Starting from a general characterization of roles, we focus on the ways in which roles are specified, we examine the formal constraints on their definitions, and propose definitional schemas motivating different kinds of roles. This classification, in addition to clarify the notion of role itself, helps us to reconsider the two standard solutions that have been proposed for the famous counting problem, and to suggest that a third mixed approach may be considered.

It has been argued that causal rules are necessary for representing both implicit side-effects of actions and action qualifications, and there have been a number different approaches for representing causal rules in the area of formal theories of actions. These different approaches in general agree on rules without cycles. However, they differ on causal rules with mutual cyclic dependencies, both in terms of how these rules are supposed to be represented and their semantics. In this paper we show that by adding one more minimization to Lin's circumscriptive causal theory in the situation calculus, we can have a uniform representation of causal rules including those with cyclic dependencies. We also demonstrate that sometimes causal rules can be compiled into logically equivalent (under a proposed semantics) successor state axioms even in the presence of cyclical dependencies between fluents.

This paper presents the formalization of a biological process, cell division, in modular action language ALM. We show how the features of ALM — modularity, separation between an uninterpreted theory and its interpretation — lead to a simple and elegant solution that can be used in answering questions from biology textbooks.

We present an ontology consisting of a theory of spatial dimension and a theory of dimension-independent mereological and topological relations in space. Though both are fairly weak axiomatizations, their interplay suffices to define various mereotopological relations and to make any necessary dimension constraints explicit. We show that models of the INCH Calculus and the Region-Connection Calculus (RCC) can be obtained from extensions of the proposed ontology.

In this work, we present a causal extension of logic programming under the stable models semantics where, for a given stable model, we capture the alternative causes of each true atom. The syntax is extended by the simple addition of an optional reference label per each rule in the program. Then, the obtained causes rely on the concept of a causal proof: an inverted tree of labels that keeps track of the ordered application of rules that has allowed deriving a given true atom.

The paper develops a formal model for interpreting vague languages based on a variant of "supervaluation" semantics. Two modes of semantic variability are modelled, corresponding to different aspects of vagueness: one mode arises where there can be multiple definitions of a term; and the other relates to the threshold of applicability of a vague term with respect to the magnitude of relevant observable values. The truth of a proposition depends on both the possible world and the "precisification" with respect to which it is evaluated. Structures representing possible worlds and precisifications are both specified in terms of primitive functions representing observable measurements, so that the semantics is grounded upon an underlying theory of physical reality. On the basis of this semantics, the acceptability of a proposition to an agent is characterised in terms of a combination of agent's beliefs about the world and their attitude to admissible interpretations of vague predicates.

We formalize actions and change in a housekeeping domain with multiple cleaning robots, and commonsense knowledge about this domain, in the action language C+. Geometric reasoning is lifted to high-level representation by embedding motion planning in the domain description using external predicates. With such a formalization of the domain, a plan can be computed using the causal reasoner CCalc for each robot to tidy some part of the house. We introduce a planning and monitoring algorithm for safe execution of these plans, so that it can recover from plan failures due to collision with movable objects whose presence and location are not known in advance or due to heavy objects that cannot be lifted alone. We illustrate the applicability of this algorithm with a simulation of a housekeeping domain.

A recent direction within belief revision theory is to develop a theory of belief change for the Horn knowledge representation framework. We consider questions related to the complexity aspects of previous work, leading to questions about Horn envelopes (or Horn LUB’s), introduced earlier in the context of knowledge compilation. A characterization is obtained of the remainders of a Horn be- lief set with respect to a consequence to be contracted, as the Horn envelopes of the belief set and an elementary conjunction corresponding to a truth assignment satisfying a certain body building formula. This gives an efficient algorithm to generate all remainders, each represented by a truth assignment. On the negative side, examples are given of Horn belief sets and consequences where Horn formulas representing the result of most contraction operators, based either on remainders or on weak remainders, must have exponential size.

List of organizers of the Logical Formalizations of Commonsense Reasoning AAAI symposium.

This work introduces Human-Agent Transfer (HAT), an algorithm that combines transfer learning, learning from demonstration and reinforcement learning to achieve rapid learning and high performance in complex domains. Using experiments in a simulated robot soccer domain, we show that human demonstrations transferred into a baseline policy for an agent and refined using reinforcement learning significantly improve both learning time and policy performance. Our evaluation compares three algorithmic approaches to incorporating demonstration rule summaries into transfer learning, and studies the impact of demonstration quality and quantity. Our results show that all three transfer methods lead to statistically significant improvement in performance over learning without demonstration.