Knowledge Centred Earth Observation KEO: A service oriented architecture for information discovery a

1
Knowledge Centred Earth Observation (KEO) A
service oriented architecture for information
discovery and exploitationAndrea
ColapicchioniAdvanced Computer Systems

• ESA-EUSC 2005 Image Information Mining Theory
  and Application to Earth Observation Frascati 5
  October, 2005

2
The KEO Vision

• "The information extraction process must be
  replaced by more automated, direct and human
  centered methods"

3
Automated, Direct and Human Centred

• Automated is a consequence of the data size /
  complexity and of the cost of the non-automatic
  process.
• Direct responds to the need to reducing the
  steps between the user and the information
• Human centred brings the focus on systems that
  could be managed also by non-EO experts via
  semantic interactions

4
KEO
5
KEO aims to

• extract relevant features from EO data
• accept EO data coming from Ground Segments as
  well as other data providers
• provide, as an output, valuable information in an
  easily accessible format

6
The aim

• support end-user in finding, through an
  interrogation process based on users natural
  language, EO products and/or services best suited
  for his needs
• provide a modularized structure for easiness of
  maintenance and upgrading
• leverage on open standards.

7
Feature Extraction
8
Feature Extraction
Feature Extraction
Non Interactive
Interactive
PIM
Neural Networks Land/sea Mask .
Classification
Object Recognition
CBIR
Manual Annotation
9
What we have?
10
The question

• How can we help users, which might belong to
  multiple domains, both EO and non-EO to easily
  locate the appropriate services and information?
• Semantic web (ontologically based) seems to be a
  promising field

11
First architecture
12
Service Oriented Architecture

• Services are
• well-defined
• self-contained
• and universally available
• business functions that respond to service
  requests from "consumers"
• At the enterprise level, this concept leads to
  the notion of a service-oriented architecture
  (SOA)

13
Flow

• Ok, we know how to provide the "building blocks",
  but how to combine them?

New image
Pre-processing
Catalogue Population
Feature Extraction
IIM Ingestion
Feature Extraction
Feature Extraction
PIM Batch
Features DB
14
Web Services Orchestration

• For years the industry has dreamt of modelling
  business processes in software and combining them
  like Tinker Toys.
• Web services orchestration is the new term for
  that old idea, serves as the platform layer in
  network computing that enables composition and
  coordination of services (the building blocks)
  into long-running transactions and collaborative
  business processes.

15
BPEL

• BPEL deals explicitly with the functional aspects
  of business processes
• control flow (branch, loop, parallel)
• asynchronous conversations and correlation
• long-running, nested units of work, faults, and
  compensation

16
BPEL Process Manager

• Integrate servicesinto collaborative and
  transactional business processes
• Native BPEL support
• Easy-to-use Visual Designer
• Binding for Web Services, JMS, JCA and User Tasks
• Unparalleled auditing and monitoring

17
First Workflow
18
Storage

• Ok, now we are able to extract feature from data
  but how to exploit the data to the user?
• Standards (OGC)
• WMS (Web Map Server)
• WFS (Web Feature Server)
• CS-W (Catalogue Server)

19
SSE

• Could we keep these services locked inside a
  monolithic application?
• SSE helps us to publish the produced services in
  a portal easily accessible from the user
• Based on web service
• Standardise the methods that a service MUST expose

20
Technologies

• Databases (Oracle, Informix, MySql, Postgres,
  PostGIS..)
• J2EE
• Python/C for image processing
• Oracle BPEL server
• Open source GIS (probably Degree)
• Ontology OWL, RDF

21
KEO as a computer
22
KEO
23
The Consortium
24

• Thanks for your attention
• For any question
• Andrea Colapicchioni
• Advanced Computer Systems
• a.colapicchioni_at_acsys.it