Most of the programs developed by AI researchers, suffer from two problems: (1) they do not take into account the fact of reasoning resources are limited; (2) and they remain silent (no answer is produced) whenever a definite answer cannot be produced.
The system described in this paper combines the use of resources with the capability of producing conditional answers. A conditional answer explicitly reveals the impediments that are responsible for the lack of a definite answer.
We present RR, an intelligent resource-bounded reasoner, whose resource manipulation is flexible enough to accommodate several resource representations and resource spending strategies. Since no commitments about the way resources are spent are made a priori, the process of consuming resources can be used to model non-omniscient, non-exhaustive reasoners.
The concepts explored by RR are being incorporated into the SNePS, the Semantic Network Processing System, and so improving (1) the way human reasoning can be modeled, and (2) its interface with the outside world.
KeywordsKnowledge Base Inference Rule Resource System Potential State Definite Answer
Unable to display preview. Download preview PDF.
- [Anderson and Belnap 75]A. Anderson and N. Belnap, Entailment: The Logic of Relevance and Necessity 1 (Princeton University Press, Princeton, NJ, 1975).Google Scholar
- [Belnap 77]N. Belnap, A useful four-valued logic, in: G. Epstein and J. Dunn, eds., Modern Uses of Multiple-Valued Logic (D. Reidel, Dordrecht, The Netherlands, 1977) 8–37.Google Scholar
- [Bobrow and Winograd 77]
- [Donlon 82]G. Donlon, Using resource limited inference in SNePS, SNeRG Technical Note 10, Department of Computer Science, State University of New York at Buffalo, Buffalo, NY, 1982.Google Scholar
- [Driankov 88]D. Driankov, Towards a many-valued logic of quantified belief, Ph.D. Thesis, Thesis 192, Department of Computer and Information Science, Linköping University, Linköping, Sweden, 1988.Google Scholar
- [Fagin and Halpern 87]
- [Garcia and Moussavi 90]O. Garcia and M. Moussavi, A six-valued logic for representing incomplete knowledge, in: Proceedings 20th International Symposium on Multiple-Valued Logic, Charlotte, NC (IEEE, 1990) 110–114.Google Scholar
- [Hintikka 75]
- [Horvitz 88]E. Horvitz, Reasoning under a varying and uncertain resource constraints, in: Proceedings AAAI-88, Saint Paul, MN (1988) 111–116.Google Scholar
- [Mamede and Martins 90a]N. Mamede and J. Martins, Expanding SNePS capabilities with LORE, in: D. Kumar, ed., Proceedings First Annual SNePS Workshop Buffalo, NY (Springer-Verlag, Berlin, 1990) 27–39.Google Scholar
- [Mamede and Martins 90b]N. Mamede and J. Martins, Bringing resources into logic, in: Proceedings 20th International Symposium on Multiple-Valued Logic, Charlotte, NC (IEEE, 1990) 220–227.Google Scholar
- [Mamede and Martins 90c]N. Mamede and J. Martins, Resource-bounded abduction, GIA Technical Report 90/4, Technical University of Lisbon, Lisbon, Portugal, 1990.Google Scholar
- [Martins 83]J. Martins, Reasoning in multiple belief spaces, Ph.D. Thesis, Technical Report 203, Department of Computer Science, State University of New York at Buffalo, Buffalo, NY, 1983.Google Scholar
- [Martins and Shapiro 88]
- [Minicozzi and Reiter 72]
- [Mutchler 86]D. Mutchler, Optimal allocation of very limited search resources, in: Proceedings AAAI-86, Philadelphia, PA (1986) 467–471.Google Scholar
- [Reiter and de Kleer 87]Reiter R. and J. de Kleer, Foundations of assumption-based truth maintenance systems: Preliminary Report, in: Proceedings AAAI-87, Seattle, WA (1987) 183–188.Google Scholar
- [Slagle, Chang and Lee 69]J. Slagle, C. Chang and R. Lee, Completeness theorem-s for semantic resolution in consequence-finding in: Proceedings IJCAI-69, Washington, D.C. (1969) 281–285.Google Scholar
- [Swartout 83]
- [Winograd 80]