Transcript vanderWerf 06 04 SERONTO

SERONTO
A Socio-Ecological
Research and
Observation Ontology
Bert van der Werf
Mihai Adamescu
Nicolas Bertrand
Hugues Boussard
Toon van Daele
Sabina Datcu
Mark Frenzel
Volker Hammen
Helena Karasti
Pirjo Kuitunen
Juraj Lieskovsky
Johannes Peterseil
Herbert Schentz
Liisa Tuominen
Minu Ayromlou
Jakub Borovec
Constatin Cazacu
Miklos Kertesz
Mandy Lane
Barbara Magagna
Sue Rennie
Katharina Schleidt
ALTER-Net
A Long Term Biodiversity, Ecosystem,
and Awareness Research Network
funded by the European Union. ALTERNet addresses major biodiversity issues
at a European scale.
The network is a partnership of 24
organizations from 17 European
countries
Work Package I6
A framework for effective information and
knowledge management
Objective: To construct a framework within
which can be built a system to manage
biodiversity data, information and knowledge
from the Network of Excellence, and to make
them available to scientists, policy makers and
the public.
Biodiversity data
• Different domains
– ecological, meteorological, chemical, geographical,
sociological, economic, etc.
• Different languages
– real language differences; scientific language differences
• Different scaling
– Global, regional, site; surveys, ...
• Different purpose
– monitoring versus experiments; exploration versus
parameter fitting
• Different storage methods
– spreadsheets, databases; local or distributed
Purpose of framework
• data conservation
• data sharing
• data quality
– provenance (ISO ...)
– scientific
• data integration
• data mining
• analysis and meta analysis
– graphically
– advanced techniques
• workflows
Components for analysis
• Dependent variables
– measurements/observations
• Independent variables
–
–
–
–
treatments
classifications in stratified sampling
measurements
often time
• Sources of variation
–
–
–
–
–
sampling structure
observer
method
design (blockings)
etc.
• Measurement scale
– distributional assumptions
– preferred analysis
• Assumptions
Aspects important for combining
data from different sources
• methods used
– relationship between methods
• units and dimensions
– conversion between units
• the entity of observation/treatment
– how to group them
– how are they selected
Approach for Framework
• Distributed databases
• Ontologies
– semantics for concepts
– hierarchical structure of concepts
– relationships/properties
– reasoning and restriction logic
– mappings and derivation
SERONTO
• Core ontology
Domain I
– observation ontology
– units and dimensions
ontology
• Domain ontologies
–
–
–
–
–
taxonomy
ecology
chemistry
sociology
etc.
Core
Domain II
Domain III
Designing the core
•
•
Repeatability:
The ontology should be capable of holding
enough meta-data that another person can
repeat the experiment or observation at
another place and time.
Transparency:
It must be possible to record and retrieve
meta-data describing what actually
happened.
Core contains:
• Structures for domain knowledge
– Base classes for derivation
– Hierarchical reference lists
• Data and meta data for observations
– What, When, Where, Who, How?
– What went wrong?
• Meta data for administration
– Project information
• Units and dimensions
• Versioning methods
Core classes:
SERONTO: basic classes
selection_
description
method
hasSample
hasMethod
physical_
thing
parameter
hasParameter
parameter_
method
hasInvestigationItem
hasValue
value_float
value_float
hasValue
value_
value_nomina
nominal
l
hasScale
hasUnit
hasParameterMethod
value_set
scale
unit
basics: example
Selection_1
Triangulation_
Method
hasSample
Real; interval;
[0,infinity)
Height
hasMethod hasParameter
Tree_1
Parmeth_1
hasScale
hasUnit
hasParameterMethod
hasInvestigationItem
Valueset_1
hasValue
hasTime
2.2
3.3
5.5
hasTime
hasTime
jan-2006
jan 2007
jan-2008
m
Method
Derived classes:
• Measurement method
• Treatment method
• Classification method
Method: simple example
Triangulation_
Method
encompasses
Measure_angle
predecessor
encompasses
predecessor
Determine_distance_
to_object
encompasses
Calculate_
height_of_object
selection_description
parameter_
method
physical_thing
has
Classification
hasPopulation
selection_description
hasSample
hasSampling
Method
physical_thing
sampling_
method
intended_sample_
size
hasIntended
SampleSize
hasTotal total_population_
PopSize
size
hasDeviation
Reason
deviation_
reason
Example
Country
hasPopulation
selection_description
hasSample
Forests
Forests
hasPopulation
selection_description
hasSample
Plot
Plots
hasPopulation
selection_description
hasSample
Tree
Tree
Other classes:
• grouping_description:
– design (treatment, measurement, layout in field,
block’s, etc.)
– membership
– “group by” functions (mean, variance, ...)
• reference lists
– species
– other nominal value lists
• actors:
– persons, institutions
• project
• etc.
Proof of concept
F-Logic / OntoStudio
Proof of Concept
JOKL
cultural
landscapes
JODI
vegetation
ECN
Summary
Database
2835
foodplain
Pythia
vegetation
Connect Databases
Import OWL Ontology
SERONTO
Results
Query
http://www5.umweltbundesamt.at/ALTERNet/
index.php?title=Main_Page