I built an Outlook Add-In called Tone Detector not so long ago that uses elements of the Social Opinion sentiment analysis API to determine the emotion of an email as you type (in real-time). You can download a FREE copy of Tone Detector here.
How does it work?
It sits in the tool-bar as you compose your email and displays a graphic to indicate positive or negative sentiment. It doesn’t save any of the content of your mail so don’t worry about the software recording your every message.
I built this to illustrate how machine learning and sentiment analysis can be used in everyday applications or products for the most trivial of use cases. I have “trained” the sentiment analysis engine with a “general” dataset. For increased accuracy, it could be trained with domain specific text thereby improving the usability of the tool.
It’s a proof of concept tool and I’m sure there are other features that could be added to improve it. Let me know if you have any more questions or have ideas as to how this tool could be extended or made more useful for you.