ContactPerson: flj@cse.buffalo.edu Remote host: mira.cse.buffalo.edu Remote ident: flj ### Begin Citation ### Do not delete this line ### %R 99-08 %U /projects/flj/tech-reports/CSE-tech-reports/says-who.ps %A Johnson, Frances L. %A Shapiro, Stuart C. %T Says Who? -- Incorporating Source Credibility Issues into Belief Revision %D July 31, 1999 %I Department of Computer Science and Engineering, SUNY Buffalo %K knowledge representation; knowledge; belief; belief revision; automatic; beliefs; source; credibility; ordering %Y Nonmonotonic reasoning and belief revision; Representations (procedural and rule-based); Semantic networks %X We discuss some belief revision theories and the implementation of some of them in SNeBR, the belief revision subsystem of the SNePS knowledge representation and reasoning system. The following guidelines are important to belief revision: - minimize information removal - retract less-plausible beliefs before more-plausible ones. Alterations to SNeBR that provide users with information to help them follow these guidelines include the incorporation of sources into the knowledge base, as well as ordering of both sources and individual propositions, allowing partially ordered propositional epistemic entrenchment. We also developed an automatic belief revision option that performs the retraction of a belief if one culprit can be singled out based on the guidelines above.