Metric SCSPs: Partial Constraint Satisfaction via Semiring CSPs Augmented with Metrics

  • Authors:
  • Aditya Ghose;Peter Harvey

  • Affiliations:
  • -;-

  • Venue:
  • AI '02 Proceedings of the 15th Australian Joint Conference on Artificial Intelligence: Advances in Artificial Intelligence
  • Year:
  • 2002

Quantified Score

Hi-index 0.00

Visualization

Abstract

The Semiring CSP (SCSP) framework is a popular and robust approach to solving partial constraint satisfaction problems which generalizes several other schemes such as fuzzy CSP, weighted CSP etc. We argue in this paper that it is useful to augment the SCSP framework such that each constraint specifies, in addition, a metric on the semiring values. The additional knowledge of distances between 'preference values' (the elements of the semiring) permits us to define a notion of parameterized solving of SCSPs where we can seek solutions with a preference value no worse than a given value.