I wish they had a site I could plop some text into as a demo. Hmph, Now that sounds like something Google would acquire and add to Google Translate. Google Emotional Translate.
In a crude fashion this isn't difficult to do: Simply look for things like emotionally charged words and phrases, certain sentence structures, or maybe even word and sentence length.
This is a fairly well studied area, search on the keyword "sentiment analysis" for more information, including a number of open source implementations. ToneCheck even seems to be using the 6 basic emotion categories that everyone else uses, although they've tweaked the names a bit.
It looks like they've crowd sourced evaluating the emotional content of words and phrases at http://toneaday.com/.
My guess, largely without evidence, is that they are primarily basing their analysis on words and phrases and are relying on a better word->emotion dictionary than most to provide a competitive advantage. There are a number of open source tools that do this sort of thing, the ones I've seen relied heavily on the dictionary approach as well.
This is a very clever idea and positioning, certainly a better way to productize this technology than the ideas that had been floating around my head.
Good find ;-) in face you're right. We're heavy on the crowd sourcing models. But.... it goes further than that. We've got patented technology that surround the connotative capabilities to truly collect and measure the emotional effects of words, phrases, punctuation, emoticons, etc. We've built an adaptive learning engine that is learning new language that's spoken today. It's also self-healing - the more the tool is used, the smarter it gets.
Thanks for your feedback! We're certainly happy to hear people's interest in our tools. We've abstracted our system into an API (called ToneAPI).. if you're interested in building emotional tools - sign up!
5 comments
[ 2.6 ms ] story [ 20.5 ms ] threadThanks for the feedback.
This is a fairly well studied area, search on the keyword "sentiment analysis" for more information, including a number of open source implementations. ToneCheck even seems to be using the 6 basic emotion categories that everyone else uses, although they've tweaked the names a bit.
It looks like they've crowd sourced evaluating the emotional content of words and phrases at http://toneaday.com/.
My guess, largely without evidence, is that they are primarily basing their analysis on words and phrases and are relying on a better word->emotion dictionary than most to provide a competitive advantage. There are a number of open source tools that do this sort of thing, the ones I've seen relied heavily on the dictionary approach as well.
This is a very clever idea and positioning, certainly a better way to productize this technology than the ideas that had been floating around my head.
Thanks for your feedback! We're certainly happy to hear people's interest in our tools. We've abstracted our system into an API (called ToneAPI).. if you're interested in building emotional tools - sign up!
Thanks!
Josh