TDT 55 Kunnskapsintensiv CBR (Knowledge-Intensive CBR)

Fall semester 2011

Theme organiser

Professor Agnar Aamodt

Theme content

A selected set of papers related to case-based reasoning in some way combined with general domain knowledge and other method components will be discussed.
The actual focus will to some extent depend on the project and interests of the students taking the course.

The course will be run as a set of seminar meetings. The first meeting will be on Thursday September 8th, 14:15 - 15:00, in Meeting Room 354, IT West.
In this start-up meeting a general overview of the course will be given, and the papers of the course will be presented - some are fixed and others need to be selected by the course group.

It is assumed that the students have a background in CBR corresponding to the CBR part of the ML+CBR course (TDT4173).
This means that all participants should - at least - make themselves familiar with the following set of papers before the course starts:

A. Aamodt and E. Plaza, 1994: Case-based reasoning; Foundational issues, methodological variations, and system approaches. AI Communications, 7(1), pgs. 39-59.

D. Aha, 1991: Case-based learning algorithms. Proceedings of the 1991 DARPA Case-Based Reasoning Workshop. Morgan Kaufmann.

B. Porter, R. Bareiss, R. Holte: Concept learning and heuristic classification in weak theory domains. Artificial Intelligence. Vol. 45, no.1-2, September 1990, pp. 229-263.

A. Aamodt: Knowledge-intensive case-based reasoning in Creek. ECCBR 2004. LNAI 3155, Spinger, 2004. pgs. 1-16.

Course structure

We will have five meetings to discuss the set of course papers. The papers will constitute the course "pensum". The students are assumed to work with the topics of the respective papers in between the meetings, and to browse the Internet or use other sources to fill in missing details. The students are also encouraged to arrange group meetings among themselves between the scheduled meetings. Each paper will be presented for discussion by two students, who summarize it, point out particular scientific issues of interest, research evaluation method applied (if any), and strong and weak points of the papers. All students have to read the papers addressed before each meeting and prepare for the discussions.

List of papers, and seminar plan:

Seminar 1, Overview - 08/9, 14:15 - 15:00

General overview. Structure and contents of the theme.

Please read, or refresh your knowledge of, the above four papers.

Seminar 2, CBR foundations - Mon 26/9, 10:15 - 12:00

• Janet Kolodner: Case-Based Reasoning (1993). Chapters 1, 2, 3, 4.
  Presenters:
  Ch1: Jan
  Slides Ch1
  Ch2: Jan B
  Slides Ch2
  Ch3: Ole
  Slides Ch3
  Ch4: Eivind
  Slides Ch4.

Seminar 3, CBR and games - Mon 10/10, 10:15 - 12:00

Poker
Presenter: Stian
Candidate paper(s):

• Jonathan Rubin, Ian Watson: Similarity-Based Retrieval and Solution Re-use Policies in the Game of Texas Hold'em.
  Proc. ICCBR 2010, pp. 465-479. Springer.

Starcraft
Presenter: Sigurd
Candidate paper(s):

• Ben Weber and Santiago Ontanon: Using Automated Replay Annotation for Case-Based Planning in Games.
  Proc. ICCBR 2010, Workshop on CBR in Computer GAmes, pp. 15-24.

Seminar 4, Other applications - Tor 27/10, 14:15 - 16:00

Trafikk
Presenter: Stian V
Candidate paper(s):

• Susan Fox David B. Leake: Modeling Case-based Planning for Repairing Reasoning Failures.
  AAAI Technical Report SS-95-05, AAAI. 1995.
Slides

Finance
Presenter: Sveinung
Candidate paper(s):

• Pei-Chann Chang et.al.: A neural network with a case based dynamic window for stock trading prediction.
  Expert Systems with Applications , 36 (2009,) pp. 6889–6898.

Oil and Gas
Presenter: Fredrick
Candidate paper(s):

• Agnar Aamodt: Case-based reasoning for advice-giving in a data-intensive environmen.
  Proc. SCAI 2008. Frontiers in Artificial Intelligence and Applications, , Volum 173. pp. 201-205. IOS Press, 2008.
  
Slides

Seminar 5, Explanations and ontologies - Tor 10/11, 14:15 - 16:00

Abduction
Presenter: Magnus
Candidate paper(s):

• David B. Leake. : Abduction, Experience, and Goals: A Model of Everyday Abductive Explanation.
  The Journal of Experimental and Theoretical Artificial Intelligence, 7:407-428, 1995. 25 pages. Springer LNAI 2689.
  
Slides

jColibri
Presenter: Oyvind
Candidate paper(s):

• Juan A. Recio-García et.al. : Ontology based CBR with jCOLIBRI.
  Applications and Innovations in Intelligent Systems,Applications and Innovations in Intelligent Systems, XIV, 2006. Springer, pp. 149-162.
  
Slides

A good general Web-link for all types of AI issues is AAAI's AI Topics pages. It has a lot of subpages, also related to our theme. Take a look at the Case-Based Reasoning page, for example. You find it under Machine Learning.

NTNU-IDI, September 2011.

Agnar Aamodt.

---