Long Term Data Base

1
Long Term Data Base Data Navigator II

• ESPON SEMINAR
• 14-15 November 2006
• Dipoli Conference Centre, Espoo, Finland
• Joël Boulier, Claude GraslandLaboratoire
  Géographie-cités (Paris) HYPERCARTE Research
  Group
• Marc Guerrien, Nicolas Lambert
• CNRS UMS RIATE (Paris) HYPERCARTE Research
  Group
• Jérôme Gensel, Bogdan Moisuc, Marlène
  Villanova-Oliver
• Laboratoire LSR-IMAG (Grenoble) HYPERCARTE
  Research Group

2
Context

• ESPON 3.2 Programme (European Commission
  2002-2006)
• Long Term (ESPON) DataBase
• Provide scenario builders quantitative inputs on
  selected topics at regional level for years 1980,
  2000 and 2020
• Try to establish a sustainable framework for the
  ESPON Database in the future ESPON II, taking
  into account various problems encountered
  (missing values, changing territorial units,
  etc.)
• HYPERCARTE Research Group
• 2 Research Labs in Geography Géographie-Cités
  and UMS RIATE
• 2 Research Labs in Computer Science LSR-IMAG and
  ID-IMAG
• Goals Advanced Methods and Tools for Spatial
  Analysis
• Involved in ESPON 3.1, 3.4.3 and 3.2
• Software
• Available HyperAtlas (application ESPON
  HyperAtlas ESPON 3.1)
• Soon to come HyperAdmin, HyperSmooth, LTDB

3
LTDB Objectives

• Objective 1
• Provide a framework for long-term storage of
  thematic and geometric data for the territorial
  units composing a given area, at different levels
  
• This implies tackling several issues
• Evolutivity ? rely on a flexible schema
• Data quality ? keep trace of the quality of the
  data
• Usability ? make it usable by other people than
  its designers, possibly as a shared resource
• Objective 2
• Provide a framework for a reliable estimation of
  missing indicator values
• To fill-up informational gaps
• To simulate past or/and future hypothetical
  situations
• This implies designing several components
• A set of generalized estimation methods
• A set of generalized estimation strategies
• A mechanism for evaluating the quality of the
  estimated data

4
ESTImate

• Postulate All statistical information managed by
  the LTDB can be described according to four
  dimensions E, S, T and I
• (E)space the spatial unit to which the
  statistical information is attached
• (S)ource the statistical institute which has
  produced the information
• (T)ime a period or an instant which dates the
  information
• (I)ndicator a thematic definition of the
  variable
• And then come more general problems
• Instability of the administrative structures
• The name and/or the borderline of E can have
  changed someday
• W. Germany E. Germany ? Germany
• Czechoslovakia ? Czech Republic Slovakia
• Côtes du Nord ? Côtes d'Armor
• Isère 1960 ? Isère 2006 and Rhône 1960 ? Rhône
  2006
• Heterogeneity of the sources
• The source S does not provide any value for the
  given (E,T,I)
• Missing values
• At time T, no value for E and I whatever S
• How to Cope with Reality?

5
LTDB General Architecture
6
LTDB Architecture Components

• Geographic Ontology a gazetteer containing names
  of geographic entities and some relations between
  them
• Indicator Ontology a classification hierarchy of
  the themes and indicators with some relations
  between them (aggregation, broader term, etc.)
• Indicator Formulae Knowledge Base a set of
  mathematical rules for calculating new indicators
  using existing ones
• Method Hierarchy a classification of estimation
  methods
• Estimation Strategy Knowledge Base a set of
  rules allowing the system to choose the most
  appropriate estimation method in a given
  situation
• Spatio-Temporal Database a relational database
  containing the whole set of geographic entities
  with their known indicator values

7
LTDB Schema
8
Estimation Methods

• E, S, T and I define a hypercube of information
  with holes (missing values)
• We need ESTImation methods
• To fill up missing values of the past
• To predict future values
• So far, (simple) ESTImations methods have been
  proposed
• Estimation methods based on one-dimension E, S,T
  or I
• Estimation methods based two (or more)
  dimensions ES, SI, ET
• The Method Hierarchy and the Estimation Strategy
  Knowledge Base will be designed to extend this
  set of methods

9
LTDB First and Future Developments

• A first prototype has been developed
• Implementation of the database schema in the open
  source POSTGRES DBMS
• Data acquisition mechanisms in Java
• The LTDB framework imports and exports data files
  in various formats (excel, dbf...)

