TDT4275 - Natural Language Interfaces (Naturlig språklig grensesnitt)
Teknologi
Natural Language Interfaces (Naturlig språklig grensesnitt)
Natural Language Interfaces (Naturlig språklig grensesnitt) given by Björn Gambäck.
Course overview
The subject comprises: What are words and languages? Grammars and syntactic analysis of natural language. Semantics and logical form. Statistical methods for language understanding. Dialogue analysis and conversational agents. Machine translation. Speech-based interfaces.
The course will be run as lectures and programming exercises. The first lecture will be Thursday January 12, at 12:15-13:00 in room KJL3 (Kjelhuset rom 252). The first meeting will be a general introduction to the subject.
Table of contents
Course Book
Examination
- Written account of the five lab assignments.
Send in the solution to a specific lab exercise no later than one month after the day the assignment was handed out (NB!).The assignments will be available in English only, unless somebody specifically requests to have them in Bokmål and/or Nynorsk at least one week before the publication date.
The answers may be given in either English, Bokmål or Nynorsk (other languages only if agreed on beforehand with the assistant). - A written exam Saturday 9.6, 09:00-12:00.
- The course grade will be assigned as follows:
exam: 55%, assignments: 45%.
Course Schedule 2012
Lectures
Lecture 1: Introduction and Overview
Introduction; Administrative details; Examination; Schedule for the course.
| Slides | To read | |||
|---|---|---|---|---|
|
pdf
| J&M Ch. 1
Gambäck: Human Language Technology: The Babel Fish |
Lecture 2: Text Summarisation
Guest Lecturer: Erwin Marsi
Text summarisation; extracts; abstracts; sentence compression.
Text fusion; reformulation; generalisation, specification.
Anaphora resolution.
| Slides | To read | Other relevant material | ||
|---|---|---|---|---|
| J&M Ch. 23.3-23.7 |
Lecture 3: BusTUC (with demo)
Guest Lecturer: Rune Sætre
An example of a successful NLP-application:
TUC (The Understanding Computer);
Natural language question answering.
| Slides | To read | Other relevant material | ||
|---|---|---|---|---|
|
| Amble.
"BusTUC - a natural language bus route oracle"
| BusTUC demo (ENG)
BussTUC demo (NOR) |
Lecture 4: NLP and NLP systems
The components of an NLP system; what do we need to build language processing systems?
| Slides | To read | |||
|---|---|---|---|---|
|
pdf
| J&M Ch. 1
Gambäck. Human Language Technology: The Babel Fish |
Lecture 5: Language and Linguistics
Language and languages; Relations between languages; Written and spoken languages; Human language and other types of communication; The structure of language.
| Slides | To read | Other relevant material | ||
|---|---|---|---|---|
|
pdf
| J&M Ch. 1
| Ethnologue:
Languages of the World Omniglot: Writing systems CIA Factbook: Languages and other facts about all countries Seyfarth et al.. 1980. "Monkey Responses to Three Different Alarm Calls", Science 211:801-803. Cheney & Seyfarth. 2005. "Constraints and preadaptations in the earliest stages of language evolution", Linguistic Review 22:135-159. |
Lecture 6: Phonology and Morphology
Phonemes; Syllables; Writing systems; Transliteration Morphology; Two-level morphology; Stemming; Morphological parsing
| Slides | To read | Other relevant material | ||
|---|---|---|---|---|
|
pdf
| J&M Ch. 3
J&M Ch. 7.1-7.3 | J&M links to NLP resources |
Lecture 7: Words, their neighbours, and POS Tagging
Words; Lexemes; Tokenization; N-grams; Sparse data; Part-of-speech; Open and closed classes; Tagsets; Methods for part-of-speech tagging; Unknown words.
| Slides | To read | Other relevant material | ||
|---|---|---|---|---|
|
pdf
| J&M Ch. 4.1-4.3
J&M Ch. 5.1-5-7 | J&M links to NLP resources |
Lecture 8: Basic Grammars
Context-free grammars; Definite Clause Grammars; Dependency Grammar.
| Slides | To read | Other relevant material | ||
|---|---|---|---|---|
|
pdf
|
Blackburn, Bos & Striegnitz, 2006-2011.
Learn Prolog Now!.
Saarbrücken, Germany.
Tapanainen & Järvinen, 1997. "A Non-Projective Dependency Parser", Proc. 5th Conf. on Applied NLP, Washington, DC. |
Lecture 9: Parsing
Parsing with context-free grammars; top-down parsing; bottom-up parsing;
Well-formed substring tables; head-first parsing; charts; LR parsing;
Human language processing.
| Slides | To read | Other relevant material | ||
|---|---|---|---|---|
|
pdf
| J&M Ch. 13
J&M Ch. 14.11 J&M Ch. 16.4 Kimball. "Seven Principles of Surface Structure Parsing in Natural Languages" | Kay. 1989. "Head-Driven Parsing",
Proc. 1st Int. Workshop on Parsing Technologies. Pittsburgh.
Aho & Ullman. 1972. Ch. 4.2: Tabular Parsing Methods. The Theory of Parsing, Translation, and Computation. Prentice-Hall. Tomita. 1986. Ch. 2: Informal Description of the Algorithm. Efficient Parsing for Natural Language. Kluwer Academic Press. |
Lecture 10: Semantics
Semantic models, representing meaning. Syntax-driven semantic analysis, quantifiers, compositionality.
| Slides | To read | Other relevant material | ||
|---|---|---|---|---|
|
pdf
| J&M Ch. 18.1-18.4 J&M Ch. 18.6 | Blackburn & Bos. 2005. Representation and Inference for Natural Language: A First Course in Computational Semantics. CSLI Publications, Stanford, California. |
Lecture 11: Discourse and Pragmatics
Discourse processing, reference resolution.
| Slides | To read | Other relevant material | ||
|---|---|---|---|---|
| pdf
| J&M Ch. 20.1-20.2
J&M Ch. 21.1-21.7 |
Lecture 12: Machine Translation
Machine Translation theory; Machine Translation systems, MT users, applications
| Slides | To read | Other relevant material | ||
|---|---|---|---|---|
| pdf
| J&M Ch. 25.1-25.2
| Systran commercial application
Various translation tools Free Internet MT services |
Lecture 13: Statistical Machine Translation
Guest Lecturer: Lars Bungum
Statistical Machine Translation theory and systems;
language modelling and decoding; MT system evaluation
| Slides | To read | Other relevant material | ||
|---|---|---|---|---|
| J&M Ch. 25.3-25.9 | Parallel texts - European Parliament |
Exam Information and Previous Exams
The exam questions will be given in English only, unless somebody specifically requests to have them in Bokmål and/or Nynorsk at least one week before the exam.
The answers may be given in either English, Bokmål or Nynorsk (other languages only if agreed on beforehand with the examiner).
- 7/8 2006: bokmål
- 19/12 2005: bokmål
- 13/8 2005: bokmål
- 6/12 2004: bokmål
- 11/12 2003: bokmål
- 14/12 2002: bokmål
- 2/8 2002: bokmål
- 11/5 2002: bokmål
- 15/5 2001: English Bokmål Nynorsk
- 14/1 2001: Bokmål
- 23/1 1993: Bokmål
- 21/8 1992: Bokmål
- 18/1 1992: Bokmål
For nærmere informasjon om emnet, kontakt faglærer Björn Gambäck.