What are MultiContext Logics?

The term "MultiContext Logics" (MC Logics) refers to a family of logics for the representation of contextual reasoning based on the two following principles:

• Principle 1 (of Locality)
  Reasoning uses only part of what is potentially available (e.g., what is known, the available inference procedures). The part being used while reasoning is what we call context (of reasoning).

• Principle 2 (of Compatibility)
  there is compatibility among the reasoning performed in different contexts.

Our interest in contextual reasoning

In general we aim at analysing both the notions of context and contextual reasoning from three different perspectives:

• Logic
  The work is aiming at defining and studying the properties of MultiContext logics. In these logics, contexts are viewed as a set of interacting formal theories, each with its own language, semantics and axiomatic system. Relations between contexts are represented as interaction between theories.
  From the model theoretic perspective, relations between contexts are formalized as constraints between sets of (local) models of different theories (see Local Models Semantics). Proof-theoretically, relations between contexts are formalized via a set of rules, called bridge rules, whose premises and conclusions belong to different theories (see MultiContext Systems).

• Computer Science and Artificial Intelligence
  We work on applying contexts to the formalization of various forms of common sense reasoning in artificial intelligence. In particular contexts have been succesfully applied in the formalization of propositional attitudes (e.g. beliefs), the qualification problem, reasoning with viewpoints, planning, ... (see Historical remarks and Uses of MultiContext Logics).

• Philosophy
  The notion of context as an individual's partial and approximate theory of the world plays an essential role in the solution of some important philosophical problems. In particular, we think of contextual logic as a serious alternative to modal logics as theoretical framework for a theory of meaning and a theory of representation. The crucial shift from modal logic to contextual logic is that the first encodes an "objective" perspective on problems -based upon the concept of possible world-, whereas the second encodes a "subjective" perspective -based upon the concept of context- (see Historical remarks and Philosophical foundations).

Relevant Publications

L. Serafini and F. Giunchiglia
ML Systems: A Proof Theory for Contexts
Journal of Logic, Language and Information Spring 2002, Volume 11, Issue 2 pp. 471-518. (pdf)

Relevant Presentations

F. Giunchiglia
Local Models Semantics, or Contextual Reasoning = Locality + Compatibility (ZIP file)
Talk at Stanford-Trento worldwide seminar.