A lot of people, myself included, believe the big money in social media monitoring will come from producing a fool-proof algorithm that can interpret sentiment online.
At the moment tools such as Scout Labs provide around 70% accuracy when it comes to understanding the tone and feelings expressed online, determining the language used to be positive, negative or neutral. This is a good start, but the fact that Scout Labs includes an option to adjust the category of sentiment that has been determined by the engine is an open acceptance that the system is still deeply flawed.
Many free social media tools such as , and provide sentiment feedback within their services, but I can’t help feeling that the returns you get from these tools are extremely basic, often with very high levels of comments being categorised as ‘neutral’ when they are anything but.
So, brainstorm (or fail!) moment. What if sentiment scoring worked in a similar format to the Facebook ‘Like’ option. Instead of the brand trying to monitor and measure all of its mentions across the net, their evangelists would act to ‘rate’ the sentiment expressed in a blog post, the sarcasm in a tweet or the tone of a wall post. It would be the Wikipedia of sentiment analysis, controlled by the people who are passionate about the brand, allowing for interpretation of human slang and potentially developing a stronger brand community.
It would only work for brands such as Apple, probably bring about another nightmare for Skittles and most likely be totally useless as Coke and Pepsi fans nuke it out to convey each brand as the most hated producer of cola.
But what if it worked!?
