1. Case study
Linked-data forLinked-data for
Integrated CatchmentIntegrated Catchment
ManagementManagement
Ian Dickinson
Epimorphics Ltd
ian@epimorphics.com
@ephemerian
Tom Guilbert
Environment Agency
tom.guilbert@
environment-agency.gov.uk
2. Agenda
context and aims
– catchment management data
– vision
integrated catchment linked-data project
conclusion
7. 1. Data & evidence,
consultations, local
knowledge, model outputs
and plans collated in to a
shared central system “Local
Community CPS”
Local Community Catchment
Planning System
Local Community Catchment
Planning System
MonitoringMonitoring Local
Knowledge
Local
Knowledge ActionsActions
2. Contents of Local
Community CPS
published as Linked
Data alongside EA
and research
datasets
3. Linked Data (machine
readable data) could be
automatically combined by
applications such as the EVO,
CCM Hub and any number of
web apps
CCM
HUB
Slide used by kind permission of Michelle Walker, Rivers Trust
9. ICM proof-of-concept project
16 weeks duration
project team:
– 1 FTE app dev
– 0.4 FTE user
research
– 0.5 FTE data
7400 water bodies
7.8m triples
agile principles
– four iterations
– 2-3 week sprints
– stakeholder
review
alpha/staging site
organizationscale
10. From data to linked open data
data modelling
extraction
transformation
publication
presentation
interpretationdownload
source data
SQL
Java
Apache Fuseki
explorer application
Elda
15. Data publishing
Baseline goal:
– provide access to the data
Practical considerations:
– Just “follow-your-nose” linked data?
– or SPARQL?
– or an API?
– ….
22. Data explorer key features
search
– by name, catchment, location, ...
show classification items
filter by properties
– e.g. classification value
map and tabular output
basic reports
download data
25. Data explorer application
Easy for
novices
to get
started
Not too
frustrating
and slow
for experienced
users
data explorer
26. Typical user enquiry
“Please show me all:
– rivers and lakes
– near Glastonbury
– that had overall ecological classification as
moderate, poor or bad
– between 2009 and 2012.”
31. Initial learnings
writing SPARQL by doing
– in context
– with feedback
hard to balance different user needs
– explore vs. guide
– real user input
download
– important
– RDF to useful CSV is hard