Bitte benutzen Sie diese Referenz, um auf diese Ressource zu verweisen:
doi:10.22028/D291-40424
Titel: | Distributed knowledge-based deduction using the team work method |
VerfasserIn: | Denzinger, Jörg |
Sprache: | Englisch |
Erscheinungsjahr: | 1991 |
Erscheinungsort: | Kaiserslautern |
DDC-Sachgruppe: | 004 Informatik |
Dokumenttyp: | Forschungsbericht (Report zu Forschungsprojekten) |
Abstract: | The team work method is a way to distribute deduction processes that are based on the generation of new facts. The key point to achieve the distribution is the use of various kinds of knowledge that is not well expressible as facts. In analogy to human project teams team work uses tactical control knowledge in form of experts, assessment knowledge in form of referees and strategic control knowledge in form of the supervisor of the team. The supervisor chooses experts that generate new facts. After a certain period of time the referees judge the generated facts and the supervisor uses the best facts to generate a better problem state for the experts. A system using the team work method has the ability to focus on the proper problems by abstraction of results and the possibility of self-tuning of the system by reflective analysis of the computed results. The general concept of team work is instantiated to the team work completion method for equational deduction. Possible experts, referees and supervisors are discussed and their usefulness for completion is demonstrated by examples. The examples will show that super-linear speed-ups have been achieved, even for challenging problems. Implementational aspects of this approach, concerning the problem of idle times of processors, are also discussed and solutions are given. |
Link zu diesem Datensatz: | urn:nbn:de:bsz:291--ds-404242 hdl:20.500.11880/37674 http://dx.doi.org/10.22028/D291-40424 |
Schriftenreihe: | SEKI-Report / Deutsches Forschungszentrum für Künstliche Intelligenz, DFKI [ISSN 1437-4447] |
Band: | 91,12 |
Datum des Eintrags: | 21-Mai-2024 |
Fakultät: | SE - Sonstige Einrichtungen |
Fachrichtung: | SE - DFKI Deutsches Forschungszentrum für Künstliche Intelligenz |
Professur: | SE - Sonstige |
Sammlung: | SciDok - Der Wissenschaftsserver der Universität des Saarlandes |
Dateien zu diesem Datensatz:
Datei | Beschreibung | Größe | Format | |
---|---|---|---|---|
SEKI-Report-SR-91-12_Denzinger_Distributed-Knowledge=based-deduction-using-the-team-work-method.pdf | 2,07 MB | Adobe PDF | Öffnen/Anzeigen |
Alle Ressourcen in diesem Repository sind urheberrechtlich geschützt.