Review Sentiment Analyzer

Turn a collection of customer reviews into a compact view of sentiment, recurring themes, complaints, and changes over time. Supply reviews directly or select an existing Apify dataset, and receive one structured analysis record ready for inspection or downstream automation.

What you can learn

• See positive, neutral, and negative review counts and percentages.
• Find frequently mentioned words as themes and recurring negative terms as complaints.
• Compare positive and negative percentages by month when reviews include dates.
• Confirm exactly how many reviews were analyzed and how many billable review blocks were used.

This Actor is useful for summarizing feedback exports, monitoring recurring service issues, comparing sentiment across batches, and preparing review data for reports or workflows.

Input

Provide exactly one review source:

• reviews: an array of review strings or objects. Objects may use text, review, or content for the review text, plus optional numeric rating and date or reviewed_at values.
• dataset_id: an accessible Apify dataset containing review strings or objects in the same shape.

Use max_items to limit analysis from 1 to 5,000 reviews and max_themes to return from 1 to 50 themes and complaint terms. proxy_configuration is retained for interface compatibility; the local analysis does not require a proxy.

Example:

{
  "reviews": [
    {"text": "Great service and friendly support", "rating": 5, "date": "2026-01-10"},
    {"text": "Slow delivery and poor communication", "rating": 2, "date": "2026-02-12"}
  ],
  "max_items": 5000,
  "max_themes": 10
}

Output

The default dataset receives one record with:

• analyzed_review_count;
• sentiment counts and percentages;
• frequency-ranked themes and complaint terms;
• month-ordered trends when usable dates are present; and
• billable_events, an auditable summary of charged-event quantities.

Percentages are rounded to two decimal places. The analysis is deterministic and uses a simple English sentiment lexicon with ratings as a fallback. It does not infer sarcasm, nuanced negation, non-English sentiment, or semantic topic clusters.

Pricing and limits

The Actor charges an apify-actor-start event when a run starts and one reviews-analyzed-100 event for each started block of 100 analyzed reviews. The input is capped at 5,000 reviews per run. Current event prices are shown on the Actor's Apify pricing page before you start a run.

Run locally

python3 -m unittest discover -s tests -v