Discovering Relative Importance of
Skyline Attributes
VLDB`09
1
Outline

Introduction

Basic Notions

P-skyline Discovery

Experiments

Conclusions
2
Introduction
Querying databases with preferences


All attributes are of the same importance in skyline preference
relations
P-skyline (P:prioritize)


3
Add the notion of attribute importance
(Cont.)
4
Basic Notions
Atomic preferences (H): total orders over (single)
attributes
Skyline preference relation (skyH): t1 preferred to t2 if




t1 equal or better than t2 in every attribute, and
t1 strictly better than t2 in at least one attribute
Skyline: the set wskyH(O) of best tuples (according to
skyH) in a set of tuples O

5
(Cont.)
Prioritized skyline (p-skyline)
p-skyline relations


6

Induced by an atomic preference relation >A ∈ H

Pareto accumulation (

Prioritized accumulation (
equally important as
)
more important than
)
(Cont.)
Pareto

7

Prioritized
(Cont.)
p-graph
represents attribute importance induced by a pskyline relation




8
Nodes: attributes
Edges: from more important to less important attributes
P-skyline Discovery
Discovery of p-skyline relations from user feedback
 Problem:
 Given a set A of relevant attributes and a set H of
atomic preferences over A, discover the relative
importance of attributes [in the form of a p-skyline
relation
], based on user feedback.

9
(Cont.)

Computing maximal

10
favoring G in O
Construct a system N of negative constraints from G and O
Experiments
11
(Cont.)
12
Conclusion

generalizing skylines to p-skylines to capture relative
attribute importance
13