Corpus datasets, 2026-11-12

Method, sources and limits are in the README. Checksums: SHA256SUMS. Files ending in _verified carry the fact-check verdicts.

• 10_product_ingredients.csv 764 KB
• 11_product_allergens.csv 201 KB
• 1_demand_products.csv 731 KB
• 1_product_strings.csv 1,890 KB
• 2_sg_menu_prices.csv 3,233 KB
• 2_sg_menu_prices_verified.csv 4,585 KB
• 3_cosmetic_ingredients.csv 310 KB
• 3_cosmetic_ingredients_verified.csv 565 KB
• 4_snack_nutrition.csv 80 KB
• 4_snack_nutrition_verified.csv 91 KB
• 5_faq_pairs.csv 7,817 KB
• 6_ingredient_glossary_cosing.csv 64 KB
• 7_discontinuation_vs_maker.csv 22 KB
• 8_snack_nutrition_maker_sourced.csv 13 KB
• 9_maker_nutrition_catalogue.csv 333 KB