Loading [MathJax]/extensions/MathZoom.js
Representing and reasoning with constrained PCP-nets | IEEE Conference Publication | IEEE Xplore

Representing and reasoning with constrained PCP-nets


Abstract:

A Probabilistic Conditional Preference network (PCP-net) provides a compact representation of preferences characterized with uncertainty. We propose to enrich the express...Show More

Abstract:

A Probabilistic Conditional Preference network (PCP-net) provides a compact representation of preferences characterized with uncertainty. We propose to enrich the expressive power of the PCP-net by adding constraints between some of the variables. We call this new model, the Constrained PCP-net (CPCP-net). We study the key preference reasoning task with the proposed CPCP-net which consists in finding the most probable optimal outcome i.e. the most probable outcome that best represents the preferences while satisfying all the constraints. In this regard, a variant of the Branch and Bound algorithm has been proposed and experimentally evaluated on CPCP-net instances, randomly generated based on the RB-model. The results of these experiments show that the new proposed solving method is capable of returning the most probable optimal outcome in a reasonable time.
Date of Conference: 05-08 October 2017
Date Added to IEEE Xplore: 30 November 2017
ISBN Information:
Conference Location: Banff, AB, Canada

References

References is not available for this document.