Syntactic Analysis in Language Technology / Syntactic parsing
Note that this page has been migrated from a previous server. There is thus a risk that not all links work correctly
- Course date: Autumn 2016
- Teachers:
- Sara Stymne: course coordinator 5LN455, lectures
- Joakim Nivre: examiner, course coordinator 5LN713/5LN717
News
- 171010: The course starts with a first lecture on Tuesday November 8, 13-15 in Turing
- 161010: A preliminary web page is up
Contents
The course gives an introduction to methods and algorithms used in automatic syntactic analysis and provides the skills that are required to apply and implement practical systems based on these methods. The course covers both phrase structure analysis and dependency analysis.
Examination 5LN455 (bachelor)
The course examination is split into four assignments and two seminars. The following grading system will be used both for the complete course and for the assignments: Fail (Underkänd, U), Pass (Godkänd, G), Pass with Distinction (Väl Godkänd, VG). The grade for the course will be G if you get at least G on all assignments and the seminars, VG if you also get the grade VG on at least two of the assignments, of which at least one should be a practical assignment (assignment 2 or 4).
Assignments
Assignment 1. Hand-in assignment about phrase structure parsing. Detailed description. Deadline: 2016-12-06.
Assignment 2. Implementation of the central component of a phrase structure parser. Detailed description. Deadline: 2016-12-06.
Assignment 3. Hand-in assignment about dependency parsing. Detailed description. Deadline: 2017-01-13.
Assignment 4. Practical parsing with MaltParser. Detailed description. Deadline: 2017-01-13.
All assignments should be handed in via Studentportalen.
Examination 5LN713/5LN717 (master)
For 5 credits, the course is examined by means of four assignments:- PCFG parsing: Implement the CKY parsing algorithm and evaluate the parser using treebank data. Detailed description Deadline: 2016-12-06.
- Dependency parsing: Implement a transition-based parsing algorithm and evaluate the parser using treebank data. Detailed description Deadline: 2017-01-13.
- Literature seminar: Actively participate in two literature seminars.
- Literature review: Summarize, analyze and critically review two scientific articles on syntactic parsing in a written report of 3-5 pages. Deadline: 2017-01-13.
For 7.5 credits, students must in addition do a project chosen together with the teacher.
In order to pass the course, a student must pass all assignments (and do the project for extra credits). In order to pass the course with distinction (Väl godkänt), a student must pass at least two assignments (or one assignment and the project) with distinction.
See also this page, for more information about the master courses.
Deadlines
Each assignment has a deadline, which will be announced well in advance. In case you fail to meet the deadline, you have the chance to re-submit the assignment for a second deadline. In case you fail to meet even the second deadline, the corresponding assignment will be graded as Fail (U). (See Språkteknologiprogrammens Policy för examination [in Swedish]).
- Ordinary deadlines
- Assignment 1+2: 16-12-06
- Assignment 3+4: 17-01-13
- Resubmission deadline
- All assignments: 17-02-03
- (If you have failed to hand in assignment 1+2 at the first deadline, they can also be handed in at the second deadline: 17-01-13)
Assignments will only be graded in connection with each deadline. If you fail to meet these deadlines, you will have to retake the course the next time it is given. In case of special circumstances, please contact your teacher before the deadline it concerns.
Seminars
There are two literature seminars during the course, which are common for bachelor and master students. In order to get a passing grade on the seminars you need to prepare for them, and be active during the seminar.The seminars will be held in smaller groups, which will be announced later.
The two seminars are obligatory. If you miss a seminar, or do not participate actively, you will have to do a complementary task. Contact Sara if this is the case.
Literature
Daniel Jurafsky and James H. Martin. Speech and Language Processing: An Introduction to Natural Language Processing, Computational Linguistics, and Speech Recognition. Second Edition, Pearson Education, 2009. (Main course book.)
Sandra Kübler, Ryan McDonald, and Joakim Nivre. Dependency Parsing. Morgan and Claypool, 2009. (Electronic edition)
Two research articles on parsing to be discussed during the literature seminars:
- Seminar 1: Mark Johnson. PCFG Models of Linguistic Tree Representations. Computational Linguistics 24(4). Pages 613-632.
- Seminar 2: Joakim Nivre and Jens Nilsson. Pseudo-Projective Dependency Parsing. Proceedings of the 43rd Annual Meeting of the Association for Computational Linguistics (ACL'05). Pages 99-106. Ann Arbor, USA.
Additional reading is required for the master student courses. The following is a starting point:
- Yoav Goldberg and Joakim Nivre. 2013. Training Deterministic Parsers with Non-Deterministic Oracles. Transactions of the ACL 1(Oct):403-414.
- Terry Koo and Michael Collins. Efficient Third-Order Dependency Parsers. Proceedings of the 48th Annual Meeting of the Association for Computational Linguistics (ACL). 2010. Pages 1-11.
- Yue Zhang and Stephen Clark. 2008. A Tale of Two Parsers: Investigating and Combining Graph-based and Transition-based Dependency Parsing. In Proceedings of the 2008 Conference on Empirical Methods in Natural Language Processing, 562-571.
Schedule
All lectures and seminars take place in room Turing (9-2042), unless otherwise noted below
Tuesday 2016-11-08 (13–15) Sara Stymne
Introduction
Reading: J&M 12.1–12.7.Thursday 2016-11-10 (13–15) Sara Stymne
The CKY algorithm part 1: Recognition
Reading: J&M 13.1–13.3, 13.4.1.Monday 2016-11-14 (13–15) Sara Stymne
The CKY algorithm part 2: Probabilistic parsing
Reading: J&M 14.1–14.2.Tuesday 2016-11-15 (13–15) Sara Stymne
Treebank grammars and parser evaluation
Reading: J&M 12.4 (again), 14.3, 14.7.Tuesday 2016-11-22 (13–15) Sara Stymne
The Earley algorithm
Reading: J&M 13.4.2.Tuesday 2016-11-29 (Group A: 13-14, group B: 14-15) Sara Stymne
Room: 4-2007
Literature seminar 1
Reading: Mark Johnson. PCFG Models of Linguistic Tree Representations and J&M 14.9Tuesday 2016-12-06
Deadline for Assignments 1 and 2Thursday 2016-12-01 (13–15) Sara Stymne
Room: 7-1013
Arc-factored dependency parsing
Reading: KMN 1-2, 4.1, 4.3.Tuesday 2016-12-06 (13–15) Sara Stymne
Collins’ and Eisner’s algorithms
Reading: KMN 4.2 (not 4.2.2).Tuesday 2016-12-13 (10–12) Sara Stymne
Transition-based dependency parsing
Reading: KMN 2.1.2, 3.1–3.3, 6.1Thursday 2017-01-12 (group D: 10-11, group C: 13-14) Sara Stymne
Literature seminar 2
Room: 9-2029
Reading: Joakim Nivre and Jens Nilsson. Pseudo-Projective Dependency Parsing
Friday 2017-01-13
Deadline for Assignments 3 and 4