No label defined (Q4040947)

The articles of this volume will not be indexed individually. Since the discovery of semantic and deductive tableaux by E. Beth, there has been a development towards a procedural treatment in logic and semantics. In computer science this has led to methods and languages for logic programming with new emphasis on predicate logic as a means for knowledge representation. In philosophical logic one observes the rise of dialogical logic introduced by P. Lorenzen and developed by Barth and Krabbe, abandoning the one-role solipsist grab of model theoretic and axiomatic logic in favour of a multi-party, argumentation oriented concept. In natural language semantics we find Hintikka's reconstruction of Wittgenstein's language games and Kamp's discourse representation theory. The procedural view of logic and semantics is directly connected with the study of partial models in natural language semantics. Partial models, on the other hand, play an important role in topics like non- monotonic reasoning which has come in the focus of interest of the AI community during the last decade. The eight papers of this volume are based on the talks given during a workshop on Dialogue Tableaux Theory, Discourse Representation Theory and Logic Programming in Stuttgart in June 1985. The titles are: Johan van Benthem: Games in Logic: A survey, Jan van Eijck: Games and Representation, A. J. M. van Hoof: On how to convince a hangman, Hans Kamp: Conditionals in DR Theory, Robert Kowalski: Logic-Based Open Systems, Eric C. W. Krabbe, Dialogue Sequents and Quick Proofs of Completeness, Drs. P. A. Smit: Argumentation Theory and Knowledge Representation and finally Gerhard Heyer: Generic Generalizations, Discourse Representation Structures and Knowledge Representation. In the first paper, van Benthem gives a survey on so-called Dialogue Games. The idea is that an opponent who has granted the premises of an inference attacks its conclusion, defended by a proponent. The rules of this game essentially define the set of theorems and therefore the logic. The main outcome of dialogue analysis in this: the inferences where the proponent has a winning strategy are precisely those of intuitionistic, or constructive logic. In Jan van Eijck's paper two approaches to the semantics of natural language that at first sign seem quite different are shown to be basically the same. The paper starts with a brief sketch of the representational approach to natural language understanding. Next, the use of game-theoretic ideas in theories of natural language understanding is discussed. It is shown how games give rise to representations, how these representations can be put to use in an account of anaphora, and how several different notions of consequence can be defined in terms of them. The conclusion of this paper is that there are no serious differences between the games approach to natural language understanding and the representational perspective. In the third paper, van Hoof uses dialogical analysis to give another solution to the Hangman Paradox using Krabbe's dialogue logics for modal sentences. Hans Kamp gives in his paper Conditionals in DR Theory a very good introduction to discourse representation theory for natural language. The kernel of DR theory are so-called Discourse Representation Structures. For a sequence of sentences S1,...,Sn, a DRS for a subsequence S1,...,Sk represents information for interpreting sentence \(Sk+1\) with respect to the first k sentences. Kamp then describes an extension of the current formalism for handling conditionals such as if, when, implies, each and every. Robert Kowalski addresses the problem to represent open systems involving cooperating and conflicting agents interacting in parallel with mutually inconsistent beliefs in logic. He argues that the methodology of logic and logic programming can be extended to meet the needs of open systems. A first step towards this goal is the concept of knowledge assimilation. But this is not sufficient: Belief systems need to be embedded within agents which process their own, resident goals, plan courses of actions to achieve their goals, and have the ability to put actions into effect. In Eric C. W. Krabbe's very short article quick inductive proofs of completeness of some dialectic systems are given by using dialogue sequents. The article of Drs. P. A. Smit is essentially an abstract of P. N. Johnson-Laird's book ``Mental Models, towards a cognitive science of language, inference and consciousness''. In addition some recommendations to refine the representation of inferential processes by a dialogical distinction of speaker and hearer are given. The last paper,written by Gerhard Heyer, is an attempt to demonstrate that it makes sense to distinguish between formal semantics of natural language, for example by means of DR theory, and knowledge representation with AI techniques, e.g. logic, frames or production rules. He gives the following reasons: - without a semantic representation language like DRS, that allows for a representation of, in particular anaphoric links in discourse, constructing a knowledge base would be clumsy, if not impossible; - there is an intuitive distinction between background knowledge, permanent domain-specific knowledge, and temporally restricted updates provided by a knowledge base that would be blurred if all knowledge were only expressed on the semantic representation level in terms of DRS's; - present-day DRT does not fulfill AI standards of knowledge representation with regard to manipulating and retrieving information about the knowledge base. Heyer substantiates his claim by way of the example of generic reference. He sketches how an AI model of definite generics in terms of frames can meaningfully supplement their analysis in terms of DRS's.