Bitte benutzen Sie diese Referenz, um auf diese Ressource zu verweisen: doi:10.22028/D291-25226
Titel: The eager bidder problem : a fundamental problem of DAI and selected solutions
Verfasser: Schillo, Michael
Fischer, Klaus
Kray, Christian
Sprache: Englisch
Erscheinungsjahr: 2001
Quelle: Kaiserslautern ; Saarbrücken : DFKI, 2001
SWD-Schlagwörter: Künstliche Intelligenz
Freie Schlagwörter: task assignment
contract net protocol
bidding agents
DDC-Sachgruppe: 004 Informatik
Dokumentart : Report (Bericht)
Kurzfassung: The contract net protocol [17] [18] is a widely used protocol in DAI, as it proved to be a flexible and low communication interaction protocol for task assignment. It is however not clear how agents participating in a contract net should allocate their resources, if a large number of contract nets is performed concurrently. If the agent allocates too many resources at an early stage, it may not get its bid accepted and therefore resources are not available for other tasks, if it allocates too late, it may have committed to do more tasks than its resources allow for. In this paper we present an ad hoc solution and two more complex strategies for solving this problem. Furthermore, we introduce a new method based on a statistic approach. We describe these mechanisms and how they deal with the concept of commitment on different levels. There is no optimal solution for every problem setting, but each has advantages and disadvantages. Our discussion concludes with criteria for the decision on which of these mechanisms is best to select for a given problem domain.
Link zu diesem Datensatz: urn:nbn:de:bsz:291-scidok-50434
Schriftenreihe: Research report / Deutsches Forschungszentrum für Künstliche Intelligenz [ISSN 0946-008x]
Band: 01-03
SciDok-Publikation: 7-Feb-2013
Fakultät: Sonstige Einrichtungen
Fachrichtung: SE - DFKI Deutsches Forschungszentrum für Künstliche Intelligenz
Fakultät / Institution:SE - Sonstige Einrichtungen

Dateien zu dieser Ressource:
Datei Beschreibung GrößeFormat 
RR_01_03_.pdf6,75 MBAdobe PDFÖffnen/Anzeigen

Alle Ressourcen in diesem Repository sind urheberrechtlich geschützt.