(Course code CTH: TDA 510 (4.0p), GU: INN 770 (5p))
5/5 Updated version of Exercise 5. Lectures on speech technology and statistics changed order (so that those who want to have speech in their lab2 have more time to do it).
26/4 Lecture and exercise on Thursday 4 May cancelled. Instead, there is a possibility to participate in the Dialogverkstad at Lindholmen. We can take the ferry ("Älvsnabben") from Rosenlund at 9.40.
26/4 An up-to-date version of Lab2 PM.
18/4 Lab1 PM now contains an example of all the phases of the work, if done by using a GF resource grammar (the last section in the PM).
11/4 Lab 1 reporting system now available at Fire. Please use this for reporting your lab!
29/3 Lectures of next week swapped: 3/4 is on translation systems and 6/4 on parsing algorithms.
23/3 New versions available of extract (1.0beta) and GF (2.5). Download or use the ones in lanctec06/bin/ (linux binaries; this gf is a late build of 2.4, with no essential difference from 2.5).
13/3 Added link to Lab times and updated the Lab 1 description.
26/2/2006 The first version of the course page. Copied from 2005. Most things need to be updated. The course starts on Thursday 16 March at 10.00 in ES52.
Applications on which we are going to work include:
Official description in Swedish: Kursplan.
An alternative is Jurafsky, D., Martin, J.M., Speech and Language Processing, Prentice Hall, 2000; paperback ("international edition") 2003. Price £37.99 at Amazon.co.uk. Ask Cremona as well.
Other material on this web page.
Lab room: 6225A on both Tue 10-12 and Wed 10-12, beginning on Tue 21 March. Here is a list of lab times.
Supervision times: by individual agreement.
Here is a preliminary lecture schema, also containing links to the lecture slides.
See the preliminary lecture and exercise schema for next week's exercises.
Students are expected to do the exercises in advance, and they can earn extra points in this way. To earn extra points, you mark on a list (circulated in the beginning of each exercise class) which exercises you have done. You may be picked to present your solution on the blackboard: the point is to make the exercise classes lively and interactive. The amount of credits one can earn in this way is 15% of the final mark (which is slightly less than the difference between 3 and 4, and 4 and 5).
Read: course slides, GF Tutorial. Questions will be problem-oriented and similar to the exercises.
There will be 0-2 questions on each of the following topics: morphology, syntax, semantics, translation, parsing algorithms, statistical methods, speech. Altogether 6 questions, so that CTH students must answer to 4 and GU students to 5.
The course is graded on the usual scales (3-4-5, G-VG). The grade is jointly determined by the exercises (the number of solved ones), and the written exam.
A renewal (omtenta) will be arranged upon individual agreement, probably as an oral exam.
Here is a copy of the Exam of 2004.
Chalmers Language Technology Group
Computational Linguistics at Gothenburg University. This course will provide prerequisites to some of their advanced courses.
The Systran Machine Translation System. You can test in on line.