The hedgehog and the fox. An argumentation-based decision support system ∗ . Maxime Morge † and Paolo Mancarella † † Dipartimento di Informatica Università di Pisa MFI, May 2007 * is supported by the Sixth Framework IST programme of the EC, under the 035200 ARGUGRID project. Maxime Morge (ARGUGRID) MFI May 2007 1 / 16
20
Embed
The hedgehog and the fox. An argumentation-based …groups.di.unipi.it › ~morge › publis › morge07mfimargo_show.pdfThe hedgehog and the fox. An argumentation-based decision support
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
The hedgehog and the fox.
An argumentation-based decision support system∗.
Maxime Morge† and Paolo Mancarella†
†Dipartimento di InformaticaUniversità di Pisa
MFI, May 2007
∗is supported by the Sixth Framework IST programme of the EC, under the 035200 ARGUGRID project.
Maxime Morge (ARGUGRID) MFI May 2007 1 / 16
Which kind of decision maker are you ? [Tetlock 06]
Hedgehog say:
◮ “it is annoying to listen to someone whocannot seem to make up his or her mind”
◮ “the most common error indecision-making is to abandon good ideastoo quickly”
Hedgehogs know one big thing, have intuitions, and never surrender.
Fox say:
◮ “when considering most conflicts, I canusually see how both sides could be right”
◮ “I prefer interacting with people whoseopinions are very different from my own”
Foxes knows many little things, interact, and change their mind.
Maxime Morge (ARGUGRID) MFI May 2007 2 / 16
Interaction-based explanation of the decision
Decision support system
decide
inform
Maxime Morge (ARGUGRID) MFI May 2007 3 / 16
Interaction-based explanation of the decision
Argumentation engine
challenge
argue
inform
Maxime Morge (ARGUGRID) MFI May 2007 3 / 16
Outline
Abstract decision structure
Concrete data structures
Argumentation framework
Interaction amongst arguments
Semantics and procedure
Summary & Future works
Maxime Morge (ARGUGRID) MFI May 2007 4 / 16
Abstract decision structure
Influence diagram [Clemen. 06]
Suitable location (g0)
Regulation (g1) Accesible (g2)
Taxes (g3) Permit (g4) Assistance (g5) Sewage (g6) Transport (g7)
Decision
Sea? Road?
Decision
Abstract value
Concrete value
Maxime Morge (ARGUGRID) MFI May 2007 5 / 16
Abstract decision structure
Influence diagram [Clemen. 06]
Suitable location (g0)
g2 ≺ g1
Regulation (g1)
g5 ≺ g4 ≺ g3 ≺ g4 + g5
Accesible (g2)
g7 ≺ g6
Taxes (g3)
a1 ≺ a2
Permit (g4)
a2 ≺ a1
Assistance (g5)
a2 ≺ a1
Sewage (g6)
a2?a1
Transport (g7)
a1 ≺ a2
Decision
London(a1) or Pisa(a2)
Sea? Road?
Sea(a2) ¬Road(a2) ≺ Road(a2)
Decision
Abstract value
Concrete value
Maxime Morge (ARGUGRID) MFI May 2007 5 / 16
Concrete data structures
Data strutures and priorities: hierarchies of conflicting rules
The theory compiles:
◮ goal rules such as R012 : g0 ← g1, g2
◮ epistemic rules such as R123 : b1 ← b2,¬b3
◮ decision rules such as R11 : g1 ← D(a1), b1
Different priorities for different rules:
◮ the priority over goal rules comes from preferences, eg R01 : g0 ← g1
has priority over R02 : g0 ← g2
◮ the priority over epistemic rules comes from probabilities,eg F1 : Road(pisa)← has priority over F2 : ¬Road(pisa)←
◮ the priority over decision rules come from expected utililies,eg R11 : g1 ← D(a1), b1 has priority over R12 : g1 ← D(a1), b2
1. hypothetical, i.e. built upon an hypothesissent(A) = hyp(A)A = 〈D(pisa), ∅, [D(pisa)]〉 or A = 〈Sea(pisa), ∅, [Sea(pisa)]〉
2. trivial, i.e. built upon an unconditional ground statementsent(A) = premise(A)A = 〈Road(pisa), [Road(pisa)], ∅〉 orA = 〈¬Road(pisa), [¬Road(pisa)], ∅〉
3. tree, i.e. built upon a top rule where all literals in the body are theconclusions of subargument s.asent(A) = ∪Ai=subarg(A)sent(Ai ) ∪ body(R)A = 〈g7, (D(pisa), Sea(pisa)), (D(pisa))〉
Maxime Morge (ARGUGRID) MFI May 2007 8 / 16
Interaction amongst arguments
Interaction: choice between different explanations
Philip E. TetlockExpert Political Judgment: How Good is It? How Can We Know?Princeton university press, 2006.
R. T. Clemen.Making Hard Decisions.Duxbury. Press, 1996.
Phan Minh Dung, Paolo Mancarella, Francesca ToniA dialectic procedure for sceptical, assumption-based argumentation,1st International Conference on Computational Models of Argument,September 2006.