A myriad of Internet sentiment analysis applications are emerging such as monitoring blogs for coverage, tone polarity and semantic categorization of pretty much anything from consumer brands, to products and services, events, generic entities and of course individuals including celebrities and politicians.

There is a shift underway from few-to-many to many-to-many publishing models resulting in an explosion of consumer-generated media. 44% of Internet users are content creators (PEW), and an increasing ratio of search engine’s top results are now user-authored content. Brand owners are losing control over their message. The release of this anti Hillary Clinton video on YouTube clearly demonstrates it is also true for politicians.

You might as well be proactive and setup your own MySpace, LinkedIn, PreFound, ZoomInfo and other social networks profiles. Andy Beal’s got a long list of advice on how to track Internet sentiment and manage reputations online. Unfortunately, after email and search, this all also sounds like an invitation for some to seed and “spam” more content, frankly making discriminating between spontaneous consumer-to-consumer communications and planted opinions a challenge.

The Clinton campaign first announced her running online and Obama has been generating much Internet noise already. Matthew Hurst has a very interesting post about how disruptive social media can be, all the way to impacting on how democracy plays out “Bloggers Driving Gonzales Story“. And the impact of consumer generated media impacts on a few more than politician. Max Kalehoff reviews how several disciplines are getting affected, from public-relations people to traditional full-service advertising agencies, the big media shops, savvy direct marketers, interactive and specialty word-of-mouth shops.

Without going into tone polarity, I thought I would check out the social media coverage for both. The capability and applications are more than interesting. Internet sentiment analysis can most definitely inform as a warning system and suggest corrective actions such as targeting an online reputation campaign at the right audience, by media, by geographic location and theme to specifically address.

Take the charts below with a grain of salt, maybe not even directional … or is it? The queries were setup as phrases “hillary clinton” and “barrack obama” and do not include miss-spellings and forms like “Senator Clinton” or “Obama”.

IceRocket reports about 420 posts per day for “hillary Clinton” while “barrack obama” is getting around 525 at a run rate of close to 50,000 posts over the past 3 months.

Google Trends can be used to monitor online buzz not so much in terms of how often both candidates have been mentioned in social media, but by counting how many times folks have been looking them up in Google. “barrack obama” is searched for more often than “hillary clinton”. Now, go figure which of all this chatter is positive and negative.


OpinMind developed a tool called “Sentimeter” that works like a polarity index. “barack obama“‘s got 76%(+) versus 24%(-) as “hillary clinton“‘s sentimenter oscillates between 52%(+) and 48%(-). Opinmind is a search engine with a twist, scouring the web for the opinions expressed by people on anything and everything. OpinMind finds what people like and don’t like, including people’s views on politics and current events. The Sentimeter™ displays the relative number of positive and negative opinions identified by Opinmind’s automated search processes.

opinmind Bloggers like:
"hillary clinton"
52%(+) opinmind.com
opinmind Bloggers like:
"barack obama"
76%(+) opinmind.com

And then there is “sucks/rocks” brought to you by two dudes who definitely don’t suck, Gary Bernhardt and Brian Beck. When you enter a search term, sucks/rocks searches the web for several positive and negative phrases using that term. The score is the fraction of positive results to the sum of positive and negative results, normalized to 10. “hillary clinton” and “barrack obama” rock about as much.

If America’s next president gets elected the “American idol” way, then Internet Sentiment Analysis has reached enough maturity to significantly augment insights from formal scientific polls.