A sentiment polarity score compresses a text’s expressed favorability into one number: +1 signals strongly positive sentiment, −1 strongly negative sentiment, and values near 0 mostly neutral language. section 2.6
This score helps process large volumes of customer feedback, filter junk or troll messages, and let chatbots judge whether an incoming or proposed response is appropriate. section 2.6
Neutral feedback
“It was OK. There were some good and some bad things.” illustrates a mostly neutral statement that could receive a small positive or negative score. section 2.6
Sentiment analysis can produce the score through:
- A rule-based algorithm composed by a human. section 2.6
- A machine-learning model based on data. section 2.6
