favorite12Besides improving performance over non-N-free KBs, a more powerful module extractor might enable the application of the combined Mod+Opt method to non-N-free DLN knowledge bases, by removing all normality concepts from KB before Opt is applied.
favorite7This is the first time such performance is reached over nonmonotonic KBs of this size: more than 20K concept names and over 30K inclusions.8 Our approach brings technology closer to practical nonmonotonic reasoning with very large KBs. Only the random dependencies introduced by synthetic DIs, combined with numerous restrictions of role ranges to normal individuals, can raise response time over 40 seconds; in most of the other cases, computation time remains below 2 seconds.
favorite2Most other logics silently neutralize the conflicts between nonmonotonic axioms with the same (or incomparable) priorities by computing the inferences that are invariant across all possible ways 2.
favorite5Both are based on specificity: the specific default properties of a concept C have higher priority than the more generic properties of its superconcepts (i.e. the effects of adding default properties to an existing, classical KB are more predictable, as the hierarchy used for determining specificity and resolving conflicts is the original, validated one, and is not affected by the new DIs (see also the related discussion in [3,4], that adopt (1)).
favorite2Given the negligible number of applications based on nonmonotonic logics deployed so far, DLN has been designed to address real-world problems and concrete knowledge engineering needs.
favorite101 Introduction Recently, a new family of nonmonotonic Description Logics (DLs), called DLN , has been introduced .