kartographie.geo.tu-dresden.de€¦ · Title: untitled Created Date: 8/21/2004 8:05:57 PM

ICA Workshop on Generalisation and Multiple representation – 20-21 August 2004 – Leicester

Multi-representation in spatial databases using the MADS conceptual

model

Christelle VangenotEPFL

Database laboratory

MADS is a conceptual model ...

ER conceptual

conceptual schema

USERS

OO DB

relationalDB

logical schema

ER model

Thematic

spatial attribute

spatial object typeRiver

name reservoir

R# name location

tributary

main

Flows

spatial attribute

topological semantics

temperature

space varying attribute

f( )

+ Space + Time15 12 17

19 17

life cycle

quality f( )

time varying attribute

synchronisation semantics

SmallBig

Objectives

Add Multi-representation facilities in the MADS conceptual model

Considering the following facets:Resolution

level of detail for spatial and thematic data

Viewpointuser perception of real world

Multi-Representation Modeling

Two strategies:

Multi-representation strategyIntegrate representations in a single data structure

Inter-representation strategyLink representations with a link with an inter-representation semantics

Stamping (Builder, 10)(Risk manager, 50)

Road

number: (1:1), integername: (1:1), stringadm.class.: (1:1), integerdpt: (1:1), integertype: (1:1), string

integer

ģ

Multi-representation strategy

number: (1:1), integername: (1:1), stringadm.class.: (1:1), integertype: (1:1), string

Road £ Road ƒnumber: (1:1), integername: (1:1), stringdpt: (1:1), integertype: (1:1), integer

Stamps(Builder, 10)(Risk manager, 50)

f( . )Route NapoléonRN17

Multi-perception relationship

Multi-perception relationship “Adjacent”:Adjacency of buildings and roads is described for red and blue stampsAdjacent relationship may contain different sets of attributes according to stamps

RoadBuilding £ £adjacent1:n 1:n

Multi-perception relationship (2)

Mono-perception relationship “Adjacent”:Adjacency of buildings and roads is described only for

Transactions having only the stamp do not see which instances of Building are linked to which instances of Road

RoadBuilding £ £1:n 1:n

rolescardinalities

1:n adjacent

Multi-perception hierarchy

Tourist Site

Name: (1,1) StrDistrict: (1,1) StrDistrict: (1,n) Int

£

PublicPrivate

Owner: (1,1) Str

MonumentMuseum

Entrance: (1,1) ¢Exhibition: (0,n) Str f(§)OpenTime: (1,n) §Description: (1,1) Str

Multi-perceptionhierarchy TouristSite

Name: (1,1) StrDistrict: (1,1) StrDistrict: (1,n) Int

£

PublicPrivate

Owner: (1,1) Str

MonumentMuseum

Entrance: (1,1) ¢Exhibition: (0,n) Str f(§)OpenTime: (1,n) §Description: (1,1) Str

Multi-perceptionhierarchy TouristSite

Name: (1,1) StrDistrict: (1,1) StrDistrict: (1,n) Int

£

PublicPrivate

Owner: (1,1) Str

MonumentMuseum

Entrance: (1,1) ¢Exhibition: (0,n) Str f(§)OpenTime: (1,n) §Description: (1,1) Str

Inter-representation strategy

Relationship with a inter-representation

semantics

number: (1:1), integername: (1:1), stringadm.class.: (1:1), integertype: (1:1), integer

Road £ Road ƒnumber: (1:1), integername: (1:1), stringdpt: (1:1), integertype: (1:1), string

Road £number: (1:1), integername: (1:1), stringadm.class.: (1:1), integertype: (1:1), integer

Road ƒ

number: (1:1), integername: (1:1), stringdpt: (1:1), integertype: (1:1), string

Road ƒRoad £

Inter-representation strategy

Several kinds of correspondences:

Adding a inter-representation semantics on relationships:Binary relationship 1:1 Aggregation 1:nMulti-associations n:m

1:n

n:m1:1

1:1 correspondence

Binary relationship with inter-representation semanticsLinks 2 objects representing the same phenomenon in different perceptions

correspond⇔

quality

0:1£Building Monument£0:1

number: (1:1), integertype: (1:1), string

period: (1:1), integername: (1:1), stringconstruction date (1:1), §

1:n correspondence

Aggregation (intrinsic inter-representation semantics)

1:n

0:1

composed¹

Road �

ƒRoad section

n:m correspondence

Multi-Association with inter-representation semantics

( {b1,b2,b3,b4,b5} , {ba,bb,bc} )

£Building1 Building2 £0:1 1:1correspond

⇔1:n 1:n

Linked Instances are two perceptions of the same real world phenomenon

Implementation

MurMur Project (ULB, IGN, Cemagref, Star)Software on top of a GIS (Oracle)

Schema editor MADS with multi-representation capabilitiesQuery editor multi-representationViewer

DB

conceptuallevel

logicallevel

schema editor query editor viewer

Future work

Consistency, derivation between representationsStating constraints between multiple-representationsDerivation rules

Multi-representation in the continuous view of spaceSpace-varying attribute referring to a spatial attribute defined at various spatial resolutions Space varying attribute for which the same value may be defined at different semantic resolution

Describing the correspondences between the continuous and discrete representation of space.

ICA Workshop on Generalisation and Multiple representation – 20-21 August 2004 – Leicester

Thank you