UML is the de-facto standard formalism for software design and analysis. To support the design of large-scale industrial applications, sophisticated CASE tools are available on the market, that provide a user-friendly environment for editing, storing, and accessing multiple UML diagrams. It would be highly desirable to equip such CASE tools with automated reasoning capabilities in order to detect relevant formal properties of UML diagrams, such as inconsistencies or redundancies. With regard to this issue, we consider UML class diagrams, which are one of the most important components of UML, and we address the problem of reasoning on such diagrams. We resort to several results developed in the eld of Description Logics (DLs), a family of logics that admit decidable reasoning procedures.

We introduce a formal proof system based on tableau methods for analyzing computations made in Answer Set Programming (ASP). Our approach furnishes declarative and fine-grained instruments for characterizing operations as well as strategies of ASP-solvers. First, the granulation is detailed enough to capture the variety of propagation and choice operations of algorithms used for ASP; this also includes SAT-based approaches. Second, it is general enough to encompass the various strategies pursued by existing ASP-solvers. This provides us with a uniform framework for identifying and comparing fundamental properties of algorithms. Third, the approach allows us to investigate the proof complexity of algorithms for ASP, depending on choice operations. We show that exponentially different best-case computations can be obtained for different ASP-solvers. Finally, our approach is flexible enough to integrate new inference patterns, so to study their relation to existing ones. As a result, we obtain a novel approach to unfounded set handling based on loops, being applicable to non-SAT-based solvers. Furthermore, we identify backward propagation operations for unfounded sets.

Abstract. Program verification is concerned with proving that a program is correct and adheres to a given specification. Testing a program, in contrast, means to search for a witness that the program is incorrect. In the present paper, we use a program logic for Java to prove the incorrectness of programs. We show that this approach, carried out in a sequent calculus for dynamic logic, creates a connection between calculi and proof procedures for program verification and test data generation procedures. Starting with a program logic enables to find more general and more complicated counterexamples for the correctness of programs.

We present a calculus that integrates equality handling by superposition into a free variable tableau calculus. We prove completeness of this calculus by an adaptation of the model generation [1, 15] technique commonly used for completeness proofs of resolution calculi.

Dynamic First Order Logic (DFOL) results from interpreting quantification over a variable v as change of valuation over the v position, conjunction as sequential composition, disjunction as nondeterministic choice, and negation as (negated) test for continuation. We present a tableau style calculus for DFOL with explicit (simultaneous) binding, prove its soundness and completeness, and point out its relevance for programming with DFOL, for automated program analysis including loop invariant detection, and for semantics of natural language. Next, we extend this to an infinitary calculus for DFOL with iteration and connect up with other work in dynamic logic.

Decomposable negation normal form (DNNF) was developed primarily for knowledge compilation. Formulas in DNNF are linkless, in negation normal form (NNF), and have the property that atoms are not shared across conjunctions. Full dissolvents are linkless NNF formulas that do not in general have the latter property. However, many of the applications of DNNF can be obtained with full dissolvents. Two additional methods — regular tableaux and semantic factoring — are shown to produce equivalent DNNF. A class of formulas is presented on which earlier DNNF conversion techniques are necessarily exponential; path dissolution and semantic factoring handle these formulas in linear time.

Mathematical logicians had developed the art of formalizing declarative knowledge long before the advent of the computer age. But they were interested primarily in formalizing mathematics. Because of the important role of nonmathematical knowledge in AI, their emphasis was too narrow from the perspective of knowledge representation, their formal languages were not sufficiently expressive. On the other hand, most logicians were not concerned about the possibility of automated reasoning; from the perspective of knowledge representation, they were often too generous in the choice of syntactic constructs. In spite of these differences, classical mathematical logic has exerted significant influence on knowledge representation research, and it is appropriate to begin this handbook with a discussion of the relationship between these fields. The language of classical logic that is most widely used in the theory of knowledge representation is the language of first-order (predicate) formulas. These are the formulas that John McCarthy proposed to use for representing declarative knowledge in his advice taker paper [176], and Alan Robinson proposed to prove automatically using resolution [236]. Propositional logic is, of course, the most important subset of first-order logic; recent

Abstract. In this paper we reduce the question of validity of a first-order intuitionistic formula without equality to generating ground instances of this formula and then checking whether the instances are deducible in a propositional intuitionistic tableaux calculus, provided that the propositional proof is compatible with the way how the instances were generated. This result can be seen as a form of the Herbrand theorem, and so it provides grounds for further theoretical investigation of computer-oriented intuitionistic calculi. 1

Abstract. We add labels to first-order clauses to simultaneously apply superpositions to several proof obligations inside one clause set. From a theoretical perspective, the approach unifies a variety of deduction modes. These include different strategies such as set of support, as well as explicit case analysis, e.g., splitting. From a practical perspective, labelled clauses offer advantages in the case of related proof obligations resulting from multiple conjectures over the same axiom set or from a single conjecture that is a large conjunction. Here we can share clauses (e.g., the axioms and clauses deduced from them, share Skolem symbols), share deduced clause variants, and transfer lemmas between the different obligations. Motivated by software verification, we have created a prototype implementation of labelled clauses that supports multiple conjectures, and we provide convincing experiments for the benefits. 1

Proving completeness of NC-resolution under a linear restriction has been elusive; it is proved here for formulas in negation normal form. The proof uses a generalization of the Anderson-Bledsoe excess literal argument, which was developed for resolution. That result is extended to NC-resolution with partial replacement. A simple proof of the completeness of regular, connected tableaux for formulas in conjunctive normal form is also presented. These techniques are then used to establish the completeness of regular, connected tableaux for formulas in negation normal form.