Sentiment Analysis

Every piece of text carries emotional weight. A product review can be positive, negative, or neutral. A tweet can be angry, excited, or flat. TextScore uses sentiment analysis to measure this emotional tone and flag content where the intensity might hurt your reach.

What Is VADER

VADER (Valence Aware Dictionary and sEntiment Reasoner) is a sentiment analysis model built specifically for social media text. Unlike older models trained on formal writing, VADER understands slang, abbreviations, emoji, and internet-era punctuation.

How the Compound Score Works

VADER produces four scores for any text: positive, negative, neutral, and compound. TextScore uses the compound score, which is a normalized sum of all word-level sentiments. It ranges from -1.0 (most negative) to +1.0 (most positive).

• +0.05 to +1.0: Positive sentiment
• -0.05 to +0.05: Neutral sentiment
• -1.0 to -0.05: Negative sentiment

What VADER Catches

• Word intensity: "Great" is positive. "Amazing" is more positive. "AMAZING" (caps) is even more so.
• Negation: "This is good" vs. "This is not good" - VADER handles the flip.
• Boosters and dampeners: "Very good" scores higher than "good." "Kind of good" scores lower.
• Punctuation: "Great!" scores slightly higher than "Great." because exclamation marks signal intensity.
• Conjunctions: "The food was great but the service was terrible" - VADER weights the sentiment shift.

Beyond the positive/negative spectrum, TextScore categorizes the dominant emotional tone of your text. This gives you more actionable information than a raw number.

• Anger: Hostile language, accusatory tone, aggressive phrasing
• Joy: Celebratory language, gratitude, excitement
• Fear: Warnings, threats, anxiety-driven language
• Sadness: Loss, disappointment, regret
• Neutral: Factual, informational, balanced

Platform Algorithms and Emotion

Platforms actively manage emotional content. Extremely negative or angry content often gets reduced distribution, even if it does not violate any specific rule. This is because high-negativity content drives users away from platforms over time.

The Sweet Spot

Mildly positive content tends to get the best distribution. It generates engagement without triggering safety filters. Scores between +0.2 and +0.6 typically perform well across platforms.

• Highly positive (+0.7 to +1.0): Can look inauthentic. "Everything is amazing" reads like spam or a fake review.
• Mildly positive (+0.2 to +0.6): Natural, engaging, well-distributed.
• Neutral (-0.05 to +0.2): Safe but may get less engagement. Good for factual content.
• Mildly negative (-0.5 to -0.05): Criticism and critical analysis. Usually fine for reach.
• Highly negative (-1.0 to -0.5): Risk of reduced distribution. May trigger review queues.

• Good (Green): Balanced tone. Your sentiment is in a range that works well across platforms. Not too hot, not too flat.
• Fair (Yellow): Leaning toward an extreme. Your content may still perform fine, but consider whether the intensity is intentional.
• Poor (Red): Extreme sentiment detected. High risk of reduced distribution or content review. Unless the strong emotion is your goal, dial it back.