A lightweight epistemic logic and its application to planning

On the Epistemic Logic of Incomplete Argumentation Frameworks

“…Epistemic logic of visibility or observability was born as a lightweight alternative to standard epistemic logic (van der Hoek, Troquard, and Wooldridge 2011;Herzig, Lorini, and Maffre 2018;Cooper et al 2021). We consider its singleagent version and adapt it to our purposes.…”

A Dynamic Epistemic Logic with Finite Iteration and Parallel Composition

“…Clearly, atom α is valid if and only if it is introspective. It is easily shown that all properties of EL-O given in (Cooper et al 2020a), in particular NP-completeness of the satisfiability problem, still hold in EL-OC. The EL-OC validities are axiomatised in Table 1; the completeness proof closely follows that of (Cooper et al 2020a).…”

“…We assume deterministic actions with conditional effects described by add-and delete-lists. Such conditional effects are crucial: when an agent performs an action then the effects on another agent's epistemic state typically depend on whether that agent sees the variables that are modified by the action (Andersen, Bolander, and Jensen 2012;Cooper et al 2020a).…”

“…The latter is directly related to the general undecidability of DEL-based planning (Bolander and Andersen 2011;Aucher and Bolander 2013;Bolander et al 2020). In previous work we have proposed a different route whose starting point is a restricted version of standard epistemic logic: boolean combinations of 'knowing-whether' operators followed by a propositional variable (Cooper et al 2016;Cooper et al 2020a). In that lightweight epistemic logic, EL-O, the existence of sequential and parallel epistemic plans can be decided in PSPACE (Cooper et al 2020b).…”

“…We illustrate our logic by means of the gossip problem where each agent i ∈ Agt knows a secret s i that none of the other agents knows and where the goal is to find a sequence of phone calls after which every agent knows every secret (Landau 1954;Knödel 1975;van Ditmarsch, van der Hoek, and Kuijer 2020). Several interesting variants of that problem can be designed, in particular where the goal is to obtain higherorder knowledge (Cooper et al 2019;Cooper et al 2020a;van Ditmarsch, Gattinger, and Ramezanian 2020).…”