Presentation is loading. Please wait.

Presentation is loading. Please wait.

NSF DataNet Initiative

Similar presentations


Presentation on theme: "NSF DataNet Initiative"— Presentation transcript:

1 NSF DataNet Initiative
Research Agenda NSF DataNet Initiative Site Visit 8 February 2010 DataSpace Some general comments plus one specific example … (v8)

2 DataSpace Research: Risk Management Spectrum
Lowest risk: Incremental improvements to starting operational Dspace/Fedora (DuraSpace) platform Guaranteed useful tool for curation of scientific data Minor risk: Inclusion of existing and operational components including those suggested by partners, such as XAM (EMC), OpenII (Google), etc. Risk: Selection, Integration Modest risk: Adapting existing and evolving research technologies & prototypes (e.g., Context Interchange) Risks: Make robust, Scaling, and Integration

3 Some Research Areas Listed in Proposal (p.10-14)
Data policy, protection and security (Abelson, Berners-Lee, Pato, Weitzner, White) Data discovery and data semantics (Madnick, Siegel, Smith) Data quality & provenance (Madnick, Abelson) Data analysis and analytics Model calibration and mediation (Woon) Operational Scientific Intelligence (Hsu) High-speed pre-processing and data consistency (Hsu) Data visualization (Karger) Workflow for scientific research and archives (Smith) Data storage (Todd, Milojicic) Legal issues with data (Wilbanks) Data interoperability, conversion, integration (Madnick, Smith)

4 Types of Semantic Differences
Representational Ontological Temporal Simple Example (snow fall) Temporal Representational Meter vs Feet Feet before 2001, Meters afterward Ontological “Snow fall” – using standard "snowboard” method or other method (e.g., liquid) “Snowboard” before 1990, liquid method afterward Given recent situation – measuring “snow fall” seems like a very relevant example. Q: How many familiar with “snowboard” method? Briefly describe: "snowboard“ method. Essentially this is a piece of wood about 16" by 16" that is painted white. The snowboard should be wiped clean every six hours or so to prevent the natural settling of fallen snow from occurring. In addition, these should be placed well away from structures and obstructions -- about 20 to 30 feet if possible -- in order to prevent drifting from inflating the totals as well. NOAA – has some differences. Another approach uses liquid precipitation (melted snow): in general an inch of water is equal to about 10 inches of snow. Snow Depth is a Different Measurement CoCoRaHS, a network of precipitation observers, describes it this way: "For example, if half the ground has 2" of old snow and the other half of the ground is already bare, the average snowdepth would be 1"."

5 COntext INterchange (COIN) Approach to Resolving Semantic Differences
Concept: Depth Modifiers: Meters Feet f() Meters Feet Specialized symbolic equation solving techniques used to dynamically create comprehensive conversion programs from small conversion components Light-weight Ontologies with Context Modifiers Shared Ontologies Conversion Creation Mediation & Transformation uses an integrated framework of abductive and constraint logic programming Context Mediator Declarative description of Source’s actual semantics Source Context Declarative description of Receiver’s desired/expected semantics Receiver Context 2 1 Select depth x 3.35 From dataset A Where id =“12AY” Select depth From dataset A Where id =“12AY” Note animation … Steps: SQL Query Gets raw data and checks on source context (Meter) Sees if matches receiver context (Feet) Sees if knows how to do conversion (Meters -> Feet) If so, converts to feet (e.g., about 55 feet) and returns data to receiver. depth Context Transformation 17 55.25 3 Source (Data set A) Receiver 10

6 Intuition – Capability for automatic determination of complex conversion programs
Dataset A context Depth notion std 1 meters Scale factor Depth units Component conversions are provided along modifier axes Composite conversions between any cubes in the space can be composed automatically std std 1000 1 meters feet

7 Research Agenda DataSpace
Will draw on extensive research experience and on-going efforts of the entire DataSpace team, including research by our corporate partners Much more details on these research efforts to date can be found in the more than 200 papers, written by the research team, and listed in the References section of the DataSpace Full Proposal Collectively the proposed research efforts represent an ambitious research agenda. We will present two more examples …


Download ppt "NSF DataNet Initiative"

Similar presentations


Ads by Google