Technical briefing

How Cyber Sentiment Works

Real-time lexical sentiment scoring

Cyber Sentiment uses the `sentiment` package to score text in real time as the user types. It maps raw score to a mood label and highlights positive/negative tokens.

Data pathway

Input Text

Lexicon

Mood Score

1) Real-time analysis

The component recalculates sentiment whenever input text changes.

Memoized analysis

const analyzer = new Sentiment();
const result = useMemo(() => analyzer.analyze(text), [text]);

2) Score to mood mapping

A threshold mapping translates numeric sentiment into more human-readable mood states.

Mood mapping

function getMood(score: number) {
  if (score >= 3) return { label: "Hyper Positive", color: "#1f7a66" };
  if (score >= 1) return { label: "Positive", color: "#2563a8" };
  if (score <= -3) return { label: "Critical", color: "#9b2c2c" };
  if (score <= -1) return { label: "Negative", color: "#a15c12" };
  return { label: "Neutral", color: "#5a6470" };
}

3) Token diagnostics

Displaying positive and negative token lists helps users understand why a score was produced.

Key takeaways

No model download required.

Interpretability comes from token-level output.

Lexicon methods are fast but context-limited.

Back to experiment Browse all experiments