Sentiment Analysis

A term worth defining carefully, because it's often used more loosely than the actual technique it describes.

What it is

Sentiment analysis is the automated process of determining the emotional tone behind a piece of text — typically classifying it as positive, negative, or neutral, and sometimes scoring it along a continuous scale or across more specific emotions (anger, joy, fear). It's a genuine natural-language-processing task, usually built on machine learning models trained against large sets of text that have been labeled with their actual sentiment by human annotators.

How it's typically done

Real sentiment analysis systems generally do one of two things: apply a trained classification model that's learned statistical patterns linking word combinations and context to sentiment labels, or use a large pre-built lexicon that scores individual words and combines those scores while accounting for negation, intensifiers, and sentence structure ("not good" isn't just "good" with a minus sign). Either approach requires substantially more machinery than checking whether a word appears on a short list.

What word-balance scoring is, by contrast

The Headline Analyzer's word-balance score is a related but meaningfully simpler technique: it checks each word in a headline against small, curated lists — power words, emotional words, common words — and reports the mix. It's word-category tagging, not sentiment analysis. It can tell you a headline contains the word "shocking," which is tagged as an emotional word; it can't tell you whether the headline reads as positive or negative overall, doesn't account for negation or sarcasm, and isn't attempting to assign a sentiment score.

Why the distinction matters

It would be easy to describe word-balance scoring as "sentiment analysis" in marketing copy, and technically inaccurate to do so. The word-balance score answers a narrower, more mechanical question — "does this headline lean on words known to trigger interest or emotional response" — which is useful for headline writing but isn't the same claim as understanding what a piece of text actually means emotionally.

Back to the full glossary — 200 terms covering case conversion, style guides, and text tools.