Multidimensional Databases

1
Multidimensional Databases

• Challenge representation for efficient storage,
  indexing querying
• Examples (time-series, images)
• New multidimensional data sets approaches
• Graphs (e.g., road networks)
• Immersidata (e.g., haptic)
• User profiles aggregation/clustering

2
Challenges

• Storing multidimensional data (matrix vs.
  relations)
• Indexing multidimensional data (R-tree)
• Queries
• Search for similar objects (similarity
  search)ICDE00,ICME00
• Spatial and temporal queries IDEAS00,ACM-GIS01,
  KAIS02
• Multidimensional data mining
• Aggregation EDBT02,PODS02
• ClusteringACM-MMj02
• Classification INFORMS02
• Finding outliers SSDBM01

3
Stock Prices
S1
Sn
4
More Similarity Search Clustering
C
Shapes ICDE99 ICME00
5
On-Line Analytical Processing (OLAP)
Market-Relation

• Multidimensional data sets
• Dimension attributes (e.g., Store, Product, Date)
  
• Measure attributes (e.g., Sale, Price)
• Range-sum queries
• Average sale of shoes in CA in 2001
• Number of jackets sold in Seattle in Sep. 2001
• Tougher queries
• Covariance of sale and price of jackets in CA in
  2001 (correlation)
• Variance of price of jackets in 2001 in Seattle

Store Location
Date
Sale
Product
Price
LA Shoes Jan. 01 21,500 85.99
NY Jacket June 01 28,700 45.99
. . .
. . .
. . .
. . .
. . .
Avg (sale)
s(d ltingt 2001)
Too Slow!
s(s ltingt CA)
s(pshoe)
Market-Relation
6
Example Solution (Pre-computation) Prefix-sum
Agrawal et. al 1997
Salary
Age Salary
100k
120k
150k
40k
55k
65k

• 50k
• 55k
• 58k
• 100k
• 130k
• 57 120k

0
25
40
Age
50
60

• Issues
• Measure attribute should be pre-selected
• Aggregation function should be pre-selected
  (sum or count)
• Updates are expensive (need re-computation)

80
Result I II III IV
7
Spatial Temporal Data
Complex Queries
ACM-GIS01, VLDB01

• Data types
• A point ltlatitude, longitude, altitudegt or ltx,
  y, zgt
• A line-segment ltx1, y1, x2, y2gt
• A line sequence of line-segments
• A region A closed set of lines
• Moving point ltx, y, tgt (e.g., car, train, )
• Changing region ltregion, value, tgt (e.g.,
  changing temperature of a county)

• Queries
• Rivers ltintersectgt Countries
• Hospitals ltingt Cities
• Taxi ltwithingt 5km of Home
• ltin the nextgt 10 min
• Experiments ltoverlapgt BrainR

Visual99
8
Spatial Temporal Data Queries

• Data types
• A point ltlatitude, longitude, altitudegt or ltx,
  y, zgt
• A line-segment ltx1, y1, x2, y2gt
• A line sequence of line-segments
• A region A closed set of lines
• Moving point ltx, y, tgt (e.g., objects, car,
  train, )
• Queries
• Molecules ltintersectgt Microbes
• Train-stations ltingt Cities
• Round objects ltwithingt 5cm of Hand ltin the nextgt
  10 s
• Number of distractions in ltsouth-eastgt of subject

9
Spatial Temporal Data Queries

• K Nearest Neighbor queries find the k nearest
  objects to a query point (5 closest hospitals to
  my car)

10
Immersidata and Mining Queries
CIKM01, UACHI01
11
Immersidata and Mining Queries
A dynamic sign, e.g., ASL colors
Subject-1
12
User Profiles Clustering Offline Processes
PPED Similarity Measure and Clustering
Favorite Features (Rock High Classical Low Po
p Low Rap High)
Voting
13
User Profiles Clustering Online Processes
Current Users Profile