Natural Language Generation

This course provides an introduction to the theory and practice of computational approaches to natural language generation. The course covers common approaches to content selection and organization, sentence planning, and realisation. It includes both symbolic approaches to generation, as well as more recent statistical and trainable techniques. It also aims to provide an understanding of evaluation methods used in this field, an understanding of key aspects of human language production, exposure to techniques and tools used to develop practical systems that can communicate with users, and insight into open research problems in applications of natural language generation, e.g., summarization, paraphrase, dialogue, multimodal discourse.

Course descriptors are here:

No tutorials are planned for this course, but there will be occasional lab sessions.

LECTURER:

Prof. Johanna D. Moore

LECTURE TIMES:

4:10-5:00 Tuesday and Friday: Lecture Theatre 3, 7 Bristo Square [map]

ASSESSMENT:

70% of the mark is allocated to the final examination.
30% of the mark is allocated to the programming assignments.

ASSIGNMENTS:

Assignment 1: Handout Fri 8 Feb. Due Thurs 28 Feb 16:00 (40% of overall coursework grade)
Assignment 2: Handout Fri 1 March. Due Thurs 28 Mar 16:00 (60% of overall coursework grade)

READINGS:

Primary textbook:

Ehud Reiter and Robert Dale, Building Natural Language Generation Systems, Cambridge University Press, 2006.

Suggested readings:

Week 1: Reiter & Dale Chap. 1
Week 2: Mark Steedman, A Very Short Introduction to Combinatory Categorial Grammar (CCG)
Week 3: Michael White. 2006. Efficient Realization of Coordinate Structures in Combinatory Categorial Grammar.Research on Language and Computation, 4(1):39-75.
Week 4: Knott, A., Oberlander, J., O'Donnell, M. and Mellish, C. (2001) Beyond elaboration: the interaction of relations and focus in coherent text. In T. Sanders, J. Schilperoord and W. Spooren (eds.) Text representation: linguistic and psycholinguistic aspects, pp181-196. Amsterdam: Benjamins. Reiter and Dale: Sections 4.1-4.3.
Week 5: Reiter and Dale: Sections 4.3-4.4.
Kathleen R. McKeown, Discourse Strategies for Generating Natural-Language Text, Artificial Intelligence 27: 1-41, 1985.
Week 6: Marilyn Walker, S. Whittaker, A. Stent, P. Maloor, J. Moore, M. Johnston, G. Vasireddy. Generation and Evaluation of User Tailored Responses in Multimodal Dialogue, Cognitive Science, 28: 811-840, 2004.
Week 7: R. Dale and E. Reiter, Computational Interpretations of the Gricean Maxims in the Generation of Referring Expressions Cognitive Science 19:233-263, 1995. Reiter and Dale: Chapter 5.
Week 8: Marilyn Walker, Amanda Stent, Francois Mairesse, Rashmi Prasad Individual and Domain Adaptation in Sentence Planning for Dialogue. Journal of Artificial Intelligence Research, 30: 413-456, 2007.
Week 9: E. Reiter and A. Belz. An Investigation into the Validity of Some Metrics for Automatically Evaluating Natural Language Generation Systems Computational Linguistics 35: 529-558, 2009.

Other useful sources of information and reading:

Mark McConville. Beginner's Guide to OpenCCG
Mark Steedman and Jason Baldridge. Combinatory Categorial Grammar. In Robert Borsley and Kersti Borjars (eds.) Constraint-based approaches to grammar: alternatives to transformational syntax. Oxford: Blackwell.
Martin Kay, Chart Generation, Proceedings of the Thirty-Fourth Annual Meeting of the Association for Computational Linguistics, 200--204, 1996.
J. D. Moore and M. E. Pollack, A Problem for RST: The Need for Multi-Level Discourse Analysis , Computational Linguistics , 18(4):537-544, 1992.
Regina Barzilay and Mirella Lapata, Modeling Local Coherence: An Entity-Based Approach, Proceedings of the 43rd Annual Meeting of the Association for Computational Linguistics, 141-148, 2005.

LECTURE TOPICS

15 Jan: No Class.
18 Jan: Introduction to NLG (Moore)
22 Jan: Hybrid Logic Dependency Semantics (Moore)
25 Jan: Introduction to CCG (Moore)
29 Jan: Lab Appleton Tower 5.05
1 Feb: Sentence realisation with OpenCCG (Moore)
5 Feb: Human language production (Moore)
8 Feb: Discourse Coherence (Moore)
12 Feb: Discourse Coherence Cont'd (Moore)
15 Feb: Content Planning 1 (Moore)
19 Feb: Innovative Learning Week. No Class.
22 Feb: Innovative Learning Week. No Class.
26 Feb: Content Planning 1 Cont'd
1 Mar: No Class
5 Mar: Content Planning 2 (Moore)
8 Mar: Sentence Planning 1: Lexical Choice (Moore)
12 Mar: Sentence Planning 2: Aggregation (Moore)
15 Mar: Sentence Planning 3: Referring Expression Generation (Moore)
19 Mar: Statistical Generation 1: Overgeneration and Ranking (Moore)
22 Mar: Statistical Generation 2: Trainable Sentence Planning (Moore)
26 Mar: Statistical Generation 3 (Moore)
29 Mar: Evaluation (Moore)

>This page is maintained by Johanna Moore

Home : Teaching : Courses