Computational Semantics and Pragmatics
Semantics and pragmatics are concerned with the study of natural
language meaning and its context of use.
The computational counterparts of these disciplines
address these issues from an explicitly empirical and computational point of view,
combining insights from linguistic theory, psycholinguistics, computational linguistics,
and artificial intelligence. The overall objective of the course is to
give a broad overview of the major topics and methodologies in computational semantics and pragmatics, exposing students to classic seminal work and current research in the field.
- Lecturer: Raquel
Fernández (ILLC, University of Amsterdam)
-
Timetable: Block 2, Wednesdays and Fridays 11-13h. Room G3.13, Science Park 904.
-
Prerequisites:
There aren't any strong prerequisites, but I'll assume some basic knowledge of semantics or pragmatics of natural language. Programming skills may be useful but are not required.
Please fill in this
intake questionnaire if you intend to follow the course.
-
Evaluation: The grade for the course will be based on regular homework exercises and participation in class.
-
Possible projects:
I'm willing to discuss the possibility of supervising small research projects as a follow up to the course.
Week 1
Week 2
- 07 Nov: First-order logic for natural language semantics in Haskell
[slides by Jan van Eijck and Christina Unger; for reference only - note that the code is different from that in CSFP]
[code: FSynF.hs, MCWPL.hs, Model.hs]
- 09 Nov: Guest lecture by Jan van Eijck: Haskell implementation of a Montague-style
fragment (sort of)
- Assignment:
You are encouraged to do this assignment in pairs,
where someone with programming experience teams up with someone with
little programming experience. Submit one set of answers per pair.
Week 3
- 14 Nov: Final session on computational formal semantics
[slides]
- 16 Nov: We started to look into lexical semantics
[slides]
- Assignment: read and prepare for discussion on Wednesday 21 the following two papers (it's a good idea to take notes while working on the papers, but there is no need to submit any report by Wednesday - be ready to discuss in class):
Week 4
- 21 Nov: Discussion of Lapata (2001) [note takers: Jetze and Ciyang] and Utt & Padó (2011) [note takers: Sanne and Philip].
- Discussion summaries will be available on Blackboard in the blog "Papers for Discussion"
- 23 Nov: Word sense disambiguation
[slides]
- Assignments:
- Homework #3, due on Monday 26 Nov, 10am. Submission via Blackboard.
- Alternative option for HW#3: do one of the exercises in Homework #3 plus exercises 3 from
chapter 6 of the
NLTK book. For those who choose this option, the homework is due on Wednesday 28, 10am. Submit via Blackboard (one PDF file and one Python file with comments).
- Readings: choose one of the following papers:
Week 5
- 28 Nov: Psychological theories of concepts and unsupervised WSD
[slides]
- 30 Nov: Distributional semantic models
[slides]
Background readings:
- Assignment: read a recent paper on distributional semantic models and prepare a presentation. This assignment can be done in pairs.
Week 6
- 5 Dec: Introduction to DSMs continued
[slides], plus two paper presentations
- 7 Dec: More on Compositionality with DSMs
Week 7
- 12 Dec: DSMs and word senses
- 14 Dec: Multi-modal DSMs
-
Assignment:
Tentative list of topics
Here is a tentative list of topics for the course. Under each topic, you will find some representative references. Note that this is only to give you an idea of what each topic is about. As the course progresses, you'll be given a more precise list of required and recommended readings for each theme.
- Montague-style compositional semantics with functional programming
- Textual entailment
- Lexical semantics
- Pragmatic inference and abductive reasoning
- Speech acts and dialogue modelling
- Generation of referring expressions