PRACTICE / REVIEWS
Review Sentiment Analysis Tool: Topics & CSV Export
Import text, get Jev suggestions, then confirm the results yourself.
Analyze review sentiment and fixed product, delivery and usability topics with Jev. Review each result, filter sentiment and export CSV. Free beta credits.
Download a CSV with 3 fictional inputs · Workflow and criteria
The template has no assigned labels. Import it here, then label manually or choose to run Jev.
JEV / PUBLIC BETA
Suggestions you can review
One credit per item. Guests get 10 daily; signed-in users get 20 total, without stacking or rollover. Resets at midnight China time (UTC+8). Network limits are shared; the site allows 200 calls per day. Manual review and exports are free.
Example labels are authored. Import your text, or clear example labels before running.
Stopping does not cancel an already sent request. Failed-item retries send a new request; uncertain network outcomes reuse the original request ID to avoid duplicate charges. Model suggestions still need human confirmation.
Travel mug · 12 authored examplesLabels are authored. Imported reviews need manual classification. Jev is not connected.
Issue distribution
Select a category to see the originals. Multiple labels may total over 100%.
Sentiment distribution
Separate from issue labels. Mixed is not neutral; not assessed is not unknown.
| ID | Review | Categories | Sentiment | Status | Action |
|---|
No matching reviews
From CSV to reviewable decisions
- Import a UTF-8 CSV text or review column. Originals remain in the current browser session.
- Choose labels manually, or confirm submission and use beta credits for Jev suggestions.
- Inspect each original, correct labels and confirm review; export CSV or save a project JSON.
Decision boundaries
Themes are fixed: product quality, delivery and usability, with multiple labels allowed. Sentiment is independently positive, negative, neutral, mixed or unknown. Missing evidence needs review. The tool does not discover clusters, authenticate reviews or generate replies.
Model suggestions are not an accuracy guarantee. Existing API smoke tests establish interoperability only; evaluate quality on your own labeled samples. Evaluation and review thresholds