AN INVESTIGATION INTO COMMONSENSE REASONING

Abstract

In this thesis, we introduce commonsense reasoning, some of its features and reasoning types. We establish the Event Calculus as a logical formalisation to handle commonsense reasoning; and introduce circumscription as a mathematical machinery to implement default reasoning.We define a framework in which we simulate a world scenario, initiated by an idea from Shin and Davis [40]. They simulate a real world scenario in which an agent moves from a location to another and fills in some buckets with liquid. They implement this in PDDL+. We develop their idea further, represent the scenario in the Event Calculus and elaborate on their formalisations weak points. We introduce a flagging system to deal with triggered events and prevent them from repeated occurrence. We show the elaboration tolerance of the Event Calculus and discuss that carrying out modifications on an already-developed framework does not need performing surgeries on the formalisation. We compare our Event Calculus formulas with PDDL+ of Shin and Davis. We show that their formalism not only does not handle many "commonsense" aspects of their own scenario, performing small changes in their scenario requires major modifications whereas in the Event Calculus representation this is not the case due to its elaboration tolerance.Later in the thesis, a method to transform Event Calculus formulas into propositional logic will be introduced that can be fed into a SAT solver for automated reasoning. The results can be transformed back into Event Calculus formulas by reverse mapping.Different automated reasoners that deal with the Event Calculus are discussed and SAT solving method is explained in more detail.

Date of Award	1 Aug 2011
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Ian Pratt-Hartmann (Supervisor)