extracting-youtube-comments-for-research

Extracting YouTube Comments for Research

Pulls YouTube video comments for sentiment analysis, question mining, and product feedback. YouTube comments are more considered than TikTok — viewers invest more time before commenting.

Prerequisites

• APIFY_TOKEN environment variable set
• Optional: Apify MCP server installed

Inputs

Workflow

Progress:
- [ ] Step 1: Scrape comments from target videos
- [ ] Step 2: Filter and clean dataset
- [ ] Step 3: Analyze by research goal
- [ ] Step 4: Extract top themes and insights
- [ ] Step 5: Deliver comment research report

Step 1: Scrape Comments

Recommended — run_actor.js (handles waiting, output, and file saving automatically):

# Quick answer (prints table to chat)
node scripts/run_actor.js \
  --actor "apidojo~youtube-comments-scraper" \
  --input '{"param": "value"}'

# Save as CSV
node scripts/run_actor.js \
  --actor "apidojo~youtube-comments-scraper" \
  --input '{"param": "value"}' \
  --output YYYY-MM-DD_results.csv --format csv

# Save as JSON
node scripts/run_actor.js \
  --actor "apidojo~youtube-comments-scraper" \
  --input '{"param": "value"}' \
  --output YYYY-MM-DD_results.json --format json

APIFY_TOKEN must be set in environment or .env file.

If Apify MCP is available:

Tool: apify:run-actor
Actor: "apidojo~youtube-comments-scraper"
Input:
{
  "startUrls": [{"url": "[VIDEO_URL_1]"}, {"url": "[VIDEO_URL_2]"}],
  "type": "comments",
  "maxComments": 500
}

REST API fallback:

curl -X POST \
  "https://api.apify.com/v2/acts/apidojo~youtube-comments-scraper/runs?token=$APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"startUrls": [{"url": "[VIDEO_URL]"}], "type": "comments", "maxComments": 500}'

Step 2: Clean Dataset

• Remove comments < 8 words (usually emoji-only or "great video!")
• Remove self-promotional comments (contain external links)
• Remove creator's own replies (match authorName to channel name)
• Apply min_likes_on_comment filter if set

Step 3: Analyze by Goal

Questions: Contains "?", "how do you", "what is", "can you" Pain points: "I struggle", "I can't", "problem is", "doesn't work" Product feedback: Product mentions + opinion signals Sentiment: Standard lexical classifier (positive/negative/neutral)

comment_importance = likeCount * 0.60 + replyCount * 10 * 0.40

Step 4: Edge Cases

• Comments disabled: Note; try different video from same channel
• Mostly non-English: Report language distribution; filter to English if needed
• Spam invasion: Filter where same username appears > 3 times
• Brigaded comment section: > 50% share coordinated theme → flag as BRIGADED

Output Format

# YouTube Comment Analysis
Videos: [N] | Comments analyzed: [N] | After filtering: [N] | Date: [DATE]

## Sentiment (if goal = sentiment)
Positive: [X%] | Negative: [X%] | Neutral: [X%]

## Top 10 Most-Liked Comments
| # | Comment (excerpt) | Likes | Replies |
|---|------------------|-------|---------|

## Key Themes
| Theme | Frequency | Avg Likes | Example |
|-------|-----------|-----------|---------|

## Most Asked Questions
1. "[question]" — [N] viewers

Troubleshooting

Few comments returned: YouTube limits access for some videos; try high-comment video from same channel. Mostly surface-level praise: Use min_likes_on_comment = 5 to filter for substantive comments. Research goal not present: Audience may not engage that way on YouTube; try Reddit or TikTok for this niche.