Conceptual Modeling of Geographic Databases Emphasis on Relationships among Geographic Databases

1
Conceptual Modeling of Geographic Databases -
Emphasis on Relationships among Geographic
Databases

• Anders Friis-Christensen
• National Survey and Cadastre

2
Outline

• Geographic data
• Conceptual foundation
• Conceptual modeling
• What and why
• Modeling geographic data
• Modeling multiple representations
• Summary

3
Geographic data (1)

• Data where spatial and temporal aspects are
  important for the intended applications
• Applications
• The spatial aspects appear as regions, lines, and
  points, and changes occur discretely across time
• Topographic, cadastral, and network applications
• The continuously changing spatial aspects
• Environmental applications and location-based
  services

4
(No Transcript)
5
Geographic data (2)

• A geographic object has
• A spatial attribute value, which specifies the
  location in space
• Its data type can be, e.g., a polygon or a point
• Thematic attribute values, which specify thematic
  properties
• Any data type (non-spatial data types)
• Temporal attribute values, which specify temporal
  properties
• Valid time, vt, which specifies when something is
  valid. E.g., the color of a building was white
  from 1991-2000
• Existence time, et, which specifies when
  something exists. E.g., a building has existed
  from 1900-present
• Transaction time, tt, which specifies when
  something is being recorded as current in a
  database. E.g., in Aug. 2000 we record that John
  Doe was the owner of a building from Mar. 2000 to
  present. tt is (Aug, 2000, today), vt (Mar, 2000,
  today)

6
Geographic data (3)

• Associations among geographic objects
• Topological, e.g., that an address point is
  inside a building
• Metric, e.g., that a building is less than 10
  meters from a road
• Part-whole associations, e.g., that a county
  consist of several municipalities
• Constraints on geographic objects
• Constraints on objects. E.g., the size of an
  building may not be smaller than 25 sqm
• Constraints on associations. Eg., one object
  should be inside another
• Most model elements imply constraints, e.g., the
  data type of attributes

7
Conceptual Data Models (1)

• The idealization of the world to be described
• A way to organize and structure data
• Common conceptual data models can
• Support an Infrastructure for Geographic data,
  e.g., based on the work by ISO TC211
• Support reuse of solutions and designs (design
  patterns)
• Solve the problem of interoperability (exchange
  and querying of data)
• Give a clear overview and understanding of a
  given application (non-technical)
• Be used as a language between users, domain
  experts and developers
• Be modified and maintained easily
• Be used as a documenting tool

8
Conceptual Data Models (2)

• Conceptual data models describe
• Object classes and their properties
• Associations among object classes
• Possible constraints on objects and attribute
  values
• Conceptual data models are
• Independent of later implementation
• Conceptual modeling notations
• Entity Relationship Model (E/R)
• The Unified Modeling Language (UML)

9
Systems Development Context
Universe of Discourse
Requirements
Conceptual Data Modeling
Conceptual Schema
DBMS- independent
Data Model Mapping
DBMS- specific
Logical Schema
Physical Design
Internal Schema
10
Example

• An example in UML

11
Modeling Approaches (1)

• It is instructive to distinguish between two
  different approaches to providing better support
  for conceptual modeling of geographic data
• One approach is to extend the base notation
• Spatio-temporal concepts are given special syntax
• This makes for more compact diagrams
• The modeling notation becomes more complex
• Another approach is to not extend base notation
• With this approach, a library of generic diagrams
  is offered
• Patterns can be identified
• Diagrams remain complex
• Existing design and transformation tools are
  applicable

12
Modeling Approaches (2)

• One sub-approach of the extension approach is to
  introduce spatio-temporal annotations into the
  base notation
• In UML, which is extensible, stereotypes may be
  defined
• A two-step design process can been advocated for
  the annotation approach
• First, a diagram is designed that models the
  universe of discourse without taking into account
  the spatial and temporal aspects of the universe
  of discourse
• Second, the diagram resulting from the first step
  is annotated with spatio-temporal annotations

13
Modeling Geographic Data

• Example using extensions

14
Modeling Geographic Data (2)

• Underlying Spatial Model

Open GIS Consortium geometry model
15
Modeling Geographic Data (3)

• DBMSs (e.g., MySQL and Oracle) have adopted and
  implemented the OGC geometry model
• This means that we can map the spatial extension
  of the conceptual model directly to a logical and
  physical model
• No standardized model has been adopted for
  temporal aspects (several different
  implementations exist)

