Enriching a CP-Net by Asymmetric Merging

TL;DR

This paper introduces an asymmetric merging method for CP-nets that enriches a user's preference structure by incorporating information from other CP-nets while preserving original preferences, improving upon symmetric merging approaches.

Contribution

The paper proposes a novel asymmetric merging technique for CP-nets that maintains original preferences and effectively fills in incomplete preference information.

Findings

Enriching CP-nets with asymmetric merging preserves original preferences.

The method effectively fills gaps in incomplete CP-nets.

Compared to symmetric merging, the approach better maintains user-specific preferences.

Abstract

Conditional ceteris paribus preference networks (CP-nets) are commonly used to capture qualitative conditional preferences. In many use cases, when the preferential structure of an agent is incomplete, information from other preferential structures (e.g. that of other users) preferences can be used to fill in the gaps. Earlier works proposed methods to symmetrically merge multiple incomplete CP-nets by means of voting semantics. However, the merged CP-net can contain preference relations that do not fit to a given user's original preference profile. This paper proposes an asymmetric merging (or enriching) method to obtain and fill-in preference relations of a user's CP-net from another CP-net in a way that preserves the original preference relations.