In the dark ages, I studied educational sciences and agogics, the science of social interaction. How people learn, how group dynamics influence behavior and how individuals interact with their environment has intrigued me ever since.
I was baffled to hear Deb Roy speak @ #LeWeb11. When his little baby boy was born, Roy and his wife took a courageous decision: they would move into a house that would record every second of their lives, sound and video. The house has sound recording equipment, and cameras all over, and is packed with the most spectacular processing and storage power MIT can provide.
Deb Roy first mission was a misleading simple one, he wanted to understand how language is assimilated by humans. Language is without a shadow of a doubt one of the defining features of the human species. By collecting busloads of data on how his child learns in the natural habitat of its home, Roy wants to track the process. Cameras record where the family moves, microphones record sounds…all this is captured, labeled and data mined. The system is able to produce the whole sequence from the baby sound “gaga” to saying “water.” The system can also find correlations between assimilating words and vocabulary, and visual impressions. Roy was able to prove that words that are more often associated with an image are quicker absorbed by the child.
By recording every single bit of data, scrubbing the data from unwanted ballast, correctly labeling it, and building intuitive learning algorithms to process, analyze and understand the patterns Roy is now able to visualize these patterns, interpret them, and close on understanding.
On top of the obvious hearing/learning connection, Roy can prove a definite social context to learning, up to a point where his analytic tools can predict how fast a word will be learned, based on the movements in the house, and the number of times the child is exposed to the word.
Armed with this amazing technology, Roy started up Bluefin Lab, convinced that the same principles of linguistic analysis, visualization and predictive interpretation can be applied to new areas, such as TV and the social web. Bluefin Lab is now not only indexing over 200 US networks, 200,000 shows and 2 million ads per month; it is also connecting all the online conversations generated by 20 million people around these indexed items, .
The associations, learnings and predictive formulas that Roy is able to distill out of his system of cognitive machines are mindboggling, and I am convinced that we’re only at the beginning of social datamining. This will in earnest change the way we deal with “influence” and “metrics”
Tell me what you did when you said it… and I’ll predict what you will do next…http://www.youtube.com/watch?v=VwgkT34g61w