16
Outline

• Geographic data
• Conceptual modeling
• Modeling geographic data
• Modeling multiple representations
• What it is and requirements
• Conceptual modeling approach to integrate data
• Multiple representation schema language
• Summary

17
Multiple Representation of Geographic Entities

• Multiple representation is when several databases
  describe the same entity
• The reasons for multiple representation vary
• Different approaches in data collection
• Different application purposes / definitions
• Varying levels of detail (multi-scale)

18
Motivation

• Integration depends on
• Definitions/semantics
• Data models/structures
• Varying levels quality

• Integration of data
• May support new potential use of data for various
  analysis purposes (e.g., integration of register
  and map data)
• Can be used to rationalize the production of data

19
Example (Multi-Scale)
Topographic Map 150,000
20
Example (Multi-Scale)
Topographic Map 110,000
21
Example (Different Definitions)
22
Multiple Representation of Geographic Entities

• Problems today
• Inconsistencies may occur among these multiple
  representations
• Less concern is given to the fact that data
  change
• Maintaining spatial data is costly
• Benefits of a solution to handle MR
• Reduce the cost of maintaining multiple
  representations of an entity
• Ensure that users are working with updated
  representations
• Possible integration of data comming from
  different sources

23
Requirements

• An effective approach to the management of MR
  among legacy systems is needed
• Several requirements exist. We need to be able
  to
• Specify consistency, matching, and restoration
  rules
• Evaluate consistency rules with respect to a
  multiply represented entity
• Match r-objects located in different
  representation databases
• Monitor the representation databases for changes
• Restore consistency if inconsistency occurs
• Translate the specification into a database
  schema
• Keep the various legacy GISs autonomous

24
Multiple Representation Schema Language

• The MRSL is based on an Integration Class
  (i-class), which abstracts the entity represented

• Introducing an i-class has several advantages
• No change in representation databases
• A simple approach to describe complex
  correspondence scenarios
• A logical abstraction of an entity
• Consistency, matching, and restoration
  requirements can be expressed in a single class

25
Integration Class in MRSL

• The i-class is the main element in the MRSL and
  contains
• Attributes which are common for the r-classes
• Consistency, matching, and restoration rules
• Operations to restore consistency and match
  objects
• A new stereotype is defined ltlti-classgtgt, which
  specifies that a class
• Is defined within an integration database schema
• Is associated to at least two representation
  classes
• Should be identified uniquely
• Has extra specification compartments for VCs,
  matching rules, restoration rules

26
Consistency Rules

• The consistency rules consist of
• Object correspondences (OCs), which specify
  existence dependencies between the i-object and
  its r-objects
• Value correspondences (VCs), which specify value
  (attribute) dependencies between the i-object and
  its r-objects

27
Consistency Rules (OC)

• A new stereotype ltltmr-associationgtgt is defined
  for the OC
• It is always binary (either connects and i-class
  with a r-class or connects two i-classes)
• Its navigation is always from the i-class to the
  r-class
• It can be a master, i.e., the associated r-class
  controls the instances
• It is represented as a dash-dotted line

28
Consistency Rules (VC)
In UML the VCs are specified in a specification
compartment and as initial values of i- attributes
29
Matching Rules

• The matching rules specify how to find
  corresponding objects from different
  representation databases
• The following matching criteria can be used
• Attribute comparison, e.g., spatial
• Global object identifier
• Manual inspection

30
Matching Process
Create i-object
31
Restoration Rules

• The restoration rules specify the restoration
  actions that need to be applied when an OC or VC
  is not satisfied
• The restoration rules follow the principle of
  event-condition-action (ECA) rules

32
Restoration Process
33
Restoration Rules
Example Restoration Rules rr1 on insert
tm then insert br immediate, rr2 on
update tm.shape if not v1 then placeInside(br.loc
ation)immediate
v1 br.location inside shape
34
Complete Schema
35
Summary

• An overview of geographic data
• Different conceptual modeling approaches
• Standardized conceptual models support
• Standardized logical models and implementations
• Integration and exchange of data
• Extensions can be used to satisfy those
  requirements posed by
• Geographic data in general
• Multiple representations of geographic entities
• Extensions are solutions to capture special
  semantics

36
Thank you for your attention!

• ???Questions???