A lot of my job as a software developer for a digital marketing company involves investigating new tools and finding out ways of analysing websites and their content, usage and trends. As such, my team are always discussing new things we can explore and coming up with new ways of looking at the web.
We’ve discussed sentiment analysis many times before, checking keywords or site content to determine whether what’s being said is in a positive of negative tone. One of our discussions today resulted in one of the guys finding uClassify’s Sentiment Classifier.
This handy service accepts either a URL or text content (no HTML) and parses the contents to return a percentage result, detailing how much of the content is positive or negative in tone. As a quick test, passing in the URL of this site has currently returned a result of 55% positive, 45% negative…a sign that perhaps I need to brighten up my writing style!
The site claims to have been based upon extensive research using tens of thousands of Amazon product reviews in conjunction with “mood training” data. I’m not sure exactly what mood training data is, but it certainly sounds the part.
The API’s free to use, you just need to sign up for a read API key.