Skip to main content

​
Overview

SENTi-radar allows you to perform real-time sentiment analysis on any topic, brand, event, or hashtag by pulling live data from multiple sources including X (Twitter), Reddit, YouTube, and news feeds. The platform combines multi-source data with AI-powered emotion detection to give you instant insights into public sentiment.
SENTi-radar analyzes up to 100+ live posts per topic from across X, Reddit, YouTube comments, and Google News RSS feeds in real-time.

​
Searching for a Topic

1

Enter your search query

Navigate to the main dashboard and locate the search bar at the top of the page. You can search for:
  • Topics: “AI regulation”, “Climate change”
  • Brands: “iPhone”, “Tesla”
  • Hashtags: “#ClimateSummit2026”
  • Events: “Food prices”, “Bitcoin”
Use the suggestion chips below the search bar for quick access to trending topics like “AI regulation”, “Climate change”, “iPhone”, “Food prices”, and “Bitcoin”.
2

Click 'Analyze' to start the analysis

Once you’ve entered your query, click the Analyze button. You’ll see a loading state with the text “Analyzing” while SENTi-radar:
  1. Fetches live posts from X (Twitter) via Scrape.do
  2. Retrieves Reddit discussions via Scrape.do
  3. Pulls YouTube comments from relevant videos
  4. Gathers news headlines from Google News RSS
  5. Analyzes all text for emotion and sentiment patterns
  6. Generates an AI-powered summary
The entire process typically takes 5-10 seconds depending on the volume of data available.
3

View the topic detail panel

After analysis completes, the topic detail panel opens automatically, displaying:
  • Overall sentiment gauge (positive/negative/neutral distribution)
  • Dominant emotion with percentage
  • Total mention volume and hourly change
  • Volatility score with visual chart
  • AI-generated summary with key takeaways
  • Emotional breakdown across 6 emotions
  • Sentiment trend over time
  • Crisis alerts (if applicable)

​
Understanding the Live Overview Panel

The Live Overview Panel displays four key KPI cards:

​
Overall Sentiment

A gauge chart showing the distribution of:
  • Positive sentiment (green): Joy, surprise, excitement
  • Negative sentiment (red): Anger, fear, sadness, disgust
  • Neutral sentiment (gray): Mixed or unclear signals

​
Dominant Emotion

The most prevalent emotion detected in the analyzed texts, shown with:
  • Emotion name (Joy, Anger, Sadness, Fear, Surprise, or Disgust)
  • Percentage of total emotion distribution
  • Number of texts analyzed
  • Data source badge (e.g., “X · Reddit · YouTube · News”)
SENTi-radar uses a keyword lexicon-based scoring system defined in TopicDetail.tsx:34-42. Each text is scanned for emotion keywords:
  • Fear: “scared”, “worried”, “panic”, “threat”, “crisis”, “uncertainty”
  • Anger: “outrage”, “furious”, “frustrat”, “scandal”, “protest”, “blame”
  • Sadness: “disappoint”, “tragic”, “loss”, “grief”, “devastat”
  • Joy: “excited”, “great”, “love”, “celebrate”, “success”, “optimis”
  • Surprise: “shocking”, “unexpected”, “stunning”, “breaking”
  • Disgust: “appalling”, “horrible”, “corrupt”, “toxic”, “vile”
Scores are aggregated across all sources and normalized to 100%.

​
Total Mentions

Displays the total conversation volume with:
  • Formatted count (e.g., “284.5K”, “1.2M”)
  • Percentage change in the last hour
  • Trending indicator (up/down arrow)

​
Volatility Score

A mini sparkline chart showing conversation volatility with:
  • Visual representation of volume spikes
  • Risk level: High or Moderate
  • Numerical volatility score (0-100)

​
AI-Generated Summary

The summary section provides:
  1. Headline: Includes emoji indicator (🔴 High Crisis, 🟡 Elevated Concern, 🟢 Positive Momentum, 🔵 Mixed Signals), dominant emotions, and risk assessment
  2. Narrative: 2-3 sentences analyzing the topic with:
    • Specific emotion percentages
    • Direct quotes from real posts
    • Evidence-based insights
  3. People’s Voice – Key Takeaways: 5 bullet points covering:
    • Most repeated sentiments
    • Public concerns or reactions
    • Data-driven observations
    • Forward-looking trends
    • Ground-level impact stories
When VITE_GEMINI_API_KEY or VITE_GROQ_API_KEY is configured, SENTi-radar streams AI-generated summaries in real-time using:
  1. Gemini 2.0 Flash (primary) - Google’s generative AI
  2. Groq Llama 3.3 70B (fallback) - High-speed open-source LLM
The summary appears word-by-word with a blinking cursor animation.
The right panel displays:

