A paper titled “Identifying emotion by keystroke dynamics and text pattern analysis” published in the journal Behavior & Information Technology outlines software designed by A.F.M. Nazmul Haque Nahin and colleagues that can recognize the emotional state of a human user with up to 87% accuracy.
The study described in the paper asked volunteers to type pre-determined text and note their emotional state at the time. They also made notes of their emotional state during times when they could type freely. By studying the keystroke dynamics and text-pattern analysis of what the subjects typed, the software tried to parse out how they were feeling out of seven options; joy, fear, anger, sadness, disgust, shame, and
It makes sense for the software to start with the big emotions and work its way up to things like longing, wistfulness, and ennui.
So far, the software does a surprisingly adept job at recognizing emotion, particularly with joy and anger. For those emotions the system gets it right 87% and 81% of the time respectively.
Having computers capable of knowing our emotions means they could be able to automatically adapt to our moods in a lot of different ways. Angry? Maybe your computer will delay sending that email until you calm down. Feeling down? Wouldn’t it be great if your computer knew it and loaded up an adorable kitten video or something?
This is early research, but the findings are exciting. What would you want a computer that knows your emotions be able to do?
- Soon cars will recognize your emotions as well
- Which GIF best expresses emotion?
- Men are better at reading emotions in men than in women