Certain extensions of Nelson's constructive logic N with strong negation have recently become important in arti.cial intelligence and nonmonotonic reasoning, since they yield a logical foundation for answer set programming (ASP). In this paper we look at some extensions of Nelson's .rst-order logic as a basis for de.ning nonmonotonic inference relations that underlie the answer set programming semantics. The extensions we consider are those based on 2-element, here-and-there Kripke frames. In particular, we prove completeness for .rst-order here-and-there logics, and their (...) minimal strong negation extensions, for both constant and varying domains. We choose the constant domain version, which we denote by QNc5, as a basis for de.ning a .rst-order nonmonotonic extension called equilibrium logic. We establish several metatheoretic properties of QNc5, including Skolem forms and Herbrand theorems and Interpolation, and show that the .rst-oder version of equilibrium logic can be used as a foundation for answer set inference. (shrink)

In this article, we compare the expressive powers of classes of normal logic programs that are obtained by constraining the number of positive subgoals in the bodies of rules. The comparison is based on the existence/nonexistence of polynomial, faithful, and modular translation functions between the classes. As a result, we obtain a strict ordering among the classes under consideration. Binary programs are shown to be as expressive as unconstrained programs but strictly more expressive than unary programs which, in turn, are (...) strictly more expressive than atomic programs. We also take propositional theories into consideration and prove them to be strictly less expressive than atomic programs. In spite of the gap in expressiveness, we develop a faithful but non-modular translation function from normal programs to propositional theories. We consider this as a breakthrough due to sub-quadratic time complexity |). Furthermore, we present a prototype implementation of the translation function and demonstrate its promising performance with SAT solvers using three benchmark problems. (shrink)

Abduction is or subsumes a process of inference. It entertains possible hypotheses and it chooses hypotheses for further scrutiny. There is a large literature on various aspects of non-symbolic, subconscious abduction. There is also a very active research community working on the symbolic (logical) characterisation of abduction, which typically treats it as a form of hypothetico-deductive reasoning. In this paper we start to bridge the gap between the symbolic and sub-symbolic approaches to abduction. We are interested in benefiting from developments (...) made by each community. In particular, we are interested in the ability of non-symbolic systems (neural networks) to learn from experience using efficient algorithms and to perform massively parallel computations of alternative abductive explanations. At the same time, we would like to benefit from the rigour and semantic clarity of symbolic logic. We present two approaches to dealing with abduction in neural networks. One of them uses Connectionist Modal Logic and a translation of Horn clauses into modal clauses to come up with a neural network ensemble that computes abductive explanations in a top-down fashion. The other combines neural-symbolic systems and abductive logic programming and proposes a neural architecture which performs a more systematic, bottom-up computation of alternative abductive explanations. Both approaches employ standard neural network architectures which are already known to be highly effective in practical learning applications. Differently from previous work in the area, our aim is to promote the integration of reasoning and learning in a way that the neural network provides the machinery for cognitive computation, inductive learning and hypothetical reasoning, while logic provides the rigour and explanation capability to the systems, facilitating the interaction with the outside world. Although it is left as future work to determine whether the structure of one of the proposed approaches is more amenable to learning than the other, we hope to have contributed to the development of the area by approaching it from the perspective of symbolic and sub-symbolic integration. (shrink)

The paper focuses on a difficult problem when formalizing knowledge: What about the possible concepts that didn’t make it into the formalization? We call such concepts the unconsidered context of the formalized knowledge and argue that erroneous and inadequate behavior of systems based on formalized knowledge can be attributed to different states of the unconsidered context; either while formalizing or during application of the formalization. We then propose an automatic strategy to identify different states of unconsidered context inside a given (...) formalization and to classify which parts of the formalization to use in a given application situation. The goal of this work is to uncover unconsidered context by observing sucess and failure of a given system in use. The paper closes with the evaluation of the proposed procedures in an error diagnosis scenario featuring a plan based user interface. (shrink)

Answer Set Programming is a new paradigm based on logic programming. The main component of answer set programming is a system that finds the answer sets of logic programs. During the computation of an answer set, systems are faced with choice points where they have to select a literal and assign it a truth value. Generally, systems utilize some heuristics to choose new literals at the choice points. The heuristic used is one of the key factors for the performance of (...) the system. A new heuristic for answer set programming has been developed. This heuristic is inspired by hierarchical planning. The notion of criticality, which was introduced for generating abstraction hierarchies in hierarchical planning, is used in this heuristic. The resulting system (CSMODELS) uses this new heuristic in a static way. CSMODELS is based on the system SMODELS. The experimental results show that this new heuristic is promising for answer set programming. A comparison of search times with SMODELS demonstrate CSMODELS’ usefulness. (shrink)