opinmind
December 17th, 2005
One of the hardest pieces of the blogging community to measure is context. How many bloggers like something as opposed to those that don’t? This is not an easy task to accomplish (I’ve had many discussions at the office about context and intent in blog posts) and one that I have not seen done successfully.
Opinmind comes incredibly close. Claiming to track “15,000,000+ opinions of 1,700,000+ bloggers”, it will break down topics into positive and negative opinions. I ran a few searches and was semi-pleased with the results.
As an example, take the search for librarians. As of this writing (see the screenshot), there was only one false result (on the negative opinions side, there is a post that says, “sorry librarians are gay..” - it was a comment on a negative blog statement on librarians).
I also like the Sentimeter, which provides a quick percentage of positive to negative opinions (31% positive, 61% negative for librarians).
Opinmind is not without flaws. I ran a search for librarian and the sentimeter almost absolutely reversed itself, giving librarian 68% of the positive opinions (does that mean that bloggers don’t like groups of librarians, but only solo?)
I learned from the blog (subscribed) that you can put terms against one another (like Googlefight) and them compare the negative and positive opinions.
I ran a bunch of odd searches (some that I won’t link to here do to their graphic nature) to see what would happen and the results weren’t as thrilling. Using programming language to measure context, intent, feelings, personality, behavior, and the like is not easy. Opinmind comes close. I’ll keep an eye out on this one for you.