Mauro Di Manzo

Natural Language driven Image Generation

In this paper the experience made through the development of a NAtural Language driven Image Generation is discussed. This system is able to imagine a static scene described by means of a sequence of simple phrases. In particular, a theory for equilibrium and support will be outlined together with the problem of object positioning.

Natural language input for scene generation

In this paper a system which understands and conceptualizes scenes descriptions in natural language is presented. Specifically, the following components of the system are described: the syntactic analyzer, based on a Procedural Systemic Grammar, the semantic analyzer relying on the Conceptual Dependency Theory, and the dictionary.

Planning via model checking in determistic domains: Preliminary report

In this paper we report on SMV and NuSmv performances on a set of “model checking problems” variously generated starting from deterministic domain descriptions. The comparison with other state-of-the-art planning systems reveals that “planning via model checking” is a promising research line.

Multi-Context Systems as a Tool to Model Temporal Evolution

Contexts are defined as axiomatic formal systems. More than one context can be defined, each one modeling/solving (part of) the problem. The (global) model/solution of the problem is obtained making contexts communicate via bridge rules. Bridge rules and contexts are the components of Multi Context systems. In this paper we want to study the applicability of multi contexts systems to reason about temporal evolution. The basic idea is to associate a context to each temporal interval in which the “model” of the problem does not change (corresponding to a state of the system). Switch among contexts (corresponding to modifications in the model) are controlled via a meta-theoric context responsible to keep_track_of the temporal evolution. In this way (i) we keep a clear distinction between the theory describing the particular system at hand and the theory necessary for predicting the temporal evolution (ii) we have simple object level models of the system states and (iii) the theorem prover can faster analize and answer to queries about a particular state. The temporal evolution of a U-tube is taken as an example to show both the proposed framework and the GETFOL implementation.

Proving Formulas through Reduction to Decidable Classes

As it is well known, it is important to enrich the basic deductive machinery of an interactive theorem prover with complex decision procedures. In the GETFOL system we have implemented a hierarchical and modular structure of procedures which can be either invoked individually or jointly with the others. At the top of the hierarchy there is a decision procedure for a set of formulas which can be reduced to the class of prenex universal-existential formulas via finitely many application of rewriting rules. In this paper we give a formal account of such a reduction process, arguing that (i) it greatly enlarges the set of formulas which can proven through a decision process and (ii) in some cases makes the resulting formula easier to prove. We also provide an extensional characterization of a class of formulas which can be reduced and thus decided. The implementation of such reducing procedure in GETFOL is also sketched.

Composing decision procedures: the approach and a case study

In this paper we address the problem of strengthening the inferential capabilities of an interactive theorem prover with complex and reusable proof procedures. We focus on the construction of proof procedures built out of decision procedures for (decidable) quantifierfree theories. The idea is to build proof procedures in a structured way. A set of deciders provides the low-level reasoning capabilities, while the high-level (i.e. strategical) reasoning procedures are to be synthesized on top of it. The main goal of the paper is to show that this approach has many advantages and is of wide applicability. As a case study we consider the synthesis of a proof procedure for the existential fragment of first order logic built on top of a prepositional decider. This procedure is particularly well suited for describing our approach, since in it there is a neat separation between the prepositional and the first order reasoning components.