CONDITIONAL PREFERENCE NETWORKS: CONSTRAINTS AND SIMILARITY

Thumbnail Image

Date

Journal Title

Journal ISSN

Volume Title

Publisher

Saudi Digital Library

Abstract

Conditional preference networks (CP-nets) are increasingly becoming popular in representing ordinal preference relations in an efficient manner. However, one of the main issues with CP-net is that both preferences and constraints influence real-life outcomes. In this thesis, we explore two objectives associated with constrained CP- nets (CCP-nets). Our first objective is to extend the underlying constraint network of the CCP-net to the weighted constraint satisfaction problem (WCSP). We call the Weighted Constrained CP-net (WCCP-net). This new model solving the WCCP- net consists of finding those solutions from the Pareto optimal set related to the CCP-net that maximizes the total cost function of the underlying WCSP. Our second objective is to aggregate CCP-nets by defining the similarity between CCP-nets and providing a learning algorithm that can compute the distance between pairs of CCP- nets based on the similarity defined. The similarity is defined based on the Hamming distance (between pairs of outcomes) as well as the number of worsening flips between outcomes. To achieve these goals, we propose two experimental methods: WCCP- net and disCCP-net. We then conduct several experiments on CCP-net instances and develop a search algorithm to provide the optimal outcomes that can accommodate all the constraints and preferences. Overall, our results reveal that the CCP-net models can be developed to optimize and aggregate user preferences and requirements.

Description

Keywords

Citation

Collections

Endorsement

Review

Supplemented By

Referenced By

Copyright owned by the Saudi Digital Library (SDL) © 2025