posted May 24, 2010, 5:32 AM by Joshua Lieberman
updated May 24, 2010, 8:06 AM
- Data Quality follow-on: UncertML
- Processing & Workflow Use Case
- User Analysis
- Additional Use Cases
- Scenario Development Issues
Josh Lieberman, Steve Browdy, Dan Cormford, Mattia Santoro
- Processing quality measures
- Yetman: WPS we have set up accepts GML simple feature polygons and returns the same polygons with estimates of the population (persons, 2005) that live within them. It also returns the mean administrative unit area within the polygon(s), a measure of the spatial precision of the input data used to generate our population surface. There is also a boolean quality warning field that has a value of 1 if the polygon(s) submitted are smaller than the mean unit size; an indication that the estimate may be unreliable.
- Dan Cornford: UncertML, UncertWeb, QA4EO
- UncertWeb: propagating uncertainty through web service chains
- Joan Maso and geoviqua project - uncertainty representation in geospatial representation, also involving ESA and potentially QA4EO
- UncertML: designed to sit underneath other standards and provide more rigorous uncertainty measures. As such, possibly more application-agnostic data quality measures.
- Working with ISO 19156 / 19157 efforts at data quality and lineage.
- UncertML did arise specifically from the issue of propagating uncertainties through processing. Lineage of uncertainties is being considered in UncertWeb, but only just started. For now, an implicit assertion that an expressed process uncertainty was derived from input uncertainties. Experimental to include a BPEL document to document the process.
- Propagation may be difficult to do. Large-scale Monte Carlo is time-consuming. There will need to be documentation in process descriptions of the methods used for propagation.
- Geoviqua: EC solicitation to improve presentation of data quality in geo-visualization, as well as collection of data quality and standards to govern it.
- Another aspect to include user feedback and convey "trust", then include all of the above in visualization.
- (Josh) GEOSS does need to address the issues of "secondary" data quality, whether cross-calibration or user feedback.
- Processing and Workflow Use Case
- Elements to update: workflow documentation, uncertainly propagation, workflow control channels.
- Additional use cases:
- Sensor tasking relates largely to Disaster Management. Need to consider ways in which it differs from invocation of other services and resources. Perhaps the time synchronization and sensitivity aspect is different.
- Both semantic use cases and processing services specialized cases need to be detailed, so we understand better where to specialize and where to define new cases.
- Nadine to review workflow deployment use case
- Dan to see about contributing processing with propagation use case, possibly as a CFP response.
- Team to look at sketching out the sensor tasking use case.