A Logic-Based Approach to Finding Explanations for Discrepancies in Optimistic Plan Execution

  • Authors:

  • Thomas Eiter;Esra Erdem;Wolfgang Faber;Ján Senko

  • Affiliations:

  • Institut für Informationssysteme, TU Wien, Favoritenstr. 9-11, A-1040 Wien, Austria. E-mail: {eiter,jan}@kr.tuwien.ac.at;Faculty of Engineering and Natural Sciences, Sabancı University Orhanli, Tuzla, Istanbul 34956, Turkey. E-mail: esraerdem@sabanciuniv.edu;Department of Mathematics, University of Calabria, 87030 Rende (CS), Italy. E-mail: faber@mat.unical.it;Institut für Informationssysteme, TU Wien, Favoritenstr. 9-11, A-1040 Wien, Austria. E-mail: {eiter,jan}@kr.tuwien.ac.at

  • Venue:
  • Fundamenta Informaticae
  • Year:
  • 2007

Quantified Score

Hi-index 0.00

Visualization

Abstract

Consider an agent executing a plan with nondeterministicactions, in a dynamic environment, which might fail. Suppose thatshe is given a description of this action domain, includingspecifications of effects of actions, and a set of trajectories forthe execution of this plan, where each trajectory specifies apossible execution of the plan in this domain. After executing somepart of the plan, suppose that she obtains information about thecurrent state of the world, and notices that she is not at acorrect state relative to the given trajectories. How can she findan explanation (a point of failure) for such a discrepancy? Ananswer to this question can be useful for different purposes. Inthe context of execution monitoring, points of failure candetermine some checkpoints that specify when to check fordiscrepancies, and they can sometimes be used for recovering fromdiscrepancies that cause plan failures. At the modeling level,points of failure may provide useful insight into the action domainfor a better understanding of the domain, or reveal errors in theformalization of the domain. We study the question above in ageneral logic-based knowledge representation framework, which canaccommodate nondeterminism and concurrency. In this framework, wedefine a discrepancy and an explanation for it, and analyze thecomputational complexity of detecting discrepancies and findingexplanations for them. We introduce a method for computingexplanations, and report about a realization of this method usingDLV^K, which is a logic-programming based system for reasoningabout actions and change.