JPL joins DARPA's Memex project
Better search and indexing for space stuff
DARPA's Memex “deep Web” search research has a new high-profile supporter: NASA's Jet Propulsion Laboratory, which is looking for new ways to catalogue the flood of data arriving from spacecraft.
Memex, which DARPA announced early in 2014, is designed to get around what the agency sees as the shortcomings with traditional Web search – ill-organised results whose usefulness relies on the skill of the searcher.
Also, it noted at the time, the so-called “deep web” – anything that isn't searchable by Google's robots, for example – goes undiscovered in ordinary search.
Announcing NASA's participation in the project, JPL's Chris Mattmann says the lab wants technologies that “understand people, places, things and the connections between them”.
Memex's ability to index things like images, videos, pop-up ads, forms and scripts – as well as boring old text – is the special sauce here.
The JPL announcement also gives some idea of where Memex has gone in the 15-plus months since it was announced: for example, it can identify (and presumably index) the same object across “many frames of a video or even different videos”.
That's clearly attractive to space research, since it would cut the workload for scientists trying to match space probes' photography to videos and spectrometer data. JPL also reckons Earth imaging science would benefit, with better monitoring of things like snowfall and soil.
JPL also says Memex is a boon to the working scientist, because it can bust useful data from the PDF prison.
The Memex software is open source, with its various packages published here. ®