4/30/2005

Teaching computers to make "Better Mistakes"

TRN has an interesting piece here about how the MIT media lab has accumulated a vocabulary usage database that helps speech recognition by reducing mistakes and boosting recognition speed.

The researchers evaluated their common sense speech recognition technique by logging the errors and dictation times of users who dictated text that contained topics covered by the Open Mind database. It prevented 17 percent of the errors and reduced dictation time by 7.5 percent. Kind of a "Social AI", if you will.

This is a concept that can be used in business intelligence as well, not just voice recognition. I see some real business opportunities here for good DB people.