Some useful 16-valued logics: How a computer network should think (Q815018)

The paper gives a further generalization of the notion of a bilattice which is due to Belnap [\textit{N. D. Belnap jun.}, ``A useful four-valued logic'', Mod. Uses Multiple-Valued Logic, 5th Int. Symp., Bloomington 1975, 5--37 (1977; Zbl 0424.03012)] and Ginsberg [\textit{M. L. Ginsberg}, ``Multivalued logics: a uniform approach to reasoning in artificial intelligence'', Comput. Intell. 4, 265--316 (1988)], and the authors give yet another example of a trilattice different from their own given in a different paper, and from a trilattice given by Lakshmanan and Sadri [\textit{L. V. S. Lakshmanan} and \textit{F. Sadri}, ``On a theory of probabilistic deductive databases'', Theory Pract. Log. Program. 1, No.~1, 5--42 (2001)]. The basic technique of generating a trilattice is rather standard: the authors take a power set of Belnap's four values as an underlying set for their trilattice. (This method is analogous to Belnap's method of taking a power set of the set of the values ``true'' and ``false''). Then the authors fix orderings on the trilattice in a trivial way, using techniques of Ginsberg and Fitting. Further, they take this (16-element) trilattice as a set of interpretations for a propositional logic and add to it some new connectives reflecting the structure of the trilattice. No calculus is given for the whole trilattice-interpreted logic. Instead, the authors make use of the First-Degree Entailment (FDE), introduced by Anderson and Belnap in 1975 [\textit{A. R. Anderson} and \textit{N. D. Belnap jun.}, Entailment. The logic of relevance and necessity. Vol. I. Princeton, NJ: Princeton University Press (1975; Zbl 0323.02030)] for four-valued logic, and claim that a fragment of their logic based on the truth ordering corresponds to FDE, which is again trivial if we take into consideration that the trilattice is taken as a power set of Belnap's four values. The dual result is given for the dual (``false'') ordering. The strongest claim of the paper is that this 16-valued logic provides a suitable formalism for analysing of how a computer network should think (comparing this with Belnap's four-valued logic, which is suitable only for understanding a single computer). There can be two objections to this. The first one is the following: it is hard to speak about a 16-valued logic, because no calculus is given for the full fragment of it. And the other objection is that the 16 truth values are capable of interpreting only a network consisting of five computers, one of which is connected to the other four computers and then connected to an outer main computer. This is a very serious restriction to both size and structure of a computer network. The authors mention that if we want to analyse the work of bigger networks, we would have to introduce further multilattices. However, the authors give no algorithm of how a multilattice can be constructed for an arbitrary computer network, for example of a bigger size or of a different structure. Although this could be a novel and interesting result.