1 / 15

Research Aamodt & Plaza, 1994

Research Aamodt & Plaza, 1994. Robin Burke CSC 594 4/7/2004. Outline. Context Authors Venue Purpose Content foundational issues. Authors. Agnar Aamodt Norway Enric Plaza Spain. Venue. AI Communications Main European Journal for Artificial Intelligence. Purpose. Overview of CBR

Télécharger la présentation

Research Aamodt & Plaza, 1994

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. ResearchAamodt & Plaza, 1994 Robin Burke CSC 594 4/7/2004

  2. Outline • Context • Authors • Venue • Purpose • Content • foundational issues

  3. Authors • Agnar Aamodt • Norway • Enric Plaza • Spain

  4. Venue • AI Communications • Main European Journal for Artificial Intelligence

  5. Purpose • Overview of CBR • Methodology • European results • Three goals • foundational issues • methodological variations • system approaches

  6. Foundational Issues • Characterizing CBR • Case representation • Issues related to each step of CBR cycle • knowledge-weak • knowledge-intensive

  7. Characterizing CBR • 4 Rs • Retrieve • Reuse • Revise • Retain

  8. Decomposition • Retrieve • identify features • search • match • select • Reuse • copy • adapt • Revise • evaluate solution • repair • Retain • integrate • index • extract

  9. Memory organization • Dynamic memory model • cognitively-based • generalized episodes • cases discriminated by feature values • Category / exemplar model • also cognitively-based • categories defined by exemplars • prototypical examples

  10. Retrieval • feature extraction • simple – use features present • complex – infer (deep) features • matching • simple – find nearby cases • complex – reason about similarity

  11. Reuse • Transformational reuse • adapt the case • emphasis on experience with rules as a guide • Derivational reuse • adapt the problem-solving process • emphasis on rule-based problem-solving with experience as a guide

  12. Revise • Evaluation • was the proposed solution successful? • simple – user critiques • complex – system generates and processes feedback • Repair • another adaptation step guided by evaluation • simple – discard failures • complex – explain failure and adjust accordingly

  13. Retain • Extract • package problem-solving episode as a case • simple – save features of problem situation and solution • complex – save explanation of why/how the solution solved the problem • Index • decide how to label the case • simple – all input features • complex – use diagnostic / predictive features • Integrate • put the case in memory • simple – just store it • complex – adjust indexing mechanism and/or background knowledge

  14. Bottom Line • Review of the state-of-the-art in 1994 • Indicates some of the design tradeoffs still important in case-based systems • Knowledge management field did not exist • discussion couched in AI terms (planning, problem-solving) • both fielded examples we would now consider as KM

  15. KM Implications • Should an organizations just store stuff? • value of identifying and organizing cases • How should retrieval work? • syntactic: keywords • semantic: reasoning about the domain • What else is needed besides cases?

More Related