For every mention we automatically determine sentiment, e.g. positive, negative, neutral, or unknown. In this article we'll focus on the different aspects of Sentiment Analysis:
The automatic sentiment detection service automatically determines the sentiment polarity class (negative, neutral, or positive) of texts (mentions, articles, etc.).
We use state of the art classification algorithms, specifically, maximum entropy classification of texts represented as “bags of words” enriched with additional features such as the source and length of texts. This basically means that we only look at the different “bags” of words in a text and ignore all syntactic information such as word order. The algorithms then compute the probability that a certain bag of words occurs in a certain polarity class (positive, negative, or neutral). The algorithms were trained on a combination of manually annotated and semi-automatically mined training examples.
We dare say our sentiment analysis has greatly improved over the couple of years and performs at least as good as other services available (but with increased coverage). More precisely, we achieve accuracy and precision of around 75% which (on average) is almost as good as humans.
You can always overwrite the automatic sentiment we have added or add sentiment whenever we didn't recognize it. You do this by hovering over the mention, clicking on the smiley, and choosing 'positive', 'negative', or 'neutral'.
Automated sentiment analysis is currently supported for Dutch, English, French, German, Greek, Italian, Spanish, Portuguese, Turkish and Arabic.
In the CX Social Insights you have analytics about the sentiment of your mentions to give you a general idea on how the brand is perceived.
Another interesting thing to know is that you can always filter on sentiment in Inbox to only view the positive/negative/neutral mentions.