Skip to the restaurant evidence
← Back to Methodology page

Restaurant Review Analysis: Coverage by Restaurant

From 10,309 Dubai restaurant listings, 2,255 with checked customer highlights were ranked by star rating, rating volume, and listing quality; 1,000 were selected.

AI-assisted analysis identified candidate dishes and dish-specific opinions in written reviews. Equivalent dish names were grouped, source text was checked, and unsupported or ambiguous votes were excluded.

This table shows reviews analyzed, reviews with accepted food evidence, and distinct positive and negative review counts for top-ranked dishes at each restaurant. The leaderboard uses a separate, fixed tally of dish votes across restaurants; adding the counts in this table will not reproduce its totals.

Loading restaurant evidence…

Scroll across for all columns →

Opening the evidence ledger…

How to read the counts

Each dish count refers to distinct reviews for that dish, not raw mentions. A review can discuss more than one dish, so adding dish counts can exceed the number of distinct reviews with top-ranked dishes.

For example, at Rosa’s Thai in Dubai Hills, 24 reviews mention Mango Sticky Rice and 14 mention Tom Yum soup. Four reviews mention both. The dish counts add to 38, but the two dishes appear in 34 distinct reviews. These counts are evidence subsets, not percentages of all diners or all dishes served.

Reviews analyzed
477,898 reviews were processed for food analysis across 1,000 restaurants.
Accepted food mentions
Distinct reviews with evidence for a conservatively accepted food or one of the 65 top-ranked dishes. This is not a complete scan of every food word.
Top-ranked dishes
Distinct reviews with extracted evidence for one of the 65 listed dishes, with positive and negative review counts shown for each dish. These are per-restaurant evidence counts. The leaderboard uses a separate, fixed tally of dish votes across restaurants; adding the counts in this table will not reproduce its totals.