Learning effect axioms via probabilistic logic programming

Rolf Schwitter*

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference proceeding contribution

5 Downloads (Pure)

Abstract

Events have effects on properties of the world; they initiate or terminate these properties at a given point in time. Reasoning about events and their e ects comes naturally to us and appears to be simple, but it is actually quite di cult for a machine to work out the relationships between events and their e ects. Traditionally, effect axioms are assumed to be given for a particular domain and are then used for event recognition. We show how we can automatically learn the structure of effect axioms from example interpretations in the form of short dialogue sequences and use the resulting axioms in a probabilistic version of the Event Calculus for query answering. Our approach is novel, since it can deal with uncertainty in the recognition of events as well as with uncertainty in the relationship between events and their effects. The suggested probabilistic Event Calculus dialect directly subsumes the logic-based dialect and can be used for exact as well as a for inexact inference.

Original languageEnglish
Title of host publicationICLP 2017
Subtitle of host publicationTechnical Communications of the 33rd International Conference on Logic Programming
EditorsRicardo Rocha, Tran Cao Son, Christopher Mears, Neda Saeedloei
Place of PublicationSaarbrücken/Wadern
PublisherDagstuhl Publishing
Pages1-15
Number of pages15
ISBN (Electronic)9783959770583
DOIs
Publication statusPublished - 1 Feb 2018
EventInternational Conference on Logic Programming (33rd : 2017) - Melbourne, Australia
Duration: 28 Aug 20171 Sep 2017

Publication series

NameOASIcs
Volume58
ISSN (Electronic)2190-6807

Conference

ConferenceInternational Conference on Logic Programming (33rd : 2017)
CountryAustralia
CityMelbourne
Period28/08/171/09/17

Bibliographical note

Version archived for private and non-commercial use with the permission of the author/s and according to publisher conditions. For further rights please contact the publisher.

Keywords

  • Effect axioms
  • Event calculus
  • Event recognition
  • Probabilistic logic programming
  • Reasoning under uncertainty

Fingerprint Dive into the research topics of 'Learning effect axioms via probabilistic logic programming'. Together they form a unique fingerprint.

  • Cite this

    Schwitter, R. (2018). Learning effect axioms via probabilistic logic programming. In R. Rocha, T. C. Son, C. Mears, & N. Saeedloei (Eds.), ICLP 2017: Technical Communications of the 33rd International Conference on Logic Programming (pp. 1-15). [8] (OASIcs; Vol. 58). Saarbrücken/Wadern: Dagstuhl Publishing. https://doi.org/10.4230/OASIcs.ICLP.2017.8