385

Document Sample
385 Powered By Docstoc
					     Mariage de la géomatique et de
      l'informatique décisionnelle:
     le programme de recherche de
la nouvelle Chaire industrielle CRSNG en
     bases de données géospatiales
              décisionnelles
               Prof. Yvan Bédard, PhD
                    titulaire de la chaire




                  10 février 2005
                         CRG
Contenu de la présentation
•   Problématique
•   Objectif global
•   Approche de recherche
•   Objectifs spécifiques
•   Méthode de recherche
•   Collaborations
•   Résultats prévus
Problem Domain
• Canadian organisations invest hundreds of millions of
  dollars annually to acquire large amounts of data
  about the land, its resources and uses.
• These data however prove difficult to use by
  managers who need:
   –   aggregated information
   –   space-time comparisons
   –   fast synthesis
   –   trends discovery
   –   correlation analysis
   –   unexpected queries
   –   interactive exploration
   –   etc
Problem Domain
• GIS data have a transactional nature
  – Transactional databases are oriented
    towards data:
    •   acquisition
    •   storing
    •   updating
    •   integrity checking
    •   minimal querying, with the help of an expert


    Ex. Normalized relational databases
Problem Domain
• Decision-makers need data with an « analytical » nature
    –   Decision-support databases are oriented towards data:
    –   aggregation
    –   summarization
    –   with several levels of granularity in time, space and themes

• Decision-support databases provide fast and intuitive ways of:
    –   navigating
    –   exploring
    –   querying
    –   intersecting instantly whatever dimensions at whatever levels of
        granularity without the help of an expert.

        Ex. MOLAP database (Multidimensional On-Line Analytical
        Processing)
  Problem Domain
Transactional              Decision-support
  Database                 Database
• Built for transactions   • Built for analysis, decisions
• Original source          • Copy or read-only data
• Detailed data            • Aggregated/summary data
• Current data only        • Current + historic data
• Normalized data          • Denormalized, redundant
  structure                  data structure
• run on RDBMS,            • run on Warehousing
  GIS, web servers,          engine, MOLAP server,
  CAD ...                    ROLAP engine ...
Problem Domain
• Several categories of Decision-Support
  Systems exist
        High
                                            OLAP
   Level of
 interactivity
                                            EIS
                  Multicriteria Querying/
                      ES        Reporting

         Low     Data Mining

                 Slow                             Fast   Response time

• Most are built on transactional databases and
  do not support the kind of interactivity,
  intuitiveness and navigability we are looking for
Problem Domain
                                           Transactional   Decisional
                                               (GIS)       (SOLAP)
• Analysis 1
                                                Easy         Easy
  geospatial - 1 theme - 1 epoch
  (detailed data)                               Fast         Fast



• Analysis 2
  geospatial - many themes - 1 epoch          Difficult      Easy
  (detailed data)                              Slow           Fast


• Analysis 3
  geospatial - many themes - many epochs Very                 Easy
  (aggregated data)                     difficult if        Seconds
                                               not
                                            impossible
                                              Days
Problem Domain
  • Non-spatial solutions (BI) are:
     – too limited
     – not efficient enough for:
        •   aggregation
        •   processing
        •   analysis
        •   querying of geospatial data
     They were not built with "geospace" in mind...
        • These solutions don’t harness the full power of:
              – geospatial data
              – spatio-temporal analysis
              – cartographic visualization
  • Today’s commercial solutions are limiting
    our ability to better support spatial decision-
    making
      Research Program Objective
       Further develop geomatics données sources pour les cubes
#3: méthode + outils d'évaluation et sélection desconcepts, methods
           and tools de gestion to the building of
#4: intégration d'un volet related des ontologies dans ISTory geospatial
           databases les cubes multidimensionnels géospatiaux
#6: méthodes pour peupler for decision-support
#9: intégration de métadonnées agrégatives dans les cubes
#7: améliorer + fonctions décisionnelles SOLAP
#12: méthode lesoutil de mise à jour en temps réel des données descriptives des cubes
                             axes des données
       • 4 researchweb jour based on team’s current
#15: méthode + outil de mise à géodécisionnels géospatiales des cubes
#8: prototypes de services
           expertise:
#18: projet intégrateur pour tester la suite de méthodes et outils développés (à définir)
#10: technologie décisionnelle mobile (ex. sur système nomade)
#21: méthode betterd'agrégationréel geospatial identifier)
            1. outil en temps spatiale des données pour peupler
#19: prototype+SOLAPstructuring (application à databases les cubes
#1: concevoir facilitating the et l'ontologie unifiée du projet (ISTory)
            2. le géodécisionnelle et LBS
#20: technologiecorpus théoriqueautomatic aggregation of geospatial
#22: nouvellesdata into decision-support information
                un environnement de spatio-temporels géodécisionnels (ex.
#2: développer analyses et opérateursdéfinition des besoins géodécisionnels géoMDX)
#23: interface better assessing the quality of the obtained agrégée
#5: enrichir3. à l'usagerpour modéliser les processus de production d'info.
             Perceptory optimal pour SOLAP
#24: fonctionsinformation fins géodécisionnelles
                matricielles pour
#13: créer méthode + outil modélisation données multidimensionnelles géospatiales
            4. developing new qualité des cubes a innovate
#11: développer des pour évaluer latechnologies orposteriori with
#14: méthode + outilméthodes d'optimisation des cubes de données géospatiales
#16: méthode existing ones qualité a priori (contraintes intégrité S-T agrégées)
             + outil d'assurance
#17: méthode + outil pour ajuster la qualité de l'info. agrégée aux cubes évolutifs
Research Program Objective
Our Vision

