Antoniou continues the generally excellent series of surveys published by the ACM for over 30 years. Each is intended for upper-level undergraduates or beginning graduate students, and describes a particular area of computer science. This tutorial focuses on the basic concepts of default reasoning, which is the logic that applies to a system that has only incomplete information at hand, and where plausible conjectures called defaults are added to the premises being assumed. If new information later becomes available, the default and any conclusions depending on its use may be retracted. Such a system is nonmonotonic. That is, if a statement s follows from a set of premises M, and M ⊆ M′, then s does not necessarily follow from M′. Only a general familiarity with the predicate calculus is assumed. The monograph is based on work beginning with that of Reiter (1980), continuing through Lukaszewicz (1988–90), and ending with the Rational Default Logic of Mikitiuk and Trusczynski (1995). A 1996 paper of Engelfriet and Treur on the semantics of default logics is included in the excellent reference list, but statement meanings are not discussed in this tutorial.
A simple example neatly explains the difference between a default logic and classical logic (p. 339). Default logic uses whereas classical has football ∧ ⌝ snow → takesPlace . In the first rule, ⌝ snow is used as a default, whereas in the second, it is a fact. Classical reasoning would prohibit any football games from being scheduled unless it were “proven” that there would be no snow, while using a default allows games to be listed assuming that there will be no snow. Although this example was well chosen, others are old chestnuts, such as the bird (tweety) series.
I was disappointed by the lack of motivation for using default logics in real-world situations, such as databases, law, and medicine, as the author claims. The tweety example is not sufficient. The author states that the purpose of this tutorial is to contribute to “the understanding of the basic concepts” (p. 358), and he refers readers to specific references discussing implementations. He notes that default reasoning has proven difficult for students at the University of Toronto to understand. I believe that the inclusion of better examples would have helped.