Nick Jones

An interesting working draft just emerged from W3C for something called EmotionML- Emotion Markup Language. The goal is to provide a framework for annotating information with human emotions. There is a clear need for some form of EmotionML, especially in the area of context where future mobile devices will provide lots of clues to our emotional state. E.g. the camera can see our expression, maybe the microphone can deduce our respiration rate, and future biometric sensors may provide information about skin conductivity, heart rate and so on. Emotional understanding will be vital for successful contextual systems. Inappropriate contextual advertising could aggravate me instead of informing me, and have a negative impact on your brand. So if you can identify that grimace on my face and feed it back in time to abort the ad with a polite apology then I’m much more likely to think you’re the sort of caring company I’d be prepared to do business with.

However, the authors of the EmotionML draft admit that they are providing a framework rather than a complete solution. They allow you to tag information with emotions, but stop short of defining the units and scales to measure them. The problem is that emotions are subjective; we can all agree what being scared feels like, but what units should we use to measure terror? If you’re a vampire maybe your unit of scariness is the Buffy, but for for guys who remember that era the same unit might measure levels of unachievable pulchritude. Years back I read a book about a UK government weapons development department who used a unit called the Pouter to measure bureaucratic obstruction, named after a particularly obdurate Navy commander. In practice they found the millipouter more appropriate for everyday use. But seriously, if we’re ever going to make context work we have to agree on at least a few common units to measure emotional response values that can be shared between applications.

So your mission for today, should you choose to accept it, is to propose some universal units for the measurement of key emotions. When you do I may be able to communicate exactly how happy I’m feeling about EmotionML.