Transcript ESIP2009ProvenanceIntegration_Fox

Semantic Provenance and
Integration
Peter Fox and Deborah L. McGuinness
Joint work with Stephan Zednick, Patrick
West, Li Ding, Cynthia Chang, …
Tetherless World Constellation
Rensselaer Polytechnic Institute
Thanks to Rob Raskin (JPL), UTEP and NASA Goddard Space Flight Center
Projects funded by NSF Office of Cyberinfrastructure and NASA Advanced
1
Information Systems Technology
Provenance
• Origin or source from which something
comes, history of ownership, intention for
use, who generated something, what is
generated for, manner of manufacture,
sense of place and time of manufacture,
production or discovery, documented in
detail sufficient to allow reproducibility
• Knowledge provenance; enrich with
ontologies and ontology-aware tools
2
Proof Markup Language
(PML)
• A new kind of linked data on the Web
World Wide Web
PML
data
Enterprise Web
PML
data
PML
data
PML
data
D
D
PML
data
PML
data
D
D
D
D
D
D
• Modularized & extensible
– Provenance: annotate provenance properties
– Justification: encodes provenance relations
– Trust: add trust annotation
• Semantic Web based
Enterprise Web
PML
data
…
Mobile Wine Agent
Selected Application Drivers
CALO
Combining
Proofs in
TPTP
Intelligence Analyst Tools
GILA
Knowledge
Provenance
In Virtual
Observatories
4
4
PML Provenance
It is about provenance concepts
• URI for identifying and addressing
• Declarative metadata
• Taxonomy
#info1 a pmlp:Information;
pmlp:hasRawString “(type TonysSpecialty SHELLFISH)” ;
pmlp:hasLanguage <http://inference-web.org/registry/LG/KIF.owl#KIF> ;
pmlp:hasFormat <http://inference-web.org//registry/FM/PDF.owl#PDF> ;
pmlp:hasPrettyString “Tonys’ Specialty is ShellFish” ;
pmlp:hasURL “http://inference-web.org/documents/tonys_fact.kif”.
Science and data
• Science is built on verifiability and
reproducibility
• As more layers are inserted between the
scientists and the origin data, or when the
data is out of the usual realm of familiarity
– Trust must be established
– Sources must be verifiable and proved
– Explanations must be given and connected
6
20080602 Fox VSTO et al.
7
Example Use Cases
• What was the cloud cover and
atmospheric seeing conditions during the
local morning of September 19, 2008 at
MLSO?
• Find all good images on September 21,
2008.
• Why are the Quicklook images from
September 21, 2008, 1900UT missing?
• Why does this image look bad?
8
Explain
Explain
9
Explain
20080602 Fox VSTO et al.
10
11
Search and structured query
Moving to faceted browse based on
PML tags (facets), using jspace
Search
Structured
Query
12
Search
20080602 Fox VSTO et al.
13
Visual browse
14
15
16
Tools
17
Implementation of PML model
• Retrospective – scraping the un-related
sources
– Needed to gain confidence and trust from the
users
– PML generated after the fact (can regenerate) which is very good at the
development stage
• Proactive – PML on the fly as data passes
through the pipeline
– Preferred but only when model is mature
18
Live demo
19
Further Information
• [email protected], [email protected]
• http://inference-web.org/
• http://tw.rpi.edu/portal/SPCDIS
• http://tw.rpi.edu/portal/MDSA
20