                                                   GeoDecisional
                                                   information
   Real time        Today’s                        anytime, anywhere
   and wireless    Mobile GIS
                                      Target

                                     Emerging
                     Today’s       technologies
   Delayed
   and wired          GIS          (Spatial data
                  (Web and LAN)       mining
                                    and SOLAP)

                   Transactional    Analytical
                     Database       Database
   Approach: Integrate Knowledge
                                from different corpus
We suggest to bring together geomatics and BI to build new
capabilities for Geospatial Decision-Support, in an IT context
and following robust software engineering concepts
                                   Geomatics
                             (GIS, spatio-temporal
                        data aggregation, web mapping)




                                  Geospatial
                                  Databases
                                     for            Information Technologies
         Software engineering                     (internet, standards, mobile)
                                   Decision
                                   Support



                             Business Intelligence
                               (Decision support,
                          Multidimensional databases)
  Approach
BI offers a strong foundation to build the data rich environment
needed for Decision-support



      Geospatial nature of data


                                        GDW
               Geospatial         GIS   GDM
                                       SOLAP
                                        DW
           Non-geospatial         DBMS  DM
                                       OLAP         Decisional nature of data
                            Not        Aggregated
                            aggregated
Approach
                                              Chair
                                             Research
         GIS




                                             Aggregate
                            Aggregate
   GIS
                                                                          OLAP

         GIS
                  Copy                  DW
                Integrate                                DM



   •         Challenges
         –     Different concepts and technologies
         –     BI community <> GIS community
         –     Merging requires knowledge in:
               •    Geomatics       • GIS                •    transactional databases
               •    analytical databases                 •    IT
               + Vision
    Approach
Team up with partners:
•Complementarity
                                   Kheops     RDDC
•Actual or past collaborations
                                                    HQ
                 Syntell                            MTQ


              Intélec

                                                    NRCan
              Holonics


                                 DVP        Laval
Research Method
•    Typically:
    1. Problem definition (including literature review)
    2. Define objectives and research method
    3. Perform research
       1. theoretical developments
       2. conceptual modeling (cf. extended UML)
       3. prototyping (cf. Agile methods)
    4. Tests and iteratively improve the results
    5. Documentation (thesis, papers, technical report, etc.)
•    Is adapted to each project and circumstances
•    Strong interaction with partners and
     collaborators
Co-directions financially
supported by the Chair
                     Sylvie
                     Daniel

         Thierry                 Mir
         Badard                Mostafavi




   CEMAGREF
                   Chair               Marc
                   _________          Gervais
                   partners


                                Michel
        Stéphane                Fortin +
         Roche                 Jacynthe
                      Jean      Pouliot
                    Brodeur
                    (RNCan)




                                                Italic = planned
Expected Results: for Science
   New knowledge, concepts, methods, and
   highly qualified personnel to:
   – better design geospatial databases for decision
     making
   – aggregate data more easily
   – build better tools to explore
     and analyze geospatial data
   – better understand the
     value of the results
Expected Results: for Science
• Long-term commitment:
  –   Better planning of projects
  –   Increased cohesion among projects and concepts
  –   Richer collaborations
  –   Sustainable interaction with partners
• Unique laboratory
  – More possibilities for testing
• More resources for:
  – diffusion of knowledge
  – technology transfer
Expected Results: for Partners
  • Working prototypes of geomatics tools
  • Functional prototypes of decision-support
    applications
Expected Results: for Canada
• Researchers undertaking an important
  research project with the geomatics industry
• Generic technology benefiting:
  – our economy (resources, energy, transportation,
    etc)
  – our quality of life (environment, defence, public
    health, etc)

  Organizations will benefit from new ways to
  « unlock » their geospatial data for decision-
  making and get better ROI
Questions ?




          From local data to global knowledge

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:2
posted:9/7/2011
language:French
pages:22