DARPA seeks algorithms to create stories from info fragments

One of the nicest aspects of blogging is when a reader points you to an interesting article you hadn’t seen. I’d like to thank reader S.E. August for this reference.

Wired’s Danger Zone blog reported last week that DARPA is looking to sensemake various forms of data by combining them into stories.

The Cognitive Edge training I took last week discussed (among many other topics) gathering narrative fragments into composite stories as a way to make sense of a situation and convey that information to others. Similar thus far. A possibly reality-defying assumption follows, though. According the Danger Zone post:

The author of this tale, however, would be a series of intelligent algorithms that can pull all of this information together, tease out its underlying meanings, and put it in a narrative that’s easy to follow.

In the Cog Edge method the sensemaking is done by a group of humans, not a computer. The assumption is that distributed cognition of a group of people can elicit meaning where a single person, or a computer, cannot. I’m not up on the latest in artificial intelligence, but I’m doubtful that an entirely computerized approach can yield anything of use.

Perhaps a partially-computerized method, where fragments were gathered (sampled?) by machine and sensemade by humans, would work better. Or if the fragments could be human-coded as they were captured the significant or related ones might be easier to isolate. I don’t know. Can any readers weigh in who are more optimistic that a totally-computerized approach might work?

A link to the DARPA RFI is here.

(Photo by cote via Flickr creative commons)

, , , ,

Share and Enjoy:
  • Print
  • Digg
  • Sphinn
  • del.icio.us
  • Facebook
  • Mixx
  • Google Bookmarks
  • Blogplay