Notes on explorable behavioral databases

README.md

XBD

We're pushing the idea of explorable behavioral datasets.

We're investigating:

• tooling
• standards
• best practices

Currently exploring strategies for the immediate capture and playback of gesture/hand-motion samples, but ultimately interested in techniques for persisting, editing, annotating, and replicating behavioral activity streams in general.

Why?

The rise of proximal sensors.

They provide a stream of behavioral data that can be ...

• captured
• visualized
• edited
• munged
• analyzed
• replicated

Relevant Domains

• motion capture

• motion analysis

• motor learning

• motor control

Tooling

In general, the plan is to build off D3 for visualization/interaction and IndexedDB as a distributed datastore.

The following modules encapsulate some key methods and techniques that we plan to utilize.

Munging

• nest - convert tables to trees

• crossfilter - fast multidimensional filtering for coordinated views

Visualization

• dc.js - reactive dimensional charting

• catcorr.js - visualize correlations across many dimensions of categorical data

Persistence and Replication

• level.js - leveldb for the browser

• dat - real-time replication and versioning for large tabular data sets

Further Reading

• The vision behind Harvest
• Replicating large datasets in HTML5
• Your API does REST, but can it SLEEP?
• Designing coordinated visualizations