Salandayia

Views:
 
Category: Entertainment
     
 

Presentation Description

No description available.

Comments

Presentation Transcript

Workflows + Ontologies: Harnessing the Grid to Advance the Geosciences: 

Workflows + Ontologies: Harnessing the Grid to Advance the Geosciences Leonardo Salayandía, Paulo Pinheiro da Silva, and Ann Q. Gates The University of Texas at El Paso Special thanks to Dr. Randy Keller Supported in part by GEON NSF

Overview: 

Overview The GRID and its challenges Ontologies, providing a consensus to the semantic description of the resources available on the GRID Workflows, a mechanism to help scientists create ontologies Workflow-Driven Ontologies, an approach to capture knowledge from scientific domains, and produce workflows from it

The GRID: 

The GRID The GRID means different things to different people i.e., different implementations The general idea: Access and share distributed resources Computational capabilities Databases Sensors …

The GRID: 

The GRID Some challenges of the GRID Security (authorization/authentication) Organization policy (Virtual Organizations) Support heterogeneous platforms Semantic compatibility of resources Provenance Data support Trust …

How can ontologies help us?: 

How can ontologies help us? There are several things that we already know about ontologies including that they can: Facilitate us to agree to a common vocabulary for terms in our fields Facilitate information integration Facilitate information exchange Support (semantic) search

Current Support for Ontology Development (in Science): 

Current Support for Ontology Development (in Science) Great achievements during the last years: Notations to represent concepts and properties about a scientific domain knowledge, e.g., OWL Tools to build ontologies, e.g., Protégé, SWOOP Tools to search for ontologies, e.g., SWOOGLE But how easy is it for scientists (not necessarily computer scientists) to create ontologies?

Current Use by Scientists: 

Current Use by Scientists Have YOU ever created an ontology? If yes, did you use a tool? Was it easy?

Workflows can be used to:: 

Workflows can be used to: Describe a “recipe” to do something; examples: a workflow to bake bread a workflow to create a map of gravity data Define formal venues for connecting GRID resources; example: Kepler executable workflow

How can workflows help us to create ontologies?: 

How can workflows help us to create ontologies? can serve as use cases, i.e., facilitate the identification of concepts and relationships can serve as a metric for assessing ontology quality

Workflow-Driven Ontologies (WDOs): 

Workflow-Driven Ontologies (WDOs) WDOs are OWL ontologies with few pre-determined scientist-defined concepts and properties in support of workflows WDOs enable the generation of workflows that automate the execution of (complex) tasks scientists can understand scientists can share, e.g., education

WDOs: Learning Concepts: 

WDOs: Learning Concepts An Example: making bread What are ingredients for making bread? Flour Dough “Derived” ingredients “Raw” ingredients Wheat Yeast

WDOs: Learning more Concepts: 

Concepts WDOs: Learning more Concepts One step in the process of making bread: “Mix flour and yeast to prepare the dough” Flour Dough Methods Derived Ingredient Raw Ingredient Wheat Yeast Mix

Learning Relationships: 

Properties Learning Relationships One step in the process of making bread: “Mix flour and yeast to prepare the dough” IsInputTo IsOuputOf Yeast Mix Flour Mix Dough Mix

A Bakery WDO : 

A Bakery WDO Bread Harvest Grind Bake Wheat Grind Dough Bake Wheat Harvest Flour Grind Bread Bake a) Select a “current” product, e.g., bread b) Find method(s) that produce current product as an output: Bake c) Find method(s) that produce the inputs for (b): Mix d) Make the input of (c) the “current” product and repeat (b) and (c)

Workflow Specification: 

Workflow Specification STILL JUST A SUGGESTION!

WDO-it! Current Capabilities: 

WDO-it! Current Capabilities Build WDOs from scratch Capture and encode dataset and method concepts Capture and encode isInputTo and isOutputOf properties Build suggested workflows from WDOs

WDO-it! Future Capabilities: 

WDO-it! Future Capabilities Harvest WDOs from generic ontologies Evaluate suggested workflows Capture process knowledge – scientist endorsement Integration with existing workflow environments Support for collaborative design of ontologies

WDO-it! In Action: 

WDO-it! In Action

Gravity Contour Map Workflow: 

Gravity Contour Map Workflow 1) Scientist selects end-result, i.e., gravity contour map 2) WDO-It! generates a suggested workflow specification 3) The scientist either endorses the workflow spec or indicates improvements

Conclusions: 

Conclusions Ontologies are needed to address the semantic match challenge of the GRID Support for Scientists to build ontologies needs to be improved WDOs are OWL Ontologies Workflow specifications can be generated from WDOs WDOs can be harvested from existing OWL ontologies Workflow specifications facilitate the evaluation of ontologies http://trust.utep.edu

authorStream Live Help