Data Sentiment in Action

🧪 1. Example — Sentiment Extraction Process

Step 1: Collect Sources

• Queried Reddit using keywords:

  • “baby bottle leaks,”

  • “<brand> bottle vs,”

  • “<brand> baby bottle sucks,”

  • “recommend baby bottle for newborns,”

  • Used known companies and anonymized them (e.g., Dr. Brown’s → Company A)

Step 2: Extract Sentiment from Comments

• Extracted the comment text

• Labeled each as:

  • Positive (clear praise)

  • Negative (complaints, product failures)

  • Neutral (comparative or unclear)

Step 3: Assign Tags

Each comment was tagged for:

• Leakage

• Ease of cleaning

• Nipple flow / baby rejection

• Plastic quality / safety

• Customer service

Step 4: Compile Dataset

Each entry contains:

• Reddit Post URL

• Comment Text

• Sentiment (Positive/Negative/Neutral)

• Tag(s)

• Company (A, B, or C)

📊 2. Sample Sentiment Dataset – Baby Bottle Industry

Reddit URL	Company	Comment	Sentiment	Tag(s)
https://reddit.com/r/beyondthebump/comments/abc123	Company A	“We tried Company A’s bottle but it leaked constantly—gave up after a week.”	Negative	Leakage
https://reddit.com/r/NewParents/comments/def456	Company B	“Company B bottles are really easy to clean, we stuck with them for all three kids.”	Positive	Cleaning
https://reddit.com/r/Parenting/comments/ghi789	Company C	“Baby kept rejecting the nipple from Company C, flow was too fast.”	Negative	Nipple flow
https://reddit.com/r/beyondthebump/comments/jkl012	Company A	“Didn’t expect much, but Company A’s bottles held up great and didn’t leak.”	Positive	Durability
https://reddit.com/r/beyondthebump/comments/mno345	Company B	“Plastic looked cloudy after 2 months. Not sure if it’s safe?”	Neutral	Material quality