​
Emotional Breakdown

A horizontal bar chart showing all 6 emotions with:
  • Color-coded bars (each emotion has a distinct color)
  • Percentage distribution
  • Live data badge showing number of texts analyzed
  • Source attribution (e.g., ”📊 Live — 127 texts”)

​
Sentiment Trend Over Time

A line chart visualizing how sentiment has evolved across recent time periods.

​
Top Voices / Pointers Panel

This section shows the most frequently mentioned phrases split into: Positive Pointers (green background):
  • Phrases associated with positive sentiment
  • Mention count for each phrase
  • Examples: “holographic display”, “take my money”, “worth the upgrade”
Negative Pointers (red background):
  • Phrases associated with negative sentiment
  • Mention count for each phrase
  • Examples: “too expensive”, “broken promises”, “corporate greed”

​
Crisis & Alerts Log

When crisis level is medium or high, you’ll see real-time alerts:

Negative spike detected

Displays when negative emotion (anger, fear, sadness) dominates public discourseShows:
  • Red pulsing indicator
  • Dominant emotion with context
  • Timestamp (“Just now”, “30 min ago”)

Sentiment Shift

Triggered by sudden increases in conversation volumeIndicates:
  • Rapid topic acceleration
  • Viral moment or breaking news
  • Potential crisis escalation

​
Data Source Indicators

SENTi-radar displays badges showing which sources were successfully queried:
  • ✓ X via Scrape.do (green badge) - X posts fetched successfully
  • ✓ Reddit via Scrape.do (green badge) - Reddit posts fetched successfully
  • ⚠️ Source unavailable (red badge) - Scrape.do quota exceeded or traffic blocked
  • ℹ️ VITE_SCRAPE_TOKEN not set (gray badge) - Scraping disabled, configure token in .env
To enable X and Reddit scraping, add VITE_SCRAPE_TOKEN=your_scrape_do_token to your .env file. Get your token at scrape.do.

​
Refreshing the Analysis

To get updated data for the same topic:
  1. Click the Refresh icon (↻) in the top-right corner of the summary panel
  2. SENTi-radar re-fetches all live sources
  3. Emotions and summary are recalculated with new data
  4. The entire panel updates in 5-10 seconds
Use this feature when monitoring a developing story or tracking sentiment shifts during live events.

​
Troubleshooting

Possible causes:
  • No VITE_SCRAPE_TOKEN configured - X and Reddit will be skipped
  • Scrape.do quota exceeded - wait for quota reset or upgrade plan
  • No YouTube API key - add VITE_YOUTUBE_API_KEY to .env
  • Query too specific - try broader terms
Solution: Check browser console for error messages and verify environment variables are set correctly.
This typically happens when:
  • All data sources fail simultaneously
  • Network timeout during API calls
  • Invalid API keys for Gemini or Groq
Solution: SENTi-radar will automatically fall back to local summary generation. Check your API keys and network connection.
Scrape.do quota exceeded: You’ve hit your API limit. Either:
  • Wait for your quota to reset (usually monthly)
  • Upgrade your Scrape.do plan
  • Analysis will continue using YouTube + News sources
Traffic blocked: The platform detected bot traffic. This is rare with Scrape.do’s residential proxies but can happen. Try:
  • Setting super: true in scraping options (uses premium proxies)
  • Waiting 10-15 minutes before retrying

​
Best Practices

  1. Use specific queries: “iPhone 18 battery life” is better than just “phone”
  2. Monitor volatility: Topics with high volatility (>80) need frequent refresh
  3. Check crisis alerts: Red/yellow alerts indicate urgent public concern
  4. Compare data sources: Cross-reference X sentiment with Reddit and YouTube for full picture
  5. Track trends over time: Use the sentiment timeline to spot acceleration or decline
  6. Export for reporting: Use the Export menu to save analysis data (see Exporting Data)

Build docs developers (and LLMs) love

Get started for freeTalk to us