Methodology
How we collect, clean, and publish the dataset. Independent project; not affiliated with Jobalots.
Sources & scope
- Publicly visible EU-facing reviews/comments related to Jobalots purchases.
- Only the EU dataset is included on this site.
Fields we publish
- date — normalized to
YYYY-MM-DD(string, no timezone shifts). - reviewer_masked — last 3 characters replaced with
***. Original usernames are never written to disk or published. - score — numeric 0..5 (steps of 0.5). Parsed from digits or star glyphs (e.g.,
★★★☆☆→ 3.0). - comment — original language; “…See more/less” tails removed.
Cleaning rules
- Trim whitespace and collapse multiple spaces.
- Remove UI tails: “…See more/less”.
- Score parsing accepts “3,5”, “3.5”, “4★”, “5 stars”, “3½”, and glyph strings.
Privacy
- Names are masked by default (last 3 characters →
***). - We honor takedown requests sent to hello@jobalots-reviews.eu.
Updates
- CSV is refreshed automatically every 48 hours by a scheduled GitHub Actions job; newest entries appear first.
- An offline tool (
tools/csv-updater2.html) is kept as a fallback for ad-hoc edits. - Site rebuilds immediately after the CSV is committed.
Limitations
- We cannot verify every claim in comments; treat scores as self-reported.
- Language detection/translation is not automated; original text is preserved.