资源描述:
《A Constrained Argumentation System for Practical Reasoning》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、AConstrainedArgumentationSystemforPracticalReasoningLeilaAmgoud1,CarolineDevred2,andMarie-ChristineLagasquie-Schiex11IRITUPS,Toulouse-France{amgoud,lagasq}@irit.fr2LERIA,Angers-Francedevred@info.univ-angers.frAbstract.Practicalreasoning(PR),whichisconcerned
2、withthegenericques-tionofwhattodo,isgenerallyseenasatwostepsprocess:(1)deliberation,inwhichanagentdecideswhatstateofaffairsitwantstoreachthatis,itsdesires;and(2)means-endsreasoning,inwhichtheagentlooksforplansforachievingthesedesires.Adesireisjustifiedifitho
3、ldsinthecurrentstateoftheworld,andfeasibleifthereisaplanforachievingit.Theagentsintentionsarethusaconsis-tentsubsetofdesiresthatarebothjustifiedandfeasible.ThispaperproposesthefirstargumentationsystemforPRthatcomputesinonesteptheintentionsofanagent,allowingth
4、ustoavoidthedrawbacksoftheexistingsystems.Theproposedsystemisgroundedonarecentworkonconstrainedargumentationsystems,andsatisfiestherationalitypostulatesidentifiedinargumentationliterature,namelytheconsistencyandthecompletenessoftheresults.Keywords:Argumentati
5、on,PracticalReasoning.1IntroductionPracticalreasoning(PR)[16],isconcernedwiththegenericquestionwhatistherightthingtodoforanagentinagivensituation.In[22],ithasbeenarguedthatPRisatwostepsprocess.Thefirststep,oftencalleddeliberation,consistsofidentifyingthedesi
6、resofanagent.Inthesecondstep,calledmeans-endreasoning,onelooksforwaysforachievingthosedesires,i.e.foractionsorplans.Adesireisjustifiedifitholdsinthecurrentstateoftheworld,andisfeasibleifithasaplanforachievingit.Theagentsintentions,whattheagentdecidestodo,isa
7、consistentsubsetofdesiresthatarebothjustifiedandfeasible.Whatisworthnoticinginmostworksonpracticalreasoningistheuseofargumentsforprovidingreasonsforchoosingordiscardingadesireasanintention.Indeed,severalargumentation-basedsystemsforPRhavebeenproposedinthelit
8、erature[3,13,15].However,inmostoftheseworks,theproblemofPRismodeledintermsofatleasttwoseparatesystems,eachofthemcapturingagivenstepoftheprocess.Suchanapproachmaysufferfromaseriousdrawback.Infact,somede