Methodology

The Groubal Community Sentiment Index tracks the global customer satisfaction of hundreds of companies through near real-time analysis of social media messages.

Our system continuously monitors updates from websites that include Facebook, Twitter and Groubal, checking for mentions of brand names. For updates that are deemed to express an opinion about a brand, a Bayesian classifier calculates the relative positive or negative sentiment of the message, using a variety of linguistic features, including unigrams and bigrams.

Sentiment analysis can never be 100% accurate: even humans face the difficulties of detecting sarcasm and irony in online messages. Even so, our tests show an impressive 90% accuracy in our data rating.

Through this sophisticated analysis of global social media data, we are highlighting the very latest international opinions of each brand, without the inevitable bias introduced by customer satisfaction surveys.

Talk to us about licensing the Groubal CSI data.