MADS Space, time and context modeling for geographic applications Stefano Spaccapietra http:lbd'epfl

1
MADSSpace, time and context modeling for
geographic applications Stefano
Spaccapietrahttp//lbd.epfl.ch
2
Geographical Applications and CS

• Information Storage and Retrieval
• Spatio-Temporal Databases
• GIS and Extended DMS
• Decision Support
• Spatio-Temporal Data Warehousing
• Data Mining, OLAP
• Interoperability - Exchange
• Ontologies

3
ST-DW
OLAP Server   

Data Acquisition
Information Extraction
External Sources
OLAP
Data Intégration
Data Querying And Analysis
Data Warehouse
Data Mining
Internal Sources
Monitoring Administration
Building Maintenance
4
Ontologies

• A means to share the meaning of data
• between humans
• between humans and computerized agents
• An explicit specification of a
  conceptualization" (Gruber)
• a shared agreed vocabulary for concepts in a
  domain
• A formal description of the meaning of those
  concepts
• Formalism / Language
• OWL
• DL Description Logic, a decidable subsets of
  First Order Logic

5
Work Done/Ongoing at EPFL-LBD

• ST-DB define a conceptual data model, MADS
• ST-DW
• Methodology for Data Integration
• Directly applicable to heterogeneous data sets
• Currently EU GeoPKDD project (2006-2008)
• Privacy-aware data mining and knowledge
  extraction
• Based on a Trajectory DB - DW
• Ontologies
• Specify a hybrid DB-DL approach
• ST ontologies

6
MADS today

• A data model for geographic applications
• space
• time
• multiple representations
• contexts,scale, viewpoints,
• A conceptual data model
• for designers
• based on Extended Entity-Relationship constructs
• A set of tools implemented on top of existing
  DBMS and GIS systems
• Oracle 9i
• MapInfo
• ArcView

7
Some applications designed with MADS

• Water resource management of the state of Vaud
  (CH)
• Electricity network (CH)
• Study of porcupines' habits in Tuscany (I)
• A multiscale cartographic database for the French
  national mapping agency, IGN (F)
• Spatio-temporal database for analysis of natural
  risks in mountainous areas (F)
• The networks of pipes for clean and used water in
  Geneva (CH)
• Study of the evolution of the course of the
  Sarine river (CH)
• Fire department database in Besançon (F)
• Management of the network of oil pipelines in
  Columbia
• ...

8
MADS Tools

• Goal Provide users with a conceptual front-end
  to existing systems
• Easy-to-use graphical tools for contextual
  spatio-temporal databases
• To visually design the database schema
• To visually query the database
• MADS tools are downloadable at
         http//cs.ulb.ac.be/mads_tools/
• These tools are research prototypes

9
MADSSchemaEditor
10
(No Transcript)
11
MADS Query Editor
12
MADS structural dimension
A MADS Schema
Extended ER
13
MADS Supports Complex Objects
14
Relationships
Buys
Customer
Product
Supplier
15
Multi-Instantiation Hierarchies
overlapping
Transport Means
Movie Accessory
disjoint
disjoint
Vehicle
Bicycle
Plane
overlapping
Collectible
Car
16
Continuous and Discrete View (Space)

• Continuous view
• captured as, e.g. a grid of cells (one value per
  cell)
• Discrete view
• captured as points, lines (series of points), or
  areas (bounded by lines)
• Real world

17
Space and attributes

• A RELATIONAL IMPLEMENTATION
• Lake (Lname, geometry, km2)
• Island (Lname, island, area)
• Depth (Lname, point, depthValue)
• Harbour (Lname, harbourName, location,
  capacity)

18
Time and attributes
19
Lifecycles

• Different possible kinds of lifecycle
• a time interval, e.g. Person
• an instant, e.g. FallingStar
• a set of intervals
• gt active and suspended states
• e.g. a faculty on temporary leave for a
  sabbatical
• The set of possible states and their properties
  are application dependent

20
Synchronization Topological Relationships

• A synchronization and topological relationship
• It constrains the lifecycles and geometries of
  the linked objects

21
Spatio-temporal concepts

• Moving and deforming objects time-varying
  spatial attributes
• thanks to orthogonality
• e.g. moving car
• function Time --gt Point
• e.g. moving and deforming hurricane
• function Time --gt Area
• Space and time-varying attributes
• Varying data type for one or several dimensions
• e.g. landOccupation varying on time and space
• function (spatial domain X time domain) --gt
  range of values
• Implementationset of (point, instant, value)

22
Multi-Perception Multi-Representation
23
Multiple Viewpoints
24
Multiple Resolution
less precise
more precise
ltN11947, Smith, squaregt
land occupation cultivated area, forest,
built-up area
ltN11947, 1987, Smith, 600m2, 2 floors, polygongt
land occupation orchard, vineyard, cereal,
field, forest
25
Multiple Geometry
River described as an area or as a line
more precise resolution
less precise resolution
26
Hierarchical value domains

• Describe the same property at different
  abstraction levels
• Hierarchical value domains for attributes
• (similar to classification hierarchies for
  objects)

land occupation
cultivated area forest
orchard
cereal
oleaginous
vineyard
pear
citrus
apple
corn
barley
colza
sunflower
27
MADS support of multiple perceptions

• Identifying the perceptions/contexts
• e.g. a geographical database for risk assessment
• 3 resolutions 1/1000, 1/5000, 1/25000
• 3 viewpoints Technician, Management, Public
• 9 contexts gt  T1,T5,T25, M1,M5,M25, P1,P5,P25
• Associate to each schema and database element the
  context ids of the contexts it belongs to
• Each real world object or link may have one or
  several representations (zero, one or more per
  context)
• Elements may be shared by different contexts
• Transactions may run using a single context or
  using multiple contexts

28
Modeling Alternatives

• Designing a multi-context database
• Integrated modeling
• all representations merged in oneobject (or
  relationship) type
• Separate modeling
• each representation one object type
• Inter-representation link
• Using a multi-context database
• Open as single-context gt normal database
• Open as multi-context gt a new kind of database
• new consistency rules

29
Integrated Modeling
Perceptions
30
Integrated Instances
Oid 37364
Road
number (11) integer name route viticole dpt
21 type Local
number 675
This part created by a transaction running in a
blue or bluered context
This part created by a transaction running in a
red or bluered context
The whole instance may be created in a single
step by a transaction running in a bluered
context
31
Separate modeling
01
01
inter-representation semantics
32
Open a MPR database
Open with red and blue perceptions
Road
all instances
number (11) integer name (11) string
f(P) admClass (11) integer dpt (11)
integer type (11) enum (local,region ,state)
(11) string
Open with red perception
Open with blue perception
all blue, redblue instances
all red, redblue instances
33
The next steps

• Using spatio-temporal data for decision support
• Spatio-temporal data warehousing
• Trajectory data warehousing
• Using support of multiple perceptions
• Personalized information services
• Context-aware information services
• "Intelligent" Location-Based Services
• Interoperability
• Semantics in data exchanges
• Ontologies for spatio-temporal applications
• Spatio-temporal ontologies

34

• Thanks for your attention

More info on         http//cs.ulb.ac.be/mad
s_tools/ And in the book
eBZ 2006