Skip to main content

๐ŸŒˆ Understanding Emotions Analysis

The BGBX emotion model is fine-tuned on the GoEmotions dataset, which is comprised of 58,000 Reddit comments labeled by human annotators. It identifies 27 distinct emotions

Written by Team
Updated yesterday

28 Emotions and Emotion Categories

Emotions are organized into 8 primary families based on famed psychologist, Robert Plutchik's Wheel of Emotions.

๐Ÿ˜Š Joy

  • Emotions reflecting happiness, pleasure, and positive engagement

  • Admiration โ€“ Finding something impressive or worthy of respect

  • Amusement โ€“ Finding something funny or being entertained

  • Excitement โ€“ Great enthusiasm and eagerness

  • Joy โ€“ Pleasure and happiness

  • Love โ€“ Strong positive emotion of regard and affection

  • Pride โ€“ Satisfaction from one's own or associated achievements

  • Relief โ€“ Reassurance following release from anxiety

๐Ÿ˜ข Sadness

  • Emotions reflecting sorrow, loss, and emotional pain

  • Sadness โ€“ Emotional pain and sorrow

  • Grief โ€“ Intense sorrow, especially from loss or death

  • Disappointment โ€“ Displeasure from unfulfilled hopes or expectations

  • Remorse โ€“ Regret or guilty feeling

  • Embarrassment โ€“ Self-consciousness, shame, or awkwardness

๐Ÿค Trust

  • Emotions reflecting confidence, care, and positive regard for others

  • Caring โ€“ Displaying kindness and concern for others

  • Gratitude โ€“ Thankfulness and appreciation

  • Acceptance โ€“ Having or expressing a favorable opinion (labeled as "Approval" in model output)

๐Ÿคข Disgust

  • Emotions reflecting revulsion and strong disapproval

  • Disgust โ€“ Revulsion aroused by something unpleasant or offensive

  • Disapproval โ€“ Having or expressing an unfavorable opinion

๐Ÿ˜จ Fear

  • Emotions reflecting anxiety, worry, and apprehension

  • Fear โ€“ Being afraid or worried

  • Nervousness โ€“ Apprehension, worry, anxiety

  • Confusion โ€“ Lack of understanding or uncertainty

๐Ÿ˜ก Anger

  • Emotions reflecting displeasure and irritation

  • Anger โ€“ Strong feeling of displeasure or antagonism

  • Annoyance โ€“ Mild anger or irritation

๐Ÿ˜ฒ Surprise

  • Emotions reflecting unexpected discoveries and awareness

  • Surprise โ€“ Feeling astonished or startled by something unexpected

  • Realization โ€“ Becoming aware of something

๐Ÿ”ฎ Anticipation

  • Emotions reflecting forward-looking desires and expectations

  • Curiosity โ€“ Strong desire to know or learn something

  • Desire โ€“ Strong feeling of wanting something or wishing for it to happen

  • Optimism โ€“ Hopefulness and confidence about the future

โšช Neutral

  • No clear emotional expression detected

Emotions as Part of Product Analyses

Within BGBX products like SmartRoom, emotions will be displayed as distributions per stakeholder per journey moment.

As part of topic analysis, emotions are measured against the stakeholder's behavior for a give moment in the healthcare journey.

Emotion badges appear under verbatims indicating the presence of emotions in the text. The emotion model measures emotions for all healthcare related sentences within a verbatim.

How Emotions Are Detected

The model analyzes conversational data and assigns probability scores for each emotion category. Multiple emotions can be detected at a time

Key Characteristics

Lexical Markers: Some emotions have strong word associations (e.g., "thanks" โ†’ gratitude, "lol" โ†’ amusement)

Context-Dependent: Many emotions require understanding context and tone

Multi-label: Text can express multiple emotions simultaneously

Intensity Clustering: Related emotions (e.g., annoyance and anger, joy and excitement) often co-occur

Did this answer your question?