years. We will concentrate on producing student research projects of high quality, which can be either reusable software tools or research of other kinds. See http://www.ai.uga.edu/mc/pronto for past work of this type.
Use of other programming languages for sufficiently sophisticated projects will also be permitted. But you must know Prolog coming in and do the Prolog-based homework.
We will take the usual topics in a different order than in the textbook. We will do tokenization and morphology first, then chunk parsing, then other types of parsing, and finally semantics.
Prerequisites:
- CSCI/ARTI 4540/6540 (Symbolic Programming) or sufficient knowledge of Prolog.
- ENGL/LING 8150 (formal syntax of English). This requirement is normally waived for AI students.
- Knowledge of English so that you can verify that your parsers, generative grammars, and speech synthesizer programming are correct. This involves distinguishing ungrammatical or mispronounced utterances from correct ones. You will be using your own knowledge of English as a source of data for analysis.
Required textbook:
- Covington, Natural Language Processing for Prolog Programmers
Prentice Hall, ISBN 0-13-629213-5
Stocked by University Bookstore.
In the University Bookstore, be sure to look in BOTH locations, CSCI 8570 and LING 8570, because it is often missing from one of them but not the other.
Amazon.com Barnes & Noble
Be sure to download and print out the corrections for this book:
http://www.ai.uga.edu/mc/NLPPP-Corrected-Pages.pdf
Optional reading (NOT ordered at UGA Bookstore):
Caution! If you order your books online, be sure to place the order at least 1 month before the course begins, because we will use the textbooks on the first day. Textbooks are also available at the University Bookstore next to the Tate Student Center on campus.
- Jurafsky and Martin, Speech and Language Processing
Prentice Hall, ISBN 0-13-095069-6
Amazon.com Barnes & Noble
- Nugues, An Introduction to Language Processing with Perl and Prolog
Springer, ISBN 9783540250319
Amazon.com Barnes & NobleRequirements:
- Attendance is expected.
- Homework will be assigned and discussed in class; all students are expected to do homework and be ready to present their answers.
- Mid-term and final examinations.
- Term project.
Academic honesty policy: The rules of the University apply to this course.
As a University of Georgia student, you have agreed to abide by the University’s academic honesty policy, "A Culture of Honesty," and the Student Honor Code. All academic work must meet the standards described in "A Culture of Honesty" found at www.uga.edu/honesty. Lack of knowledge of the academic honesty policy is not a reasonable explanation for a violation. Questions related to course assignments and the Academic Honesty policy should be directed to the instructor.
Note: All graded work is done individually; although projects are part of a larger effort, each project is done by a student working alone. Student projects in this course are published on the Web, so it is extremely important for them to be ready for the scrutiny of the entire scholarly world.