Escolar Documentos
Profissional Documentos
Cultura Documentos
Outline
Motivation
& Preliminaries
Monochromatic Reverse Top-k Queries
Bichromatic Reverse Top-k Queries
Threshold-based
Algorithm
Materialized Views
Experimental
Evaluation
Conclusions & Future Work
2
amount of
available data
Users prefer to retrieve
a limited set of k
ranked data objects
that best match their
preferences (top-k
queries)
3
Top-k Query
Weight w[i]:
the perspective of
manufacturers:
it is important that a
product is returned in
the highest ranked
positions for as many
user preferences as
possible
estimate the impact of
a product compared to
their competitors
products
advertise a product to
potential customers
sales representative
Which customers
would be interested?
top-k query:
different versions
Monochromatic:
sales representative
no knowledge of user
preferences
Bichromatic:
Query point q=p3, k=1: empty result set for the bichromatic query
Outline
Motivation
& Preliminaries
Monochromatic Reverse Top-k Queries
Bichromatic Reverse Top-k Queries
Threshold-based
Algorithm
Materialized Views
Experimental
Evaluation
Conclusions & Future Work
9
mRTOP1(q)
1
Solution space
10
mRTOP1(q)
Monochromatic
Outline
Motivation
& Preliminaries
Monochromatic Reverse Top-k Queries
Bichromatic Reverse Top-k Queries
Threshold-based
Algorithm
Materialized Views
Experimental
Evaluation
Conclusions & Future Work
14
Nave
approach:
for each weighting vector process the top-k query
test if query point q is in the top-k list
15
Threshold-based Algorithm
sort
(RTA):
top-k
sets
evaluate
16
top-2 query
for w1
p9
p8
Set
threshold based
on w2
fw2(q)
> threshold
discard w2
Refine
Buffer: p1, p2
threshold for
p10
p5
p1
p6
p4
w3
p
2
w1
w2
p7
p3
w3
W=[ w1, w2, w3 ]
17
Materialized Views
Threshold-based Algorithm
(RTA)
reduce
Views
find
Materialized Views
Grid-based
space
w1, w2, w3
partitioning
cell
Ci
lower
upper
We store
Materialized Views
Given
a point q enclosed
in cell Ci
all weighting vectors
in RTOPk(CiU) belong
to the result set of q
only weighting
vectors in
w1, w2, w3
RTOPk(CiL) - RTOPk(CiU)
have to be examined
Materialized views can
be generalized for
arbitrary k<K values
w1, w2, w3 , w4
20
Outline
Motivation
& Preliminaries
Monochromatic Reverse Top-k Queries
Bichromatic Reverse Top-k Queries
Threshold-based
Algorithm
Materialized Views
Experimental
Evaluation
Conclusions & Future Work
21
Experimental Setup
Comparison
of top-k evaluations
22
23
Outline
Motivation
& Preliminaries
Example of Reverse Top-k Queries
Monochromatic Reverse Top-k Queries
Bichromatic Reverse Top-k Queries
Threshold-based
Algorithm
Materialized Views
Experimental
Evaluation
Conclusions & Future Work
26
Thank you!
Related work:
Akrivi Vlachou, Christos Doulkeridis, Yannis Kotidis, Kjetil Nrvg: "Reverse
Top-k Queries"
Akrivi Vlachou, Christos Doulkeridis, Kjetil Nrvg, Yannis Kotidis: "Identifying
the Most Influential Data Objects with Reverse Top-k Queries"