Program models and semi-public environments

Abstract

We develop a logic for reasoning about semi-public environments , i.e. environments in which a process is executing, and where agents in the environment have partial and potentially different views of the process. Previous work on this problem illustrated that it was problematic to obtain both an adequate semantic model and a language for reasoning about semi-public environments. We here use program models for representing the changes that occur during the execution of a program. These models serve both as syntactic objects and as semantic models, and are a modification of action models in Dynamic Epistemic Logic, in the sense that they allow for ontic change (i.e. change in the world or state). We show how program models can elegantly capture a notion of observation of the environment. The use of these models resolves several difficulties identified in earlier work, and admit a much simpler treatment than was possible in previous work on semi-public environments.

Program Semantics and Verification Technique for AI-Centred Programs

“…To this end, interpreted systems were introduced [31], whereby agents are associated with private shares of the possible worlds called local states; worlds' indistinguishability is then "sunk" at the level of local states. Alternatively, others looked at how epistemic logic with imperfect information could be expressed via direct notions of visibility (or observability) of propositional variables, e.g., [37,20,6].…”

“…A line of research in DEL, the so called semi-public environments, also builds agents' indistinguishability relations from the observability of propositional variables [37,6,20]. The work of Grossi [19] explores the interaction between knowledge dynamics and non-deterministic choice/sequential composition.…”

“…On the one hand, it goes without saying that evaluating such knowledge-centric properties considered in a partial observability setting are of paramount importance for AI-based decision making [3]. On the other hand, it is known that logics of knowledge, also called epistemic logics [16], have been well-explored in computer-science since Hintikka [22], and even in the context of multi-agent systems [31] and under partial observability [37,20,6]. In this space, our innovation focuses in turn on new methods for automatically verifying epistemic properties, but -unlike most of our predecessors-we concentrate on verification methods not for abstract systems, but rather analyses of concrete programs (over an arbitrary first-order domain), as well as requirements richer than what went before us.…”

Program Semantics and a Verification Technique for Knowledge-Based Multi-Agent Systems

“…A line of research in DEL, the so called semi-public environments, also builds agents' indistinguishability relations from the observability of propositional variables [38,8,22]. The work of Grossi [21] explores the interaction between knowledge dynamics and non-deterministic choice/sequential composition.…”

“…Then, knowledge evaluated over interpreted systems hinges closely on these local states: an agent cannot distinguish two global states if her local states within these global states are equal. Several lines also looked at how epistemic logic with imperfect information could be expressed via notions of visibility (or observability) of propositional variables, e.g., [38,22,8].…”