10
LTDB First and Future Developments

• Short Term
• Estimation methods hierarchy using AROM (an
  Object-Based Knowledge Representation System)
• Indicator formulae knowledge base in AROM
• Estimation strategy knowledge base with AROMTasks
• Test and validation through an incremental
  approach start with an example with a small set
  of indicators and some estimation methods, adding
  more indicators later
• Mid term
• LTDB as a research project of the HYPERCARTE
  Research Group, will be integrated into the
  HyperAtlas and HyperAdmin software
• In the case of the ESPON HyperAtlas, this will
  allow the visualization (by simply moving a
  cursor on a time line, for instance) of the
  evolution of the ratio of two indicators through
  past, present, but also future time
• Long term
• in order to perform simulations that validate
  different scenarios, LTDB will integrate
  estimation methods relying on different
  parameters which convey tendencies, hypothesis
  and assumptions corresponding to these scenarios

11
Data Navigator II

• General objective produce a handbook on data
  acquisition and harmonization, with a focus on
  the themes investigated by the ESPON Program
• Applied research
• This project can be seen as an application and a
  validation test for the LTDB structure through
  European Databases (ESPON DB,)

12
Work in Progress

• Three workpackages have been determined
• WP1 Use and practices of data collection in
  ESPON I
• Short survey on practices of some TPG's (IGEAT)
• Problem of national data collections (TIGRIS)
• WP2 Choice of data model for ESPON II
• Practical example of data integration between
  environmental and socio economic data
  (Géographie-cités LSR-IMAG)
• Choice of the best solution for data modeling and
  data integration (Géographie-cités LSR-IMAG)
• WP3 Handbook for data collection
• Practical rules for harmonization (time and
  space, thematic harmonization)
• Practical rules for the use of national sources
• Recommendations for ESPON II one or two
  databases

13
Integration of environmental and socio economic
data

• Question How many m2 of
• forest are accessible
• for a European citizen?

NUTS23_99 CLC00_forest
14
Integration of environmental and socio economic
data
Forest area per inhabitant in 2000
15
Integration of environmental and socio economic
data
Potential of forest area per inhabitant within
a 10 km radius in 2000
16
How far do we get?

• ESPON Database in its current structure is a
  repository for a huge set of European indicators
• Long Term Database relies on a structured schema
  designed for the import of different kinds of
  indicators (different sources, different grids,
  different census times,)
• Two different approaches (from philosophical
  technical points of view)
• During the development of LTDB, we have
  experienced that to extract and import some
  indicators from a data source is not a trivial
  task (ESPON Database included)

17
Coupling LTDB and ESPON Database?

• The acquisition process (values from the ESPON
  Database) has to be automated
• A wrapper dedicated to the ESPON Data Base
  which will enable the import of incomplete sets
  of indicators from the ESPON Database into LTDB
• This wrapper exploits a meta description of the
  schema of the imported source of data (in this
  case ESPON Database)
• To update the LTDB with ESPON data, a complete
  description of the structuration of the ESPON
  Database is needed
• A cooperation between authors of LTDB and ESPON
  Databases will be beneficial for both tools in
  the future

18
Long Term Data Base Data Navigator

• Thank You for Your Attention
• Questions?
• Joël Boulier, Claude Graslandjoel.boulier,
  claude.grasland_at_parisgeo.cnrs.fr
• Marc Guerrien, Nicolas Lambert
• marc.guerrien, nicolas.lambert_at_ums-riate.org
• Jérôme Gensel, Bogdan Moisuc, Marlène
  Villanova-Oliver
• jerome.gensel, bogdan.moisuc, marlene.villanova_at_
  imag.fr

19
LTDB Schema
Temporal name
Temporal object with an internal identifier
Proximity/similarity measure matrix
Temporal value of some indicator for some GU
Indicators
Temporal code system
Reliability of the source
Temporal spatial representation
Temporal splitting and merging of GUs
Composition relation between GUs, depends on the
hierarchy
Code depending on a code system
Database or process genealogy of a value
Temporal hierarchical organization of GUs
Providing organism
20
Open Questions

• Evolution of the conceptual model of the LTDB?
• Terminology (semantic units, spatial inclusion or
  aggregation, )
• Partial space inclusion what does it mean?
• Data importation from different sources?
• Data from different sources will be imported
• Needs an interface for facilitating the
  importation of heterogeneous data (tools
  developped for HyperAdmin could serve)
• Sources?
• ESPON database (BBR)
• Corine Land Cover
• United Nations