Requirements

Last updated: 1.24.12

The following outlines requirements to support event-response research for the Distributed Autonomy project and integration of its components into the ODSS. These requirements are also intended to support research on folksonomy classification based on tagging, and machine learning to identifying key features that best describe data and significant events from past and future CANON field experiments.

This is very much a work-in-progress and intended as starting point for discussion.

Visualization

• Tool to assist human in the loop labeling/tagging of events
  Display trends to inform users on prediction. What kind of trends ? What kinds of predictions ? How to best visualize these ?
• User login capability in ODSS
• STOQS spatial temporal query for past experiments to mine data for model development and to ground truth data

Data Format and Query

• Spatial temporal query of satellite images (SST, MODIS, etc.) to mine data for model development and to ground truth data
• Data shall be stored in a format compatible with algorithm development (Matlab ?), as necessary for building models.
• Data shall be time-stamped and spatially referenced in a standard format as necessary for e.g. interpolation, subsampling, etc.

Data Storage

Database capable of storing:

Annotator (user)

• properties (name, title, etc.)
• probability (e.g. probability provided correct label)

Class attributes

• tags
• labels (binary, multivalued, or continuous)
• confidence (e.g. definitely, guessing, probably, percent)

Images

Model parameters