Friday, June 29, 2007

And even more plenary...

Interesting start; Katy Borner is talking about how the amount of knowledge out there has grown enormously and yet our brains are not getting any bigger, and there's not a good way to extract it automatically. How can we create infrastructures to help us? Google is good, but doesn't show patterns and trends and outliers.

One approach is to create maps. Example of Co-Authorship of IEEE papers from 2004 shows different patterns between say colleagues in a lab or a professor at a university, and also shows that these are all local networks--most connections are to people physically close.

She and her colleagues developed maps of science which they use to educate kids about how science works. They develop them by searching different databases. But now onto her main topic (!).

  • We need software glue to interlink datasets and algorithms written in different language using different data formats.
  • The smaller the glue or 'CI Shell' the more likely it can be maintained.
  • Dataset and algorithm 'plugins' are provided by application holders/community users
  • Applications resemble custom 'fillings.'
Cyberinfrastructure Shell (CIShell) is an empty shell that support many functions. She then goes into detail about some tools like Network Work Bench and one she didn't mention but I found online.

No comments: