Skip to main content

​
Overview

SENTi-radar classifies every analyzed text into one of six core emotions: Joy, Anger, Sadness, Fear, Surprise, and Disgust. This goes beyond simple positive/negative sentiment to reveal the emotional drivers behind public reactions.

​
The Six Emotions

Joy

Happiness, excitement, celebration, optimism, and positive breakthroughs

Anger

Outrage, frustration, condemnation, and demands for action

Sadness

Disappointment, grief, despair, and expressions of loss

Fear

Worry, anxiety, threats, risks, and uncertainty

Surprise

Shock, unexpected revelations, and unbelievable moments

Disgust

Revulsion, moral outrage, and strong disapproval

​
How Emotion Classification Works

1

Text Collection

Gather all text data from social media posts, comments, news headlines, and video descriptions.
2

Keyword Matching

Scan text against a curated lexicon of emotion keywords. Each emotion has 20-30 associated words and phrases.
3

Score Calculation

Count keyword matches for each emotion and calculate percentage distribution across all six emotions.
4

Normalization

Ensure percentages sum to exactly 100% and rank emotions by dominance.

​
Emotion Keyword Lexicon

The classification engine uses a comprehensive keyword dictionary: 
// Source: TopicDetail.tsx (lines 35-42)
const EMOTION_KEYWORDS: Record<string, string[]> = {
  fear: [
    'fear', 'scared', 'worried', 'panic', 'threat', 'risk', 
    'dangerous', 'crisis', 'collapse', 'shortage', 'anxiety', 
    'alarm', 'uncertainty', 'instability', 'catastroph', 'turmoil'
  ],
  anger: [
    'anger', 'angry', 'outrage', 'furious', 'rage', 'frustrat', 
    'unacceptable', 'scandal', 'corrupt', 'condemn', 'protest', 
    'exploit', 'injustice', 'blame', 'backlash', 'fury'
  ],
  sadness: [
    'sad', 'disappoint', 'tragic', 'loss', 'suffer', 'grief', 
    'regret', 'devastat', 'despair', 'victim', 'casualt', 'death', 
    'pain', 'mourn', 'unfortunate', 'heartbreak', 'sorrow'
  ],
  joy: [
    'happy', 'excited', 'great', 'amazing', 'love', 'excellent', 
    'fantastic', 'celebrate', 'breakthrough', 'success', 
    'innovation', 'optimis', 'hopeful', 'wonderful', 'proud'
  ],
  surprise: [
    'shocking', 'unexpected', 'unbelievable', 'stunning', 
    'incredible', 'reveal', 'bombshell', 'breaking', 
    'unprecedented', 'wtf', 'omg', 'cant believe'
  ],
  disgust: [
    'disgust', 'appalling', 'horrible', 'corrupt', 'toxic', 
    'vile', 'sickening', 'revolting', 'gross', 'nauseating', 
    'shameful', 'pathetic'
  ]
};
Keywords use partial matching (e.g., “frustrat” matches “frustrate”, “frustrated”, “frustrating”) to capture variations.

​
Scoring Algorithm

The emotion scoring process: 
// Source: TopicDetail.tsx (lines 146-167)
function scoreEmotions(texts: string[]): EmotionData[] {
  const allText = texts.join(' ').toLowerCase();
  const scores: Record<string, number> = {};
  
  // Count keyword matches for each emotion
  for (const [emotion, words] of Object.entries(EMOTION_KEYWORDS)) {
    scores[emotion] = words.reduce((sum, w) => {
      const re = new RegExp(w.replace(/[.*+?^${}()|[\]\\]/g, '\\$&'), 'gi');
      return sum + (allText.match(re)?.length || 0);
    }, 0);
  }
  
  // Calculate percentages
  const total = Object.values(scores).reduce((a, b) => a + b, 0) || 1;
  const emotions: EmotionData[] = Object.entries(scores)
    .map(([emotion, score]) => ({
      emotion,
      percentage: Math.round((score / total) * 100),
      count: score,
    }))
    .sort((a, b) => b.percentage - a.percentage);
  
  // Normalize to exactly 100%
  const sum = emotions.reduce((s, e) => s + e.percentage, 0);
  if (sum !== 100 && sum > 0) emotions[0].percentage += (100 - sum);
  
  return emotions;
}

​
Emotion Breakdown Panel

The Emotion Breakdown component visualizes the distribution:
Each emotion appears as:
  • Color-coded indicator dot
  • Emotion label (Joy, Anger, etc.)
  • Animated progress bar (fills from 0 to percentage over 0.5s)
  • Percentage value in monospace font
Emotions are sorted from highest to lowest percentage.

​
Dominant Emotion Card

The dashboard displays the dominant (highest percentage) emotion prominently: 
// Source: TopicDetail.tsx (lines 757-764)
const topEmotion = [...displayEmotions]
  .sort((a, b) => b.percentage - a.percentage)[0];

<div className="text-4xl font-bold capitalize">
  {topEmotion?.emotion}
</div>
<div className="text-xl font-mono mt-1">
  ({topEmotion?.percentage}%)
</div>
The dominant emotion card includes a data source badge (e.g., “X · Reddit · YouTube”) when using live data.

​
Interpreting Emotion Data

​
High-Emotion Scenarios

Joy + Surprise (60%+)

Product launches, breakthroughs, positive announcementsExample: “45% joy, 25% surprise → Praise: holographic display (62%), battery life (24%)”

Anger + Fear (60%+)

Crises, controversies, policy backlashExample: “48% anger, 22% fear → Outrage at: corporate profiteering (52%), government inaction (33%)”

Sadness + Disgust (50%+)

Scandals, tragedies, moral failuresExample: “42% anger, 25% sadness → Key frustrations: broken promises (55%), insufficient targets (30%)”

Balanced Distribution

Complex, multi-faceted issues with divided opinionsExample: “35% fear, 28% anger, 18% surprise → Mixed reactions across stakeholder groups”

​
Emotion Shifts Over Time

Monitor how emotions evolve by comparing snapshots:
  • Fear → Anger: Issue escalation (inaction fueling frustration)
  • Surprise → Joy: Positive reception after initial shock
  • Anger → Sadness: Resignation setting in after prolonged controversy

​
Use Cases

​
Brand Crisis Management

// Example: Tracking iPhone launch emotions
{
  joy: 45%,        // "take my money", "holographic display"
  surprise: 25%,   // "this changes everything"
  anger: 15%,      // "too expensive", "no USB-C"
  sadness: 8%,     // price disappointment
  fear: 5%,
  disgust: 2%
}
Insight: While overwhelmingly positive, the 15% anger around pricing (especially “no USB-C” memes) needs addressing in marketing.

​
Policy Response Tracking

// Example: Climate Summit emotions
{
  anger: 42%,      // "broken promises", "greenwashing"
  sadness: 25%,    // "too little too late"
  fear: 18%,       // island nations, future concerns
  disgust: 10%,    // corporate hypocrisy
  surprise: 3%,
  joy: 2%
}
Insight: Dominant anger + sadness signals deep public frustration. Requires transparent action, not more rhetoric.

​
Advanced: Theme-Aware Context

The system detects topic themes (geopolitical, tech, health, etc.) and uses theme-specific interpretation: 
// Source: TopicDetail.tsx (lines 44-116)
const TOPIC_THEMES: Record<string, { keywords: string[]; templates: string[] }> = {
  geopolitical: {
    keywords: ['war', 'tension', 'iran', 'ukraine', 'missile', 'sanction'],
    templates: [
      'Escalation fears are driving market volatility and public anxiety',
      'Diplomatic channels remain under pressure',
      'Defense and security discussions dominate'
    ]
  },
  tech: {
    keywords: ['phone', 'launch', 'ai', 'chip', 'samsung', 'apple'],
    templates: [
      'Early adopters are buzzing — design, specs, and pricing are most discussed',
      'Comparisons with competitors are driving heated debates'
    ]
  }
  // ... 5 more themes
};
Theme detection enhances AI-generated summaries by providing domain-specific context.

​
Data Quality Indicators

The emotion panel shows data quality metrics:
  • Text count: “127 texts analyzed”
  • Data sources: “Live from X · Reddit · YouTube · News”
  • Refresh status: Spinner during analysis, checkmark when complete

Build docs developers (and LLMs) love

Get started for freeTalk to us