I was recently pointed in the direction of IBM’s Watson API which allows everyday designer/developers like me access the power of the Watson supercomputer’s ability to understand and interpret natural language.
One particular Watson API service which will be useful to my project is the ‘Tone Analyzer’.
Up until now, I have been using the Moriarty score in prototypes and testing as well as a similar system for the font moriarty.ttf. However, the system was a little crude – it didn’t take in the full picture. For example, “Not bad”, a phrase which means OK/reasonable, would get a 10/10 Moriarty score due to the presence of the word bad in a 2 word statement.
Watson adds an entirely new layer of natural language processing and understanding which I would not be able to recreate myself.
Tone Analyzer is able to intelligently interpret the emotional tone of a portion of text under three headings:
It assigns each a score between 0 and 1. A score above .2 is indicative, .5 is medium and .7 is high.
I have been taking these scores and mapping them to create score out of 100.
You can try this out here:
This is a breakthrough for my project however will require me to rethink a large portion of the information design and UI