In this article
Sentiment analysis is performed using Deep Learning neural network techniques, which learn the domain of your text in order to provide more accurate sentiment analysis. This sentiment configuration provides sentiment on a 5-point scale. It cannot be tuned at present, but generally provides good results out of the box.
We have another sentiment configuration which uses probabilistic scoring for determining sentiment. This works on an 11-point scale (-5 to +5). Sentiment tuning can be performed on this configuration to ensure that the sentiment analysis takes into account your business’s nuances (for example, “outstanding” could be interpreted as ‘excellent’ or ‘not yet resolved’ depending on the type of business), how your customers talk (formal vs informal) etc.
Our sentiment configurations are currently in English and Italian. We also have the ability to send other language comments to Google Translate to translate the verbatims into English for analysis using our English sentiment engine. Speak to your Account Manager if this is of interest.