Methodology
Trustworthy food data needs transparent rules. These are ours.
How ratings work
We separate rating types: user ratings, editorial notes, and source mentions (e.g. a Michelin Bib Gourmand listing) are stored and displayed separately, never blended into one opaque number. In this early preview, rankings are neutral placeholders — real rankings require verified data and community input.
source_score20%user_rating_score20%editorial_score10%popularity_score10%verification_score15%queue_score5%value_score10%freshness_score10%Verification statuses
Every hawker centre, stall, dish, menu item and photo carries one of four statuses:
- VerifiedConfirmed against an official source or an on-site visit.
- Partially verifiedCore facts confirmed, details (e.g. prices, hours) still being checked.
- UnverifiedCollected from public knowledge or a single source; not yet confirmed.
- Demo dataPlaceholder created to develop the product. Not real data.
Source reliability
Each record links to its sources. Official datasets score highest, then manual visits, established guides and media, then community submissions — which are moderated before publishing.
No fake reviews
We never generate, buy or paraphrase reviews. Review counts stay at zero until real people write real reviews under moderation. Sponsored placements, when they exist, will always be labelled.